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 |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
6ffc7ce3-6ef2-408b-bf45-00cf11af7ece | interactive-acquisition-of-fine-grained | 2305.03461 | null | https://arxiv.org/abs/2305.03461v1 | https://arxiv.org/pdf/2305.03461v1.pdf | Interactive Acquisition of Fine-grained Visual Concepts by Exploiting Semantics of Generic Characterizations in Discourse | Interactive Task Learning (ITL) concerns learning about unforeseen domain concepts via natural interactions with human users. The learner faces a number of significant constraints: learning should be online, incremental and few-shot, as it is expected to perform tangible belief updates right after novel words denoting unforeseen concepts are introduced. In this work, we explore a challenging symbol grounding task--discriminating among object classes that look very similar--within the constraints imposed by ITL. We demonstrate empirically that more data-efficient grounding results from exploiting the truth-conditions of the teacher's generic statements (e.g., "Xs have attribute Z.") and their implicatures in context (e.g., as an answer to "How are Xs and Ys different?", one infers Y lacks attribute Z). | ['Subramanian Ramamoorthy', 'Alex Lascarides', 'Jonghyuk Park'] | 2023-05-05 | null | null | null | null | ['implicatures'] | ['natural-language-processing'] | [ 4.86657470e-01 8.44007075e-01 -1.46855563e-01 -7.64368236e-01
-5.94267488e-01 -7.26015210e-01 5.40962458e-01 8.33743691e-01
-3.80531341e-01 1.06313288e+00 1.51490018e-01 -6.69446111e-01
-2.84729540e-01 -5.75900137e-01 -9.21825588e-01 -3.25249672e-01
-3.77425253e-01 5.37562072e-01 2.07286730e-01 -2.41276696e-01
3.33185464e-01 1.56134352e-01 -1.95261455e+00 3.95662040e-01
9.67755258e-01 7.68832266e-01 2.13700861e-01 3.88326555e-01
-4.89989161e-01 1.17369270e+00 -8.53312969e-01 -7.05688059e-01
-2.23903149e-01 -1.97329566e-01 -1.27420449e+00 -6.66552186e-02
8.96612704e-01 -6.73707902e-01 3.94817650e-01 1.44590342e+00
2.08358169e-02 3.40136439e-01 4.86221939e-01 -1.44632447e+00
-5.93467057e-01 1.11439824e+00 -6.85388371e-02 1.02190085e-01
1.00491285e+00 3.18307072e-01 1.35925555e+00 -6.71545565e-01
5.80748796e-01 1.53247857e+00 4.55265552e-01 8.23864520e-01
-1.36716318e+00 -6.44379616e-01 7.68822372e-01 4.09416080e-01
-1.15096903e+00 -2.10520715e-01 4.71071482e-01 -5.34388661e-01
1.01949406e+00 6.04505420e-01 7.84440577e-01 1.20660210e+00
-6.46402538e-02 1.05315208e+00 1.15686738e+00 -6.23706102e-01
5.43286622e-01 4.92262453e-01 3.61031264e-01 7.83412635e-01
4.74262536e-01 2.08697647e-01 -1.14877725e+00 -1.06174327e-01
4.77519035e-01 -3.11150461e-01 -7.81845003e-02 -4.69308227e-01
-1.08707249e+00 6.30118370e-01 1.15343496e-01 1.90147445e-01
-1.61506012e-01 2.09739178e-01 1.91845924e-01 5.86368561e-01
2.67889351e-01 8.08851480e-01 -8.73331368e-01 -2.11093172e-01
-4.96387422e-01 6.68223500e-01 8.22553754e-01 1.41131711e+00
9.35479701e-01 -1.96865454e-01 1.01151213e-01 3.43001395e-01
3.36032391e-01 1.23895936e-01 5.29105186e-01 -1.02630913e+00
2.54200727e-01 2.94115543e-01 4.40507621e-01 -7.59557724e-01
-3.15555185e-01 -2.16277212e-01 -1.05759256e-01 2.69842595e-01
5.74630678e-01 -3.18847477e-01 -6.67567551e-01 2.03225255e+00
7.28802502e-01 4.22913224e-01 2.42103621e-01 6.84356749e-01
1.03677380e+00 1.44133553e-01 6.69119060e-01 -3.82897973e-01
1.36045718e+00 -1.48721099e-01 -9.56941366e-01 -4.75524366e-01
1.06467640e+00 -2.62948096e-01 1.57486391e+00 6.96590245e-01
-1.08619392e+00 -4.33675408e-01 -9.48288321e-01 -3.70009750e-01
-6.28455460e-01 -5.63844860e-01 1.05972540e+00 5.82527637e-01
-6.32562399e-01 4.98822123e-01 -5.75206757e-01 -2.75617957e-01
3.67023498e-01 1.56140342e-01 -2.54189163e-01 -5.35718724e-03
-1.38215601e+00 1.04610980e+00 8.38400304e-01 -3.68650079e-01
-1.23790622e+00 -1.04429340e+00 -1.02089798e+00 1.85156390e-01
1.01146591e+00 -5.34575939e-01 1.93060660e+00 -1.12727511e+00
-1.46258318e+00 1.19152367e+00 -1.99738946e-02 -5.43857694e-01
6.10987008e-01 -8.13857794e-01 -3.52714062e-01 -1.08142406e-01
2.12969244e-01 7.76125550e-01 7.53526211e-01 -1.25317371e+00
-9.96604860e-01 -3.92706096e-01 6.42803907e-01 3.24174672e-01
-1.38588727e-01 1.07819080e-01 3.97371203e-01 -4.63917524e-01
3.34328502e-01 -6.87342942e-01 -4.38328497e-02 5.54366969e-02
-5.70844114e-01 -6.59876883e-01 7.14068115e-01 -2.14464784e-01
1.12310386e+00 -2.16363144e+00 -2.35860765e-01 1.21015050e-01
3.51010412e-01 2.39164196e-02 3.54061961e-01 2.38398418e-01
-3.73262703e-01 2.04790413e-01 -5.55770397e-02 1.91410974e-01
2.99655318e-01 4.68331903e-01 -7.09735096e-01 6.12697825e-02
1.25896022e-01 6.20134473e-01 -1.37223327e+00 -6.55226469e-01
1.40912354e-01 -6.93880320e-02 -6.11116171e-01 5.77817857e-01
-9.32542562e-01 1.65450588e-01 -2.38182336e-01 4.10692096e-01
4.38012183e-01 -1.52788013e-01 2.89186239e-01 7.44226500e-02
1.05153427e-01 7.21711159e-01 -1.50137901e+00 1.74458337e+00
-2.65206844e-01 5.56925833e-01 -2.31746525e-01 -8.03882778e-01
6.32662416e-01 2.83889562e-01 -3.17320317e-01 -2.22957850e-01
-2.82982796e-01 2.41016150e-02 -6.12269435e-03 -9.87445474e-01
1.55336395e-01 -5.56601524e-01 1.82544328e-02 2.65159488e-01
4.02429491e-01 -7.07107067e-01 1.49719387e-01 5.90952992e-01
7.92715669e-01 2.70283699e-01 7.52455235e-01 -5.63894212e-01
3.62781316e-01 -1.07797824e-01 3.82818401e-01 1.21276760e+00
2.92956345e-02 -1.51335806e-01 7.73699105e-01 -6.55547857e-01
-2.96375632e-01 -1.31517005e+00 -4.45976853e-02 1.65418303e+00
1.76691994e-01 -5.35293639e-01 -5.27929187e-01 -9.19173956e-01
9.53330696e-02 1.55979228e+00 -6.74908400e-01 -2.22790495e-01
-2.06452191e-01 8.04865062e-02 2.85648614e-01 4.52335775e-01
2.77047753e-01 -1.00460088e+00 -1.15586424e+00 2.07964197e-01
-3.82671878e-02 -8.67583692e-01 -7.91989714e-02 6.82806969e-01
-1.01924872e+00 -1.07136202e+00 7.16771781e-02 -5.76474369e-01
6.63055837e-01 -1.29260063e-01 1.59502459e+00 3.21726233e-01
-2.11172760e-01 6.31128669e-01 -1.96632013e-01 -9.94735956e-01
-9.24336985e-02 -6.06046319e-01 1.85926333e-02 -2.19514638e-01
7.65809417e-01 -6.24950469e-01 -1.94060162e-01 -1.10588185e-01
-8.38675678e-01 4.02037501e-01 1.73472404e-01 5.53904712e-01
3.31520438e-01 2.13489488e-01 3.86857331e-01 -1.56328630e+00
8.17300558e-01 -4.95638669e-01 -6.41929209e-01 6.40342653e-01
-6.17458224e-01 2.69849420e-01 3.75455678e-01 -7.37014353e-01
-1.36871111e+00 -2.02924043e-01 3.20178479e-01 1.18262460e-02
-6.73193693e-01 5.28581917e-01 -3.96577895e-01 3.16946805e-01
9.93827939e-01 8.32111463e-02 -5.57341874e-01 -2.43093476e-01
6.25451207e-01 1.37041971e-01 7.78110445e-01 -1.55420542e+00
5.95721126e-01 1.95797414e-01 -1.39229819e-01 -8.20993006e-01
-1.51819682e+00 -2.70135462e-01 -3.90862703e-01 -3.76426995e-01
5.96007347e-01 -8.36647272e-01 -1.24101782e+00 6.96683228e-02
-1.16652453e+00 -6.24972939e-01 -5.26286066e-01 3.35934281e-01
-7.31808066e-01 1.39129668e-01 -1.83533713e-01 -1.05499673e+00
2.74108917e-01 -7.40789533e-01 6.25913620e-01 2.90161699e-01
-7.27078438e-01 -1.15226841e+00 -3.81525546e-01 1.21266022e-02
1.35617658e-01 2.75972456e-01 1.32553792e+00 -1.20992768e+00
-3.16215903e-01 3.00665647e-01 1.14048146e-01 2.43768189e-02
-4.07404872e-03 -4.50822376e-02 -9.56847608e-01 -7.36172795e-02
1.92382455e-01 -9.39073443e-01 2.62132168e-01 -1.28295749e-01
1.69567549e+00 -9.33914125e-01 -6.48931861e-02 2.23166168e-01
1.20712256e+00 1.30089626e-01 1.96176022e-01 9.79670733e-02
2.66749859e-01 1.01884794e+00 8.69452119e-01 5.52770972e-01
3.15705776e-01 2.63265729e-01 4.93918836e-01 4.30751055e-01
2.25539938e-01 -6.55643106e-01 1.34300470e-01 -2.68993348e-01
2.16215074e-01 -5.71195930e-02 -1.16364801e+00 4.54405040e-01
-1.79087353e+00 -9.40060079e-01 -2.25876067e-02 2.29496121e+00
1.43358076e+00 6.59227490e-01 -2.08007663e-01 4.38939855e-02
3.93088907e-01 -6.37813732e-02 -7.85440624e-01 -3.16575289e-01
2.08840489e-01 3.63766193e-01 -2.42797226e-01 9.08080697e-01
-8.61188769e-01 1.18314767e+00 5.96914530e+00 5.73817074e-01
-7.72904575e-01 -9.81128737e-02 4.88803834e-01 -1.80077795e-02
-7.14413643e-01 2.12818012e-01 -8.04514110e-01 1.17910527e-01
6.63008213e-01 -3.58950466e-01 1.25108093e-01 9.52805161e-01
-2.79974908e-01 -4.45606619e-01 -1.95617390e+00 7.47689188e-01
-6.30973130e-02 -1.01745665e+00 2.53542304e-01 -5.37904859e-01
5.49585998e-01 -6.70292974e-01 2.61238277e-01 5.66636264e-01
1.05660510e+00 -1.04815269e+00 1.00776482e+00 3.43560070e-01
7.25720525e-01 -6.15252972e-01 2.11054727e-01 6.72704220e-01
-4.75191802e-01 -1.79224342e-01 -1.24565557e-01 -4.66877460e-01
-3.94999087e-01 5.43344975e-01 -1.15162146e+00 -5.83639182e-02
6.69557989e-01 3.84635746e-01 -5.13570309e-01 4.99227077e-01
-1.06172264e+00 5.68136394e-01 -2.75591642e-01 -2.43874624e-01
2.20017329e-01 6.41427636e-02 3.76302183e-01 1.32750177e+00
1.64742731e-02 6.69706821e-01 1.18984200e-01 9.77435112e-01
1.00180887e-01 -1.18833005e-01 -6.67907894e-01 1.36792719e-01
6.72549486e-01 8.59264016e-01 -3.36188078e-01 -8.96980524e-01
-3.30146730e-01 6.65050745e-01 3.99504542e-01 5.24792075e-01
-4.60671484e-01 -2.12516144e-01 7.96652436e-01 7.00132772e-02
-1.11225739e-01 9.44404602e-02 -1.86851785e-01 -1.06424773e+00
-3.04089099e-01 -1.14726770e+00 7.80797780e-01 -1.09742224e+00
-1.22102261e+00 -7.83753246e-02 4.46892262e-01 -6.52773440e-01
-3.08642060e-01 -6.54703617e-01 -6.82944715e-01 5.06853580e-01
-1.22927165e+00 -6.93894446e-01 -1.48764417e-01 7.30745971e-01
6.61956012e-01 4.01312619e-01 1.05223775e+00 -4.60999787e-01
1.97371654e-02 4.58250046e-01 -7.16780543e-01 -1.83949620e-01
4.87796634e-01 -1.92425871e+00 1.66867554e-01 7.08608150e-01
3.91811788e-01 1.06210911e+00 1.49526858e+00 -8.11838984e-01
-1.04496443e+00 -5.83759308e-01 1.04910493e+00 -6.10172093e-01
7.91660070e-01 -4.30908948e-01 -1.40773642e+00 1.22242367e+00
1.53860196e-01 -6.46803528e-02 1.20644414e+00 5.06722748e-01
-8.30877841e-01 1.35575697e-01 -1.31569397e+00 6.89143002e-01
1.19191933e+00 -7.76923895e-01 -1.37261391e+00 8.06013882e-01
9.66479897e-01 -6.20988369e-01 -4.56064135e-01 1.89747214e-01
3.31677616e-01 -9.44308639e-01 8.06525171e-01 -1.52722847e+00
3.88148248e-01 -1.01989910e-01 -1.85870096e-01 -1.25373030e+00
7.43875355e-02 -8.95236015e-01 -4.65285808e-01 8.93412292e-01
3.00549746e-01 -4.36152756e-01 7.08580256e-01 1.21216047e+00
-3.57520878e-02 -5.53082764e-01 -8.66724730e-01 -7.66646802e-01
-2.59453878e-02 -7.27621317e-01 8.11884463e-01 1.28793919e+00
6.08756542e-01 2.90540576e-01 6.49974421e-02 4.39535439e-01
7.60748625e-01 -1.55184465e-02 5.74405909e-01 -1.45143247e+00
-2.34253615e-01 -1.87646076e-01 -9.26583931e-02 -1.21560216e+00
3.33561689e-01 -6.39799237e-01 1.90422595e-01 -9.83895481e-01
1.65193658e-02 -3.99826586e-01 -2.16951296e-01 7.12678552e-01
-5.32513022e-01 -6.53135717e-01 1.35284692e-01 -3.79889786e-01
-1.04325557e+00 2.05662534e-01 9.38233018e-01 2.59268023e-02
4.10455950e-02 1.00067109e-02 -8.05045605e-01 1.05545986e+00
6.73714638e-01 -4.80395168e-01 -7.57123351e-01 -3.54683667e-01
9.25223172e-01 3.24452341e-01 3.95917743e-01 -8.25056553e-01
4.68867838e-01 -7.05906868e-01 1.84870020e-01 -2.76728123e-01
7.56580383e-02 -1.01574647e+00 -1.26907200e-01 4.51905817e-01
-1.13899446e+00 -1.78257570e-01 3.59698445e-01 5.40572524e-01
-7.43262097e-02 -4.91097063e-01 4.08201486e-01 -4.28474873e-01
-1.03611708e+00 -8.36831704e-02 -2.22988904e-01 4.52263266e-01
1.05879581e+00 -7.18120812e-03 -2.76118696e-01 -4.44786817e-01
-9.81228292e-01 2.97180265e-01 -1.37839735e-01 4.01501656e-01
8.08315516e-01 -9.07843709e-01 -5.52844465e-01 1.87474385e-01
5.70894718e-01 5.77899992e-01 5.80141172e-02 1.03939682e-01
-1.71438292e-01 2.40172714e-01 -1.04938552e-01 -3.87914211e-01
-1.44380748e+00 7.47469902e-01 3.06184322e-01 1.01898380e-01
-3.94568920e-01 1.55064416e+00 3.06046009e-01 -3.70063245e-01
8.74590397e-01 -6.27358258e-01 -6.03516996e-02 1.58517420e-01
7.34151542e-01 -9.95589513e-03 -1.01851607e-02 2.08768830e-01
-1.05446398e-01 -6.32014945e-02 -4.29388583e-01 -1.18761562e-01
1.13698292e+00 -3.56238782e-02 -7.86663741e-02 1.00724757e+00
7.46790290e-01 -1.22600354e-01 -1.22227204e+00 -6.85704172e-01
5.63192070e-01 -7.70720899e-01 -2.84801602e-01 -1.28946364e+00
-7.72391260e-02 8.71545196e-01 3.16411078e-01 3.48375261e-01
8.08671832e-01 3.86444449e-01 -1.04009598e-01 9.52564955e-01
4.42439348e-01 -1.12209344e+00 3.49178851e-01 2.79029012e-01
9.76635873e-01 -1.37183154e+00 2.92077750e-01 -3.65693688e-01
-4.75265861e-01 1.12230158e+00 1.16483951e+00 5.24004459e-01
3.05771500e-01 9.80879366e-03 -2.54377839e-03 -5.44034481e-01
-1.21125901e+00 2.85554547e-02 -3.61973606e-02 6.14409626e-01
5.60580432e-01 4.26339746e-01 3.76715362e-02 3.34320545e-01
-5.62638581e-01 -2.32219413e-01 6.66069269e-01 1.06176567e+00
-8.04211557e-01 -6.06601179e-01 -2.13773787e-01 4.87915486e-01
-2.83383548e-01 -2.69583493e-01 -5.33022046e-01 7.84677088e-01
3.70516598e-01 9.21380281e-01 -1.26451358e-01 6.66580275e-02
2.09810898e-01 3.91154379e-01 6.12357914e-01 -1.14099097e+00
-5.66729963e-01 -6.97855413e-01 6.13849610e-02 -6.39516711e-01
-2.73967177e-01 -6.68849289e-01 -1.39778447e+00 -2.03369409e-02
-2.97505260e-01 2.51847029e-01 5.48596859e-01 1.24103630e+00
-2.35160455e-01 1.83796838e-01 5.18301800e-02 -6.34735972e-02
-8.99669349e-01 -7.76444793e-01 -4.33860302e-01 6.30596161e-01
6.71268642e-01 -6.41972482e-01 -5.71408391e-01 1.20144807e-01] | [9.932011604309082, 7.60526180267334] |
dccac65b-5eb6-4c1c-871d-8b16ad1ea0fb | demystifying-oversmoothing-in-attention-based | 2305.16102 | null | https://arxiv.org/abs/2305.16102v1 | https://arxiv.org/pdf/2305.16102v1.pdf | Demystifying Oversmoothing in Attention-Based Graph Neural Networks | Oversmoothing in Graph Neural Networks (GNNs) refers to the phenomenon where increasing network depth leads to homogeneous node representations. While previous work has established that Graph Convolutional Networks (GCNs) exponentially lose expressive power, it remains controversial whether the graph attention mechanism can mitigate oversmoothing. In this work, we provide a definitive answer to this question through a rigorous mathematical analysis, by viewing attention-based GNNs as nonlinear time-varying dynamical systems and incorporating tools and techniques from the theory of products of inhomogeneous matrices and the joint spectral radius. We establish that, contrary to popular belief, the graph attention mechanism cannot prevent oversmoothing and loses expressive power exponentially. The proposed framework extends the existing results on oversmoothing for symmetric GCNs to a significantly broader class of GNN models. In particular, our analysis accounts for asymmetric, state-dependent and time-varying aggregation operators and a wide range of common nonlinear activation functions, such as ReLU, LeakyReLU, GELU and SiLU. | ['Ali Jadbabaie', 'Zihui Wu', 'Amir Ajorlou', 'Xinyi Wu'] | 2023-05-25 | null | null | null | null | ['graph-attention'] | ['graphs'] | [ 1.89602554e-01 6.70762658e-01 1.66419193e-01 9.49635133e-02
1.54507488e-01 -5.72144806e-01 5.94031394e-01 2.05695629e-01
-8.88351500e-02 6.35112464e-01 5.28712682e-02 -6.87694252e-01
-4.17477310e-01 -7.17287779e-01 -8.03365767e-01 -9.68036056e-01
-6.63591504e-01 -1.51015241e-02 1.08857252e-01 -6.84853077e-01
-4.55660522e-02 7.42766738e-01 -9.27757084e-01 -4.40225661e-01
9.22473550e-01 6.81956232e-01 -3.66393358e-01 9.57321465e-01
1.94150820e-01 1.03635406e+00 -2.89368957e-01 -3.70098472e-01
3.74783307e-01 -4.11255896e-01 -7.60023892e-01 -1.23303525e-01
7.56107152e-01 1.08636186e-01 -7.86801875e-01 1.46952772e+00
3.53780836e-01 4.92243767e-01 6.77627861e-01 -1.45952713e+00
-1.35935366e+00 1.03913796e+00 -4.85700548e-01 6.99592054e-01
-3.63052815e-01 1.98136911e-01 1.17422283e+00 -5.83033919e-01
5.30652225e-01 1.23663759e+00 1.19391108e+00 4.73883718e-01
-1.58170772e+00 -3.88135195e-01 5.56826413e-01 -1.73961133e-01
-1.29671311e+00 -8.73154774e-02 9.07464087e-01 -4.04671133e-01
1.00003386e+00 3.64272892e-01 7.05881178e-01 8.97725701e-01
4.71634895e-01 5.52237153e-01 6.47959054e-01 -1.34328336e-01
9.91980806e-02 -2.20065936e-01 6.14237487e-01 8.98873568e-01
6.71432555e-01 5.15404940e-02 4.18360494e-02 2.53065787e-02
1.08178926e+00 -1.00831218e-01 -5.07459462e-01 -5.41422129e-01
-9.36858892e-01 8.73300314e-01 8.80477846e-01 4.90177929e-01
-4.80044097e-01 6.06724262e-01 3.72306943e-01 5.84996283e-01
6.84690714e-01 5.92760086e-01 -2.44319350e-01 3.54026765e-01
-5.86183727e-01 5.06919846e-02 8.25742304e-01 8.50974858e-01
7.43963003e-01 7.00169802e-01 -7.46018067e-02 2.06623241e-01
7.75415674e-02 2.76224971e-01 1.34389058e-01 -9.21273232e-01
3.25473279e-01 5.79091251e-01 -2.29367673e-01 -1.34174812e+00
-7.91483283e-01 -8.27640474e-01 -1.56161034e+00 1.81603193e-01
6.17021024e-01 -3.10863763e-01 -7.66772509e-01 2.16908741e+00
5.52244671e-02 2.70370960e-01 -7.70096928e-02 8.86884093e-01
6.71544254e-01 5.54773390e-01 -8.90596807e-02 -1.55472398e-01
7.84132242e-01 -8.04243267e-01 -8.37649167e-01 9.93961096e-02
6.71364903e-01 -4.85467911e-02 1.03034496e+00 1.63149551e-01
-1.23181295e+00 -2.27144942e-01 -9.57181633e-01 -2.20929727e-01
-6.65151894e-01 -3.55822980e-01 8.18058848e-01 5.65775931e-01
-1.86679709e+00 9.75578725e-01 -8.77971947e-01 -5.58021963e-01
4.00520414e-01 5.96743345e-01 -3.07711184e-01 3.13231468e-01
-1.42744434e+00 7.74475873e-01 2.54592091e-01 6.63895190e-01
-6.64901674e-01 -8.63893509e-01 -9.54440713e-01 2.33864471e-01
2.33843684e-01 -7.15707541e-01 8.27951670e-01 -1.21409905e+00
-1.33760631e+00 4.95480180e-01 4.33896244e-01 -1.08383882e+00
5.43477774e-01 -4.20008972e-02 -1.59629375e-01 5.47382385e-02
-3.72398555e-01 2.04738677e-01 8.05404007e-01 -9.05462861e-01
2.65078932e-01 -2.37745374e-01 4.82672453e-01 1.37591571e-01
-4.53808516e-01 -3.61857980e-01 1.91762045e-01 -7.59084940e-01
1.29285166e-02 -9.26339209e-01 -6.00034654e-01 -2.02768192e-01
-5.89898050e-01 -1.08430289e-01 7.55576372e-01 -4.94311690e-01
1.33104992e+00 -1.94147325e+00 6.69208109e-01 3.23806852e-01
9.41112638e-01 2.65516758e-01 -2.12636501e-01 5.74100077e-01
-5.21127343e-01 4.32070941e-01 -3.60986054e-01 -4.40344661e-02
8.33629370e-02 2.36381173e-01 -2.36154079e-01 9.32772160e-01
4.19096053e-01 1.29941213e+00 -9.56083238e-01 -1.25153929e-01
1.07130289e-01 7.10790336e-01 -6.40014470e-01 -3.95844191e-01
-1.41254246e-01 1.71806991e-01 -2.72103786e-01 3.14977735e-01
5.60943842e-01 -5.85856199e-01 5.97434975e-02 -6.66681528e-02
-1.19119503e-01 -2.44940400e-01 -9.46409941e-01 1.17308640e+00
-5.90321496e-02 9.34487760e-01 4.97543633e-01 -1.47371638e+00
5.79540253e-01 1.52109340e-01 4.92294610e-01 -2.04607204e-01
3.84539366e-01 -1.03073232e-01 3.44593436e-01 -2.77830571e-01
5.69787979e-01 -7.74144009e-02 -7.46919662e-02 4.23807383e-01
1.49394855e-01 1.86382726e-01 3.19219679e-01 6.33692563e-01
1.44595277e+00 -3.37211788e-01 2.19406858e-01 -1.00447452e+00
5.83672047e-01 -4.43725288e-01 7.12977052e-02 9.25583243e-01
-3.43315601e-01 1.20282687e-01 1.08912206e+00 -4.21253145e-01
-1.08690619e+00 -1.08499432e+00 1.33368507e-01 1.18682897e+00
5.46806790e-02 -2.71129340e-01 -9.48803067e-01 -4.10060644e-01
6.29556403e-02 2.45955274e-01 -1.15176177e+00 -6.48770511e-01
-6.10978723e-01 -8.35770130e-01 6.68053389e-01 4.38836455e-01
3.62941265e-01 -1.03273785e+00 -8.36249739e-02 8.04410577e-02
2.76708156e-01 -7.84305215e-01 -6.35330200e-01 1.90189645e-01
-1.05248606e+00 -9.65340912e-01 -9.85267699e-01 -7.88665414e-01
8.88306856e-01 8.74576569e-02 1.18321848e+00 3.03984046e-01
-1.03521861e-01 4.79361802e-01 -1.47441328e-02 -1.54326811e-01
-5.14143705e-01 3.75823230e-01 1.59016952e-01 1.63495839e-01
-1.84037387e-01 -7.92416751e-01 -3.97031218e-01 -9.19767767e-02
-1.12443388e+00 -3.14628296e-02 4.11579907e-01 7.12175369e-01
1.43780977e-01 1.00023896e-01 6.11471653e-01 -1.00700474e+00
1.21202385e+00 -5.56280911e-01 -5.64793825e-01 9.51870382e-02
-6.95480883e-01 4.50369686e-01 8.85419130e-01 -7.06227124e-01
-4.98345524e-01 -2.79708028e-01 2.97486126e-01 -5.55107594e-01
3.30674469e-01 7.65704691e-01 2.30715707e-01 -4.42928135e-01
7.44529665e-01 1.14782043e-01 2.36179084e-01 5.32885864e-02
6.35508716e-01 -1.30776435e-01 5.00944793e-01 -3.29310030e-01
9.58870471e-01 4.89261121e-01 3.27131003e-01 -1.22911942e+00
-5.66806078e-01 -2.55944803e-02 -4.87015456e-01 -3.52092385e-01
6.75292075e-01 -3.89283717e-01 -8.41083527e-01 5.34424186e-01
-1.00001049e+00 -8.02313268e-01 -4.85913277e-01 1.58032849e-01
-4.10977393e-01 5.35261691e-01 -1.13026619e+00 -1.09221542e+00
-4.72256660e-01 -6.97683096e-01 7.14675963e-01 9.36163664e-02
-3.29362266e-02 -1.59423387e+00 -1.01321787e-02 -3.71614963e-01
7.13558555e-01 4.55001384e-01 9.34002817e-01 -5.80511034e-01
-4.48209465e-01 -2.02970609e-01 -3.81964892e-01 3.11461359e-01
-1.84237778e-01 2.33161062e-01 -4.16273147e-01 -4.17588174e-01
-1.37153968e-01 1.64809808e-01 1.02157450e+00 8.38519156e-01
7.88444936e-01 -5.75889826e-01 -2.20402107e-02 6.32552266e-01
1.38568115e+00 -1.75449133e-01 5.52821219e-01 -9.39462334e-02
1.06054151e+00 4.79693830e-01 -4.02424306e-01 9.88159925e-02
1.68877557e-01 9.88146663e-02 7.07552612e-01 -3.17328542e-01
2.13646591e-02 8.44594240e-02 3.26354951e-01 9.73367453e-01
-4.77616191e-01 -4.43659902e-01 -9.20440376e-01 5.17269433e-01
-1.90270758e+00 -9.82558072e-01 -3.63749534e-01 1.91916859e+00
3.68790060e-01 1.52242169e-01 2.58890480e-01 1.42603323e-01
1.01370776e+00 4.31864500e-01 -6.67380214e-01 -5.31959355e-01
-3.46081883e-01 4.62029427e-02 8.91245484e-01 8.56685698e-01
-1.04887462e+00 9.72258329e-01 6.78399086e+00 3.90293807e-01
-1.28390336e+00 -1.98315121e-02 5.80232561e-01 9.22056735e-02
-4.41535980e-01 -1.71255931e-01 -2.35582292e-01 1.98212832e-01
9.25876021e-01 -3.88177693e-01 7.72695005e-01 3.90666008e-01
2.29937494e-01 5.32341719e-01 -9.44315493e-01 5.97436547e-01
-3.33522186e-02 -1.31419098e+00 -8.98290202e-02 2.36539677e-01
7.95969248e-01 2.42647156e-01 3.93614143e-01 4.44859892e-01
7.01228499e-01 -1.20163369e+00 4.91556078e-01 6.64989591e-01
4.04898435e-01 -5.45961499e-01 5.62920392e-01 4.06310931e-02
-1.32270908e+00 -9.58885029e-02 -3.21172565e-01 -3.26077729e-01
4.68306839e-02 5.27813077e-01 -3.97827148e-01 3.92420799e-01
4.03169692e-01 1.02299738e+00 -6.11198187e-01 6.68224752e-01
1.53275922e-01 6.72043025e-01 -4.08942163e-01 -2.38869920e-01
5.30841470e-01 -5.65488696e-01 7.87294984e-01 1.02692568e+00
1.92415103e-01 -7.67826289e-02 -2.41615012e-01 1.03826511e+00
-3.21940742e-02 5.22119887e-02 -8.64854336e-01 -4.70448673e-01
-1.52561992e-01 1.22791290e+00 -9.20329750e-01 -2.37535298e-01
-2.65185326e-01 7.52917230e-01 6.49916232e-01 9.17280674e-01
-9.18154955e-01 -4.39553946e-01 5.29708922e-01 1.94555327e-01
2.71340907e-01 -4.64525491e-01 -1.84140205e-01 -1.23469746e+00
-1.10767610e-01 -3.74989331e-01 4.51505154e-01 -6.32931888e-01
-1.23403811e+00 6.73982382e-01 -1.72308803e-01 -7.62765825e-01
8.46484676e-02 -6.67800069e-01 -8.38613749e-01 6.86846137e-01
-1.19255221e+00 -1.01261997e+00 -1.40213296e-01 5.44839799e-01
2.86535248e-02 1.36782289e-01 2.56027192e-01 3.32835615e-01
-8.94736171e-01 4.39707249e-01 3.44754048e-02 1.91882432e-01
4.22787778e-02 -1.62140000e+00 6.46309614e-01 1.10862017e+00
-1.76591817e-02 8.84914994e-01 1.08232558e+00 -6.87321126e-01
-1.49116492e+00 -1.13976443e+00 5.27613163e-01 -2.31006414e-01
1.32752037e+00 -4.25060451e-01 -1.18873978e+00 8.31478179e-01
2.05064297e-01 3.95917177e-01 2.43195780e-02 4.84910309e-02
-2.96211064e-01 9.99371409e-02 -8.35776389e-01 9.29089427e-01
1.31719816e+00 -4.00470942e-01 -3.55707332e-02 2.82425344e-01
8.30429316e-01 -2.54559666e-01 -9.40495014e-01 2.03209728e-01
3.03696930e-01 -7.79719889e-01 9.58644867e-01 -1.09720254e+00
2.13182420e-01 1.24782173e-05 1.10007420e-01 -1.48272312e+00
-5.71493566e-01 -1.03769910e+00 -5.04025221e-01 6.31589234e-01
4.83664066e-01 -9.59626436e-01 7.44418263e-01 4.96128410e-01
-3.20319563e-01 -6.79992735e-01 -7.32488334e-01 -8.65118444e-01
5.70982277e-01 -1.29475772e-01 2.29448631e-01 1.09494948e+00
1.41329825e-01 2.47133151e-01 -3.25197279e-01 2.80652612e-01
6.28428400e-01 -2.45884866e-01 3.57006192e-01 -1.47807240e+00
-2.66708732e-02 -9.84098136e-01 -6.55925453e-01 -7.38676608e-01
3.82931650e-01 -1.23406458e+00 -1.68233916e-01 -1.46214771e+00
-2.64091611e-01 -1.78874910e-01 -4.41302180e-01 1.83873296e-01
-2.36610183e-03 2.32219398e-01 1.72225311e-02 -5.39487600e-02
-7.05631435e-01 3.28670859e-01 1.40997303e+00 -9.10602659e-02
-1.70169443e-01 -8.96253735e-02 -8.62762749e-01 6.34623885e-01
9.39717531e-01 -8.59294385e-02 -6.20298445e-01 -2.17391223e-01
7.59722173e-01 -6.59722686e-02 6.09549284e-01 -8.27527404e-01
2.72194445e-01 1.66835040e-02 -1.94507837e-01 -2.55506597e-02
1.55198351e-02 -6.63234890e-01 8.05967376e-02 6.09012485e-01
-5.55087924e-01 2.47327745e-01 1.76386327e-01 8.67122710e-01
1.80307999e-01 1.89403370e-01 7.77597308e-01 -7.49311000e-02
-3.67978513e-01 7.11173534e-01 -7.34683871e-01 3.30885857e-01
8.68543983e-01 -5.04714511e-02 -4.83697981e-01 -7.64910161e-01
-1.07536793e+00 1.06177084e-01 3.08475018e-01 1.28195554e-01
2.64786720e-01 -1.32804799e+00 -6.02019370e-01 2.23678470e-01
-2.55196631e-01 -1.38165638e-01 4.61600244e-01 1.30458021e+00
-6.34225011e-01 2.42213190e-01 -2.48773415e-02 -3.89064878e-01
-8.89264047e-01 7.03629434e-01 7.77288437e-01 -2.64912903e-01
-7.74771869e-01 8.29031348e-01 3.70099425e-01 -3.88668180e-01
1.68529943e-01 -9.98116493e-01 -2.99373697e-02 -5.73056266e-02
8.65080953e-02 4.45455581e-01 -7.10063875e-02 -6.18964374e-01
-5.36948331e-02 2.99440980e-01 8.33681449e-02 1.87699854e-01
1.26166916e+00 -1.95088059e-01 -3.61892462e-01 5.05916417e-01
1.12760127e+00 -3.15713942e-01 -1.36490846e+00 -1.40928775e-01
-9.99861062e-02 4.91361797e-01 6.94026127e-02 -1.20312534e-01
-1.33917248e+00 8.88132930e-01 1.59030437e-01 1.34620869e+00
8.69151771e-01 8.49995539e-02 5.43777823e-01 5.06506741e-01
-1.01473480e-01 -9.20768023e-01 -1.01938181e-01 7.57220805e-01
1.03975523e+00 -8.20362508e-01 -1.65762573e-01 -3.51874620e-01
-2.02466518e-01 9.00642872e-01 4.80389297e-01 -7.71204233e-01
9.90882754e-01 3.21040899e-01 -3.48967373e-01 -4.17190462e-01
-7.07171679e-01 -1.17066935e-01 2.46869311e-01 6.18207633e-01
3.86606604e-01 -1.09387115e-02 -9.52491537e-02 3.48223656e-01
-2.25817829e-01 -3.19849283e-01 9.31052983e-01 4.70507026e-01
-2.42540330e-01 -3.91793162e-01 -4.72330563e-02 5.65334558e-01
-5.53866446e-01 -3.11856776e-01 -5.51679134e-01 1.04921484e+00
-4.96258855e-01 3.87413114e-01 1.80103019e-01 -2.37525359e-01
1.64455250e-01 -1.10725395e-01 4.14640695e-01 -2.99544424e-01
-6.53462529e-01 -2.52287298e-01 -1.36475161e-01 -3.36465210e-01
-4.67328966e-01 -4.82807189e-01 -9.56940949e-01 -6.24333620e-01
-5.00323057e-01 2.57402901e-02 2.12301731e-01 6.84203446e-01
3.35926235e-01 8.65437627e-01 4.99938689e-02 -9.29204404e-01
-5.55308521e-01 -9.51636374e-01 -8.76783729e-01 3.19887877e-01
8.52773130e-01 -5.14640391e-01 -8.97577465e-01 -1.94998711e-01] | [6.832408905029297, 6.025649070739746] |
0166eb9f-0a5e-42d7-8db1-389e07bad192 | xview3-sar-detecting-dark-fishing-activity | 2206.00897 | null | https://arxiv.org/abs/2206.00897v4 | https://arxiv.org/pdf/2206.00897v4.pdf | xView3-SAR: Detecting Dark Fishing Activity Using Synthetic Aperture Radar Imagery | Unsustainable fishing practices worldwide pose a major threat to marine resources and ecosystems. Identifying vessels that do not show up in conventional monitoring systems -- known as ``dark vessels'' -- is key to managing and securing the health of marine environments. With the rise of satellite-based synthetic aperture radar (SAR) imaging and modern machine learning (ML), it is now possible to automate detection of dark vessels day or night, under all-weather conditions. SAR images, however, require a domain-specific treatment and are not widely accessible to the ML community. Maritime objects (vessels and offshore infrastructure) are relatively small and sparse, challenging traditional computer vision approaches. We present the largest labeled dataset for training ML models to detect and characterize vessels and ocean structures in SAR imagery. xView3-SAR consists of nearly 1,000 analysis-ready SAR images from the Sentinel-1 mission that are, on average, 29,400-by-24,400 pixels each. The images are annotated using a combination of automated and manual analysis. Co-located bathymetry and wind state rasters accompany every SAR image. We also provide an overview of the xView3 Computer Vision Challenge, an international competition using xView3-SAR for ship detection and characterization at large scale. We release the data (\href{https://iuu.xview.us/}{https://iuu.xview.us/}) and code (\href{https://github.com/DIUx-xView}{https://github.com/DIUx-xView}) to support ongoing development and evaluation of ML approaches for this important application. | ['Jared Dunnmon', 'David Kroodsma', 'Daniel Kuster', 'Nirav Patel', 'Bryce Goodman', 'Ritwik Gupta', 'Tsu-ting Tim Lin', 'Fernando Paolo'] | 2022-06-02 | null | null | null | null | ['holdout-set', 'decision-making-under-uncertainty', 'decision-making-under-uncertainty'] | ['computer-vision', 'medical', 'reasoning'] | [ 2.14986168e-02 -1.65822297e-01 4.01151001e-01 -4.35483485e-01
-9.36941803e-01 -1.12335372e+00 4.21751648e-01 7.84719661e-02
-6.44922316e-01 3.96380246e-01 9.46966633e-02 -3.54612678e-01
-1.39053538e-01 -8.88280869e-01 -5.54036677e-01 -6.77826643e-01
-7.12388635e-01 1.59298465e-01 2.36706108e-01 -3.48515272e-01
8.87585580e-02 7.88785338e-01 -1.28621984e+00 9.74483863e-02
4.15222734e-01 9.74217594e-01 3.31981361e-01 9.09272969e-01
3.37799251e-01 3.17560017e-01 -3.43012810e-01 -8.23518112e-02
9.27275181e-01 -4.56489831e-01 -2.98742503e-01 -4.13469039e-02
8.13789666e-01 -3.56067330e-01 -2.66818821e-01 1.27861595e+00
5.45715511e-01 1.65181868e-02 4.74730045e-01 -5.80750942e-01
-2.65883982e-01 3.83825600e-01 -7.01552391e-01 8.26511681e-01
-2.45009780e-01 4.85234618e-01 1.11079860e+00 -9.32663500e-01
5.84955096e-01 7.49813557e-01 9.73006129e-01 2.76257932e-01
-1.05653119e+00 -7.09663033e-01 3.59835140e-02 -1.67141616e-01
-1.50427687e+00 -7.83309460e-01 2.81011343e-01 -7.48935759e-01
4.99069512e-01 5.52584231e-01 8.61330509e-01 4.06552196e-01
2.52996795e-02 4.88993675e-01 1.18271041e+00 1.42829530e-02
1.21959418e-01 -3.45038235e-01 -2.25149035e-01 6.51852727e-01
4.44207966e-01 3.40128988e-01 -4.53797281e-01 -2.16193274e-01
9.92758334e-01 9.72084235e-03 -4.98205453e-01 1.11693650e-01
-1.13940418e+00 1.06445265e+00 4.64578420e-01 -1.42501026e-01
-3.35042775e-01 9.29040089e-03 2.05381572e-01 5.33391893e-01
3.85608315e-01 8.05100620e-01 -5.41800439e-01 9.98911187e-02
-1.07679093e+00 6.43035471e-02 6.59052730e-01 6.95801973e-01
8.85334790e-01 6.22283995e-01 5.52953720e-01 7.37338424e-01
4.53384995e-01 9.93006289e-01 -1.03216775e-01 -9.11604941e-01
1.21731438e-01 4.11080629e-01 1.95771694e-01 -9.08960879e-01
-7.56712317e-01 -6.02180362e-01 -7.22816885e-01 6.11066520e-01
3.50075752e-01 -7.70212233e-01 -7.57960320e-01 1.06558359e+00
1.45338863e-01 2.80117653e-02 2.94853717e-01 1.13395274e+00
1.16850543e+00 7.13075757e-01 -1.83799416e-01 -1.97639793e-01
1.54629385e+00 -4.88512278e-01 -1.56226873e-01 -6.53532028e-01
4.93997604e-01 -7.06176698e-01 6.06138587e-01 1.28578573e-01
-6.08809590e-01 -1.17552660e-01 -9.72328663e-01 5.87855160e-01
-3.36267561e-01 3.33643466e-01 5.78999817e-01 2.83277363e-01
-8.51386666e-01 3.31348121e-01 -1.13274908e+00 -2.67960995e-01
6.95625186e-01 6.37467057e-02 -6.05106652e-01 2.86993086e-01
-7.69181252e-01 8.14664006e-01 -3.35151941e-01 6.52808070e-01
-1.46868157e+00 -7.74003506e-01 -1.16927254e+00 -4.37575221e-01
3.51381898e-02 3.15178096e-01 1.16646945e+00 -9.25773025e-01
-6.83247983e-01 1.01165426e+00 3.53656739e-01 -5.41457772e-01
4.40355480e-01 -3.66512269e-01 -5.62281013e-01 4.71084595e-01
2.24703789e-01 4.28219289e-01 4.05913353e-01 -1.28315520e+00
-9.85565782e-01 -4.08589691e-01 -1.17646478e-01 1.61939695e-01
5.95400259e-02 4.67478037e-01 -7.75437336e-03 -7.14558244e-01
1.96054965e-01 -7.95231521e-01 -3.38511288e-01 4.49200660e-01
-1.56427115e-01 5.62434673e-01 7.22807407e-01 -9.49542761e-01
6.59655094e-01 -2.15250134e+00 -1.90302610e-01 -1.49802733e-02
-1.36023045e-01 3.28355372e-01 -1.55253395e-01 6.50238276e-01
1.85202554e-01 4.74445336e-02 -6.22976601e-01 -5.45324497e-02
-2.98306853e-01 2.42916018e-01 -3.15536171e-01 1.12142801e+00
9.19514894e-02 2.87957370e-01 -7.89668560e-01 -1.13076173e-01
1.76139727e-01 1.67321220e-01 9.83197987e-03 1.51643664e-01
3.64519767e-02 7.29154527e-01 -4.22680348e-01 1.25994325e+00
8.88443112e-01 2.11214900e-01 7.79631510e-02 -2.35717043e-01
-1.01881468e+00 -3.45932513e-01 -1.15745103e+00 1.29091752e+00
4.64042164e-02 1.14134943e+00 8.83997917e-01 -7.40093827e-01
1.07182658e+00 6.29316345e-02 2.17536822e-01 -5.55690110e-01
-5.21350242e-02 6.76212385e-02 -1.84246637e-02 -6.58295810e-01
3.15226644e-01 -3.88440043e-01 -1.11472838e-01 1.08880423e-01
-2.79728830e-01 -1.91510245e-01 7.15582371e-02 -1.73970863e-01
8.35889995e-01 8.93664509e-02 2.76470572e-01 -6.03264451e-01
1.06311940e-01 8.01831186e-01 9.01369929e-01 6.45395577e-01
-1.45790443e-01 5.91167271e-01 2.70725787e-01 -9.55669522e-01
-8.75842869e-01 -9.13597226e-01 -6.10355854e-01 9.98361945e-01
1.20900132e-01 2.05064446e-01 -2.21475303e-01 -3.75179857e-01
1.38082623e-01 3.01748663e-01 -7.70134449e-01 4.40912932e-01
-4.86248314e-01 -1.09187043e+00 9.63194430e-01 4.55558538e-01
4.54259098e-01 -8.46068680e-01 -1.17653608e+00 2.49525979e-02
1.39454573e-01 -9.20396209e-01 -3.63300219e-02 1.50477543e-01
-7.52468050e-01 -1.08944893e+00 -4.64863926e-01 -5.58400214e-01
6.42538786e-01 3.45320106e-01 9.96200621e-01 1.74462393e-01
-7.22559631e-01 2.04945177e-01 -6.20723844e-01 -5.95119774e-01
-1.01439029e-01 -4.97881621e-01 1.84217736e-01 3.98264751e-02
1.31404221e-01 -3.86074066e-01 -8.98499370e-01 4.34752941e-01
-6.51872754e-01 -3.71907294e-01 6.77584887e-01 6.36018813e-01
5.54086447e-01 -3.52012187e-01 3.68591994e-01 -4.96072263e-01
-3.78973275e-01 -4.70697343e-01 -1.24066269e+00 -5.25373779e-02
1.06861405e-01 -5.29304624e-01 2.51369327e-01 3.37724835e-02
-5.51082194e-01 3.91027033e-01 -2.50228018e-01 -2.70633161e-01
-4.79371816e-01 8.40304673e-01 2.92226821e-01 -4.36429590e-01
8.97770703e-01 3.27835113e-01 3.51517856e-01 -8.11110377e-01
-1.14022821e-01 8.27605605e-01 6.61046147e-01 1.09721079e-01
1.11626148e+00 1.03755641e+00 -1.01461746e-01 -1.44139767e+00
-9.21137214e-01 -6.68811023e-01 -5.59599519e-01 -3.27760339e-01
8.81281495e-01 -1.45906401e+00 -2.60682166e-01 3.58038753e-01
-4.56584305e-01 -6.47290468e-01 -8.71846899e-02 9.05812502e-01
7.80182034e-02 3.39743465e-01 -3.45649391e-01 -9.32407677e-01
-4.03188586e-01 -1.02046776e+00 9.28945661e-01 3.84472251e-01
2.60903299e-01 -9.65260208e-01 2.57708192e-01 2.08859906e-01
4.96419191e-01 6.87678397e-01 -1.38872880e-02 -6.71818495e-01
-4.09614623e-01 -2.53775686e-01 -2.06662223e-01 4.15861547e-01
9.79211405e-02 2.36258283e-01 -8.28409851e-01 -6.35306120e-01
-4.81826216e-01 -9.60794240e-02 1.02951002e+00 4.33568954e-01
2.50269711e-01 -2.83848912e-01 -3.92007418e-02 8.98594797e-01
1.60723448e+00 1.43289998e-01 4.71509844e-01 6.91869318e-01
1.78279817e-01 7.45827794e-01 9.38443482e-01 8.56478333e-01
1.88977152e-01 4.41642880e-01 1.03623998e+00 -4.40003812e-01
1.36301473e-01 4.07462120e-01 3.62237841e-01 3.23191918e-02
-5.14452875e-01 2.25480497e-01 -1.11893320e+00 8.47699702e-01
-1.26612163e+00 -1.19100118e+00 -5.36483526e-01 2.35647058e+00
4.72247809e-01 -4.15575296e-01 -1.16205566e-01 -6.09588265e-01
4.32946086e-01 2.95450002e-01 -2.83481210e-01 9.47279930e-02
-3.98294985e-01 -9.73418076e-03 1.24964201e+00 5.42617142e-01
-1.65030992e+00 8.94906521e-01 5.29843521e+00 7.59493932e-02
-1.30378819e+00 2.74052471e-03 2.03370556e-01 1.55857233e-02
9.22887474e-02 1.69378117e-01 -9.95966196e-01 4.50355895e-02
6.61974847e-01 1.97522506e-01 6.56557009e-02 7.48893499e-01
6.04907513e-01 -1.85080111e-01 -3.54222685e-01 5.54428458e-01
4.66668159e-02 -1.67013454e+00 -3.44739735e-01 1.13701031e-01
4.48726207e-01 8.29702318e-01 -3.24580342e-01 -7.83021227e-02
3.70345086e-01 -8.32238317e-01 9.98062074e-01 6.00012779e-01
9.93460715e-01 -4.01669443e-01 9.58063126e-01 5.13386242e-02
-1.30982184e+00 -1.62845358e-01 -6.40104234e-01 -2.68092960e-01
2.83066899e-01 2.13347092e-01 -4.26290214e-01 5.18360198e-01
1.30438757e+00 8.37618113e-01 -5.27857006e-01 1.44595683e+00
-1.78130060e-01 7.37015843e-01 -4.15420562e-01 2.09825918e-01
4.36779290e-01 -4.01819646e-01 9.56583917e-01 1.34415567e+00
5.61701953e-01 4.24559534e-01 2.18641654e-01 2.81289577e-01
1.98125586e-01 -1.82153750e-02 -6.31360054e-01 3.10114827e-02
3.84686500e-01 1.47910190e+00 -5.59715569e-01 4.34482507e-02
-6.64055884e-01 2.61029422e-01 -2.65170217e-01 1.56237349e-01
-3.99178356e-01 -3.51332158e-01 1.23450720e+00 1.81742728e-01
4.53359872e-01 -6.19531214e-01 8.67473055e-03 -9.47132587e-01
-3.98950398e-01 -7.21043766e-01 5.80704868e-01 -5.63576937e-01
-1.16267335e+00 8.00284922e-01 8.47548153e-03 -1.97233582e+00
3.90370101e-01 -6.03468955e-01 -6.39004230e-01 7.21833408e-01
-1.89316928e+00 -1.26227152e+00 -7.80936599e-01 6.91947713e-02
6.26253843e-01 -4.33112770e-01 9.82586324e-01 1.54189125e-01
-3.76889616e-01 -6.20709173e-02 4.00859743e-01 6.88845515e-01
5.46255350e-01 -1.06392145e+00 2.44266078e-01 1.05624604e+00
2.53443625e-02 1.73194975e-01 9.79135990e-01 -7.51975596e-01
-1.80308211e+00 -1.49734080e+00 2.10440785e-01 -2.05216780e-01
1.12580752e+00 -7.92975351e-02 -7.45674968e-01 1.14607561e+00
1.59536585e-01 3.90338987e-01 9.62826371e-01 -4.49046880e-01
3.41578089e-02 -3.53271186e-01 -9.61514354e-01 2.06548184e-01
4.70400006e-01 7.93725252e-02 -5.84234059e-01 7.22769678e-01
2.61929724e-02 -3.83922964e-01 -1.15238702e+00 5.15187562e-01
5.08075893e-01 -6.61528230e-01 7.47500658e-01 -3.81390274e-01
7.38277882e-02 -8.45409751e-01 -4.28053230e-01 -1.11312604e+00
-2.23364010e-01 -4.85820383e-01 6.76273823e-01 7.59087205e-01
7.12090790e-01 -6.58823371e-01 5.38675666e-01 -7.75582865e-02
-6.86486959e-01 -1.43278146e-03 -1.12294245e+00 -9.39150870e-01
4.18207683e-02 -1.53711244e-01 4.50174995e-02 1.03974593e+00
-4.61757630e-01 -1.87595397e-01 -5.10253251e-01 1.26862824e+00
1.09711361e+00 5.08200407e-01 8.64237785e-01 -1.41773880e+00
-1.95413500e-01 -2.16301262e-01 -6.00258350e-01 -6.11146927e-01
-5.58262944e-01 -4.82418597e-01 2.41391331e-01 -1.67384732e+00
-3.15541744e-01 -4.14485335e-01 3.50410014e-01 9.49111044e-01
4.74371523e-01 9.23706830e-01 -1.01942375e-01 5.20834565e-01
-1.09947592e-01 2.66853780e-01 7.77537644e-01 2.41565704e-02
-2.80476548e-02 7.72844329e-02 -4.58565593e-01 7.40397036e-01
8.57446671e-01 -4.59394366e-01 4.95713890e-01 -8.32534432e-01
1.44490406e-01 1.64076611e-01 5.53170323e-01 -1.04943454e+00
3.21647942e-01 -4.58307087e-01 2.42478907e-01 -5.08531034e-01
3.95467758e-01 -5.71453512e-01 3.77527326e-01 6.75478280e-01
1.75566971e-01 9.36705023e-02 4.31829721e-01 4.55767989e-01
-3.69426489e-01 -4.90779847e-01 1.23774040e+00 -5.86937487e-01
-1.19735229e+00 2.81332523e-01 -7.52616465e-01 -3.91199514e-02
1.14645553e+00 -8.18979517e-02 -7.12191582e-01 -1.34370998e-01
-7.96312094e-01 6.05339706e-01 5.56249619e-01 4.85663116e-02
6.75007045e-01 -3.71566951e-01 -1.50368059e+00 2.29790330e-01
3.94370317e-01 8.17732513e-02 6.69771016e-01 1.06323588e+00
-1.44000399e+00 -2.46852651e-01 -4.11707997e-01 -4.43354845e-01
-1.38360584e+00 -1.98616669e-01 6.29280210e-01 5.70365131e-01
-9.78070796e-01 1.03006148e+00 6.45697862e-02 -2.18773276e-01
-3.35116446e-01 5.52324951e-02 -4.45063919e-01 3.35795373e-01
1.06616950e+00 -2.43406426e-02 -1.48088157e-01 -9.91222382e-01
-4.41823572e-01 6.78781331e-01 3.60232264e-01 4.88642417e-02
1.94954252e+00 -9.09633487e-02 -1.47751480e-01 9.49724987e-02
6.32918656e-01 2.48544842e-01 -1.59639263e+00 -1.15382135e-01
-1.86612591e-01 -6.40803397e-01 3.03567916e-01 -6.46531641e-01
-1.30798495e+00 8.42348337e-01 7.06677258e-01 1.68994769e-01
8.42237115e-01 2.07703859e-01 1.16066232e-01 3.27282518e-01
8.07295591e-02 -6.82902634e-01 -4.75952476e-01 3.89319420e-01
1.09999776e+00 -1.37116265e+00 4.50157493e-01 -5.57979494e-02
-9.48704481e-01 1.27609479e+00 2.89489836e-01 -1.61739096e-01
6.05702698e-01 4.59247231e-01 8.00090551e-01 -6.09354854e-01
-3.38407040e-01 -5.40734053e-01 -1.64738998e-01 6.07837379e-01
-4.84178215e-02 2.16765717e-01 -7.17603974e-03 3.57315809e-01
-1.70677781e-01 -7.12461472e-01 8.53671134e-01 1.31204653e+00
-8.69065762e-01 -6.26610696e-01 -4.96392608e-01 5.35626471e-01
-4.16195750e-01 -2.77893782e-01 -1.10417552e-01 7.93192983e-01
-1.21912040e-01 9.22229826e-01 2.80577123e-01 -8.02782550e-02
3.70479137e-01 -5.31525314e-01 -2.34319076e-01 -7.10629761e-01
-6.04751825e-01 2.37164438e-01 5.96808076e-01 -1.71298906e-01
-6.84333205e-01 -1.24023628e+00 -1.10997283e+00 6.47983626e-02
-2.22669542e-01 5.68470180e-01 7.95100451e-01 5.73749125e-01
1.32958442e-01 1.28095159e-02 6.81966126e-01 -1.07375360e+00
-5.80099225e-01 -1.13151848e+00 -1.08374131e+00 -3.65947217e-01
4.41269636e-01 -5.50650954e-01 -8.21397483e-01 2.94318557e-01] | [8.41639518737793, -1.2212144136428833] |
61ee5075-faf5-4660-948f-d1a942138206 | zero-shot-entity-linking-with-less-data | null | null | https://aclanthology.org/2022.findings-naacl.127 | https://aclanthology.org/2022.findings-naacl.127.pdf | Zero-shot Entity Linking with Less Data | Entity Linking (EL) maps an entity mention in a natural language sentence to an entity in a knowledge base (KB). The Zero-shot Entity Linking (ZEL) extends the scope of EL to unseen entities at the test time without requiring new labeled data. BLINK (BERT-based) is one of the SOTA models for ZEL. Interestingly, we discovered that BLINK exhibits diminishing returns, i.e., it reaches 98% of its performance with just 1% of the training data and the remaining 99% of the data yields only a marginal increase of 2% in the performance. While this extra 2% gain makes a huge difference for downstream tasks, training BLINK on large amounts of data is very resource-intensive and impractical. In this paper, we propose a neuro-symbolic, multi-task learning approach to bridge this gap. Our approach boosts the BLINK’s performance with much less data by exploiting an auxiliary information about entity types. Specifically, we train our model on two tasks simultaneously - entity linking (primary task) and hierarchical entity type prediction (auxiliary task). The auxiliary task exploits the hierarchical structure of entity types. Our approach achieves superior performance on ZEL task with significantly less training data. On four different benchmark datasets, we show that our approach achieves significantly higher performance than SOTA models when they are trained with just 0.01%, 0.1%, or 1% of the original training data. Our code is available at https://github.com/IBM/NeSLET. | ['L Venkata Subramaniam', 'Alexander Gray', 'Salim Roukos', 'Pavan Kapanipathi', 'Dinesh Garg', 'Saswati Dana', 'Dinesh Khandelwal', 'G P Shrivatsa Bhargav'] | null | null | null | null | findings-naacl-2022-7 | ['type-prediction'] | ['computer-code'] | [-2.69225508e-01 4.98846620e-01 -4.91934180e-01 -1.86011359e-01
-7.91094899e-01 -6.54052436e-01 4.24332440e-01 4.89679277e-01
-7.72132218e-01 1.01126373e+00 -1.45674974e-01 -2.56153286e-01
2.29732871e-01 -9.84909058e-01 -1.08383465e+00 -2.72161812e-01
3.14666033e-02 6.05914652e-01 5.48954189e-01 -3.80313724e-01
-2.54975230e-01 3.13920528e-03 -1.23846340e+00 3.08169603e-01
1.01637805e+00 8.23732197e-01 1.65265307e-01 4.21451956e-01
-3.35666925e-01 7.29996800e-01 -5.18003523e-01 -7.58737922e-01
1.37594700e-01 3.08683440e-02 -1.07629025e+00 -5.59228539e-01
2.70459771e-01 9.60030854e-02 -2.03574985e-01 9.83161747e-01
4.66136605e-01 8.41111392e-02 4.39537138e-01 -1.40913117e+00
-7.35764444e-01 8.29464495e-01 -5.17466366e-01 2.99541622e-01
2.87820220e-01 -1.61770135e-01 1.22368598e+00 -1.16449809e+00
8.00615132e-01 9.06494617e-01 9.71427917e-01 4.56377447e-01
-1.20620012e+00 -6.68142021e-01 6.02159910e-02 1.54434636e-01
-1.46720290e+00 -4.56006885e-01 2.61388838e-01 -3.93892616e-01
1.26318491e+00 5.11646830e-02 1.94492042e-01 9.67877448e-01
-2.27449492e-01 9.61716354e-01 8.98050070e-01 -4.61402178e-01
7.34021664e-02 3.68411392e-01 5.17579079e-01 7.81489491e-01
4.51077044e-01 -2.16482282e-01 -2.98519522e-01 -1.16165534e-01
4.40921336e-01 -3.01609665e-01 -1.80338874e-01 -8.96196514e-02
-1.12230122e+00 5.65435588e-01 5.79002857e-01 3.83063316e-01
-1.47500530e-01 6.67187124e-02 4.05825257e-01 3.92124861e-01
5.73078513e-01 6.18364632e-01 -8.53019416e-01 -8.97663236e-02
-7.60719657e-01 5.81047870e-02 1.02967906e+00 1.32091439e+00
9.19155836e-01 -3.06403965e-01 -1.22638494e-01 9.45071518e-01
-4.18873690e-02 4.02048558e-01 3.89532536e-01 -5.73118985e-01
7.08721161e-01 7.98194230e-01 6.25119731e-02 -7.57163346e-01
-5.69042325e-01 -5.22127986e-01 -5.97168446e-01 -3.20625812e-01
6.22587025e-01 -4.71298993e-01 -8.79366457e-01 1.97128105e+00
3.66973996e-01 3.32250267e-01 2.72313237e-01 4.49181914e-01
1.13090611e+00 6.43944860e-01 2.89758205e-01 1.17026838e-02
1.41267776e+00 -1.06855929e+00 -6.09701276e-01 -4.37860221e-01
1.13484263e+00 -5.12522757e-01 1.02258539e+00 -7.03267604e-02
-9.86111045e-01 -3.59541625e-01 -7.96149313e-01 -2.44234785e-01
-7.40182459e-01 1.64074212e-01 6.38092339e-01 4.51230168e-01
-8.38838041e-01 5.60043454e-01 -7.83598721e-01 -3.37387949e-01
3.67723644e-01 3.55459422e-01 -5.26200414e-01 -1.11209348e-01
-1.70077634e+00 1.01975882e+00 7.33913302e-01 -8.07297751e-02
-3.12143832e-01 -1.04957044e+00 -9.31115329e-01 1.98112041e-01
7.20717311e-01 -6.68226361e-01 1.25808430e+00 -6.32388353e-01
-8.74690115e-01 8.70234072e-01 -3.34700465e-01 -4.83800828e-01
3.60807091e-01 -4.39895153e-01 -4.89464045e-01 -2.11439297e-01
3.03079337e-01 7.56840110e-01 2.27085352e-01 -1.19588816e+00
-8.85667443e-01 -5.67667633e-02 3.03572893e-01 -9.79386568e-02
-3.99787545e-01 -1.46026000e-01 -7.18542635e-01 -5.24465859e-01
-2.88755596e-01 -1.01309419e+00 -2.59728488e-02 -2.98749238e-01
-4.56776619e-01 -4.99998063e-01 3.94926906e-01 -6.47480786e-01
1.43483961e+00 -2.02407789e+00 7.10696429e-02 -3.21866423e-02
3.32943469e-01 4.70938951e-01 -2.15777725e-01 3.86515230e-01
-1.00377016e-02 3.77964348e-01 -2.56076574e-01 -2.83835500e-01
-5.85643724e-02 1.83452800e-01 1.37586743e-02 -1.00266427e-01
3.13303113e-01 1.28359973e+00 -1.12932932e+00 -3.95230025e-01
-4.15309191e-01 1.18522689e-01 -6.13302708e-01 1.25815282e-02
-1.23688601e-01 1.43253043e-01 -3.42601836e-01 6.34435892e-01
4.96392071e-01 -5.55212975e-01 2.61814117e-01 -3.20560426e-01
2.09121220e-02 4.20500368e-01 -1.06213248e+00 1.52300191e+00
-4.79021341e-01 5.08182108e-01 -4.16444361e-01 -8.36424232e-01
7.27838159e-01 4.48538244e-01 2.36259267e-01 -6.16466761e-01
-8.53605643e-02 4.20119792e-01 2.13579144e-02 -5.32601535e-01
5.20100296e-01 -8.59342664e-02 -4.04095948e-01 1.94712058e-01
2.82099962e-01 4.76219952e-01 4.46328938e-01 2.95130998e-01
1.29787493e+00 -4.99148183e-02 5.24040997e-01 2.59088799e-02
2.28084221e-01 1.45845845e-01 9.89041269e-01 7.20736623e-01
-7.66983256e-02 1.23334810e-01 5.68038285e-01 -2.24327087e-01
-1.02191281e+00 -9.28103864e-01 -1.54533029e-01 1.15919316e+00
1.56277061e-01 -5.17213047e-01 -7.10229039e-01 -1.02114904e+00
3.56599927e-01 9.10721600e-01 -6.87030792e-01 -1.90990672e-01
-6.01718366e-01 -8.84611130e-01 8.84499550e-01 6.66929662e-01
6.47828877e-01 -1.15544534e+00 -1.47294030e-01 1.10359654e-01
-4.26320702e-01 -1.33704555e+00 -1.75829634e-01 1.49786219e-01
-5.78499198e-01 -9.29258585e-01 -4.69741732e-01 -8.14769089e-01
7.04598069e-01 -7.86063522e-02 1.28787124e+00 7.58727938e-02
4.62058224e-02 -6.83497190e-02 -5.20629823e-01 -2.66842276e-01
-2.61547655e-01 5.58111191e-01 5.26137650e-02 -2.48634487e-01
6.66740239e-01 -5.08139491e-01 -1.56358883e-01 1.44830003e-01
-5.89677513e-01 1.74362957e-01 7.61595964e-01 9.83540773e-01
3.98374677e-01 -1.82735249e-01 1.10025251e+00 -1.50068891e+00
3.63507599e-01 -9.59104300e-01 -3.61774623e-01 4.35428292e-01
-6.65344596e-01 2.00977132e-01 5.55508971e-01 -5.10893881e-01
-9.89555478e-01 -1.46062136e-01 -1.40482232e-01 -3.90885860e-01
-1.50549099e-01 9.83765602e-01 -1.06833808e-01 2.25903854e-01
5.75509131e-01 -1.45606875e-01 -4.96274590e-01 -7.65034914e-01
4.24035370e-01 6.17604733e-01 6.68627024e-01 -6.62555575e-01
7.62754023e-01 -5.26135825e-02 -2.58528531e-01 -2.75814086e-01
-1.31643188e+00 -5.30136228e-01 -9.27651107e-01 1.78898990e-01
5.31416655e-01 -9.39091384e-01 -5.98653555e-01 2.44376704e-01
-1.02093315e+00 -4.81860816e-01 -2.53960520e-01 3.05113107e-01
-1.60653919e-01 1.26462787e-01 -7.11272478e-01 -4.55093771e-01
-3.68016511e-01 -7.61899889e-01 7.95974612e-01 6.47785366e-02
-1.20752625e-01 -1.07083368e+00 8.40280727e-02 3.47323388e-01
1.82729885e-01 2.09471792e-01 1.06390142e+00 -1.13590455e+00
-4.29143220e-01 -3.80617470e-01 -3.36843759e-01 6.48387000e-02
2.36549526e-01 -3.74106169e-01 -9.42123592e-01 -3.03091079e-01
-4.39349681e-01 -5.06789505e-01 8.96725595e-01 -9.99786630e-02
7.93320000e-01 -2.97964126e-01 -6.16097748e-01 5.16015947e-01
1.34425044e+00 1.49029121e-01 3.52697790e-01 5.57104886e-01
8.85936797e-01 4.76570904e-01 7.81361639e-01 1.30714267e-01
7.99004972e-01 6.98211551e-01 5.34985811e-02 -1.84877560e-01
-1.48980051e-01 -3.93851995e-01 1.94408491e-01 9.76365089e-01
-1.40870348e-01 -3.64056736e-01 -1.31204462e+00 7.85793185e-01
-2.02296805e+00 -8.82002831e-01 -1.71294436e-01 2.09702516e+00
1.38290238e+00 2.57868379e-01 1.48108928e-02 -1.99574053e-01
7.48434842e-01 -2.82027632e-01 -7.02572763e-01 -8.81021097e-02
-9.54584107e-02 8.15720782e-02 7.44814456e-01 4.25550401e-01
-1.29640937e+00 1.28441727e+00 5.53617430e+00 8.05015564e-01
-8.68780136e-01 3.68267387e-01 4.49814677e-01 -8.53952318e-02
-2.01395005e-01 3.55351381e-02 -1.26910830e+00 6.92642629e-01
1.09410286e+00 -4.85982478e-01 1.65768057e-01 7.87309408e-01
-2.14262232e-01 -3.65786031e-02 -1.23451567e+00 6.85582280e-01
-1.10209644e-01 -1.26850617e+00 -8.74662995e-02 -3.53925675e-03
6.37865186e-01 3.29325378e-01 -3.30443084e-01 8.84294748e-01
5.15072763e-01 -8.63219082e-01 5.67666829e-01 4.52783704e-01
9.04321611e-01 -6.85474336e-01 1.09665108e+00 6.17374301e-01
-1.21480417e+00 1.50487700e-03 -2.88027823e-01 6.61530644e-02
2.55369931e-01 6.55994534e-01 -9.06779408e-01 8.23860049e-01
6.32738531e-01 6.86110556e-01 -7.96835303e-01 1.00927401e+00
-3.53649735e-01 6.47594690e-01 -3.11648339e-01 9.96193588e-02
9.37536955e-02 2.43184298e-01 4.78534132e-01 1.45667398e+00
2.06267253e-01 2.24940166e-01 3.65372986e-01 7.06542492e-01
-5.87415457e-01 3.91141977e-03 -6.15411520e-01 6.45884201e-02
8.68025601e-01 1.36683500e+00 -2.86893725e-01 -5.56896746e-01
-7.54493833e-01 8.43944907e-01 9.64221716e-01 2.99307257e-01
-7.82231271e-01 -5.52581489e-01 3.03914845e-01 -1.13890179e-01
3.13635439e-01 2.47616414e-02 -1.22798942e-01 -1.33370113e+00
1.57562755e-02 -4.97668058e-01 5.66243529e-01 -6.01945877e-01
-1.43870616e+00 6.02127671e-01 -5.34450728e-03 -8.41877878e-01
-3.36776257e-01 -4.41544056e-01 -3.22307527e-01 8.53900015e-01
-1.70480597e+00 -1.16926849e+00 -2.90200830e-01 2.79975384e-01
3.40343028e-01 -1.39513180e-01 7.73754179e-01 6.87268674e-01
-9.17177916e-01 1.08192480e+00 1.10656403e-01 6.35834396e-01
9.18416083e-01 -1.47823405e+00 8.25746834e-01 7.19309151e-01
-4.94551798e-03 6.54748559e-01 4.71573979e-01 -8.03636014e-01
-1.04699707e+00 -1.36307824e+00 1.45263946e+00 -7.15057254e-01
8.79114151e-01 -3.93377274e-01 -1.32744014e+00 1.17924583e+00
-3.76546942e-02 1.97095200e-01 8.24104190e-01 5.81773520e-01
-4.19129521e-01 1.42908171e-01 -9.73784447e-01 4.26113576e-01
1.15628624e+00 -3.34654123e-01 -9.09042716e-01 1.78868011e-01
9.09395754e-01 -4.69764084e-01 -1.31661093e+00 5.94222784e-01
3.40442568e-01 -3.99720550e-01 7.85974681e-01 -7.99340427e-01
4.42784756e-01 -2.47137263e-01 -8.40025693e-02 -1.42912602e+00
-3.84472340e-01 -1.62049577e-01 -4.42764193e-01 1.57359886e+00
9.95528996e-01 -7.44483948e-01 6.19571209e-01 7.69766033e-01
-2.10177168e-01 -1.02485967e+00 -6.76262558e-01 -1.05177915e+00
1.96509689e-01 -1.68354437e-01 5.14616311e-01 1.42262769e+00
1.90476328e-01 6.05698586e-01 -8.59837979e-02 2.28030041e-01
4.06096071e-01 1.73056841e-01 6.70466542e-01 -1.32741356e+00
-1.93598881e-01 -8.13253820e-02 -1.51258081e-01 -7.84175217e-01
4.45771664e-01 -1.38572621e+00 -1.75087333e-01 -1.48696613e+00
3.83586556e-01 -9.26061332e-01 -4.63047087e-01 1.24967027e+00
-6.05612338e-01 1.24292724e-01 3.01230669e-01 3.45183760e-01
-5.70774555e-01 3.84472817e-01 8.12189400e-01 -1.06027894e-01
-2.31980845e-01 -1.03320472e-01 -6.70878470e-01 7.05993533e-01
7.82390535e-01 -7.28118360e-01 -2.11506516e-01 -5.87907135e-01
4.83953446e-01 -2.05209255e-02 2.40112454e-01 -6.07691169e-01
4.62419689e-01 1.33033797e-01 1.46614149e-01 -4.23050702e-01
2.23333910e-01 -4.24866199e-01 1.31935105e-01 2.60929823e-01
-4.16762918e-01 4.96953279e-02 4.17910427e-01 3.83880705e-01
-1.71450555e-01 -4.84025508e-01 5.11294067e-01 -1.22927586e-02
-1.04000676e+00 1.98696166e-01 1.50297701e-01 4.01467890e-01
8.57704639e-01 -1.31541202e-02 -6.95317209e-01 9.29445103e-02
-9.62596357e-01 5.46193659e-01 4.11453545e-01 4.61254656e-01
1.75435185e-01 -1.30732596e+00 -6.33736372e-01 -4.18751128e-03
2.48839438e-01 1.43705979e-01 -3.53141222e-03 9.58794594e-01
-9.01898518e-02 5.28608322e-01 1.74190046e-03 -2.61072904e-01
-1.21197224e+00 6.07237995e-01 1.84939474e-01 -5.51147878e-01
-5.70899904e-01 9.09261942e-01 2.41385251e-01 -6.80140913e-01
7.03224679e-04 -1.48231223e-01 -3.30979556e-01 2.02033490e-01
1.94530711e-01 3.11043471e-01 4.80382964e-02 -4.31376696e-01
-3.99021924e-01 2.89022595e-01 -4.25283104e-01 1.10405378e-01
1.34075999e+00 7.43643641e-02 -1.02673799e-01 7.05398321e-01
1.03986835e+00 1.56318694e-01 -8.85080278e-01 -6.04657888e-01
5.83896697e-01 -1.44437283e-01 1.07374135e-02 -9.82355237e-01
-9.57321167e-01 5.20610809e-01 6.65279403e-02 1.84153214e-01
8.55453312e-01 3.00153911e-01 1.01680458e+00 7.04082847e-01
6.56923234e-01 -9.15588379e-01 -2.83502251e-01 7.20254362e-01
5.45541048e-01 -1.37455499e+00 -1.22843586e-01 -6.79481626e-01
-6.51258647e-01 6.43997550e-01 7.93288231e-01 1.12586513e-01
5.02556682e-01 3.93116951e-01 -2.60075539e-01 -2.28061646e-01
-1.06964147e+00 -4.31112945e-01 2.28887722e-01 3.58766675e-01
4.99306738e-01 -8.93252250e-03 -2.22031996e-01 9.51074541e-01
-2.43565857e-01 8.79878551e-02 3.69613379e-01 8.59887838e-01
-4.56434429e-01 -1.05302680e+00 1.18595995e-01 5.40972054e-01
-5.59813499e-01 -5.32210708e-01 -2.70948231e-01 1.05620241e+00
1.44767731e-01 7.09820688e-01 6.44024462e-02 -2.03946710e-01
5.29925466e-01 3.77566725e-01 1.50109395e-01 -8.54931712e-01
-5.52517474e-01 -2.58704960e-01 5.09028852e-01 -3.29932153e-01
-2.54564345e-01 -6.24519110e-01 -1.45709300e+00 -4.12115663e-01
-4.01454598e-01 2.76843369e-01 2.79611140e-01 1.01281464e+00
6.22585714e-01 4.78612006e-01 4.37461138e-01 -2.90796340e-01
-1.48557335e-01 -1.03472126e+00 -3.12929839e-01 3.12644720e-01
1.64070979e-01 -8.64834666e-01 -2.19338670e-01 1.04153417e-01] | [9.523014068603516, 8.72318172454834] |
7d646187-44c3-41fb-a5e2-1d93880602be | 3d-point-cloud-pre-training-with-knowledge | 2212.08974 | null | https://arxiv.org/abs/2212.08974v1 | https://arxiv.org/pdf/2212.08974v1.pdf | 3D Point Cloud Pre-training with Knowledge Distillation from 2D Images | The recent success of pre-trained 2D vision models is mostly attributable to learning from large-scale datasets. However, compared with 2D image datasets, the current pre-training data of 3D point cloud is limited. To overcome this limitation, we propose a knowledge distillation method for 3D point cloud pre-trained models to acquire knowledge directly from the 2D representation learning model, particularly the image encoder of CLIP, through concept alignment. Specifically, we introduce a cross-attention mechanism to extract concept features from 3D point cloud and compare them with the semantic information from 2D images. In this scheme, the point cloud pre-trained models learn directly from rich information contained in 2D teacher models. Extensive experiments demonstrate that the proposed knowledge distillation scheme achieves higher accuracy than the state-of-the-art 3D pre-training methods for synthetic and real-world datasets on downstream tasks, including object classification, object detection, semantic segmentation, and part segmentation. | ['Xiaoshui Huang', 'Wanli Ouyang', 'Jiebo Luo', 'Zhenfei Yin', 'Yuanhan Zhang', 'Yuan YAO'] | 2022-12-17 | null | null | null | null | ['concept-alignment', 'point-cloud-pre-training'] | ['computer-vision', 'computer-vision'] | [ 1.46088138e-01 3.98673236e-01 -3.11530113e-01 -4.60445344e-01
-6.00469172e-01 -5.20399034e-01 4.46814060e-01 4.17697839e-02
-1.48816854e-01 5.85762784e-02 -3.33541572e-01 -2.85148352e-01
3.05713475e-01 -6.74736738e-01 -1.25373733e+00 -2.84470141e-01
2.83990920e-01 9.16433752e-01 3.75937223e-01 2.04876199e-01
4.01708633e-01 7.56324410e-01 -1.55878878e+00 -4.41874340e-02
9.04868782e-01 1.16777670e+00 7.20990658e-01 2.70669639e-01
-7.71396637e-01 3.28678221e-01 -3.38562399e-01 -8.50666538e-02
5.29396534e-01 -6.93055987e-03 -6.74012840e-01 5.17617106e-01
5.91766953e-01 -5.01146436e-01 -2.42476657e-01 1.15211093e+00
2.70530015e-01 -1.35741383e-01 8.31207752e-01 -1.16682160e+00
-8.49417090e-01 5.87889887e-02 -8.87808621e-01 -1.36224721e-02
7.11714476e-02 7.58213699e-02 6.23157144e-01 -1.05853331e+00
5.52684546e-01 1.49961936e+00 6.50169253e-01 6.16014123e-01
-9.88287508e-01 -6.31192863e-01 4.21603262e-01 9.47662145e-02
-1.31775510e+00 5.68532906e-02 1.12048745e+00 -6.07742012e-01
1.20143795e+00 -4.96598125e-01 8.94910932e-01 7.01993704e-01
-4.55899268e-01 1.34823060e+00 9.40844059e-01 -4.40797478e-01
2.49174945e-02 8.78475979e-02 5.26426993e-02 7.03291476e-01
1.64644212e-01 1.62689090e-01 -1.42728150e-01 2.34447613e-01
1.23367083e+00 1.01721317e-01 -7.02675432e-02 -8.77180874e-01
-1.03321052e+00 5.53193927e-01 7.79727399e-01 1.66469719e-02
-3.83659929e-01 1.39981970e-01 9.01618749e-02 -2.51175985e-02
6.91421330e-01 3.10353875e-01 -9.56354499e-01 1.70138255e-01
-7.93845177e-01 3.66284177e-02 4.53600258e-01 1.49628043e+00
1.05901456e+00 -8.74757618e-02 8.10926780e-02 7.32309759e-01
5.85712135e-01 8.78902197e-01 2.78751403e-01 -8.70036840e-01
5.64835072e-01 9.49341416e-01 -1.55435950e-01 -6.14445210e-01
-1.44084811e-01 -3.65385771e-01 -5.40804446e-01 3.36999625e-01
1.11834966e-01 2.82618374e-01 -1.59424686e+00 1.31884885e+00
4.91209954e-01 7.52639771e-01 1.53333634e-01 1.01904702e+00
1.07319117e+00 5.19831955e-01 -4.49702591e-02 1.15057983e-01
9.31632042e-01 -1.04426801e+00 -1.06519267e-01 -3.74534577e-01
2.69538194e-01 -6.81183755e-01 7.30309665e-01 -8.88614208e-02
-1.01098907e+00 -1.13776326e+00 -9.38094020e-01 -2.36495256e-01
-3.74979645e-01 1.43109024e-01 5.95003963e-01 2.30510518e-01
-6.48849845e-01 3.20439011e-01 -1.04015887e+00 -3.07141453e-01
1.15214896e+00 2.98814029e-01 -3.71626437e-01 -3.78017396e-01
-5.29613137e-01 8.86635303e-01 6.76733196e-01 -1.08489454e-01
-1.18268096e+00 -1.14041722e+00 -9.89582896e-01 -1.22395098e-01
3.85227382e-01 -9.77018595e-01 1.37136149e+00 -8.42540026e-01
-1.45717144e+00 1.51081431e+00 1.48382768e-01 -3.63199621e-01
2.62897253e-01 -5.32366931e-01 3.85918498e-01 2.70491302e-01
3.40123445e-01 1.26166844e+00 9.22160685e-01 -1.77604318e+00
-5.79598665e-01 -6.39402866e-01 5.11228330e-02 3.97139966e-01
4.03401464e-01 -4.44584608e-01 -1.02076256e+00 -2.83159941e-01
5.89731812e-01 -8.12576294e-01 -3.71627271e-01 2.10906699e-01
-5.09606123e-01 -3.27291518e-01 1.03337920e+00 -2.44949922e-01
-1.36505924e-02 -2.05271220e+00 5.01063354e-02 6.10038824e-02
2.97693014e-02 5.05854070e-01 -3.47989261e-01 -1.32074118e-01
-8.61577392e-02 -7.70384073e-02 -3.42538178e-01 -5.01507759e-01
-2.09163968e-02 4.69491422e-01 -4.01787758e-01 2.26403743e-01
8.45173836e-01 1.15059149e+00 -8.61816347e-01 -6.30092859e-01
7.07688987e-01 2.89703697e-01 -5.69375277e-01 4.85896826e-01
-7.49372184e-01 7.17606604e-01 -8.24032366e-01 8.85867298e-01
1.11406040e+00 -3.79197627e-01 -4.71737415e-01 -2.13902801e-01
5.21216579e-02 1.99185491e-01 -6.97623849e-01 2.58806372e+00
-3.55327576e-01 3.80574048e-01 -2.08877638e-01 -1.40588379e+00
1.13051403e+00 1.40563637e-01 6.17082596e-01 -5.51222205e-01
1.98458567e-01 1.58251777e-01 -3.64018738e-01 -6.21204376e-01
5.16070193e-03 -8.08666721e-02 1.32106632e-01 1.35729939e-01
5.85845888e-01 -1.21295226e+00 -3.81601155e-01 3.00391708e-02
6.04502916e-01 6.44616187e-01 -3.62069905e-02 2.60298401e-01
4.20403153e-01 3.74929875e-01 4.68548417e-01 6.24601066e-01
-1.96456105e-01 8.94212425e-01 1.87891394e-01 -8.02881569e-02
-1.10978198e+00 -1.18232679e+00 -2.48231634e-01 2.23359108e-01
7.06863165e-01 -4.54940982e-02 -6.33820057e-01 -9.88368511e-01
4.42498207e-01 5.87327063e-01 -4.46056873e-01 -1.77868560e-01
-4.46457744e-01 -2.08253384e-01 3.37491840e-01 8.47920954e-01
9.09604788e-01 -6.60583794e-01 -5.11536419e-01 -6.74378797e-02
1.71422407e-01 -1.68439794e+00 -7.16128200e-02 2.07076609e-01
-1.42613733e+00 -1.32979167e+00 -6.29087985e-01 -1.05969667e+00
9.80909407e-01 7.82005429e-01 1.33317769e+00 -1.47891184e-02
-2.14364767e-01 7.05141246e-01 -3.53077650e-01 -8.78319621e-01
-1.17066443e-01 -4.14226204e-02 -1.88768759e-01 -4.14312780e-01
8.48653197e-01 -5.62700868e-01 -4.06420082e-01 1.91734776e-01
-5.09550214e-01 4.02759731e-01 1.08430135e+00 6.48760140e-01
1.26598179e+00 -1.96559042e-01 3.02721441e-01 -6.39722109e-01
-2.22590193e-01 -2.87244409e-01 -6.98070288e-01 6.37880638e-02
-3.70529175e-01 5.72564080e-02 5.51938117e-02 -2.36013979e-01
-1.10286760e+00 5.12833774e-01 -2.42148504e-01 -1.35717654e+00
-6.55854881e-01 2.56006241e-01 -3.46720517e-01 -1.12572141e-01
1.66134670e-01 2.77592123e-01 8.77614245e-02 -8.56521368e-01
6.73008144e-01 5.93072355e-01 6.96480930e-01 -6.88579321e-01
1.06031084e+00 5.59413612e-01 5.92921749e-02 -6.60985351e-01
-1.20104814e+00 -7.31931269e-01 -1.30327260e+00 1.15122288e-01
1.15135956e+00 -1.35467696e+00 -3.24217677e-01 5.93022764e-01
-1.51235604e+00 -2.56815225e-01 -4.36766148e-01 6.32101834e-01
-8.66000235e-01 2.20750198e-01 -1.82153404e-01 -5.75951874e-01
-1.52961984e-01 -9.53138113e-01 1.63234925e+00 3.38044971e-01
5.60001314e-01 -8.33574593e-01 -4.05428074e-02 6.56156003e-01
-2.04908893e-01 2.25224093e-01 1.09076023e+00 -6.61700845e-01
-1.11692357e+00 -2.05317378e-01 -6.79808259e-01 6.90071583e-01
1.40746385e-01 -2.56150633e-01 -1.01978326e+00 5.87131493e-02
1.62865490e-01 -5.02946615e-01 8.29898059e-01 4.43123281e-01
1.46024346e+00 3.15428346e-01 -6.91230655e-01 7.05014348e-01
1.47857141e+00 -5.73261157e-02 3.13372403e-01 -7.21804798e-02
1.07916892e+00 3.82932812e-01 7.82923162e-01 4.31614295e-02
4.72368777e-01 4.19979692e-01 8.42133820e-01 -1.44556701e-01
-3.73921305e-01 -6.91466987e-01 -2.19498560e-01 8.28907549e-01
-1.28690235e-03 2.40406394e-01 -1.04020929e+00 6.05413914e-01
-1.78039706e+00 -6.67400241e-01 5.26763648e-02 1.86350679e+00
7.03015149e-01 4.11408633e-01 -2.35602841e-01 -1.00246735e-01
6.22296631e-01 -1.64103970e-01 -9.82943118e-01 2.98966289e-01
1.48627702e-02 2.31655791e-01 4.67409521e-01 1.56556889e-01
-1.22423279e+00 1.32205403e+00 5.68212271e+00 6.64177954e-01
-1.04194069e+00 3.34158987e-02 3.03465962e-01 1.93003014e-01
-1.39227033e-01 -1.49729205e-02 -8.95283937e-01 2.25419149e-01
2.67522305e-01 1.10639215e-01 -2.65893221e-01 1.21153033e+00
-2.68716086e-02 1.18496567e-01 -1.46922970e+00 1.36787021e+00
2.79119819e-01 -1.36701608e+00 3.01158935e-01 -3.71515267e-02
1.02543676e+00 4.28737432e-01 -1.16696343e-01 4.04376239e-01
1.30534068e-01 -8.49357903e-01 6.42889917e-01 4.78124022e-01
6.03612542e-01 -5.13497889e-01 6.34232283e-01 5.59936047e-01
-9.48800385e-01 1.43820986e-01 -5.73851526e-01 8.52276385e-02
1.45598739e-01 7.05708385e-01 -1.15528357e+00 6.56542897e-01
6.26781762e-01 1.23974133e+00 -4.94024277e-01 1.27140582e+00
-4.99640584e-01 4.99983579e-01 -4.90429670e-01 2.76645631e-01
4.75938588e-01 -1.12160556e-01 5.52047491e-01 7.76708841e-01
2.19781920e-01 3.33328485e-01 3.06771964e-01 1.23201716e+00
-2.04095468e-01 -3.78745198e-01 -7.74245977e-01 -4.28015552e-02
4.34420258e-01 8.63629580e-01 -6.43294215e-01 -4.00217563e-01
-7.42426813e-01 8.90308917e-01 2.55751520e-01 2.90850937e-01
-6.55374825e-01 -2.77251065e-01 8.70824158e-01 -1.06412813e-01
8.33330750e-01 -5.85865319e-01 -4.17814523e-01 -1.12873530e+00
-8.04682970e-02 -2.65572220e-01 -1.14994720e-01 -1.24104714e+00
-1.46311450e+00 1.63728833e-01 2.29972228e-01 -1.34223056e+00
-3.44227850e-02 -9.48594928e-01 -5.20930827e-01 8.29421163e-01
-2.04636955e+00 -1.53692746e+00 -5.14061332e-01 6.04455292e-01
8.77692223e-01 -1.04803830e-01 5.09730041e-01 9.61181894e-02
-9.91047770e-02 2.14154497e-01 -1.31139517e-01 3.92157316e-01
3.09077770e-01 -1.34098959e+00 7.32346117e-01 5.25363982e-01
3.17400247e-01 2.10496947e-01 4.79561612e-02 -6.48920894e-01
-1.45758021e+00 -1.32089818e+00 1.94500059e-01 -8.06309521e-01
1.36823729e-01 -3.39899659e-01 -9.87251759e-01 5.60114503e-01
-1.31391466e-01 2.06294835e-01 4.87150788e-01 -1.49453893e-01
-4.27077264e-01 1.11137629e-01 -1.07764709e+00 1.41032591e-01
1.45018303e+00 -6.06230140e-01 -1.17378986e+00 3.90407652e-01
1.18757284e+00 -7.83652127e-01 -7.89923728e-01 7.84165561e-01
7.38035291e-02 -4.66809750e-01 1.39622521e+00 -7.02426016e-01
4.93417472e-01 -4.60266203e-01 -2.69818395e-01 -1.24119127e+00
3.84196751e-02 1.07565187e-02 -2.95216054e-01 1.04749060e+00
1.16862468e-01 -1.06238900e-02 1.16380620e+00 2.93364584e-01
-7.55526900e-01 -6.10391259e-01 -8.11271727e-01 -7.43370771e-01
2.83449024e-01 -4.94776905e-01 6.71358526e-01 8.63310158e-01
-6.29104555e-01 5.83617508e-01 3.11473638e-01 4.75347400e-01
8.83686602e-01 6.73631370e-01 1.10012579e+00 -1.45436549e+00
-1.24085508e-01 -3.59507501e-01 -8.69994938e-01 -1.89978552e+00
4.83695090e-01 -1.08411884e+00 2.60672927e-01 -1.76712191e+00
9.02204737e-02 -7.43888199e-01 -9.37655866e-02 4.89826143e-01
-1.61139995e-01 1.15975432e-01 3.46362561e-01 3.05650830e-01
-5.52120626e-01 8.25187743e-01 1.83292258e+00 -5.22142351e-01
-2.61126369e-01 1.84946299e-01 -3.94573033e-01 8.06732357e-01
4.66178179e-01 -4.18637872e-01 -6.80133939e-01 -9.67804134e-01
-3.06615889e-01 -1.46042287e-01 6.93166912e-01 -1.02783811e+00
1.95714831e-01 -1.49287477e-01 5.85174263e-01 -1.40949738e+00
5.03345013e-01 -1.01125240e+00 -4.53267455e-01 1.02000743e-01
2.07872363e-03 -6.18732035e-01 5.18722236e-01 8.36251199e-01
-3.59500438e-01 -2.83721656e-01 6.49774730e-01 -4.55519170e-01
-1.05626726e+00 7.37088144e-01 2.44567588e-01 1.06309786e-01
9.91164029e-01 -4.89391148e-01 -5.84356412e-02 2.24099413e-01
-6.85263157e-01 5.35944581e-01 4.85003591e-01 6.74240649e-01
9.00228739e-01 -1.27446580e+00 -4.84394222e-01 4.88062114e-01
3.11667681e-01 1.23416519e+00 2.15004846e-01 4.51772332e-01
-4.83461618e-01 5.42873621e-01 -4.04219806e-01 -1.27756810e+00
-8.26193333e-01 7.99750686e-01 3.80075723e-01 3.67987663e-01
-7.56704330e-01 9.70622718e-01 3.99543256e-01 -7.54338861e-01
3.07668179e-01 -7.77931631e-01 6.21386953e-02 -4.45665210e-01
4.73346412e-02 -2.49454513e-01 3.18340678e-03 -5.07774293e-01
-4.07368928e-01 1.38730860e+00 -3.10698897e-02 1.50181890e-01
1.32720351e+00 -3.71541642e-02 2.48971283e-01 3.56994331e-01
1.37331247e+00 -7.02918589e-01 -1.65960383e+00 -4.54184234e-01
-1.57609180e-01 -7.07622349e-01 1.51619375e-01 -6.20451450e-01
-1.08017623e+00 1.32853520e+00 6.10029221e-01 -2.02903166e-01
9.02517676e-01 4.65321898e-01 6.88267052e-01 6.58459663e-01
4.02592659e-01 -9.67329144e-01 3.60065758e-01 7.02033281e-01
6.89489186e-01 -1.68412614e+00 -4.44850437e-02 -7.37007797e-01
-4.54958439e-01 8.90979707e-01 1.06572998e+00 -4.18959528e-01
1.06132019e+00 -2.64793575e-01 9.17963311e-02 -3.34620059e-01
-4.72692102e-01 -6.45837545e-01 2.24723086e-01 1.17489409e+00
-1.92652926e-01 -3.05084258e-01 4.96103376e-01 5.46533585e-01
5.38819842e-02 1.31013259e-01 5.36689572e-02 9.03126717e-01
-4.85483319e-01 -9.46930826e-01 -1.58605918e-01 1.40521064e-01
1.17537491e-01 1.34708226e-01 -4.87967253e-01 8.58983934e-01
4.16441888e-01 5.13022304e-01 5.15764534e-01 -2.62631088e-01
4.47451025e-01 5.23596704e-02 9.61833239e-01 -9.89721119e-01
9.80190095e-03 -6.91172257e-02 -3.48682225e-01 -3.48793685e-01
-8.27465117e-01 -4.94134367e-01 -1.51067960e+00 3.06033880e-01
-5.52157521e-01 7.86971673e-02 1.00788033e+00 1.08853829e+00
5.88717639e-01 4.64519352e-01 6.69275463e-01 -1.39155507e+00
-3.69287550e-01 -9.13261712e-01 -3.07432294e-01 4.04936135e-01
2.65074849e-01 -1.04170215e+00 -2.46446449e-02 2.61444241e-01] | [8.103740692138672, -3.2538599967956543] |
17c314d2-35a2-4da3-93fc-4ce106266b91 | score-based-conditional-generation-with-fewer | 2307.04081 | null | https://arxiv.org/abs/2307.04081v1 | https://arxiv.org/pdf/2307.04081v1.pdf | Score-based Conditional Generation with Fewer Labeled Data by Self-calibrating Classifier Guidance | Score-based Generative Models (SGMs) are a popular family of deep generative models that achieves leading image generation quality. Earlier studies have extended SGMs to tackle class-conditional generation by coupling an unconditional SGM with the guidance of a trained classifier. Nevertheless, such classifier-guided SGMs do not always achieve accurate conditional generation, especially when trained with fewer labeled data. We argue that the issue is rooted in unreliable gradients of the classifier and the inability to fully utilize unlabeled data during training. We then propose to improve classifier-guided SGMs by letting the classifier calibrate itself. Our key idea is to use principles from energy-based models to convert the classifier as another view of the unconditional SGM. Then, existing loss for the unconditional SGM can be adopted to calibrate the classifier using both labeled and unlabeled data. Empirical results validate that the proposed approach significantly improves the conditional generation quality across different percentages of labeled data. The improved performance makes the proposed approach consistently superior to other conditional SGMs when using fewer labeled data. The results confirm the potential of the proposed approach for generative modeling with limited labeled data. | ['Hsuan-Tien Lin', 'Si-An Chen', 'Paul Kuo-Ming Huang'] | 2023-07-09 | null | null | null | null | ['image-generation'] | ['computer-vision'] | [ 3.20466727e-01 2.31503800e-01 -1.31731242e-01 -4.14422303e-01
-1.07618058e+00 -4.43072796e-01 7.76445448e-01 -2.45373473e-01
-1.36259928e-01 8.93582046e-01 -1.05703697e-01 2.85313819e-02
1.72265336e-01 -9.14220631e-01 -7.01548398e-01 -1.09764946e+00
4.99792129e-01 3.64079803e-01 7.71733299e-02 5.84723540e-02
2.33415127e-01 2.31701687e-01 -1.71789241e+00 1.88131362e-01
1.41014767e+00 8.42304766e-01 4.23979670e-01 6.84020102e-01
4.43268009e-02 7.47104526e-01 -5.92393935e-01 -4.83387053e-01
1.84375539e-01 -9.25512671e-01 -3.66950661e-01 1.58821225e-01
2.26260304e-01 -8.22499469e-02 3.27139884e-01 1.03411520e+00
4.76614892e-01 7.54991174e-02 1.04565012e+00 -1.40135574e+00
-6.41901672e-01 4.66413051e-01 -4.35756922e-01 -2.24352568e-01
-4.59544659e-02 3.93210985e-02 9.20161247e-01 -9.74517226e-01
5.22026837e-01 1.08158851e+00 6.24976277e-01 7.57850230e-01
-1.24478114e+00 -5.93305111e-01 1.76149711e-01 -1.02216899e-02
-1.46486592e+00 -2.35299796e-01 8.99880946e-01 -5.03435075e-01
5.67632735e-01 1.85965568e-01 5.99193037e-01 9.64182079e-01
1.65666655e-01 8.36610675e-01 1.29262269e+00 -8.00717652e-01
4.59712297e-01 6.37493432e-01 -2.11811755e-02 7.85382152e-01
2.77749956e-01 2.07278103e-01 -4.79788721e-01 -4.21712995e-02
4.65212792e-01 -3.63609642e-01 -1.59514412e-01 -4.08069134e-01
-7.27460027e-01 1.09525025e+00 3.90868872e-01 1.76932886e-01
-3.86829406e-01 3.12249035e-01 -6.57662526e-02 -1.35408267e-01
7.07673669e-01 2.48860151e-01 -8.00932758e-03 9.31032971e-02
-1.38244176e+00 1.26535490e-01 5.48916996e-01 9.50407147e-01
9.31944549e-01 3.99021417e-01 -3.63966078e-01 8.88246357e-01
5.66853404e-01 5.54656982e-01 4.02008533e-01 -7.85550535e-01
3.01843405e-01 5.91140866e-01 -7.04444125e-02 -6.07340991e-01
4.63335030e-02 -8.73458922e-01 -6.80159867e-01 3.38883728e-01
2.00943097e-01 -2.21083745e-01 -1.24600160e+00 1.94847953e+00
2.26645336e-01 2.31664628e-01 1.39313325e-01 6.60191119e-01
5.57960033e-01 6.76071048e-01 2.40489617e-01 -2.37980247e-01
5.96260607e-01 -9.44787979e-01 -4.72509235e-01 -8.78313407e-02
6.11511111e-01 -6.76688194e-01 1.07895792e+00 3.99510354e-01
-9.12407219e-01 -7.42765725e-01 -1.23954523e+00 2.41558358e-01
-2.73720026e-01 5.33955514e-01 6.35571480e-01 1.14746451e+00
-1.20522869e+00 4.44130599e-01 -9.99741137e-01 -2.43458644e-01
3.27606618e-01 2.19523221e-01 2.43933573e-01 -9.84463096e-02
-9.16734040e-01 7.68690407e-01 4.65147227e-01 3.27584334e-02
-1.07568705e+00 -5.49045265e-01 -7.19808280e-01 -8.67283195e-02
-1.77886412e-01 -6.66976392e-01 9.68298554e-01 -1.06121898e+00
-1.36702859e+00 5.06612718e-01 -2.46749625e-01 -3.56362730e-01
5.85688055e-01 -1.25904322e-01 -1.06723405e-01 3.87775563e-02
8.08288679e-02 8.86403918e-01 8.65514755e-01 -1.75624847e+00
-4.28849995e-01 -1.16598472e-01 -7.51673430e-02 3.48810256e-01
-2.49858782e-01 -6.30613446e-01 -1.55105993e-01 -5.56353271e-01
2.57447332e-01 -1.00883710e+00 -1.29950076e-01 -3.09092104e-01
-6.03986323e-01 -9.19948593e-02 7.17985511e-01 -3.33624691e-01
1.16213250e+00 -1.79308045e+00 2.50818785e-02 2.34639436e-01
-1.70264751e-01 1.54888481e-01 2.54920628e-02 3.08619201e-01
-3.08423433e-02 1.42181709e-01 -4.50832844e-01 -5.45798779e-01
-1.36732444e-01 2.60723501e-01 -2.04156056e-01 1.68464527e-01
4.21766430e-01 9.80839431e-01 -8.03246021e-01 -5.62333882e-01
3.24327976e-01 7.41586387e-01 -7.21798956e-01 1.84866086e-01
-2.41057158e-01 5.13011634e-01 -3.69867027e-01 5.27744353e-01
7.83243716e-01 -1.63725778e-01 1.39688984e-01 -1.72981232e-01
1.69414833e-01 -2.34952476e-02 -1.13113725e+00 1.56946671e+00
-5.86465538e-01 3.68391246e-01 -3.47254217e-01 -1.13246286e+00
1.00846946e+00 2.14010850e-01 2.66808212e-01 -2.00192377e-01
-9.97266993e-02 2.93414056e-01 -5.00473790e-02 -5.92763461e-02
3.06122243e-01 -4.17365968e-01 1.43491432e-01 3.37909162e-01
3.91014159e-01 -3.22805256e-01 8.02660137e-02 3.56759697e-01
5.33379376e-01 6.28586590e-01 1.97769403e-02 -2.36550108e-01
4.46878403e-01 -1.77487746e-01 4.72094119e-01 6.85043454e-01
1.30554631e-01 8.24254692e-01 1.15874209e-01 7.07675368e-02
-1.02099979e+00 -1.23152745e+00 -1.22593619e-01 9.15979445e-01
6.41827509e-02 -1.23221926e-01 -1.16291201e+00 -7.08640516e-01
-4.40462440e-01 1.07923543e+00 -6.17009163e-01 -3.42961282e-01
-1.64429501e-01 -1.41507220e+00 3.77633423e-01 5.78519821e-01
6.35850012e-01 -7.23069251e-01 -4.85821456e-01 2.31424302e-01
-3.56445044e-01 -7.38997579e-01 -1.27854377e-01 3.20751876e-01
-1.04991078e+00 -8.54657292e-01 -7.14815021e-01 -6.12618566e-01
1.07223725e+00 -1.18717775e-01 1.06758749e+00 1.20686047e-01
-2.96707563e-02 3.35997671e-01 -4.28338617e-01 -2.84673214e-01
-8.36103141e-01 1.44580379e-01 -1.17695980e-01 1.72087520e-01
1.78321674e-01 -5.48223376e-01 -6.18015468e-01 2.56977320e-01
-7.76276886e-01 3.53267759e-01 6.15127861e-01 9.35159802e-01
6.56279147e-01 -6.90722391e-02 9.99294817e-01 -1.00173283e+00
4.79385674e-01 -4.02483195e-01 -4.10166502e-01 3.55292886e-01
-1.10486031e+00 3.28706533e-01 5.53260505e-01 -3.33069265e-01
-1.52266204e+00 2.10885778e-01 -3.45319211e-01 -3.26919377e-01
-3.22378390e-02 3.51927280e-01 -2.72742927e-01 -1.26059568e-02
7.09022820e-01 4.51303065e-01 -2.76106805e-01 -2.85857230e-01
4.53701437e-01 4.91203964e-01 3.73425066e-01 -7.43904233e-01
7.05423832e-01 3.56248230e-01 1.48133889e-01 -3.97403777e-01
-7.24140346e-01 -1.48282766e-01 -5.22701085e-01 -5.48137009e-01
8.67811739e-01 -7.77103961e-01 -1.25498772e-01 4.72865731e-01
-7.88751900e-01 -2.23395243e-01 -3.10041279e-01 4.52162743e-01
-6.46647394e-01 2.26280525e-01 -4.10531253e-01 -1.24478853e+00
-3.05857241e-01 -1.19621193e+00 1.13633597e+00 3.49591792e-01
1.02856137e-01 -1.27136099e+00 1.00869372e-01 3.06199610e-01
4.39316362e-01 4.23890620e-01 8.44632447e-01 -5.43764591e-01
-4.71074849e-01 -4.38274860e-01 1.35731265e-01 7.09366798e-01
2.21091777e-01 2.28785992e-01 -1.24412811e+00 -2.65526712e-01
5.42648733e-02 -4.69816118e-01 1.03973615e+00 6.06630623e-01
8.48599255e-01 -6.13243543e-02 -3.07166249e-01 5.43712676e-01
1.68116438e+00 2.74325281e-01 6.83514893e-01 -8.11967775e-02
7.54032552e-01 3.25351566e-01 5.78699648e-01 2.72118419e-01
2.71229327e-01 4.74423736e-01 4.70832825e-01 -1.63723916e-01
-2.95745105e-01 -5.86119354e-01 4.38756496e-01 9.49221671e-01
-2.74525970e-01 -4.91418898e-01 -6.29234612e-01 4.17972744e-01
-1.73997760e+00 -9.51510131e-01 -2.70748645e-01 2.32315445e+00
8.18792045e-01 6.36564568e-02 -1.90117553e-01 3.48864228e-01
8.33005548e-01 -1.65372610e-01 -4.58301783e-01 -2.44314745e-01
-1.00263245e-01 3.20160896e-01 3.05208683e-01 6.15331054e-01
-1.04906976e+00 7.42443740e-01 6.86042547e+00 1.03331673e+00
-1.08167160e+00 2.66060621e-01 8.85982096e-01 2.90746409e-02
-5.01721203e-01 1.31718844e-01 -9.55567777e-01 6.74935341e-01
9.01750088e-01 -8.02364349e-02 1.34562880e-01 9.68588710e-01
9.99406651e-02 -3.31810057e-01 -1.07899857e+00 6.44320369e-01
2.46209249e-01 -1.05957592e+00 1.67744905e-01 9.39048007e-02
1.25656521e+00 -3.27109098e-01 2.81945050e-01 2.88874328e-01
4.56277311e-01 -9.38222051e-01 7.41711736e-01 5.65203309e-01
6.60201013e-01 -8.06692660e-01 7.18682647e-01 5.49655855e-01
-1.18158853e+00 2.33622700e-01 -4.29726928e-01 1.81693181e-01
1.03757149e-02 6.91247284e-01 -9.29230571e-01 6.18883431e-01
2.26922646e-01 4.11463112e-01 -7.46681333e-01 8.19735587e-01
-4.42161411e-01 9.71358597e-01 -2.80249774e-01 1.59947425e-01
1.19955279e-01 -3.33113641e-01 2.86611944e-01 1.14029717e+00
6.15839064e-01 -1.90139800e-01 7.67383426e-02 1.17734110e+00
2.07619637e-01 -4.54705693e-02 -4.76304293e-01 1.96829572e-01
4.10669953e-01 1.27260649e+00 -9.38964427e-01 -5.20920873e-01
-1.36753038e-01 9.91976321e-01 1.38957575e-01 5.03164768e-01
-1.05484223e+00 1.14308499e-01 -3.39794494e-02 4.20958102e-02
1.38520017e-01 1.14343055e-01 -5.09619474e-01 -1.06072462e+00
-8.22322294e-02 -3.21395546e-01 2.90652037e-01 -9.43216741e-01
-1.14020991e+00 6.62725568e-01 2.43159220e-01 -1.34850490e+00
-7.14915216e-01 -2.00978503e-01 -6.88585281e-01 9.93126810e-01
-1.37915504e+00 -1.42661595e+00 -4.01027352e-01 4.14535731e-01
6.34838283e-01 -5.21050915e-02 7.90865064e-01 1.37366652e-01
-4.65356201e-01 7.59793818e-01 1.18216775e-01 -2.46596172e-01
5.39176166e-01 -1.39888322e+00 -6.72290921e-02 1.20775414e+00
2.31814370e-01 4.94374067e-01 6.71163976e-01 -7.08740652e-01
-8.84886205e-01 -1.20835149e+00 6.26745999e-01 -4.84577596e-01
-9.10309181e-02 -3.22731107e-01 -7.12908447e-01 4.73902732e-01
1.62248611e-01 -1.48912758e-01 6.95612550e-01 -1.95130512e-01
-8.75174180e-02 -2.75942646e-02 -1.34945965e+00 1.53499097e-01
7.63400555e-01 -2.41418019e-01 -2.64558107e-01 2.16054291e-01
2.78772920e-01 -1.64310113e-01 -5.57902873e-01 4.19245660e-01
3.32840860e-01 -1.09773993e+00 6.76000714e-01 -1.39148995e-01
5.26620746e-01 -4.19993341e-01 -2.50714988e-01 -1.59585273e+00
-6.10109232e-02 -1.44941658e-01 -2.01597679e-02 1.45173860e+00
6.40494764e-01 -6.46502197e-01 9.60368454e-01 6.33663714e-01
-2.92172164e-01 -7.73366451e-01 -8.34090889e-01 -8.58935237e-01
3.28919619e-01 -4.96325344e-01 2.67772585e-01 7.17347264e-01
-2.90009767e-01 7.11714700e-02 -3.61812860e-01 4.32828404e-02
7.95772135e-01 1.32326111e-01 4.75878984e-01 -1.16642261e+00
-4.00148064e-01 -1.86310932e-01 -2.96453565e-01 -8.35010111e-01
-3.47119942e-02 -1.08352244e+00 3.67661685e-01 -1.75889111e+00
4.48752403e-01 -7.59181142e-01 -2.07686201e-01 4.79357660e-01
-5.16226888e-01 5.37293911e-01 1.47688389e-01 1.89926997e-01
-2.17756048e-01 6.08378828e-01 1.16417944e+00 -2.21859501e-03
-1.45366743e-01 6.83120117e-02 -6.84778869e-01 6.19696915e-01
8.24158609e-01 -5.15093982e-01 -8.96792352e-01 -8.42948630e-02
1.96263582e-01 -3.89294565e-01 4.90283400e-01 -1.20432496e+00
1.00900374e-01 -8.06894675e-02 5.89632869e-01 -4.78365093e-01
3.49250108e-01 -5.58805168e-01 5.71200430e-01 4.23239380e-01
-2.97479421e-01 -4.64148521e-01 -1.49090543e-01 4.75767046e-01
-2.21252576e-01 -4.04176563e-01 1.06520760e+00 -2.88448799e-02
-3.95390511e-01 3.76507011e-03 -2.79827416e-01 -6.54786974e-02
1.04574621e+00 -4.96067792e-01 -8.72002393e-02 -5.10339499e-01
-7.89735794e-01 -1.58239871e-01 5.33742726e-01 1.89955875e-01
5.51675916e-01 -1.56125152e+00 -6.44216537e-01 2.43100360e-01
-7.87863955e-02 -1.39891237e-01 -8.64581845e-04 7.02057183e-01
-2.80667871e-01 2.65179396e-01 1.27101094e-02 -7.88377881e-01
-8.98340166e-01 3.75445157e-01 4.45464075e-01 -2.61626393e-01
-2.20057800e-01 9.60227013e-01 4.39852566e-01 -2.66577393e-01
-5.15907072e-02 -2.27197811e-01 -2.16616448e-02 7.61230737e-02
4.94233705e-02 2.90533692e-01 2.13524312e-01 -5.80329061e-01
-2.67159522e-01 5.08532345e-01 1.83079273e-01 -3.43485415e-01
1.11356938e+00 -7.29629323e-02 3.16563159e-01 4.29847926e-01
1.06114376e+00 -1.64595455e-01 -1.45406914e+00 1.99724570e-01
-4.69470084e-01 -3.38157415e-01 1.54558659e-01 -1.03700852e+00
-1.33800340e+00 1.02446640e+00 7.69859612e-01 8.05853226e-04
1.36276138e+00 -1.49342299e-01 3.26963753e-01 -1.27708539e-01
5.70515275e-01 -1.17054617e+00 1.40189484e-01 -3.84193473e-02
6.75303876e-01 -1.15342689e+00 -9.03172046e-02 -5.37346423e-01
-7.77093709e-01 8.72351110e-01 7.08001256e-01 -1.64461061e-01
4.77614731e-01 1.72604606e-01 4.12824564e-02 2.15062704e-02
-6.38975263e-01 -1.80453166e-01 4.41266358e-01 7.67244875e-01
5.24887860e-01 1.35428533e-01 -4.96011585e-01 5.37982941e-01
-1.83369190e-01 -4.10092361e-02 4.26597446e-01 8.33502412e-01
-3.90735149e-01 -1.17906022e+00 -2.71774232e-01 5.82875729e-01
-2.28229970e-01 -2.20921516e-01 -2.58771330e-01 5.69761455e-01
4.01468426e-01 1.17885506e+00 -1.26341686e-01 -3.49108636e-01
-9.06751156e-02 4.82925117e-01 5.78998029e-01 -6.66295171e-01
-3.39646190e-01 2.93447495e-01 -2.19412223e-01 -3.13847631e-01
-7.24999428e-01 -6.72279060e-01 -1.03400517e+00 5.75231649e-02
-8.66563261e-01 2.81241328e-01 8.36916149e-01 7.25882053e-01
2.19264984e-01 5.07023275e-01 7.46803164e-01 -8.09837639e-01
-4.20298129e-01 -1.03569996e+00 -4.20421600e-01 2.12319523e-01
1.35240126e-02 -8.31125557e-01 -5.77854574e-01 4.24451500e-01] | [11.37212085723877, -0.11953268200159073] |
50425178-c505-45c4-b0e5-4d494eb94c43 | spontaneous-facial-micro-expression | 1608.02255 | null | http://arxiv.org/abs/1608.02255v1 | http://arxiv.org/pdf/1608.02255v1.pdf | Spontaneous Facial Micro-Expression Recognition using Discriminative Spatiotemporal Local Binary Pattern with an Improved Integral Projection | Recently, there are increasing interests in inferring mirco-expression from
facial image sequences. Due to subtle facial movement of micro-expressions,
feature extraction has become an important and critical issue for spontaneous
facial micro-expression recognition. Recent works usually used spatiotemporal
local binary pattern for micro-expression analysis. However, the commonly used
spatiotemporal local binary pattern considers dynamic texture information to
represent face images while misses the shape attribute of face images. On the
other hand, their works extracted the spatiotemporal features from the global
face regions, which ignore the discriminative information between two
micro-expression classes. The above-mentioned problems seriously limit the
application of spatiotemporal local binary pattern on micro-expression
recognition. In this paper, we propose a discriminative spatiotemporal local
binary pattern based on an improved integral projection to resolve the problems
of spatiotemporal local binary pattern for micro-expression recognition.
Firstly, we develop an improved integral projection for preserving the shape
attribute of micro-expressions. Furthermore, an improved integral projection is
incorporated with local binary pattern operators across spatial and temporal
domains. Specifically, we extract the novel spatiotemporal features
incorporating shape attributes into spatiotemporal texture features. For
increasing the discrimination of micro-expressions, we propose a new feature
selection based on Laplacian method to extract the discriminative information
for facial micro-expression recognition. Intensive experiments are conducted on
three availably published micro-expression databases. We compare our method
with the state-of-the-art algorithms. Experimental results demonstrate that our
proposed method achieves promising performance for micro-expression
recognition. | ['Su-Jing Wang', 'Guoying Zhao', 'Xiaoyi Feng', 'Xiaohua Huang', 'Xin Liu', 'Matti Pietikainen'] | 2016-08-07 | null | null | null | null | ['micro-expression-recognition'] | ['computer-vision'] | [ 3.00193071e-01 -8.65048468e-01 -4.45376813e-01 -5.61990023e-01
-4.54125911e-01 -1.01185858e-01 3.37714136e-01 -5.44221282e-01
-3.63533318e-01 4.71402675e-01 -5.33955060e-02 2.85048604e-01
-1.22478805e-01 -6.14892244e-01 -3.85698885e-01 -1.39019549e+00
1.32078335e-01 -4.68551219e-01 -9.47640017e-02 -1.70120195e-01
1.04043417e-01 8.04376960e-01 -1.69096589e+00 4.64802712e-01
6.79776371e-01 1.49404812e+00 -6.74678758e-02 1.44910917e-01
-4.45953161e-01 7.81165361e-01 -4.65464056e-01 -1.31619930e-01
-1.75682784e-04 -7.25386858e-01 -1.93672001e-01 2.51267135e-01
2.25236371e-01 -6.48658276e-02 -1.66235343e-02 1.35334253e+00
3.98983330e-01 -3.52644846e-02 7.14667857e-01 -1.25497365e+00
-7.04101741e-01 -3.50991249e-01 -1.18196678e+00 5.02790153e-01
4.57408249e-01 -1.67552248e-01 4.34072107e-01 -1.05634081e+00
6.07091844e-01 1.18109095e+00 5.63656807e-01 7.28584826e-02
-9.16669667e-01 -1.03842258e+00 -1.93241537e-02 7.03654170e-01
-1.97150505e+00 -6.90237939e-01 1.04870641e+00 -3.31974119e-01
6.44086599e-01 3.12505305e-01 7.25517631e-01 8.42526734e-01
3.88862491e-01 7.14542508e-01 1.47812843e+00 -3.99710953e-01
-1.46559596e-01 3.49763036e-02 -8.38658512e-02 9.21106637e-01
-5.61969936e-01 -2.19305024e-01 -5.98966658e-01 -2.10237131e-01
7.56188035e-01 1.56515509e-01 -2.10401386e-01 -5.50840162e-02
-7.26279497e-01 4.30828452e-01 -3.19243297e-02 8.16978455e-01
-5.00062287e-01 -2.74213552e-01 4.26258177e-01 4.24073964e-01
7.06027627e-01 -4.00464207e-01 -2.22083792e-01 -6.18556142e-01
-8.07188511e-01 -8.78906101e-02 2.77582020e-01 5.64794898e-01
1.10505533e+00 3.09914649e-02 -8.50321725e-02 1.30980229e+00
8.21103454e-02 7.20652878e-01 6.17644668e-01 -7.58386195e-01
2.86196709e-01 7.45342910e-01 -3.02217543e-01 -1.71027732e+00
-1.90964997e-01 2.04463899e-02 -9.51395273e-01 5.13848551e-02
2.47781873e-01 -3.60473767e-02 -4.70730156e-01 1.71979427e+00
3.63003582e-01 4.37179208e-01 -5.70579506e-02 9.41504180e-01
4.22241181e-01 8.33985031e-01 1.23022944e-01 -9.07130599e-01
1.18286812e+00 -4.82555032e-01 -1.14440000e+00 4.34144825e-01
6.23294830e-01 -7.10388064e-01 9.29786146e-01 2.98422724e-01
-5.64752042e-01 -4.53995377e-01 -8.24925900e-01 2.63357848e-01
-2.41901234e-01 4.53006059e-01 5.21827221e-01 6.94132507e-01
-8.10976326e-01 2.01527447e-01 -6.83510542e-01 -3.12159836e-01
3.48396122e-01 4.77559090e-01 -8.27443838e-01 1.67406514e-01
-1.03693330e+00 5.84576011e-01 -2.35184535e-01 4.47032481e-01
-4.30893525e-02 -2.85956502e-01 -8.76504719e-01 -2.72364706e-01
1.94482207e-01 2.20954850e-01 7.90217519e-01 -1.63486862e+00
-1.91156602e+00 8.93102586e-01 -7.85082400e-01 3.78419459e-01
1.31562486e-01 2.96198159e-01 -8.65727067e-01 4.23233837e-01
1.33016214e-01 -5.94719406e-03 9.76257026e-01 -6.62976444e-01
-4.20083702e-01 -8.08619738e-01 -5.36532044e-01 1.72509626e-01
-7.05610394e-01 5.40896595e-01 -4.39206094e-01 -5.91052353e-01
2.67886937e-01 -7.76609063e-01 3.04409415e-01 1.78033054e-01
1.01533078e-01 -4.23801810e-01 1.49222386e+00 -5.45878589e-01
1.22041392e+00 -2.51093650e+00 1.14747640e-02 4.71822202e-01
-2.75986791e-01 1.00462288e-01 -1.84226573e-01 -1.05901338e-01
-2.50276804e-01 6.34541037e-03 -2.62619346e-01 -1.09823056e-01
-3.50594819e-01 2.38380760e-01 -1.81640580e-01 8.43616843e-01
1.68177545e-01 7.76023269e-01 -4.54398304e-01 -9.81108665e-01
1.49743214e-01 5.24142683e-01 -7.24153891e-02 7.12756440e-02
4.00394380e-01 5.49341381e-01 -8.97336841e-01 8.51997674e-01
1.02753460e+00 2.25097299e-01 -6.90639913e-02 -3.76816809e-01
-2.50872046e-01 -6.31699920e-01 -8.82936835e-01 1.49405098e+00
-5.51131666e-01 4.98064548e-01 4.44945991e-01 -1.28550792e+00
1.19969058e+00 3.78695935e-01 8.68761659e-01 -1.11319840e+00
4.43889380e-01 2.50841171e-01 -3.94834936e-01 -9.18090224e-01
-8.75784010e-02 -4.98420179e-01 1.02053374e-01 2.57826030e-01
-2.91970116e-03 3.28049421e-01 4.79907133e-02 -4.06704873e-01
6.70683563e-01 2.22488478e-01 2.82814503e-01 -3.56757671e-01
1.04000640e+00 -5.08165240e-01 8.59160185e-01 -4.79958355e-02
-4.76761490e-01 2.53170639e-01 5.72395086e-01 -3.93785834e-01
-4.23331559e-01 -7.73314714e-01 -4.48064327e-01 1.12914956e+00
2.70586461e-01 -2.93622404e-01 -6.39546752e-01 -6.87002182e-01
-2.54498005e-01 7.02160373e-02 -8.20694983e-01 -4.89124376e-03
-5.60210586e-01 -8.83156717e-01 7.35964179e-01 3.59961003e-01
8.33383799e-01 -1.01726329e+00 -3.91523212e-01 -4.72862571e-02
-1.77604347e-01 -1.06404257e+00 -5.47345281e-01 -3.79482388e-01
-5.84861696e-01 -8.53193820e-01 -8.54169309e-01 -8.74982655e-01
5.85530162e-01 2.02239007e-01 2.39414454e-01 -9.09007415e-02
-4.20928448e-01 3.10173690e-01 -3.24071258e-01 -2.96266079e-02
7.89463446e-02 -4.10017788e-01 1.88630044e-01 9.27582383e-01
7.41458595e-01 -8.67069244e-01 -3.46222311e-01 5.20110250e-01
-8.13149393e-01 -3.64550471e-01 6.85567617e-01 6.48603499e-01
9.21923399e-01 3.83718908e-02 4.24875498e-01 -3.38464707e-01
5.11534631e-01 -4.20102060e-01 -2.84283310e-01 3.53032649e-01
-1.27588004e-01 -8.27221051e-02 7.42411792e-01 -6.80418968e-01
-1.50146425e+00 9.14235264e-02 -2.79538333e-01 -6.49285734e-01
-1.56972200e-01 5.33736050e-01 -4.12940443e-01 -5.40198863e-01
3.19648355e-01 7.83378959e-01 4.19142872e-01 -4.19675797e-01
-3.66074555e-02 8.84947777e-01 3.53158444e-01 -6.52328908e-01
3.28626633e-01 7.16823339e-01 1.87589422e-01 -1.39675462e+00
-3.54107231e-01 -4.84886616e-01 -5.89668870e-01 -1.90889195e-01
9.37691092e-01 -6.72967851e-01 -8.93438399e-01 7.00861812e-01
-8.72299671e-01 7.94826970e-02 1.57106802e-01 2.85459667e-01
-4.66745913e-01 8.49678755e-01 -6.10299766e-01 -9.12016869e-01
-3.47857811e-02 -1.12391007e+00 1.16529047e+00 3.58317465e-01
2.68446468e-02 -7.58051813e-01 1.16965994e-01 2.69502290e-02
4.49785471e-01 3.31204802e-01 5.38771629e-01 1.86406806e-01
-1.33031026e-01 -1.69346616e-01 -3.40633482e-01 3.26297075e-01
5.39495528e-01 2.80556500e-01 -8.53299022e-01 1.71294615e-01
3.18347633e-01 -3.41698855e-01 4.89127368e-01 2.66713709e-01
1.49519932e+00 -3.35978061e-01 -3.46173823e-01 8.59307408e-01
1.03735828e+00 5.34952641e-01 6.51678562e-01 1.28513843e-01
4.71396506e-01 7.75819302e-01 9.16479766e-01 6.67487681e-01
-4.73755598e-02 1.05457306e+00 -2.19471604e-01 -7.45071769e-02
5.09869933e-01 -4.02568914e-02 5.84408224e-01 7.59098232e-01
-4.29577023e-01 3.95099312e-01 -4.20737833e-01 3.31180990e-01
-1.78046870e+00 -1.10893011e+00 2.56322324e-01 1.78741026e+00
9.01022732e-01 -5.40136099e-01 -1.40097395e-01 -2.18762122e-02
6.09390080e-01 3.52480918e-01 -3.15096200e-01 -5.06443143e-01
-6.37835681e-01 2.93214649e-01 -3.27638425e-02 9.45940316e-02
-1.09783006e+00 7.80432701e-01 5.04728937e+00 1.51406598e+00
-1.65974057e+00 3.36787909e-01 7.67583013e-01 1.08150179e-02
7.86713809e-02 -3.26844096e-01 -5.33205032e-01 4.66403782e-01
6.23724341e-01 -2.97159702e-01 -7.99786747e-02 8.74582589e-01
3.70292842e-01 -1.33083478e-01 -4.36021447e-01 1.55515897e+00
3.23986173e-01 -8.42301130e-01 4.82136831e-02 9.79711935e-02
5.43979406e-01 -6.13603771e-01 2.62473702e-01 6.24845736e-02
-6.66839719e-01 -9.79882836e-01 3.30432147e-01 6.61671698e-01
1.15945709e+00 -8.77441585e-01 7.22317398e-01 1.83659047e-01
-1.47027647e+00 2.48828214e-02 -4.04356003e-01 -1.64389402e-01
1.09428458e-01 4.36776102e-01 -1.06626026e-01 4.40212369e-01
7.24015594e-01 1.00201666e+00 -4.06045526e-01 3.85501027e-01
-6.33449294e-04 4.43117440e-01 -5.03021955e-01 -1.28013670e-01
1.48773089e-01 -8.36315513e-01 3.51599634e-01 1.39426959e+00
5.08253574e-01 5.60644627e-01 -9.17968620e-03 6.64072514e-01
2.89882630e-01 7.70853341e-01 -8.24029624e-01 -1.50654957e-01
-1.00517295e-01 1.31141520e+00 -2.91042417e-01 -1.30168155e-01
-6.23177826e-01 1.23017716e+00 -1.89845636e-02 4.66695040e-01
-7.92493165e-01 -4.68114942e-01 8.73452127e-01 -8.71411860e-02
-5.49937561e-02 -2.87758946e-01 -6.32606670e-02 -1.22030473e+00
2.78951943e-01 -7.38807321e-01 2.60336876e-01 -5.85033953e-01
-1.12808299e+00 5.50758839e-01 1.86031107e-02 -1.07650757e+00
-2.26322874e-01 -6.52025342e-01 -6.20374560e-01 7.66818821e-01
-1.28225672e+00 -1.28350425e+00 -4.74722654e-01 1.08757651e+00
2.12460086e-01 -1.17137656e-01 7.96056449e-01 2.98276663e-01
-7.03989744e-01 8.31785381e-01 8.25657025e-02 2.53135473e-01
6.92255557e-01 -4.56001759e-01 -6.90936029e-01 6.81367517e-01
1.18589021e-01 6.88789248e-01 2.62181431e-01 -3.82164329e-01
-1.37091863e+00 -7.10479915e-01 7.65562773e-01 8.99643302e-02
5.20806432e-01 -1.80880994e-01 -9.16705608e-01 4.56578583e-01
-3.71015705e-02 2.84517378e-01 8.91502500e-01 -1.57023281e-01
-1.73893631e-01 -6.59190238e-01 -1.12903702e+00 5.47061741e-01
1.07813764e+00 -8.32548082e-01 -2.91668862e-01 9.79236439e-02
3.42305116e-02 -2.19058525e-02 -9.89009678e-01 7.00298548e-01
9.42283154e-01 -1.13536656e+00 6.41401350e-01 -1.63362086e-01
1.48923039e-01 -4.22007501e-01 -2.58028895e-01 -7.30762601e-01
-1.32953838e-01 -4.62996364e-01 2.33408198e-01 1.16501808e+00
-1.13481032e-02 -7.15512097e-01 7.53071368e-01 3.57913762e-01
3.60252827e-01 -9.26775813e-01 -1.32186627e+00 -6.78029060e-01
-2.95258641e-01 -3.38425517e-01 3.61957729e-01 9.58914220e-01
3.53861511e-01 3.91284041e-02 -3.59138191e-01 -2.30467111e-01
3.59436244e-01 4.25555855e-01 7.23855436e-01 -6.58135593e-01
7.49808475e-02 -5.28661132e-01 -8.56138289e-01 -1.04742539e+00
6.26228929e-01 -5.35925269e-01 -7.57642686e-02 -7.39542723e-01
4.05456632e-01 -1.89982384e-01 -4.90396202e-01 5.59197128e-01
1.14911459e-01 6.42277420e-01 -1.62008852e-01 1.95490241e-01
-4.05864477e-01 1.02802312e+00 1.30097401e+00 -1.55851886e-01
-2.60408133e-01 -1.82440415e-01 -2.33770967e-01 6.79331124e-01
6.44103229e-01 -9.71003473e-02 -3.23159069e-01 6.11356981e-02
-2.35078931e-01 1.00382939e-01 2.39562422e-01 -8.65002990e-01
2.05128193e-01 -4.74887222e-01 4.90311027e-01 -2.23585293e-01
6.46643877e-01 -6.35141671e-01 -7.61940842e-03 4.97140996e-02
1.51002675e-01 1.21800356e-01 2.87838221e-01 5.41304886e-01
-8.22829306e-01 2.13247955e-01 8.96446586e-01 1.56736761e-01
-1.09424651e+00 5.83533168e-01 -5.87967396e-01 -4.88548607e-01
1.22194850e+00 -5.45432866e-01 1.30590364e-01 -3.89905840e-01
-7.37755656e-01 -2.19988167e-01 3.61047983e-01 4.43979859e-01
8.00877631e-01 -1.39691269e+00 -5.26874602e-01 5.74697495e-01
2.92121261e-01 -6.99230850e-01 6.33101225e-01 1.40067947e+00
-3.14810187e-01 1.89807624e-01 -4.99060571e-01 -7.76210308e-01
-1.79733682e+00 2.28317052e-01 4.13178861e-01 1.16616294e-01
-3.64300460e-01 6.70625448e-01 4.72702891e-01 -6.08837828e-02
-3.55652273e-01 -6.87414110e-02 -4.69334722e-01 1.55568972e-01
6.56909466e-01 1.28660440e-01 -1.66574448e-01 -1.32783103e+00
-5.72514415e-01 1.32319081e+00 1.78791672e-01 -6.25015423e-02
1.03461850e+00 -2.81072736e-01 -7.10781753e-01 4.88315314e-01
1.84047675e+00 2.26538450e-01 -9.60042953e-01 -2.21264288e-01
-7.08236173e-02 -8.20914209e-01 -9.60541442e-02 -2.19670072e-01
-1.11749732e+00 1.04281282e+00 7.75680006e-01 -3.55496556e-01
1.71625316e+00 -1.26442567e-01 5.59600234e-01 1.93153054e-01
4.94493008e-01 -1.01774168e+00 1.19729400e-01 2.87943363e-01
9.46447074e-01 -8.40902269e-01 -2.85572916e-01 -5.43007970e-01
-6.66531324e-01 1.22920549e+00 6.68328583e-01 3.56143266e-02
8.95812154e-01 3.99331719e-01 3.64642620e-01 -7.01491460e-02
-3.02886873e-01 1.28884673e-01 8.03667232e-02 3.60152930e-01
3.80270034e-01 -5.50309382e-02 -5.55849016e-01 6.06736898e-01
1.49193242e-01 3.16379279e-01 -9.64718238e-02 8.67793322e-01
-2.87017435e-01 -9.86736178e-01 -3.98945302e-01 2.74449766e-01
-6.89086258e-01 2.95086265e-01 -2.35172406e-01 5.24520218e-01
3.22534025e-01 7.75404811e-01 9.88246202e-02 -4.73774314e-01
2.58322060e-01 2.75760591e-01 6.26686633e-01 1.81730266e-03
8.98726657e-02 2.79264361e-01 -2.70014346e-01 -8.38366807e-01
-8.29240620e-01 -7.05933571e-01 -1.19944489e+00 -1.46434307e-01
-4.09184620e-02 2.74659127e-01 4.14541125e-01 9.36719179e-01
5.17564893e-01 -1.72013387e-01 7.46374488e-01 -7.74928033e-01
-1.77651972e-01 -8.90144706e-01 -1.05988884e+00 6.30725384e-01
1.56556979e-01 -9.43651617e-01 -2.77203977e-01 1.32826686e-01] | [13.640531539916992, 1.7484341859817505] |
d1824562-3587-4697-b541-d61439a09359 | deep-contextualized-pairwise-semantic | 1909.09490 | null | https://arxiv.org/abs/1909.09490v1 | https://arxiv.org/pdf/1909.09490v1.pdf | Deep Contextualized Pairwise Semantic Similarity for Arabic Language Questions | Question semantic similarity is a challenging and active research problem that is very useful in many NLP applications, such as detecting duplicate questions in community question answering platforms such as Quora. Arabic is considered to be an under-resourced language, has many dialects, and rich in morphology. Combined together, these challenges make identifying semantically similar questions in Arabic even more difficult. In this paper, we introduce a novel approach to tackle this problem, and test it on two benchmarks; one for Modern Standard Arabic (MSA), and another for the 24 major Arabic dialects. We are able to show that our new system outperforms state-of-the-art approaches by achieving 93% F1-score on the MSA benchmark and 82% on the dialectical one. This is achieved by utilizing contextualized word representations (ELMo embeddings) trained on a text corpus containing MSA and dialectic sentences. This in combination with a pairwise fine-grained similarity layer, helps our question-to-question similarity model to generalize predictions on different dialects while being trained only on question-to-question MSA data. | ['Hussein T. Al-Natsheh', 'Hesham Al-Bataineh', 'Wael Farhan', 'Ahmad Mustafa', 'Haitham Seelawi'] | 2019-09-19 | null | null | null | null | ['question-similarity'] | ['natural-language-processing'] | [-1.50629237e-01 -3.77864651e-02 2.14085639e-01 -3.55745107e-01
-1.23159826e+00 -8.28164816e-01 7.99439371e-01 5.50792336e-01
-4.87652868e-01 5.08596659e-01 3.73087853e-01 -1.96052343e-01
-1.66333854e-01 -8.70231926e-01 -3.82454842e-01 -3.04012954e-01
2.13967860e-01 8.46902192e-01 6.74375296e-01 -1.23058093e+00
5.92100799e-01 2.25626528e-01 -1.54412127e+00 4.83071744e-01
1.48590422e+00 7.25820482e-01 2.91122701e-02 5.80938637e-01
-3.76172662e-01 8.29170525e-01 -7.23438144e-01 -1.06872153e+00
-2.71295249e-01 -7.70101488e-01 -1.49028528e+00 -2.19204038e-01
8.59399199e-01 4.05698409e-03 -1.11885868e-01 8.28145683e-01
4.40033108e-01 1.78536043e-01 8.15462649e-01 -8.34283471e-01
-1.20045912e+00 5.55271566e-01 -1.61622360e-01 3.72774124e-01
8.09819818e-01 -3.33419114e-01 1.69255555e+00 -1.17023945e+00
7.63917565e-01 1.45381749e+00 6.98758245e-01 4.98420686e-01
-7.23776281e-01 -1.67507648e-01 -2.72805631e-01 1.00143969e+00
-1.24447834e+00 -2.31971815e-01 4.51274216e-01 -1.61157399e-02
8.04666579e-01 4.19531703e-01 -5.96732050e-02 8.08819056e-01
5.03624380e-02 6.94370985e-01 1.10260952e+00 -7.61140943e-01
2.73414165e-01 9.84229594e-02 4.23420936e-01 6.89364612e-01
4.00647148e-02 -8.67803276e-01 -3.50847036e-01 -3.53769034e-01
-4.95227128e-02 -4.01391685e-01 -4.10068929e-01 5.19656725e-02
-1.06637692e+00 1.22264862e+00 5.20437837e-01 8.27917397e-01
-5.75229712e-02 -2.77012080e-01 2.32430667e-01 7.59245872e-01
2.40744174e-01 7.86512792e-01 -2.75241405e-01 -7.29086772e-02
-6.56255007e-01 8.19315553e-01 1.27885985e+00 5.74899375e-01
6.71423733e-01 -4.83411103e-01 -5.60914129e-02 1.38934386e+00
3.41921329e-01 5.96882105e-01 9.83720303e-01 -8.88269722e-01
6.39718652e-01 8.80638003e-01 -9.15843025e-02 -1.17170012e+00
-4.69993800e-01 1.54768089e-02 -3.62113237e-01 -4.47982579e-01
9.45617735e-01 7.00921193e-02 -2.99853891e-01 1.45550895e+00
4.33407694e-01 -3.35503161e-01 2.31623605e-01 8.96134973e-01
1.22901773e+00 8.07756484e-01 -3.51888537e-01 3.23089212e-01
1.95460570e+00 -1.10606802e+00 -6.74128175e-01 -3.46957058e-01
5.71389079e-01 -1.16455507e+00 1.42562675e+00 -5.65983728e-02
-1.00006282e+00 -2.33759522e-01 -1.13869286e+00 -5.52425742e-01
-7.09875286e-01 -7.91749507e-02 -9.03669298e-02 1.00413406e+00
-9.36980784e-01 4.12298501e-01 -3.17718089e-01 -1.00638068e+00
1.26582682e-01 -2.77528800e-02 -4.18528289e-01 -5.76597154e-01
-1.45181835e+00 1.50770772e+00 8.54434520e-02 -2.78565109e-01
-5.09008110e-01 -5.24219930e-01 -8.52351964e-01 1.58339798e-01
2.86478609e-01 -2.49035135e-01 1.14352596e+00 -7.03492939e-01
-1.30728340e+00 1.27845073e+00 -2.05168724e-01 -3.78570497e-01
-3.68883014e-02 -1.46402285e-01 -6.45064950e-01 5.47035813e-01
4.19724494e-01 3.96533519e-01 6.34407640e-01 -1.10519087e+00
-4.78248984e-01 -6.27447367e-01 2.80951053e-01 1.95559606e-01
-4.54526424e-01 4.75398630e-01 -2.25873157e-01 -7.68622637e-01
2.47263059e-01 -9.87571061e-01 3.95570248e-02 -8.15631077e-02
1.80362672e-01 -6.29443228e-01 7.22654223e-01 -1.29984248e+00
1.09075010e+00 -1.66039765e+00 1.83614001e-01 -3.12067359e-03
-6.63862154e-02 2.28511587e-01 -5.17974794e-01 7.24030495e-01
3.30927819e-01 -1.39244214e-01 -7.10260391e-01 6.88207969e-02
1.45421103e-01 2.69832939e-01 -2.33748838e-01 2.41531461e-01
7.37693191e-01 9.90782320e-01 -1.00103951e+00 -5.09302616e-01
-4.63120699e-01 1.09099142e-01 -4.39448595e-01 2.71813750e-01
-3.59846920e-01 5.92382625e-02 -1.19204380e-01 7.73040354e-01
5.97261965e-01 -1.64806172e-01 4.15856749e-01 1.61230639e-01
3.45182627e-01 5.97529411e-01 -9.86676395e-01 1.63274336e+00
-5.30293047e-01 4.94753033e-01 -3.96639667e-03 -1.03315520e+00
1.09302509e+00 2.39468470e-01 -9.02913213e-02 -9.07817483e-01
1.98160797e-01 7.47574449e-01 2.08775610e-01 -7.33093917e-01
8.44959617e-01 -8.00766572e-02 -3.06952775e-01 8.64496529e-01
3.45593095e-01 -5.42237818e-01 5.72119951e-01 2.68669128e-01
9.59100127e-01 -2.21354991e-01 3.81513923e-01 -6.73689008e-01
1.24708760e+00 6.73367605e-02 -4.71856110e-02 4.63217765e-01
-2.99742073e-01 9.25351501e-01 5.10181427e-01 -2.14952841e-01
-5.72230160e-01 -1.25420868e+00 -1.45697191e-01 1.61915660e+00
1.79075167e-01 -3.80329549e-01 -1.00969267e+00 -1.02742672e+00
1.88232157e-02 7.32129931e-01 -7.06118345e-01 -5.90126142e-02
-1.11537957e+00 -7.02840328e-01 8.47435355e-01 5.47271185e-02
3.34115893e-01 -9.46123421e-01 -2.04041481e-01 3.34426135e-01
-6.40272141e-01 -1.08732867e+00 -4.92516190e-01 -3.12994719e-01
-4.93565977e-01 -1.50159788e+00 -1.01178420e+00 -1.29111445e+00
3.53624113e-02 1.74637154e-01 1.56667066e+00 1.50272995e-01
-3.90692614e-02 5.56989610e-01 -8.98197889e-01 -2.23449379e-01
-7.37643480e-01 4.26132828e-01 -3.42403442e-01 1.00783788e-01
5.46101153e-01 -2.73615479e-01 -2.38491371e-01 3.78900737e-01
-1.08510828e+00 -9.48405981e-01 1.24389753e-01 8.70134711e-01
8.53454769e-02 -8.19167674e-01 1.17740428e+00 -7.12074935e-01
9.95722115e-01 -9.18701470e-01 -1.38569966e-01 7.96874166e-01
-2.22762942e-01 1.48537263e-01 7.25420475e-01 -3.16817909e-02
-9.45329309e-01 -5.82369089e-01 -6.78601086e-01 5.72112381e-01
9.56563428e-02 3.38024318e-01 -2.31962711e-01 -5.48367947e-02
1.16497087e+00 1.77948758e-01 2.60981083e-01 -5.05288363e-01
6.76642179e-01 1.14803672e+00 4.95463699e-01 -6.13542914e-01
5.02739668e-01 3.73917282e-01 -2.77304024e-01 -9.54455733e-01
-8.77084792e-01 -6.02922082e-01 -5.23735642e-01 3.18784788e-02
8.25474620e-01 -6.24695241e-01 -5.12162268e-01 3.31546456e-01
-1.18122709e+00 2.31147349e-01 -3.30610424e-02 -4.04160358e-02
-2.60028958e-01 6.46750271e-01 -6.76798522e-01 -1.61338940e-01
-3.73174757e-01 -9.13552880e-01 9.23017859e-01 2.63175458e-01
-4.32768345e-01 -1.20253110e+00 3.73868346e-01 9.32480752e-01
6.21612906e-01 -1.48932725e-01 1.29935539e+00 -1.33743310e+00
-8.56394991e-02 1.07628606e-01 -2.07247049e-01 4.18006897e-01
2.86109895e-01 -4.58996356e-01 -8.38596642e-01 -1.78245753e-01
2.73141623e-01 -6.56005979e-01 6.45831347e-01 -4.60570604e-01
6.22633636e-01 -1.69319525e-01 2.95729011e-01 -3.72925758e-01
1.11256123e+00 -4.74651217e-01 5.05558193e-01 4.99925971e-01
4.14253771e-01 1.06925476e+00 5.97597957e-01 1.08792931e-01
1.22247124e+00 6.86028302e-01 4.70871061e-01 6.74083292e-01
-3.35404366e-01 2.35012859e-01 4.21639264e-01 1.29015434e+00
3.63087714e-01 -9.39207003e-02 -1.20519030e+00 1.04012156e+00
-1.58549476e+00 -7.61697769e-01 -5.57829857e-01 1.81959617e+00
1.13784635e+00 -6.79849207e-01 9.20456499e-02 1.92972824e-01
6.97556615e-01 1.86705887e-01 -7.94483125e-02 -8.21924567e-01
-5.15234888e-01 7.26526499e-01 -2.87695378e-01 5.56551397e-01
-9.69377756e-01 9.47663009e-01 5.58174515e+00 8.29783499e-01
-7.71595359e-01 5.43936312e-01 2.83349812e-01 3.80539089e-01
-5.75294077e-01 -2.16544643e-01 -5.56901336e-01 3.94366831e-01
1.17527640e+00 -1.40412182e-01 1.93679050e-01 4.99235004e-01
-4.34420437e-01 -4.79209907e-02 -1.05236340e+00 7.59638131e-01
1.03223169e+00 -1.37920570e+00 3.22097957e-01 -5.91146588e-01
7.77648926e-01 9.66441259e-02 -1.59741685e-01 4.50036258e-01
1.38477668e-01 -1.23420370e+00 4.21789914e-01 2.44654622e-02
2.37522736e-01 -7.32595384e-01 1.05084932e+00 2.23197445e-01
-8.48823309e-01 -9.69770476e-02 -4.14861113e-01 7.52863213e-02
-1.08760305e-01 7.96383247e-03 -8.04123461e-01 8.01808178e-01
8.62850904e-01 4.91830140e-01 -1.14495122e+00 7.86481619e-01
-2.96560109e-01 6.29521251e-01 -7.21736178e-02 -4.48513150e-01
5.44900119e-01 -1.90402508e-01 2.64332652e-01 1.20686817e+00
2.16982186e-01 -2.48052314e-01 -1.50536314e-01 5.68191946e-01
-3.80433232e-01 8.22203755e-01 -1.47609621e-01 1.25637963e-01
4.43435818e-01 1.18592155e+00 -3.71526271e-01 -8.22733492e-02
-5.32024324e-01 1.22868097e+00 5.57170451e-01 -1.74192220e-01
-4.98867214e-01 -9.54762697e-01 5.10434628e-01 -2.74681263e-02
3.49973470e-01 -3.79633866e-02 -6.20879158e-02 -1.27988172e+00
2.26500869e-01 -1.40883780e+00 8.39563012e-01 -5.58305800e-01
-1.89306772e+00 7.02424586e-01 -5.13753057e-01 -9.72949803e-01
-5.40585756e-01 -8.19740534e-01 -5.33489048e-01 9.27049637e-01
-1.77840292e+00 -1.34058249e+00 -1.92952394e-01 7.32557654e-01
5.55452943e-01 -3.26087475e-01 1.03892314e+00 3.63441974e-01
-3.19657885e-02 8.55438769e-01 5.12542784e-01 2.84887403e-01
1.21942306e+00 -1.37504709e+00 4.40527320e-01 6.65440142e-01
4.69171017e-01 4.75411654e-01 3.51905376e-01 -9.47623327e-02
-1.27540624e+00 -8.87556434e-01 1.60722923e+00 -9.06413138e-01
1.05775714e+00 -1.79076776e-01 -1.51302707e+00 3.11562359e-01
6.43834770e-01 -2.17787817e-01 9.79681194e-01 9.98308137e-02
-8.36614490e-01 -4.08508768e-03 -1.27206612e+00 6.90516353e-01
6.00192606e-01 -7.25960612e-01 -1.51128626e+00 5.67417204e-01
5.81149340e-01 -1.15818195e-01 -1.08784389e+00 5.49045876e-02
2.70811558e-01 -9.98165369e-01 8.64506006e-01 -7.51575828e-01
6.79604232e-01 -6.49997294e-02 -5.59067786e-01 -1.28137183e+00
8.55769515e-02 -2.37410784e-01 1.55800164e-01 1.32799911e+00
4.73654836e-01 -7.09894001e-01 5.37510633e-01 6.71505108e-02
-1.02386929e-01 -6.90558612e-01 -1.26741552e+00 -8.07301044e-01
1.02226615e+00 -8.87661576e-02 7.09368229e-01 1.16740298e+00
2.37494960e-01 6.44052684e-01 1.67529464e-01 1.10519186e-01
9.70448032e-02 3.52439851e-01 3.61003041e-01 -1.16657150e+00
3.98738161e-02 -6.51226223e-01 -3.61297101e-01 -8.39675903e-01
5.18824339e-01 -1.22248530e+00 2.27383971e-02 -1.58949065e+00
-2.02174574e-01 -2.44538650e-01 1.76024780e-01 -7.47413486e-02
-6.53053641e-01 5.70138633e-01 3.41149300e-01 8.67220610e-02
-6.98135734e-01 6.98243022e-01 9.75774944e-01 -2.46546015e-01
3.15841019e-01 -3.74039710e-01 -8.29149187e-01 6.11245215e-01
8.11293542e-01 -5.06581545e-01 7.17337802e-02 -7.00149179e-01
3.50171089e-01 -2.95360625e-01 1.11946926e-01 -8.37966383e-01
1.93214640e-01 1.66658297e-01 -1.46537989e-01 -2.76369065e-01
2.19258189e-01 -4.91292298e-01 -9.66460288e-01 5.95266223e-01
-2.25336686e-01 4.09711987e-01 -8.73111710e-02 1.89187363e-01
-5.63469827e-01 -8.74748886e-01 7.76395738e-01 -1.38233230e-01
-6.54208064e-01 -2.34755158e-01 -5.81005812e-01 1.03786099e+00
7.37519264e-01 1.48525283e-01 -5.90152025e-01 -5.25440991e-01
-3.44469070e-01 4.52106923e-01 3.23075712e-01 9.88240480e-01
4.52086926e-01 -1.30091548e+00 -1.41877830e+00 -1.51743889e-01
6.14248097e-01 -5.58029890e-01 2.31129736e-01 4.92154896e-01
-9.93439019e-01 4.28740293e-01 -2.61004329e-01 -4.27615136e-01
-1.13055336e+00 2.80064661e-02 3.05576980e-01 -3.86194259e-01
1.59740776e-01 9.09948766e-01 -6.01839662e-01 -1.33406234e+00
-2.22000360e-01 -9.07093883e-02 -7.32420206e-01 5.42618632e-01
7.37221062e-01 7.27520883e-01 4.51158762e-01 -1.00205326e+00
-6.39814377e-01 7.13749230e-01 -1.28118798e-01 -2.41322026e-01
1.08315051e+00 -2.48729929e-01 -8.09864461e-01 3.10058743e-01
1.31451917e+00 4.42189366e-01 -1.46796927e-01 -5.34514785e-01
6.66904151e-01 -2.51060039e-01 -7.19669521e-01 -7.95921147e-01
-4.40237135e-01 1.02804720e+00 4.96371269e-01 3.87266457e-01
6.07225299e-01 4.43183661e-01 1.15005803e+00 9.34489846e-01
2.38329563e-02 -1.10346043e+00 3.98388773e-01 1.24781728e+00
1.34692824e+00 -1.47527230e+00 -3.78250182e-01 -3.76500160e-01
-6.71283305e-01 1.13849008e+00 4.62920249e-01 -2.91215241e-01
4.78811055e-01 -6.36160314e-01 3.12249213e-01 -7.88467675e-02
-3.59300613e-01 -3.72538507e-01 5.24910867e-01 6.73251987e-01
5.60277402e-01 -2.59626567e-01 -7.11958289e-01 8.64859939e-01
-5.26071310e-01 -6.42580867e-01 7.67987013e-01 8.42661440e-01
-6.44511938e-01 -1.42252851e+00 -6.27490819e-01 3.76273215e-01
-6.58945143e-01 -2.60509998e-01 -6.09418154e-01 7.14531183e-01
-4.34009060e-02 1.43086934e+00 2.16637686e-01 2.53440603e-03
3.99281800e-01 3.30418617e-01 5.34661949e-01 -6.97784066e-01
-1.28393257e+00 -9.50267911e-01 2.66290635e-01 -2.96432763e-01
-2.47544467e-01 -7.50288129e-01 -1.19747007e+00 -3.03654522e-01
-9.35762376e-02 4.51670855e-01 4.87934411e-01 1.21059513e+00
3.14802855e-01 -4.83416533e-03 6.69344485e-01 -1.72733426e-01
-9.32493150e-01 -1.15796089e+00 -3.35178196e-01 7.45501161e-01
1.13444582e-01 -1.88921720e-01 -2.20274508e-01 -1.86169505e-01] | [11.334060668945312, 8.088408470153809] |
39e13f4a-d126-49b9-ace4-183093ce263f | weakly-supervised-mapping-of-natural-language | 2112.06311 | null | https://arxiv.org/abs/2112.06311v3 | https://arxiv.org/pdf/2112.06311v3.pdf | Weakly Supervised Text-to-SQL Parsing through Question Decomposition | Text-to-SQL parsers are crucial in enabling non-experts to effortlessly query relational data. Training such parsers, by contrast, generally requires expertise in annotating natural language (NL) utterances with corresponding SQL queries. In this work, we propose a weak supervision approach for training text-to-SQL parsers. We take advantage of the recently proposed question meaning representation called QDMR, an intermediate between NL and formal query languages. Given questions, their QDMR structures (annotated by non-experts or automatically predicted), and the answers, we are able to automatically synthesize SQL queries that are used to train text-to-SQL models. We test our approach by experimenting on five benchmark datasets. Our results show that the weakly supervised models perform competitively with those trained on annotated NL-SQL data. Overall, we effectively train text-to-SQL parsers, while using zero SQL annotations. | ['Daniel Deutch', 'Jonathan Berant', 'Tomer Wolfson'] | 2021-12-12 | null | https://aclanthology.org/2022.findings-naacl.193 | https://aclanthology.org/2022.findings-naacl.193.pdf | findings-naacl-2022-7 | ['text-to-sql'] | ['computer-code'] | [ 1.54532492e-01 7.70089567e-01 -1.66780710e-01 -8.99533272e-01
-1.32343733e+00 -7.29346335e-01 5.33185601e-01 4.83561367e-01
-2.48563021e-01 2.95851260e-01 3.26948553e-01 -8.37331116e-01
2.80975640e-01 -1.14295495e+00 -1.01605058e+00 4.05605108e-01
4.06403810e-01 8.82686496e-01 5.78922868e-01 -4.16291356e-01
8.47745389e-02 2.14269146e-01 -1.34839988e+00 8.34244430e-01
8.88138652e-01 9.50819194e-01 1.02277324e-01 8.53294790e-01
-7.71461546e-01 1.71355867e+00 -5.04844189e-01 -8.78059566e-01
-1.08949088e-01 -2.05009207e-01 -1.38036418e+00 -2.41494134e-01
3.88634384e-01 -3.50275099e-01 -4.79713567e-02 6.97532833e-01
-6.18947893e-02 -2.82840371e-01 2.01472372e-01 -8.55765700e-01
-6.03250206e-01 8.96862566e-01 3.29067409e-01 -3.44742477e-01
9.63016987e-01 3.81591357e-02 1.65981138e+00 -9.28229392e-01
7.04005897e-01 1.36936605e+00 3.90410811e-01 5.93248904e-01
-1.36972106e+00 1.76332816e-01 -1.46402851e-01 -2.63966352e-01
-1.01193917e+00 -5.82077205e-01 5.60551286e-01 -2.99012184e-01
1.47443712e+00 5.05939484e-01 -7.52249211e-02 7.25333929e-01
2.19240263e-02 8.47368002e-01 7.34307230e-01 -6.08330250e-01
1.62137613e-01 5.27051270e-01 3.41391206e-01 8.48222136e-01
-1.06364384e-01 -2.33392298e-01 -5.43185472e-01 -4.07955796e-01
4.49127764e-01 -2.59201944e-01 1.03873096e-01 -2.63615310e-01
-9.20663953e-01 9.64682877e-01 1.70069903e-01 1.30890802e-01
-2.18183264e-01 4.49552387e-02 5.61385095e-01 6.89449131e-01
2.57661879e-01 8.18585694e-01 -8.85704875e-01 -5.06199747e-02
-4.47994530e-01 3.41702819e-01 1.45572364e+00 1.34013975e+00
9.14055526e-01 -3.37835729e-01 -1.40566036e-01 7.46355474e-01
3.13084126e-01 5.76008022e-01 1.54075354e-01 -1.23386550e+00
8.32507372e-01 1.02067530e+00 1.44152567e-01 -6.52154267e-01
-1.78339124e-01 1.10829920e-01 -2.39727557e-01 -4.40523535e-01
2.72661030e-01 9.22486186e-03 -4.23324257e-01 1.44370103e+00
2.15815306e-01 -6.07026458e-01 6.47460103e-01 4.86526251e-01
9.41871703e-01 5.81794679e-01 2.39788264e-01 -3.40139046e-02
1.47534192e+00 -7.49397337e-01 -5.44475496e-01 -3.68791610e-01
1.05184662e+00 -5.89968204e-01 1.78979957e+00 2.56921589e-01
-1.15969074e+00 -7.37283051e-01 -5.84530652e-01 -4.64706928e-01
-2.79805094e-01 2.44454727e-01 4.94003415e-01 4.49124008e-01
-9.98366833e-01 1.37246430e-01 -9.17501628e-01 -4.50799525e-01
6.72335625e-02 2.69222468e-01 -2.59734601e-01 -1.18391939e-01
-1.01142716e+00 7.92621076e-01 5.26928067e-01 -2.45133355e-01
-7.56708145e-01 -6.38190746e-01 -1.03560019e+00 1.25650242e-01
8.79930913e-01 -5.84270298e-01 1.86338031e+00 -4.73108053e-01
-1.53323054e+00 9.58670557e-01 -5.40322483e-01 -6.24687910e-01
1.09154351e-01 -4.42924619e-01 -1.42122924e-01 2.48779684e-01
2.96020210e-01 5.73054910e-01 2.49037832e-01 -1.15769613e+00
-5.26380777e-01 -3.26149404e-01 5.15923381e-01 -2.28316486e-01
-9.12871808e-02 4.02087390e-01 -4.83088166e-01 -1.69678718e-01
2.06053168e-01 -4.90878701e-01 -2.62090653e-01 -2.68498123e-01
-6.77543044e-01 -7.46216297e-01 5.53459883e-01 -5.12933612e-01
1.03480530e+00 -1.86467683e+00 -2.58604527e-01 6.84902593e-02
1.46203026e-01 1.99838933e-02 -1.70073986e-01 7.38852739e-01
7.44031817e-02 2.56881058e-01 -1.76660761e-01 -1.15348086e-01
2.91248441e-01 7.20717549e-01 -9.33541715e-01 -5.06950319e-01
9.75045085e-01 1.15828562e+00 -7.28251219e-01 -7.82851160e-01
-9.45265871e-03 -1.60862058e-01 -8.29930544e-01 1.15670729e+00
-1.18874252e+00 2.90590554e-01 -7.98707545e-01 6.08951211e-01
2.20631450e-01 -3.06470126e-01 4.52581018e-01 -1.01939499e-01
1.98293343e-01 9.55238938e-01 -8.42076838e-01 1.47679925e+00
-7.31717348e-01 2.20862076e-01 -2.05279037e-01 -9.90325570e-01
1.23205256e+00 4.08346564e-01 -3.09577561e-04 -8.64053130e-01
-3.66965413e-01 4.18271780e-01 -4.87869918e-01 -1.00844610e+00
4.14662898e-01 -1.79359764e-01 -6.28996134e-01 4.95152742e-01
3.97187233e-01 -6.29232287e-01 3.28578740e-01 4.44368303e-01
1.28529429e+00 2.66502231e-01 2.67353684e-01 1.88621767e-02
8.84677529e-01 3.75748008e-01 2.06719920e-01 9.50866878e-01
4.48864281e-01 2.47272909e-01 1.12421489e+00 -5.26665390e-01
-8.69914412e-01 -1.11511457e+00 1.25330672e-01 1.56337726e+00
-4.33574766e-01 -7.44346142e-01 -6.89340353e-01 -1.03402174e+00
-2.62720525e-01 1.08415258e+00 -8.87705684e-02 6.50868863e-02
-8.83597672e-01 2.21379474e-01 7.51671255e-01 6.64736152e-01
2.69511994e-02 -1.31593156e+00 -6.92338288e-01 4.23423529e-01
-1.48444161e-01 -1.60483932e+00 -3.74290049e-02 4.86729443e-01
-9.53367770e-01 -1.15401185e+00 2.17571512e-01 -7.76380479e-01
5.02348483e-01 -2.99602181e-01 1.81737709e+00 2.60746300e-01
1.57743961e-01 5.55257261e-01 -4.99272287e-01 -4.96019006e-01
-1.01968825e+00 3.57575446e-01 -5.96210301e-01 -3.43535841e-01
6.17380977e-01 -1.56691641e-01 1.25771547e-02 2.15059713e-01
-1.33112097e+00 7.06056133e-04 6.13367140e-01 5.30736864e-01
6.45474195e-01 -3.15052539e-01 6.26215458e-01 -1.60833955e+00
7.06917286e-01 -2.27175102e-01 -7.97317088e-01 7.34256029e-01
-3.90436411e-01 7.40531862e-01 1.06536782e+00 1.00551769e-01
-1.22306657e+00 1.98582530e-01 -5.55466175e-01 1.33914560e-01
-4.50322092e-01 8.14920187e-01 -4.32576329e-01 3.77406001e-01
8.40339899e-01 4.15991172e-02 -3.20659667e-01 -4.48584110e-01
6.81131005e-01 6.79526925e-01 6.60507679e-01 -1.22079599e+00
8.09503853e-01 3.95673215e-02 -2.53006011e-01 -4.57996249e-01
-1.20022106e+00 -3.64401549e-01 -5.76676965e-01 4.13816214e-01
9.00749624e-01 -7.58268595e-01 -6.77570403e-01 -1.86732590e-01
-1.45068061e+00 -4.11738366e-01 -5.99279642e-01 4.58423905e-02
-6.41782641e-01 1.83416545e-01 -6.72681153e-01 -8.44492376e-01
-2.61927545e-01 -9.63618398e-01 1.30007231e+00 -6.31867200e-02
-3.75981152e-01 -9.03153300e-01 2.03925386e-01 5.55323422e-01
3.83622825e-01 -2.82492973e-02 1.57465649e+00 -1.35648751e+00
-7.70369351e-01 -3.19262445e-01 -2.64269680e-01 4.34914857e-01
-2.18522213e-02 -8.27791691e-02 -8.88214290e-01 3.50742221e-01
1.96543455e-01 -1.08868337e+00 2.99602389e-01 -4.35403883e-01
1.18554854e+00 -6.32494450e-01 3.72439884e-02 1.26266535e-02
1.46060276e+00 -1.16070181e-01 5.09423018e-01 -5.37461713e-02
4.34112728e-01 1.03186035e+00 7.41764605e-01 2.03727975e-01
8.75536382e-01 3.95015031e-01 3.56166869e-01 9.12651941e-02
3.60973328e-01 -8.09138656e-01 2.47338936e-01 8.52137506e-01
5.56088567e-01 -5.72684593e-02 -1.16654325e+00 4.45065886e-01
-1.60108149e+00 -4.76293832e-01 -3.58493030e-01 1.85231102e+00
1.41733277e+00 3.19075286e-01 -1.00430191e-01 -2.59551942e-01
6.36091232e-02 3.38631682e-02 -4.06414986e-01 -7.50257432e-01
9.00752395e-02 8.69843602e-01 4.48275432e-02 6.18959963e-01
-8.69883537e-01 1.19653583e+00 5.91525030e+00 2.25740418e-01
-8.20050299e-01 -4.24420722e-02 4.57882464e-01 5.35248220e-01
-7.58554876e-01 4.68023598e-01 -9.06731725e-01 -2.78454512e-01
1.49549985e+00 6.13036416e-02 3.29807967e-01 1.08267248e+00
1.25842886e-02 -8.12589377e-03 -1.74467945e+00 3.93767029e-01
-1.23353139e-01 -1.28928792e+00 3.37145090e-01 -5.98623753e-01
2.93882072e-01 -4.69448883e-03 -4.39671010e-01 7.71755397e-01
6.08128369e-01 -1.10226440e+00 5.48507512e-01 6.57383800e-01
5.22031963e-01 -3.60910982e-01 8.68705511e-01 6.43338025e-01
-1.14821064e+00 -3.11415177e-02 -5.53208411e-01 1.02616116e-01
-9.12888348e-02 4.08171326e-01 -1.09098959e+00 7.70510793e-01
5.50447762e-01 2.62861997e-01 -1.00848401e+00 1.10604614e-01
-4.69909638e-01 5.73791444e-01 -2.48568386e-01 -1.35313407e-01
1.24597000e-02 4.14958224e-02 -1.85000956e-01 1.15556347e+00
-7.26180896e-03 1.43191934e-01 3.60903829e-01 1.16623068e+00
-2.99974740e-01 2.49688670e-01 -6.91510499e-01 -3.66572708e-01
3.10525149e-01 1.03152156e+00 -1.29263401e-01 -6.43476725e-01
-8.87734830e-01 5.96231461e-01 5.10429561e-01 2.92987257e-01
-4.23838377e-01 -5.32006383e-01 5.18109202e-02 3.19636583e-01
2.26112753e-01 -5.72842211e-02 -2.82937944e-01 -1.25058067e+00
5.62500477e-01 -1.45511556e+00 5.19102991e-01 -9.79318738e-01
-1.38238525e+00 7.42033601e-01 2.21797321e-02 -6.87372506e-01
-6.80662513e-01 -6.32631779e-01 -4.35940325e-01 8.36259961e-01
-1.57929718e+00 -1.11441004e+00 -1.49428770e-01 4.22387898e-01
3.66471857e-01 7.76788816e-02 1.18041682e+00 -2.97601838e-02
-9.00852010e-02 3.01341116e-01 -7.06091523e-01 4.89936262e-01
5.51853359e-01 -1.46492791e+00 5.64131200e-01 5.40980339e-01
3.93792033e-01 1.06364441e+00 5.39139926e-01 -4.93475467e-01
-1.86474943e+00 -1.26476395e+00 1.44931996e+00 -8.93851817e-01
8.36090803e-01 -5.07248282e-01 -1.31571448e+00 1.06605673e+00
3.19878191e-01 9.83671993e-02 8.26575160e-01 1.23341121e-01
-5.95226109e-01 -2.51338333e-01 -9.31739986e-01 3.09753895e-01
5.95660210e-01 -1.17457330e+00 -1.12814581e+00 5.48599362e-01
1.42293096e+00 -3.64185452e-01 -1.18692482e+00 4.28367496e-01
1.30319417e-01 -9.99825120e-01 7.34384656e-01 -1.12586260e+00
5.96554935e-01 -1.99739024e-01 -5.67713678e-01 -5.02964318e-01
6.37502551e-01 -4.75194126e-01 -3.54303122e-02 1.26265347e+00
7.30677843e-01 -5.50630689e-01 9.39675093e-01 1.04400754e+00
-1.17105134e-01 -7.95314312e-01 -5.24255693e-01 -3.85462105e-01
9.19518024e-02 -7.45026946e-01 6.85678244e-01 5.53438902e-01
7.04053342e-02 8.13844025e-01 3.01553965e-01 1.72657996e-01
1.94205910e-01 2.47780666e-01 1.15517950e+00 -1.26763821e+00
-6.59946263e-01 9.24768522e-02 1.26047775e-01 -1.48655879e+00
3.63281429e-01 -9.32857573e-01 2.11500600e-01 -1.62536061e+00
4.15627472e-02 -4.27684426e-01 3.27495813e-01 7.34170258e-01
-2.27254525e-01 -5.40821552e-01 -1.05281395e-03 3.91174965e-02
-8.36139143e-01 2.67800003e-01 8.47551227e-01 4.74881753e-02
-1.59458593e-01 2.00669751e-01 -8.08895946e-01 5.65783739e-01
6.45670176e-01 -7.51358390e-01 -5.65128624e-01 -7.10162342e-01
7.16214716e-01 6.49682879e-01 3.11857939e-01 -6.86251938e-01
3.47721428e-01 -1.84201971e-01 -2.57460535e-01 -6.87643707e-01
-3.42972167e-02 -6.80311203e-01 -3.11268926e-01 2.08309501e-01
-9.13075089e-01 2.71313071e-01 -6.26698807e-02 2.33588308e-01
-7.28413165e-01 -5.52080333e-01 3.37955892e-01 -3.88350040e-01
-2.65857071e-01 -4.84536812e-02 -2.68169433e-01 7.62450457e-01
4.62604880e-01 4.33285981e-01 -3.80588323e-01 -4.84206498e-01
-6.03426516e-01 3.46941292e-01 2.53539681e-01 1.79601237e-01
7.48861492e-01 -8.96270335e-01 -5.34846544e-01 3.77965182e-01
5.31988323e-01 3.53733569e-01 -4.65366542e-01 2.89044410e-01
-7.25415409e-01 1.10572779e+00 3.98475885e-01 -6.67852402e-01
-7.81286120e-01 5.60130239e-01 2.01901361e-01 -7.57446289e-01
-4.99547683e-02 6.90126538e-01 -5.66763896e-03 -1.37723053e+00
3.14181775e-01 -9.37091112e-01 -6.85891928e-03 -5.33830225e-01
1.44905180e-01 -3.01719964e-01 2.32024148e-01 -8.54153559e-02
-2.12752849e-01 3.73037189e-01 2.14167312e-02 -2.16562614e-01
1.34907496e+00 2.90787846e-01 -5.62753022e-01 4.51202899e-01
1.26024187e+00 1.90129325e-01 -7.13295043e-01 -7.24860311e-01
1.02500796e+00 -1.57849073e-01 -6.97495222e-01 -5.82180142e-01
-4.88475710e-01 1.05083430e+00 -1.23588800e-01 6.60983443e-01
1.07590389e+00 5.84276974e-01 8.56668293e-01 1.33467114e+00
4.32344824e-01 -7.26556897e-01 4.67877954e-01 7.40337551e-01
7.59851098e-01 -1.27784753e+00 -4.51375008e-01 -5.03046453e-01
-7.00694084e-01 1.29922998e+00 8.16386878e-01 1.75672304e-02
2.38618091e-01 3.82840544e-01 3.15974534e-01 -3.95717174e-01
-1.24579680e+00 -1.96963266e-01 1.12034082e-01 4.68050510e-01
7.91037202e-01 -1.92135394e-01 1.34861186e-01 6.55585706e-01
-4.19279814e-01 7.51100630e-02 3.93861085e-01 1.32088900e+00
-6.13548100e-01 -1.85175419e+00 -1.45585284e-01 5.78860402e-01
-5.73316038e-01 -3.07517081e-01 -6.19828641e-01 6.79430246e-01
-3.92372966e-01 1.28655279e+00 -5.92311583e-02 -1.70152083e-01
8.30351472e-01 4.78267878e-01 2.47339725e-01 -1.27863288e+00
-9.20723975e-01 -3.51898104e-01 6.51466429e-01 -7.07002878e-01
-2.21850768e-01 -2.76356280e-01 -1.51871455e+00 1.17001541e-01
-8.51608962e-02 5.96660256e-01 3.82591814e-01 9.54138398e-01
3.58175218e-01 2.03116164e-01 6.34404838e-01 3.34075034e-01
-1.06332707e+00 -7.97417104e-01 1.25032291e-01 3.42640579e-01
1.26720145e-01 2.85993099e-01 9.41202417e-02 3.28704000e-01] | [10.011574745178223, 7.837662220001221] |
8d03f52d-e037-412b-8956-a487972110d3 | uni-perceiver-moe-learning-sparse-generalist | 2206.04674 | null | https://arxiv.org/abs/2206.04674v2 | https://arxiv.org/pdf/2206.04674v2.pdf | Uni-Perceiver-MoE: Learning Sparse Generalist Models with Conditional MoEs | To build an artificial neural network like the biological intelligence system, recent works have unified numerous tasks into a generalist model, which can process various tasks with shared parameters and do not have any task-specific modules. While generalist models achieve promising results on various benchmarks, they have performance degradation on some tasks compared with task-specialized models. In this work, we find that interference among different tasks and modalities is the main factor to this phenomenon. To mitigate such interference, we introduce the Conditional Mixture-of-Experts (Conditional MoEs) to generalist models. Routing strategies under different levels of conditions are proposed to take both the training/inference cost and generalization ability into account. By incorporating the proposed Conditional MoEs, the recently proposed generalist model Uni-Perceiver can effectively mitigate the interference across tasks and modalities, and achieves state-of-the-art results on a series of downstream tasks via prompt tuning on 1% of downstream data. Moreover, the introduction of Conditional MoEs still holds the generalization ability of generalist models to conduct zero-shot inference on new tasks, e.g., video-text retrieval and video caption. Code and pre-trained generalist models shall be released. | ['Jifeng Dai', 'Xiaogang Wang', 'Hongsheng Li', 'Xiaohua Wang', 'Wenhai Wang', 'Xizhou Zhu', 'Jinguo Zhu'] | 2022-06-09 | null | null | null | null | ['video-text-retrieval'] | ['computer-vision'] | [ 2.71658659e-01 7.65579846e-03 -8.12240317e-02 -5.12558818e-01
-5.55594444e-01 -3.00374091e-01 6.47965670e-01 -2.75273234e-01
-6.85692430e-01 4.63299870e-01 4.03147079e-02 -6.47699609e-02
-1.60144374e-01 -3.41997743e-01 -7.87146270e-01 -5.86256444e-01
3.30725372e-01 3.10435832e-01 4.53650564e-01 -1.67789832e-01
2.10524470e-01 8.68019536e-02 -1.65447927e+00 7.94446528e-01
9.50194716e-01 9.81432319e-01 8.73160422e-01 5.95134914e-01
-4.80317883e-02 6.12299144e-01 -7.48059392e-01 -5.43439388e-01
4.15345579e-02 1.18321776e-02 -6.55893922e-01 -7.97862634e-02
2.48183280e-01 -3.16122591e-01 -2.86820024e-01 9.66452301e-01
7.42390633e-01 3.30650836e-01 9.01350141e-01 -1.44511700e+00
-1.00228035e+00 5.75176716e-01 -4.10268873e-01 2.85047680e-01
3.18923481e-02 2.08567202e-01 8.00800145e-01 -1.12570655e+00
2.47047022e-01 1.38285899e+00 3.81901920e-01 8.54690611e-01
-1.04726863e+00 -6.06002331e-01 6.51084960e-01 3.60081464e-01
-1.29276645e+00 -7.68031359e-01 4.33131576e-01 -2.48868227e-01
1.29805243e+00 4.76107933e-02 -9.47334468e-02 1.62636244e+00
3.71797591e-01 9.87898111e-01 8.85182858e-01 -2.78062165e-01
2.24134818e-01 2.40019843e-01 3.35981734e-02 4.08478081e-01
5.53128123e-02 -1.68229669e-01 -9.54523027e-01 2.40954161e-01
5.86787283e-01 2.23554090e-01 -2.28388637e-01 3.44809666e-02
-1.37504208e+00 4.67430234e-01 2.63468236e-01 2.01320097e-01
-2.93515116e-01 -3.45705301e-02 4.22581732e-01 2.56084174e-01
5.08465350e-01 4.04610962e-01 -6.94131434e-01 8.82570371e-02
-8.64790738e-01 -2.79863417e-01 7.64958262e-01 1.25087583e+00
5.59062898e-01 -3.79090868e-02 -6.08529389e-01 1.05215955e+00
4.10961837e-01 4.75376308e-01 6.63344562e-01 -1.01746619e+00
6.46501839e-01 3.99055988e-01 -5.57876416e-02 -6.45781577e-01
-4.62038517e-01 -6.01407409e-01 -9.09232736e-01 -1.74626902e-01
1.43927246e-01 -2.49054253e-01 -1.21618557e+00 1.91449630e+00
-3.68864015e-02 7.23735709e-03 3.62167507e-01 1.02630460e+00
9.60067511e-01 8.47722232e-01 4.85943943e-01 -7.35054016e-02
1.61156309e+00 -1.40428329e+00 -9.01020885e-01 -8.19399536e-01
3.93889606e-01 -6.65846169e-01 1.10463750e+00 4.09168869e-01
-1.12646401e+00 -9.55642939e-01 -1.06213057e+00 -2.19932750e-01
-5.50272584e-01 3.80059421e-01 6.12774312e-01 4.42540377e-01
-1.25164986e+00 2.84270376e-01 -7.28546023e-01 -5.10024190e-01
2.41053462e-01 2.77065396e-01 -2.84360528e-01 -1.50284693e-01
-1.31600213e+00 1.01640713e+00 6.13610446e-01 1.87941447e-01
-1.14617741e+00 -4.84911680e-01 -6.63917243e-01 4.45240945e-01
5.65468252e-01 -1.07735300e+00 1.33871114e+00 -9.91040885e-01
-1.64456117e+00 6.79317415e-01 -3.15918475e-01 -1.62038118e-01
1.48475438e-01 -3.87437314e-01 -5.57533860e-01 1.96247518e-01
-1.00330196e-01 1.17850888e+00 1.03234184e+00 -1.25825858e+00
-6.11043036e-01 -2.65082181e-01 3.62147838e-01 5.01999021e-01
-7.72492945e-01 6.22281479e-03 -8.50590169e-01 -5.62511444e-01
-5.64649999e-02 -7.96008110e-01 1.76878363e-01 3.85533161e-02
-2.11473092e-01 -4.19251531e-01 6.96224630e-01 -4.55970585e-01
1.09022248e+00 -2.33234143e+00 3.79248947e-01 -4.59022224e-01
1.51258498e-01 3.99175435e-01 -5.41624904e-01 3.04580420e-01
1.29529566e-01 9.56605524e-02 2.80372258e-02 -7.10239470e-01
6.38755783e-02 3.09176534e-01 -2.64431369e-02 5.59306610e-03
2.75651097e-01 9.37319219e-01 -7.34489739e-01 -4.05130327e-01
3.71040180e-02 3.76918018e-01 -4.98576194e-01 3.53654951e-01
-3.49492431e-01 2.06439316e-01 -3.11548322e-01 6.27248824e-01
6.20979130e-01 -6.08601332e-01 1.30256847e-01 -4.42290813e-01
2.33295530e-01 1.29618451e-01 -9.42422807e-01 2.04920244e+00
-5.88821173e-01 4.71857011e-01 3.72957319e-01 -1.18954504e+00
7.10168958e-01 5.33160746e-01 -2.31046099e-02 -7.63910353e-01
8.56107697e-02 4.21788990e-02 1.78235471e-01 -6.75854206e-01
2.96119332e-01 -4.95467745e-02 -1.52774912e-03 2.22570881e-01
5.81066906e-01 1.85121730e-01 1.73098311e-01 2.99060255e-01
1.01095915e+00 2.51539171e-01 1.03511535e-01 -2.01833531e-01
3.97823155e-01 -2.39963487e-01 5.88577509e-01 1.07711172e+00
-4.22273546e-01 4.08742934e-01 7.93150067e-02 -1.47323340e-01
-7.06614912e-01 -9.50161338e-01 4.17645928e-03 1.94607067e+00
3.45012069e-01 -3.30924511e-01 -7.09981859e-01 -4.94565606e-01
-1.83630392e-01 7.26612508e-01 -5.19059598e-01 -5.30405939e-01
6.80481270e-02 -9.40517128e-01 6.15175486e-01 6.95855439e-01
7.74915278e-01 -1.05266964e+00 -4.80353445e-01 8.17648694e-02
-4.92712289e-01 -1.43323028e+00 -2.12182403e-01 3.07424843e-01
-7.99866498e-01 -5.76865196e-01 -8.04203272e-01 -6.75004244e-01
5.18456101e-01 6.25178456e-01 1.08218062e+00 -1.32725701e-01
1.27848715e-01 4.48277622e-01 -5.43966532e-01 -6.26976788e-01
-1.55843273e-01 3.02889079e-01 8.90450105e-02 7.78333470e-02
4.71789867e-01 -4.08433646e-01 -6.78269207e-01 6.57188535e-01
-1.06334853e+00 3.37073416e-01 9.51699317e-01 8.65954757e-01
2.06952080e-01 -7.31248707e-02 7.71036863e-01 -6.83645546e-01
6.39596879e-01 -6.48200750e-01 -1.74782515e-01 5.43210864e-01
-4.91554022e-01 9.71851051e-02 5.61243296e-01 -7.15920985e-01
-1.62787366e+00 -2.09267184e-01 6.17702827e-02 -4.84284043e-01
-3.03429067e-01 5.80186963e-01 -3.73174638e-01 2.37395838e-01
6.33187771e-01 2.07255378e-01 -1.96278825e-01 -5.27950883e-01
2.81101942e-01 7.80718565e-01 2.87220836e-01 -6.29392385e-01
4.17882383e-01 3.64289522e-01 -3.14001113e-01 -5.25383174e-01
-1.19019556e+00 -4.82709676e-01 -3.23554903e-01 -2.14340270e-01
1.05394757e+00 -1.38486183e+00 -6.44951642e-01 6.34032905e-01
-1.45837605e+00 -4.11942214e-01 2.24346668e-01 5.63497186e-01
-3.67001981e-01 1.40601024e-01 -8.45502734e-01 -7.17434347e-01
-2.05807954e-01 -1.25322938e+00 1.21398997e+00 4.56623256e-01
-2.80570909e-02 -9.47382212e-01 -4.16416049e-01 6.53573096e-01
6.75941706e-01 -6.44963920e-01 8.67329836e-01 -8.39929819e-01
-4.67023104e-01 1.18558109e-01 -4.77953434e-01 5.39378703e-01
-1.00756988e-01 -1.57808676e-01 -1.51484776e+00 -3.84313226e-01
9.34791341e-02 -5.30776858e-01 1.09332466e+00 3.80087346e-01
1.31105065e+00 4.63176928e-02 -5.38188040e-01 5.38105428e-01
1.09757400e+00 2.48190045e-01 5.96965969e-01 1.75437361e-01
4.74027097e-01 7.05826938e-01 4.66401428e-01 3.05935949e-01
5.12697101e-01 5.42634666e-01 4.56807882e-01 1.00788539e-02
-5.16700037e-02 5.94075583e-02 6.39973283e-01 8.93502355e-01
-1.85379460e-01 -5.83491862e-01 -7.79186070e-01 2.95258820e-01
-2.03565240e+00 -8.33151281e-01 1.77472562e-01 1.85265708e+00
6.72628760e-01 2.56796151e-01 -3.33478123e-01 -2.65717357e-01
7.37424433e-01 8.86186510e-02 -7.84809530e-01 -2.80672371e-01
-1.41915187e-01 -2.84581155e-01 1.90431744e-01 1.21282898e-01
-1.08353186e+00 9.61617529e-01 6.49673557e+00 1.06120884e+00
-8.79358530e-01 5.40376306e-01 5.00101745e-01 -1.40777901e-01
-9.01573449e-02 -2.05327392e-01 -1.03031063e+00 4.62650985e-01
1.11092818e+00 2.30760500e-02 4.81335759e-01 7.93311119e-01
9.15059298e-02 -1.55303389e-01 -1.28704309e+00 9.87422645e-01
3.50906134e-01 -8.93831909e-01 1.39037058e-01 -1.91697925e-01
6.30723894e-01 2.05503806e-01 1.41426414e-01 9.04320896e-01
1.30505577e-01 -8.72685552e-01 6.87297702e-01 7.64477968e-01
5.98930717e-01 -2.51652867e-01 7.73456335e-01 7.32877493e-01
-8.98098171e-01 -3.25334996e-01 -6.81636333e-01 -1.28081366e-01
1.21325344e-01 5.81009924e-01 -4.39318180e-01 6.86818659e-01
8.01979184e-01 4.29499716e-01 -6.42277181e-01 9.19225574e-01
-2.13274151e-01 4.59222883e-01 -1.82665303e-01 7.56014362e-02
8.16738382e-02 2.17434585e-01 4.40160513e-01 1.34316409e+00
3.90648603e-01 -1.50744040e-02 8.10723901e-02 7.66553938e-01
-1.89697683e-01 -1.97831392e-01 -4.41881597e-01 -3.45205739e-02
5.69497585e-01 1.30789208e+00 -5.30881703e-01 -5.63110709e-01
-6.93188667e-01 1.28551066e+00 5.08194923e-01 8.41915548e-01
-8.92487288e-01 -1.76760331e-01 4.85253394e-01 -2.16253638e-01
2.18419492e-01 -1.26257047e-01 -1.04126282e-01 -1.28676128e+00
7.33955391e-03 -8.56865525e-01 4.15857852e-01 -1.18601966e+00
-1.49619401e+00 6.41901970e-01 1.47908211e-01 -9.74891782e-01
-9.14766639e-02 -9.33684707e-01 -3.82205158e-01 8.12670112e-01
-1.62172425e+00 -1.00667906e+00 -3.42517287e-01 6.71548426e-01
9.51529264e-01 -4.30316389e-01 8.63370717e-01 3.94373536e-01
-8.73540521e-01 5.63841105e-01 2.28839973e-03 -3.12736854e-02
1.23510206e+00 -8.46411109e-01 4.68706936e-02 7.86715984e-01
-5.78031056e-02 7.18329370e-01 5.63702047e-01 -3.74591947e-01
-1.34500825e+00 -9.50346887e-01 6.23027623e-01 -4.94314402e-01
4.11445916e-01 -5.38679957e-01 -9.08676565e-01 7.78104007e-01
4.88714486e-01 -4.16758247e-02 7.46039212e-01 3.37088168e-01
-5.53140104e-01 -1.87114149e-01 -8.69285822e-01 5.83950281e-01
1.27751780e+00 -6.64346576e-01 -7.15549529e-01 4.17800814e-01
9.92609560e-01 -1.51308686e-01 -7.07931340e-01 5.56856513e-01
4.46753800e-01 -1.02029836e+00 9.00884748e-01 -6.04716420e-01
5.46893060e-01 -1.90171167e-01 -1.91568702e-01 -1.46927714e+00
-6.08566046e-01 -2.86508709e-01 -3.51528972e-01 1.14973974e+00
6.10405207e-01 -6.39353216e-01 3.64759803e-01 6.54282510e-01
-3.43499243e-01 -4.44109857e-01 -7.59904385e-01 -9.79935050e-01
-2.09498644e-01 -3.61831069e-01 1.85610533e-01 6.99554324e-01
4.17930484e-02 9.02992010e-01 -4.78634477e-01 1.98069021e-01
3.83019179e-01 -3.43142152e-01 5.05688906e-01 -1.15933073e+00
-4.75281447e-01 -3.14058572e-01 -1.68828685e-02 -1.43580699e+00
2.49356568e-01 -8.03225338e-01 2.26660639e-01 -1.64597833e+00
4.90295798e-01 -1.10524781e-01 -6.17918074e-01 5.92114985e-01
-5.30087590e-01 -3.33773904e-02 3.81389648e-01 4.12464023e-01
-1.13492584e+00 5.98934710e-01 1.31057775e+00 -2.24794239e-01
-1.56984199e-02 -7.44213387e-02 -8.28571141e-01 7.20310092e-01
7.00981975e-01 -3.51846665e-01 -7.99226999e-01 -1.06980443e+00
3.10741693e-01 -7.49045908e-02 2.43754566e-01 -1.14283383e+00
5.37902832e-01 -9.10820961e-02 4.69555527e-01 -3.01566362e-01
6.90711379e-01 -9.05975103e-01 8.32988974e-03 1.20780163e-01
-4.44348395e-01 -5.60417250e-02 3.87218028e-01 7.25767493e-01
-3.63826782e-01 -3.97561938e-01 5.39490640e-01 -1.83908835e-01
-1.05530596e+00 1.06066495e-01 -5.96057594e-01 -1.88646168e-01
9.13760066e-01 -1.11061580e-01 -7.84379900e-01 -4.07176167e-01
-8.38621318e-01 3.71951848e-01 -2.45660767e-02 6.57899678e-01
6.14065886e-01 -9.91475344e-01 -7.20251977e-01 -4.59261015e-02
3.82305324e-01 -9.61480364e-02 6.73377454e-01 9.25112665e-01
1.31986991e-01 6.41160369e-01 -2.23085612e-01 -6.10656619e-01
-8.17763269e-01 8.14596415e-01 1.86612174e-01 -2.11552322e-01
-1.12306319e-01 1.04507160e+00 7.45104551e-01 -3.54908884e-01
3.99407089e-01 -1.55203223e-01 -2.83235818e-01 1.67637765e-01
6.09641790e-01 1.07595772e-01 1.19168051e-01 -1.76902920e-01
-3.96492004e-01 1.36777163e-01 -2.82292157e-01 -1.54575640e-02
9.70048308e-01 -4.63369071e-01 1.17957853e-01 5.58202028e-01
9.92328644e-01 -5.77867270e-01 -1.45065904e+00 -4.17611092e-01
-2.23419815e-01 3.33506754e-03 1.11579455e-01 -1.11658037e+00
-9.91807520e-01 1.17998111e+00 3.74277830e-01 8.01894441e-02
1.28887141e+00 5.73235527e-02 5.39455533e-01 7.45662212e-01
3.10640097e-01 -1.41201234e+00 2.00051248e-01 7.93390632e-01
7.38590121e-01 -1.52638721e+00 -2.50457525e-01 -4.86804783e-01
-9.39096749e-01 8.11394513e-01 9.92828369e-01 2.45407864e-01
5.08658350e-01 2.98621058e-01 -1.62769370e-02 -3.11020333e-02
-1.35193396e+00 -2.95777231e-01 4.16258395e-01 7.18377650e-01
4.53709185e-01 -9.35018882e-02 -2.39757091e-01 1.04323637e+00
4.70546961e-01 1.62229165e-01 5.61844222e-02 8.13786089e-01
-3.95866364e-01 -7.99232006e-01 -2.37131864e-01 3.88927639e-01
-2.80444413e-01 -3.85262549e-01 -2.76540965e-02 5.38726509e-01
1.50724724e-01 1.23026037e+00 1.20838448e-01 -3.97529632e-01
3.59766632e-01 4.60776478e-01 3.24027151e-01 -7.42120445e-01
-4.42481965e-01 7.89299160e-02 1.79388359e-01 -4.61861581e-01
-7.05103159e-01 -9.57196057e-02 -8.12546074e-01 -7.83931985e-02
-3.00084293e-01 -2.01426968e-02 7.00990617e-01 1.14728534e+00
8.11350226e-01 8.51559103e-01 1.13828257e-01 -9.90584850e-01
-6.99532628e-01 -1.37634075e+00 -3.84203851e-01 3.83416563e-01
-4.92956042e-02 -8.71699333e-01 -4.33017492e-01 3.21537763e-01] | [10.276668548583984, 1.6039178371429443] |
222c8bef-6f7b-4801-9d5b-c0d7cd1bae57 | few-shot-fine-grained-action-recognition-via | 2108.06647 | null | https://arxiv.org/abs/2108.06647v1 | https://arxiv.org/pdf/2108.06647v1.pdf | Few-Shot Fine-Grained Action Recognition via Bidirectional Attention and Contrastive Meta-Learning | Fine-grained action recognition is attracting increasing attention due to the emerging demand of specific action understanding in real-world applications, whereas the data of rare fine-grained categories is very limited. Therefore, we propose the few-shot fine-grained action recognition problem, aiming to recognize novel fine-grained actions with only few samples given for each class. Although progress has been made in coarse-grained actions, existing few-shot recognition methods encounter two issues handling fine-grained actions: the inability to capture subtle action details and the inadequacy in learning from data with low inter-class variance. To tackle the first issue, a human vision inspired bidirectional attention module (BAM) is proposed. Combining top-down task-driven signals with bottom-up salient stimuli, BAM captures subtle action details by accurately highlighting informative spatio-temporal regions. To address the second issue, we introduce contrastive meta-learning (CML). Compared with the widely adopted ProtoNet-based method, CML generates more discriminative video representations for low inter-class variance data, since it makes full use of potential contrastive pairs in each training episode. Furthermore, to fairly compare different models, we establish specific benchmark protocols on two large-scale fine-grained action recognition datasets. Extensive experiments show that our method consistently achieves state-of-the-art performance across evaluated tasks. | ['Annan Li', 'Sheng Liu', 'Yunhong Wang', 'Jiahao Wang'] | 2021-08-15 | null | null | null | null | ['action-understanding', 'fine-grained-action-recognition'] | ['computer-vision', 'computer-vision'] | [ 5.40877104e-01 -4.35831696e-01 -4.65312302e-01 -1.99525312e-01
-8.40827823e-01 -2.36401856e-02 9.01795566e-01 -1.90800205e-01
-3.26893032e-01 7.87010849e-01 8.12362373e-01 4.76051927e-01
-2.48511001e-01 -5.55781245e-01 -6.00915372e-01 -7.97183633e-01
1.28198683e-01 1.01349786e-01 6.48333371e-01 -1.10770576e-01
4.22951669e-01 4.03616428e-01 -2.00441456e+00 9.42801416e-01
7.73833692e-01 1.18960106e+00 2.34091908e-01 4.32959735e-01
-7.26144090e-02 1.11380029e+00 -5.80115914e-01 2.82743536e-02
2.10244194e-01 -7.95609415e-01 -7.36941695e-01 3.60704511e-01
6.46349907e-01 -4.94595855e-01 -4.44316924e-01 8.51741970e-01
4.39967752e-01 4.90341395e-01 6.42749310e-01 -1.31997657e+00
-6.72395885e-01 3.31065714e-01 -6.45312548e-01 7.97428191e-01
3.79435897e-01 5.22964597e-01 9.46124673e-01 -8.98525596e-01
5.15016198e-01 1.40316534e+00 4.79213119e-01 6.71776891e-01
-9.37442422e-01 -5.47661185e-01 5.61649501e-01 7.55396903e-01
-1.11015856e+00 -5.13415813e-01 7.05448925e-01 -4.41676736e-01
1.22585285e+00 -3.33762318e-02 7.58398414e-01 1.56561089e+00
3.81162286e-01 9.25086498e-01 1.16219866e+00 -1.70091480e-01
4.61538553e-01 -5.39771676e-01 -7.46490806e-02 4.08835262e-01
5.36531396e-03 2.73168385e-01 -7.85446405e-01 1.77875146e-01
7.95507014e-01 4.98427570e-01 -1.23266056e-01 -4.37865973e-01
-1.38822150e+00 5.85191250e-01 2.32408658e-01 6.55300438e-01
-7.56158650e-01 2.37795755e-01 6.23104930e-01 3.37386057e-02
4.17858332e-01 2.41928160e-01 -3.04132938e-01 -8.08377922e-01
-9.06702459e-01 1.20679140e-01 2.51415581e-01 6.00458682e-01
5.64779103e-01 2.39740461e-01 -9.14687932e-01 7.83658147e-01
-1.29200846e-01 2.23792583e-01 9.40381050e-01 -8.25081229e-01
3.98325831e-01 7.32285857e-01 -2.42961422e-02 -8.92354131e-01
-1.65069342e-01 -1.63365737e-01 -9.00359571e-01 1.78158730e-01
3.47738564e-01 3.14872712e-01 -1.00099874e+00 1.40837991e+00
2.01006114e-01 5.06425440e-01 -2.13097841e-01 9.38099623e-01
6.11579597e-01 5.28859437e-01 4.93119955e-01 -3.01991343e-01
1.30375981e+00 -1.30565917e+00 -5.78683734e-01 -2.62657344e-01
4.04625714e-01 -2.61005521e-01 1.25527704e+00 2.28293434e-01
-8.55386138e-01 -9.70695615e-01 -8.55975807e-01 1.61401093e-01
-4.47844505e-01 -1.46505922e-01 5.27280152e-01 2.85856277e-01
-6.68035269e-01 5.75596452e-01 -8.01587224e-01 -4.62206274e-01
8.92339051e-01 -8.55594352e-02 -5.27772963e-01 -3.54653388e-01
-1.06080246e+00 7.42172122e-01 4.62960005e-01 -7.38753229e-02
-1.18933117e+00 -6.83234274e-01 -7.96107411e-01 1.66737467e-01
5.50700545e-01 -6.19051814e-01 1.06314754e+00 -1.02899539e+00
-1.40391195e+00 5.78711212e-01 -7.55525902e-02 -5.02902806e-01
3.80636007e-01 -2.79519111e-01 -5.31353712e-01 3.65355462e-01
2.70921558e-01 6.16922021e-01 1.22138751e+00 -6.95966005e-01
-9.60552275e-01 -3.62525642e-01 2.49029189e-01 1.58512995e-01
-3.01702440e-01 -7.75702372e-02 -2.43813634e-01 -1.05869663e+00
-2.17793435e-01 -5.99520147e-01 -1.10905081e-01 -4.05522101e-02
-1.53937424e-02 -3.88410985e-01 9.13513064e-01 -2.09214032e-01
1.30278432e+00 -2.26215744e+00 1.94320932e-01 -5.45907617e-01
5.17872982e-02 4.90032375e-01 -4.64715272e-01 4.59722906e-01
-1.04591012e-01 -2.83024698e-01 -4.30500023e-02 -1.97458323e-02
1.15925379e-01 2.93677390e-01 -4.02794898e-01 2.83580393e-01
4.57573324e-01 1.21220255e+00 -1.10176861e+00 -4.77216750e-01
6.29591703e-01 4.25690532e-01 -4.54112649e-01 1.86444268e-01
-1.89493179e-01 4.72173780e-01 -7.12430537e-01 8.08929861e-01
2.77649105e-01 -2.59693027e-01 -1.55322999e-01 -4.62598801e-01
-3.15339677e-02 -2.04532906e-01 -1.01561499e+00 1.79246438e+00
-3.02204132e-01 2.56980360e-01 -5.04494727e-01 -1.25821769e+00
6.65937424e-01 1.42606869e-01 6.88223124e-01 -1.33043838e+00
1.36009613e-02 5.97290345e-04 -1.71744674e-01 -6.76324069e-01
3.25878173e-01 -4.28388745e-01 -1.95667520e-01 4.16292161e-01
2.61001766e-01 2.15660870e-01 2.95516014e-01 3.86987478e-02
1.39895535e+00 4.42889601e-01 7.38353014e-01 1.40813380e-01
5.15498161e-01 -1.26062766e-01 6.81526721e-01 7.99908817e-01
-8.44988048e-01 5.53822994e-01 8.33670050e-02 -6.56059682e-01
-5.92349172e-01 -9.29400265e-01 2.44388968e-01 1.58637071e+00
2.74403840e-01 -5.01059830e-01 -6.41229093e-01 -9.23534751e-01
-1.11416951e-01 5.41200340e-01 -1.08363390e+00 -4.85962808e-01
-5.43383300e-01 -4.23642874e-01 4.38379109e-01 8.88999939e-01
8.67987156e-01 -1.66896570e+00 -1.08280289e+00 3.22779655e-01
-3.00386369e-01 -1.06511855e+00 -4.60086256e-01 2.05602366e-02
-8.44268441e-01 -1.20376873e+00 -9.83506918e-01 -4.12231863e-01
2.19897360e-01 6.48837328e-01 1.09674335e+00 -2.90799409e-01
-6.16932631e-01 6.89425290e-01 -8.35138559e-01 -6.12130575e-02
7.10642859e-02 -3.83476436e-01 3.36270668e-02 3.18287909e-01
7.77431190e-01 -5.30989707e-01 -7.12278366e-01 4.28390056e-01
-9.36088443e-01 -6.24123998e-02 9.41784859e-01 1.04002953e+00
8.82669508e-01 7.11242482e-02 6.84115648e-01 -4.66149271e-01
3.96383464e-01 -3.81964862e-01 -1.74888922e-03 2.82288253e-01
-1.49717852e-01 1.97028946e-02 6.02335215e-01 -4.94770765e-01
-1.11726713e+00 -7.30740651e-02 1.10701762e-01 -6.89808846e-01
-6.85000241e-01 1.17546760e-01 -1.92101404e-01 1.19493514e-01
5.01622319e-01 6.21099591e-01 -3.44130635e-01 -4.80840385e-01
4.12234306e-01 4.00585681e-01 4.32915241e-01 -4.81869400e-01
3.44051868e-01 6.13793969e-01 -1.78772673e-01 -7.18062997e-01
-1.12940753e+00 -5.50477684e-01 -8.42237175e-01 -4.31899965e-01
1.11170542e+00 -8.86361897e-01 -4.40394133e-01 6.75080121e-01
-6.37998939e-01 -3.60758483e-01 -7.29123712e-01 3.95486891e-01
-1.04735780e+00 3.79511148e-01 -2.87674844e-01 -6.07587755e-01
-8.81548747e-02 -1.07863951e+00 1.53786659e+00 8.46318230e-02
-2.84478843e-01 -6.72949314e-01 2.72424310e-01 5.04250526e-01
3.26389402e-01 3.47184896e-01 5.53043664e-01 -4.08135355e-01
-5.51107883e-01 -6.82616532e-02 -2.93811142e-01 2.56604016e-01
4.24156874e-01 -3.45944285e-01 -8.91416490e-01 -2.13258654e-01
-4.81582247e-02 -6.96739078e-01 1.11041200e+00 5.11863291e-01
1.50726366e+00 -1.54539868e-01 -3.31301302e-01 2.31842905e-01
1.06521809e+00 2.32371613e-01 7.98534214e-01 3.75188649e-01
7.72335291e-01 4.04580057e-01 1.19522512e+00 8.45602214e-01
1.58985913e-01 9.57210600e-01 3.74730498e-01 2.95043588e-01
-4.43049341e-01 -3.26269388e-01 4.13361311e-01 2.50595301e-01
-3.91358733e-01 -4.36446816e-03 -4.90976304e-01 5.57856858e-01
-2.16602087e+00 -1.68648005e+00 4.38418210e-01 1.84680521e+00
5.30354679e-01 1.91740006e-01 3.56311470e-01 1.33021668e-01
5.19808948e-01 7.04911649e-01 -7.93989778e-01 -1.41306028e-01
-7.25202709e-02 1.96682274e-01 -6.58112019e-02 -1.44128993e-01
-1.34794068e+00 9.85351324e-01 5.38083887e+00 1.26429975e+00
-1.06196654e+00 2.21949816e-01 4.61435169e-01 -3.50953877e-01
9.29345414e-02 -4.31054980e-01 -6.55222774e-01 5.96206069e-01
7.24974990e-01 1.53799178e-02 1.32256880e-01 9.15697396e-01
4.14329797e-01 -1.11651689e-01 -1.18539178e+00 1.20972240e+00
2.99562633e-01 -1.43823481e+00 3.19192588e-01 -1.04775093e-01
8.36090386e-01 -6.94922134e-02 -1.05539158e-01 7.29100168e-01
1.91129115e-03 -9.40155566e-01 6.79214358e-01 7.45474517e-01
9.35321867e-01 -5.68204880e-01 3.58148396e-01 3.64249498e-01
-1.51292861e+00 -5.20642102e-01 -3.92535597e-01 -3.48830909e-01
2.06325114e-01 2.78572589e-01 -1.69357121e-01 4.29301262e-01
8.76002908e-01 1.34050155e+00 -5.90613008e-01 9.53963220e-01
6.43748567e-02 3.52758259e-01 2.43277341e-01 1.09475620e-01
6.90203309e-01 1.86764240e-01 2.80525088e-01 1.13634348e+00
1.87219098e-01 3.85037482e-01 3.42290401e-01 3.48345101e-01
2.48098046e-01 -2.08381683e-01 -5.96345007e-01 -2.38576606e-01
1.26786232e-01 9.88229036e-01 -6.62726104e-01 -5.16501069e-01
-5.35059392e-01 1.27974486e+00 2.94103533e-01 2.68718719e-01
-8.24489236e-01 -1.39581218e-01 8.37853849e-01 -3.81327458e-02
8.10705841e-01 -5.85934520e-03 8.52752179e-02 -1.28766537e+00
-7.53533989e-02 -1.20120394e+00 7.41678596e-01 -7.03262150e-01
-1.35666966e+00 4.67961133e-01 3.27781364e-02 -1.66584229e+00
-4.08436447e-01 -4.34888184e-01 -5.77992857e-01 4.09785420e-01
-1.38218665e+00 -1.23639691e+00 -6.24550998e-01 8.76200140e-01
1.37849545e+00 -1.42337903e-01 8.66823554e-01 2.88619548e-01
-4.06223506e-01 5.04811823e-01 -1.33335263e-01 -1.33679286e-01
5.91452658e-01 -8.29647362e-01 2.69856334e-01 7.92783856e-01
1.70718506e-01 2.14468792e-01 4.52142566e-01 -7.07420111e-01
-1.31203604e+00 -1.28196526e+00 5.66078186e-01 -5.95594466e-01
6.28074050e-01 -1.01559028e-01 -8.88010800e-01 4.05368537e-01
-1.26524493e-01 4.74182785e-01 5.89231491e-01 -1.24534540e-01
-5.36816478e-01 -1.27748504e-01 -9.64506924e-01 5.25999904e-01
1.47804451e+00 -5.30727506e-01 -9.31377470e-01 1.38626814e-01
4.27401662e-01 1.11790039e-01 -7.76323915e-01 5.35919011e-01
7.44737208e-01 -1.27028823e+00 1.07332397e+00 -8.86957586e-01
5.36772788e-01 -3.47451091e-01 -4.38194662e-01 -1.35565555e+00
-8.43179464e-01 -5.20319082e-02 -5.71595728e-01 9.26261365e-01
-3.34074378e-01 -3.14347148e-01 6.87656879e-01 1.61312163e-01
-2.48226970e-01 -9.06356454e-01 -9.95174706e-01 -9.39665198e-01
-1.78917870e-01 -3.69493335e-01 4.04217035e-01 6.42641366e-01
6.08589267e-03 1.86633229e-01 -7.27225602e-01 -3.16173673e-01
4.95534480e-01 5.12285531e-01 7.43083894e-01 -1.05911052e+00
-4.45711881e-01 -5.49879730e-01 -7.88867891e-01 -1.00734317e+00
2.12611675e-01 -2.27670684e-01 2.36565024e-01 -1.56310701e+00
5.33454657e-01 1.53354973e-01 -7.19623685e-01 5.69560707e-01
-2.69678742e-01 6.79444075e-01 3.42790693e-01 1.46900892e-01
-1.19924164e+00 9.68977332e-01 1.43442619e+00 -3.72563779e-01
2.71359552e-02 -1.97175443e-01 -5.56264102e-01 7.09697008e-01
4.13948029e-01 -1.92778915e-01 -5.52323818e-01 -7.04654604e-02
-4.10043359e-01 -8.89490098e-02 5.20309865e-01 -1.39115214e+00
-4.18812521e-02 -5.29748857e-01 5.35314560e-01 -6.44625366e-01
4.89326268e-01 -7.26018190e-01 -2.95802653e-02 4.47229862e-01
-3.46043259e-01 -4.55048054e-01 1.09715156e-01 9.32196677e-01
-5.03202260e-01 2.25783527e-01 8.23621094e-01 -3.13906282e-01
-1.68921936e+00 6.45274639e-01 -3.57723027e-01 2.97510386e-01
1.30263388e+00 -7.32825220e-01 -3.40565294e-01 -1.08393013e-01
-6.78849041e-01 -9.02341455e-02 3.06925565e-01 8.95465791e-01
7.65786409e-01 -1.62754142e+00 -5.87455988e-01 2.66759187e-01
6.82923913e-01 -5.58128357e-01 1.01270306e+00 9.92990911e-01
2.38554418e-01 5.46731889e-01 -7.66049802e-01 -5.30737102e-01
-1.09305286e+00 7.99573362e-01 1.49610952e-01 -2.23220110e-01
-9.30728197e-01 7.54251122e-01 7.01797605e-01 2.00582981e-01
1.84594870e-01 -3.51215392e-01 -4.38565344e-01 3.08022112e-01
1.04758716e+00 5.77169478e-01 -2.30291620e-01 -8.96618962e-01
-5.55432737e-01 6.97692633e-01 -2.57403613e-03 3.84048253e-01
1.45607412e+00 -1.56984940e-01 4.61127937e-01 6.80760264e-01
1.01940215e+00 -5.60012639e-01 -1.88463712e+00 -2.29612291e-01
-1.99055195e-01 -8.86999667e-01 -1.36133671e-01 -8.35469007e-01
-8.23717177e-01 1.05244160e+00 7.49270260e-01 3.78465489e-03
1.39513493e+00 5.60504943e-02 8.30000699e-01 2.05761924e-01
6.59889638e-01 -1.26501548e+00 7.16203213e-01 5.94843507e-01
9.20810163e-01 -1.38606513e+00 -5.66705577e-02 -6.49442896e-03
-9.13102329e-01 8.75828564e-01 8.03595960e-01 -1.14731960e-01
4.22889054e-01 -1.94941938e-01 -1.88728228e-01 -2.33639315e-01
-8.17026913e-01 -4.15695727e-01 4.85555112e-01 8.49032223e-01
1.28226623e-01 -5.27869314e-02 -1.22247390e-01 6.88946128e-01
4.74165320e-01 4.29117233e-01 1.76130667e-01 9.96601582e-01
-6.95070148e-01 -6.85631454e-01 -3.86687070e-02 6.32770121e-01
-3.31939965e-01 6.42881244e-02 -2.72732347e-01 6.30480349e-01
2.53088444e-01 7.80526459e-01 1.95155323e-01 -3.57728153e-01
4.66941178e-01 2.75642946e-02 6.60873055e-01 -6.73135281e-01
-2.40970612e-01 -6.82576671e-02 -7.73837930e-03 -1.37605047e+00
-7.96491563e-01 -7.18154788e-01 -9.85347450e-01 4.77392673e-02
2.11057559e-01 -2.10992917e-01 -1.54925892e-02 1.31152439e+00
5.93225777e-01 8.71787131e-01 4.21960920e-01 -1.33708417e+00
-5.60827374e-01 -9.92785275e-01 -7.81493485e-01 6.32144570e-01
2.77620912e-01 -1.26941407e+00 -2.48478457e-01 5.05096130e-02] | [8.428170204162598, 0.7417348623275757] |
0df1af25-207d-4f3d-b0e4-053f53aa49c3 | cross-language-speech-emotion-recognition | 2306.13804 | null | https://arxiv.org/abs/2306.13804v2 | https://arxiv.org/pdf/2306.13804v2.pdf | Cross-Language Speech Emotion Recognition Using Multimodal Dual Attention Transformers | Despite the recent progress in speech emotion recognition (SER), state-of-the-art systems are unable to achieve improved performance in cross-language settings. In this paper, we propose a Multimodal Dual Attention Transformer (MDAT) model to improve cross-language SER. Our model utilises pre-trained models for multimodal feature extraction and is equipped with a dual attention mechanism including graph attention and co-attention to capture complex dependencies across different modalities and achieve improved cross-language SER results using minimal target language data. In addition, our model also exploits a transformer encoder layer for high-level feature representation to improve emotion classification accuracy. In this way, MDAT performs refinement of feature representation at various stages and provides emotional salient features to the classification layer. This novel approach also ensures the preservation of modality-specific emotional information while enhancing cross-modality and cross-language interactions. We assess our model's performance on four publicly available SER datasets and establish its superior effectiveness compared to recent approaches and baseline models. | ['Junaid Qadir', 'Siddique Latif', 'Syed Aun Muhammad Zaidi'] | 2023-06-23 | null | null | null | null | ['emotion-recognition', 'emotion-classification', 'graph-attention', 'emotion-classification', 'speech-emotion-recognition'] | ['computer-vision', 'computer-vision', 'graphs', 'natural-language-processing', 'speech'] | [ 6.58732802e-02 4.21656445e-02 5.15141338e-02 -5.42663515e-01
-9.65229452e-01 -3.65893006e-01 6.02971733e-01 3.84979397e-01
-5.33381104e-01 4.59521353e-01 4.96733189e-01 1.55486371e-02
2.37158060e-01 -1.97753549e-01 -4.74760920e-01 -2.91531593e-01
-8.45284984e-02 1.34665132e-01 -2.27634564e-01 -4.51475829e-01
-2.34794661e-01 3.07656139e-01 -1.29371130e+00 7.36272156e-01
7.95998216e-01 1.17788863e+00 -1.55228510e-01 9.17299330e-01
-1.65369526e-01 9.62280750e-01 -4.09969151e-01 -9.07478392e-01
-3.66643131e-01 -3.61758858e-01 -9.36129272e-01 -4.85959053e-02
1.26767009e-01 9.00340900e-02 -2.62985796e-01 6.99074328e-01
7.02389479e-01 1.70321360e-01 5.00528038e-01 -1.50371087e+00
-9.04068470e-01 5.30422151e-01 -4.74565715e-01 -2.13088080e-01
6.60541177e-01 -1.75629944e-01 1.32336545e+00 -1.03732240e+00
4.02829915e-01 1.20722806e+00 7.09428489e-01 8.53496134e-01
-1.30315721e+00 -4.43328589e-01 3.82239550e-01 1.54609978e-01
-1.17564976e+00 -6.46775723e-01 1.03562927e+00 7.18472227e-02
1.47857046e+00 2.87124485e-01 5.68456829e-01 1.66851592e+00
9.36299339e-02 1.18535924e+00 1.11239779e+00 -4.46025878e-01
-1.37199312e-01 4.21844333e-01 2.02115215e-02 8.46661985e-01
-7.42108047e-01 -2.21017554e-01 -1.00628161e+00 -1.18065543e-01
1.28503323e-01 -3.43494087e-01 -4.27852958e-01 -2.77291507e-01
-1.14988995e+00 7.01451957e-01 3.03817779e-01 6.29971862e-01
-3.91930997e-01 -1.27798885e-01 9.71380472e-01 5.36428154e-01
6.30064189e-01 3.04024458e-01 -6.58537388e-01 -4.01235968e-01
-6.42782390e-01 -3.51561517e-01 7.63005793e-01 7.33327568e-01
2.50214964e-01 4.27804291e-02 -3.29608113e-01 1.38341653e+00
3.84962469e-01 2.17743218e-01 4.49654341e-01 -3.70103240e-01
4.96221930e-01 7.89834201e-01 -5.19588172e-01 -8.90940428e-01
-4.52594787e-01 -5.04420042e-01 -9.32146907e-01 -1.22461855e-01
-2.76351899e-01 -1.38104439e-01 -6.65987968e-01 2.17700720e+00
9.82494131e-02 -5.21629602e-02 5.29522181e-01 7.02133954e-01
1.04298270e+00 6.67471468e-01 6.23752534e-01 -1.11322897e-02
1.45171309e+00 -1.09029460e+00 -9.17123497e-01 -1.82038784e-01
7.65316606e-01 -5.90298414e-01 1.19829869e+00 3.30436021e-01
-1.12941957e+00 -3.60765129e-01 -7.86344767e-01 -3.96975100e-01
-6.80852592e-01 3.19142044e-01 6.29561484e-01 6.27313375e-01
-1.23326695e+00 -1.48486663e-02 -5.06380200e-01 -4.21877027e-01
3.48680705e-01 5.50738394e-01 -9.00361896e-01 1.40415087e-01
-1.44588387e+00 1.08046365e+00 1.73857853e-01 9.70607251e-02
-5.51274955e-01 -6.51083946e-01 -1.24309564e+00 2.36556068e-01
6.63132146e-02 -7.82299995e-01 1.03643870e+00 -1.40426695e+00
-1.91579843e+00 8.74484837e-01 -2.72850782e-01 -2.25534379e-01
1.09499201e-01 -1.79236382e-01 -7.48098254e-01 2.73986548e-01
-3.40979069e-01 9.14179802e-01 7.81096637e-01 -1.25518250e+00
-4.55069214e-01 -2.26016670e-01 1.07095413e-01 4.12769228e-01
-8.28315496e-01 3.34783226e-01 -5.55719614e-01 -5.52821457e-01
-3.86164814e-01 -7.29587078e-01 7.04904273e-02 -2.79244125e-01
-2.32009277e-01 -3.65995318e-01 8.93851578e-01 -7.46177852e-01
1.17044747e+00 -2.34234977e+00 6.93590522e-01 2.04303786e-02
1.05758362e-01 1.62151620e-01 -5.43040454e-01 6.20917261e-01
-2.14026883e-01 6.63951039e-02 -1.70085490e-01 -1.20858574e+00
3.69009763e-01 2.25335568e-01 1.90060996e-02 1.08688854e-01
8.18925023e-01 1.19344056e+00 -6.11737788e-01 -4.83115166e-01
3.57184529e-01 1.13316190e+00 -5.93327940e-01 4.90468621e-01
9.71748086e-04 4.78476793e-01 -1.39439851e-01 6.27951264e-01
4.38388646e-01 -3.73123661e-02 2.60564357e-01 -4.49152946e-01
9.71045569e-02 1.17095247e-01 -5.51674664e-01 1.99898100e+00
-1.02419937e+00 5.72903872e-01 2.39631921e-01 -1.01577234e+00
7.81844616e-01 6.79681540e-01 4.71178234e-01 -8.91351521e-01
4.73660529e-01 -6.33016825e-02 -2.95824766e-01 -5.53751647e-01
6.73411012e-01 -3.61773431e-01 -5.27775884e-01 2.22288296e-01
5.52089572e-01 2.94054151e-01 -1.92461222e-01 4.54509914e-01
7.87304699e-01 1.06552720e-01 1.74144253e-01 3.72079536e-02
8.68160427e-01 -4.89503622e-01 2.07077071e-01 2.01432750e-01
-2.61177182e-01 2.72849053e-01 5.17091751e-01 1.32392734e-01
-5.21027088e-01 -7.29496479e-01 1.68353662e-01 1.60498726e+00
-1.48889884e-01 -6.73034191e-01 -5.79515636e-01 -9.64461386e-01
-3.40018243e-01 6.03852093e-01 -8.31465721e-01 -4.79921550e-01
-7.02840509e-03 -6.12251580e-01 8.47109497e-01 6.29297137e-01
3.36191148e-01 -9.87604797e-01 -3.52395356e-01 3.56615172e-03
-5.85222960e-01 -1.47212291e+00 -3.33814651e-01 3.88306051e-01
-2.60224193e-01 -6.00722194e-01 -5.44526637e-01 -7.29579806e-01
4.86614704e-01 -1.49943262e-01 1.23417926e+00 -1.69032827e-01
-4.35084384e-03 8.63894343e-01 -7.48545825e-01 -2.92542875e-01
-2.71995366e-01 2.42429301e-01 -1.78330794e-01 4.17561293e-01
3.09450269e-01 -3.44258249e-01 -1.91063151e-01 -1.53621897e-01
-8.68854761e-01 1.53769359e-01 5.48279226e-01 1.15900385e+00
8.68557468e-02 -2.60401726e-01 7.06975341e-01 -2.76223481e-01
8.85283351e-01 -5.61132312e-01 1.46629393e-01 4.30197448e-01
-1.16417274e-01 2.31375292e-01 5.71898937e-01 -4.23190355e-01
-1.29313958e+00 -1.29061773e-01 -3.85335296e-01 -4.36980665e-01
-2.68174350e-01 6.82589531e-01 -3.89802903e-01 -1.66623741e-01
-7.82185197e-02 8.13625902e-02 -1.35332361e-01 -3.10002029e-01
4.17176843e-01 8.68298352e-01 3.92429650e-01 -6.32387459e-01
1.28600135e-01 1.40914291e-01 -3.35455716e-01 -7.28056431e-01
-7.45229125e-01 -3.15849543e-01 -4.91285264e-01 -3.65990132e-01
1.12534654e+00 -1.01306868e+00 -1.02674866e+00 3.88599575e-01
-1.11891985e+00 -2.49945208e-01 3.99533063e-02 4.43729520e-01
-4.30402994e-01 3.17355305e-01 -8.15792143e-01 -1.16025460e+00
-5.31364918e-01 -1.07624292e+00 1.49730015e+00 7.33285025e-02
-2.86264658e-01 -1.37197185e+00 4.88509461e-02 4.78130668e-01
4.44030851e-01 1.85765345e-02 8.42756212e-01 -6.08913481e-01
1.09885700e-01 -1.82115752e-02 -2.72665203e-01 4.12614197e-01
-1.89376980e-01 -7.84763470e-02 -1.26156294e+00 -2.65171826e-01
-4.89217848e-01 -8.07800949e-01 8.56273174e-01 -6.89393878e-02
1.05213082e+00 9.91412848e-02 -1.01520635e-01 4.40408468e-01
1.05851531e+00 -6.63507730e-02 4.82809663e-01 1.34465724e-01
6.96355104e-01 6.78002477e-01 3.43040615e-01 4.09836471e-01
8.94164503e-01 7.22071350e-01 3.62560153e-01 -4.88184512e-01
-1.66249946e-02 -1.53997734e-01 7.21130073e-01 1.16236472e+00
1.64016113e-01 -5.32059252e-01 -6.59649670e-01 6.60649538e-01
-1.87566054e+00 -9.11845148e-01 1.62980095e-01 1.77360821e+00
7.23467052e-01 -3.62772733e-01 6.76607341e-02 8.60252753e-02
2.38301262e-01 1.09971717e-01 -8.93530548e-02 -9.61470068e-01
-4.56891328e-01 3.09305698e-01 -2.18654305e-01 5.59409738e-01
-1.20723403e+00 1.07492518e+00 6.05134869e+00 6.05108559e-01
-1.18948638e+00 2.38614455e-01 5.78640938e-01 -3.31180245e-02
-3.91642421e-01 -5.38698018e-01 -3.32995206e-01 4.40822057e-02
1.07349133e+00 1.84000716e-01 3.26670617e-01 4.05412495e-01
-1.78843662e-01 1.39598608e-01 -9.88596380e-01 1.11092520e+00
5.07849276e-01 -7.30949402e-01 -5.50092645e-02 -2.61034966e-01
2.73298204e-01 3.71673405e-02 1.92182496e-01 7.76360691e-01
1.37034073e-01 -1.05537987e+00 6.65281832e-01 4.72838968e-01
7.25241780e-01 -1.19438851e+00 8.84238660e-01 3.41840088e-02
-1.30874741e+00 -5.78057021e-02 2.45592400e-01 2.17920586e-01
2.58447498e-01 1.20120883e-01 -5.46095610e-01 1.08411276e+00
7.19829738e-01 8.46444011e-01 -5.27361214e-01 1.74202621e-01
-9.60142314e-02 2.46582076e-01 -1.52156636e-01 8.35574884e-03
3.41349006e-01 -8.25977791e-03 3.73433590e-01 1.83845007e+00
1.80545390e-01 -1.86881006e-01 9.75997224e-02 2.99566478e-01
-2.63018310e-01 5.05588591e-01 -5.36061704e-01 -3.80319208e-01
2.04021521e-02 1.51279211e+00 -8.51038024e-02 -2.64384657e-01
-7.80749619e-01 1.53096199e+00 8.97087157e-01 3.27503234e-01
-1.06775951e+00 -2.92070031e-01 8.28504443e-01 -7.52303481e-01
2.31700972e-01 -1.88734472e-01 7.18783215e-03 -1.52521753e+00
-1.15690723e-01 -9.94234204e-01 6.40947342e-01 -9.69991088e-01
-1.59879541e+00 1.08350098e+00 -4.25461799e-01 -8.05471599e-01
-5.56021258e-02 -7.06705391e-01 -2.85328031e-01 8.02325785e-01
-1.72050667e+00 -1.94528699e+00 -1.62910428e-02 9.43444252e-01
3.75779182e-01 -2.43509158e-01 1.25333786e+00 5.19793212e-01
-6.70951307e-01 1.02695143e+00 -4.12839592e-01 1.15482323e-01
8.33180547e-01 -1.21273613e+00 -1.63568512e-01 6.18414044e-01
1.16943546e-01 4.68102545e-01 4.19626057e-01 -3.65671515e-01
-1.54452419e+00 -9.21552539e-01 1.21061349e+00 -3.79094422e-01
8.83889973e-01 -6.71483994e-01 -8.67530763e-01 6.46887898e-01
9.09604669e-01 -1.51376855e-02 1.23664689e+00 5.56963146e-01
-7.88191676e-01 1.06661201e-01 -8.85172188e-01 6.32674515e-01
8.56359482e-01 -1.07045436e+00 -4.42879409e-01 -2.52604753e-01
5.90448678e-01 -2.03025758e-01 -1.28891623e+00 4.76526916e-01
7.08056688e-01 -8.72340918e-01 8.60354304e-01 -8.98983777e-01
4.80381399e-01 1.19841404e-01 -3.69466871e-01 -1.58159149e+00
-1.34483799e-01 -5.58092535e-01 -1.24193057e-01 1.62844777e+00
5.99273264e-01 -5.22607028e-01 9.53810886e-02 5.47758341e-01
-3.33323061e-01 -8.60683322e-01 -1.00368500e+00 -2.86892325e-01
-1.93552569e-01 -7.17116833e-01 3.05006057e-01 1.22875583e+00
6.61022782e-01 8.06929469e-01 -8.24779987e-01 2.32971758e-01
2.79250771e-01 -6.71718782e-03 4.97272193e-01 -8.97672594e-01
-2.28218064e-01 -5.92475116e-01 -3.62832725e-01 -5.41431606e-01
8.94625723e-01 -1.13630390e+00 -1.92599386e-01 -1.30438173e+00
3.98936540e-01 -4.66591381e-02 -7.64352500e-01 7.63392210e-01
-2.26248160e-01 3.80007535e-01 2.26373628e-01 -5.80354691e-01
-1.00505924e+00 1.16612458e+00 8.99075031e-01 -5.66371419e-02
-7.22918194e-04 -7.52177596e-01 -7.25996494e-01 3.44535679e-01
7.97218978e-01 -1.27897739e-01 -2.92197734e-01 -4.71993804e-01
1.60746410e-01 -3.64068523e-02 5.19367099e-01 -6.69017732e-01
1.17182106e-01 2.69621581e-01 4.18301284e-01 -1.94445595e-01
8.04543197e-01 -1.03819716e+00 -1.44722566e-01 -5.64582832e-02
-5.91139138e-01 1.29239723e-01 6.52987123e-01 4.05972242e-01
-5.14527142e-01 3.60535830e-01 6.80906773e-01 3.10292721e-01
-6.37615561e-01 2.76397169e-01 -4.84040260e-01 -1.17866479e-01
8.09872985e-01 2.29043305e-01 -2.07679138e-01 -5.45911074e-01
-9.25724745e-01 4.03568447e-01 1.98198959e-01 8.94308209e-01
6.29439414e-01 -1.54057503e+00 -6.19237483e-01 2.88810074e-01
6.13955200e-01 -8.87062073e-01 6.10273421e-01 1.25559139e+00
2.90615588e-01 2.95336872e-01 -2.04347029e-01 -4.39792216e-01
-1.86261046e+00 3.56207490e-01 5.08985996e-01 -4.88202512e-01
-2.20164239e-01 9.00406480e-01 9.23150778e-03 -7.60966837e-01
2.64975995e-01 5.68055063e-02 -2.98657209e-01 2.20484570e-01
3.86735409e-01 -2.02817187e-01 -1.43923657e-02 -1.07342219e+00
-5.85806012e-01 4.48255658e-01 5.17973341e-02 -3.90036106e-01
1.40137625e+00 -4.41982299e-01 -1.54666916e-01 7.02778459e-01
1.47062671e+00 8.26885775e-02 -8.80267918e-01 -2.36835629e-01
-8.32663253e-02 -7.04759434e-02 2.98230201e-01 -1.18593776e+00
-1.08314872e+00 1.15689588e+00 3.95988166e-01 5.64566590e-02
1.51893210e+00 1.82636529e-01 7.03850746e-01 2.82857399e-02
3.28108743e-02 -1.16697240e+00 1.50509834e-01 6.72113240e-01
1.00289476e+00 -1.41840148e+00 -6.18398249e-01 -3.02240878e-01
-1.24257600e+00 9.89791334e-01 6.67600572e-01 4.42555577e-01
3.32549870e-01 4.87990439e-01 2.37536103e-01 -1.97826400e-01
-1.23192954e+00 -3.94674659e-01 7.16865182e-01 4.32304531e-01
9.97385144e-01 8.13656896e-02 1.07651474e-02 9.16030049e-01
1.20461471e-01 -2.25613758e-01 -2.44667426e-01 7.10900426e-01
2.15660721e-01 -1.34655523e+00 1.19136624e-01 -1.71128452e-01
-6.08983934e-01 -2.25109324e-01 -7.21964419e-01 6.45307004e-01
-6.10869080e-02 1.15084338e+00 -1.11837067e-01 -6.21087492e-01
4.56803769e-01 5.45079947e-01 5.74449539e-01 -9.05939415e-02
-1.13865006e+00 1.21641338e-01 5.65078676e-01 -7.69060135e-01
-6.95991457e-01 -5.24182260e-01 -1.11425960e+00 -1.73323303e-02
-3.93801257e-02 2.47334450e-01 7.93733954e-01 9.67884302e-01
8.08795452e-01 8.79886687e-01 5.57414651e-01 -7.35211313e-01
5.66182844e-02 -9.17637467e-01 -2.43013099e-01 5.82753241e-01
4.53063816e-01 -5.93250394e-01 -1.49588749e-01 -2.28958298e-02] | [13.250535011291504, 5.2648115158081055] |
acdd8759-3eea-4c5f-9d53-0890e4e8b122 | neural-headline-generation-with-sentence-wise | 1604.01904 | null | http://arxiv.org/abs/1604.01904v2 | http://arxiv.org/pdf/1604.01904v2.pdf | Neural Headline Generation with Sentence-wise Optimization | Recently, neural models have been proposed for headline generation by
learning to map documents to headlines with recurrent neural networks.
Nevertheless, as traditional neural network utilizes maximum likelihood
estimation for parameter optimization, it essentially constrains the expected
training objective within word level rather than sentence level. Moreover, the
performance of model prediction significantly relies on training data
distribution. To overcome these drawbacks, we employ minimum risk training
strategy in this paper, which directly optimizes model parameters in sentence
level with respect to evaluation metrics and leads to significant improvements
for headline generation. Experiment results show that our models outperforms
state-of-the-art systems on both English and Chinese headline generation tasks. | ['Yu Zhao', 'Ayana', 'Maosong Sun', 'Zhiyuan Liu', 'Shiqi Shen'] | 2016-04-07 | null | null | null | null | ['headline-generation'] | ['natural-language-processing'] | [ 2.53074318e-01 4.17202324e-01 -5.22879601e-01 -4.35539633e-01
-9.26598012e-01 -1.06255986e-01 5.95789731e-01 6.85640723e-02
-5.52369416e-01 1.14384937e+00 6.79679871e-01 -3.88934225e-01
1.95993632e-01 -8.80673528e-01 -5.47717988e-01 -2.28168592e-01
4.05885190e-01 4.33135986e-01 -7.11789504e-02 -3.08652610e-01
3.09833765e-01 5.79643250e-02 -9.00129318e-01 2.43279248e-01
1.10408843e+00 7.51552165e-01 3.71013254e-01 7.27161944e-01
-2.00356200e-01 9.04714465e-01 -9.80152786e-01 -7.11432278e-01
-2.13177755e-01 -5.49453616e-01 -8.26883018e-01 -1.62753955e-01
5.08535445e-01 -3.75125378e-01 -3.58528167e-01 9.97450411e-01
6.99843884e-01 8.41785669e-02 9.27355409e-01 -6.98791623e-01
-9.75018740e-01 9.60196674e-01 -3.45578969e-01 9.05956924e-02
1.48747429e-01 -1.54452682e-01 1.36951232e+00 -9.54259634e-01
4.78813618e-01 1.22986460e+00 5.04029930e-01 7.75158584e-01
-1.11573577e+00 -4.70654905e-01 2.81519920e-01 -1.07301464e-02
-1.05264139e+00 -4.55741167e-01 8.13413143e-01 -1.49833411e-01
1.18834293e+00 1.37379944e-01 3.24670553e-01 1.25394607e+00
2.23335192e-01 1.32846141e+00 6.95966721e-01 -6.80363476e-01
-1.65240660e-01 4.43735331e-01 2.63350636e-01 4.93661672e-01
6.74884394e-02 3.67183648e-02 -4.95176286e-01 8.21264312e-02
5.97084284e-01 -4.92165297e-01 -5.31578422e-01 3.08907360e-01
-1.04760420e+00 1.19139922e+00 3.96312922e-01 2.03799620e-01
-3.05662006e-01 3.45561914e-02 5.33338964e-01 1.04693137e-02
6.98259771e-01 7.97483563e-01 -4.85678941e-01 -1.42694905e-01
-1.07666969e+00 2.45229349e-01 8.35564852e-01 1.09150422e+00
3.62949014e-01 2.94044375e-01 -7.17742205e-01 1.16166592e+00
2.42755353e-01 3.92682195e-01 8.14578474e-01 -4.81141686e-01
1.01870811e+00 3.15052927e-01 6.64407611e-02 -1.05176151e+00
-3.72223884e-01 -8.66313815e-01 -1.13310683e+00 -5.00979304e-01
4.99969050e-02 -5.51348209e-01 -6.14158154e-01 1.52421951e+00
-2.73225069e-01 -2.90766031e-01 1.26899436e-01 6.05552375e-01
8.05075526e-01 1.05766726e+00 -1.49881616e-01 -4.51280266e-01
8.27779412e-01 -1.50378299e+00 -1.09419930e+00 -2.91872323e-01
7.07682848e-01 -7.46382892e-01 1.39553809e+00 1.91065580e-01
-1.28810513e+00 -7.26281583e-01 -1.04950964e+00 -1.65716391e-02
-2.35825092e-01 4.59458828e-01 3.27394515e-01 5.82147062e-01
-9.87932801e-01 5.58709323e-01 -4.60984170e-01 -1.36472329e-01
1.28146842e-01 1.47368252e-01 1.78899169e-01 5.19139469e-01
-1.59022880e+00 8.96003067e-01 8.81681502e-01 1.82728246e-01
-4.72695023e-01 -5.95346451e-01 -8.63524675e-01 1.30786806e-01
8.70374292e-02 -8.61308992e-01 1.64514697e+00 -4.61060077e-01
-1.89347839e+00 2.96330363e-01 -3.63070667e-01 -8.08740079e-01
5.53980052e-01 -6.18310392e-01 -5.27879417e-01 -2.31693573e-02
-1.31312862e-01 9.85275745e-01 7.39174068e-01 -1.03974497e+00
-7.50917017e-01 3.19276035e-01 -6.52214065e-02 3.00911188e-01
-9.69233811e-01 2.25145742e-02 -5.33806264e-01 -8.71435165e-01
-4.34522033e-01 -6.09285355e-01 -3.62136602e-01 -6.57270610e-01
-8.01317692e-01 -5.45125246e-01 5.28383374e-01 -7.98458755e-01
1.92068028e+00 -1.38129079e+00 -1.12551145e-01 -1.78518221e-01
-9.08983946e-02 4.44611102e-01 -1.75624162e-01 4.74953026e-01
1.97773203e-01 2.87864357e-01 -8.46685003e-03 -5.99517703e-01
1.64765462e-01 -2.36583471e-01 -7.41686940e-01 -1.49852484e-01
2.98693627e-01 1.11973143e+00 -7.69385934e-01 -8.08065832e-01
-6.88354596e-02 2.45778665e-01 -7.48630583e-01 4.76072431e-01
-3.50695640e-01 2.02788319e-02 -4.44452494e-01 2.09241703e-01
3.28897923e-01 -3.80381465e-01 -1.36930361e-01 -7.31330663e-02
-7.25157112e-02 6.77817762e-01 -4.02650625e-01 1.36952317e+00
-8.63579392e-01 6.83359683e-01 -5.83675563e-01 -5.93077242e-01
1.16708994e+00 4.08200324e-01 -1.67372212e-01 -5.37991107e-01
-1.05060907e-02 1.11387428e-02 -2.29516611e-01 -1.89497679e-01
1.20465553e+00 9.62688550e-02 -6.03709519e-02 3.88541102e-01
-2.45910715e-02 -7.60403499e-02 3.84977669e-01 2.27782413e-01
4.31730986e-01 1.23773791e-01 1.16251156e-01 -2.05162470e-03
5.64432979e-01 -3.32014978e-01 5.11283517e-01 1.08513701e+00
1.75263509e-01 7.96389699e-01 4.06819701e-01 -1.01189896e-01
-9.87587750e-01 -8.02387059e-01 -3.15762102e-03 1.07860100e+00
-2.58215249e-01 -5.23004591e-01 -8.98005366e-01 -8.49710405e-01
-4.79541957e-01 1.24111533e+00 -4.62879479e-01 -2.50995755e-01
-9.10701275e-01 -8.15762520e-01 6.71108782e-01 5.80468237e-01
4.73656356e-01 -1.32534266e+00 -1.91165775e-01 4.98992801e-01
-3.61770958e-01 -1.09811735e+00 -8.32799554e-01 -1.43560663e-01
-9.29891586e-01 -3.87921512e-01 -1.04760945e+00 -8.37354362e-01
5.90212703e-01 -8.19317177e-02 1.29447138e+00 9.22976211e-02
9.34719592e-02 -1.84952185e-01 -3.06777000e-01 -4.79559988e-01
-5.96152484e-01 7.96595216e-01 -1.18944198e-01 -2.40033016e-01
1.49401054e-01 -8.29619542e-02 -5.05127609e-01 1.63691083e-03
-7.75229812e-01 3.55037868e-01 7.75417745e-01 1.19969320e+00
2.75982976e-01 -8.81105289e-02 1.08784068e+00 -1.10168695e+00
1.39303684e+00 -2.29067534e-01 -5.99606395e-01 6.67246521e-01
-1.03838742e+00 2.18930081e-01 9.83027995e-01 -2.03602478e-01
-1.43523026e+00 -4.22469974e-01 -2.78398991e-01 3.48497219e-02
6.13104068e-02 8.90777230e-01 7.97829777e-02 6.31479919e-01
5.19522667e-01 4.48345393e-01 -3.35701525e-01 -5.13631046e-01
4.91252035e-01 7.56089687e-01 4.69240844e-01 -3.48803222e-01
6.20402753e-01 -2.04385847e-01 -4.25938725e-01 -6.47602797e-01
-1.26326036e+00 -1.57742918e-01 -7.12628961e-01 -2.34785542e-01
7.68069804e-01 -7.11713433e-01 -3.44995737e-01 5.32965362e-01
-1.46794808e+00 -1.55090824e-01 -6.53039292e-02 4.76515114e-01
-4.61383075e-01 2.87048787e-01 -8.39441121e-01 -8.49507809e-01
-9.94761229e-01 -7.81794429e-01 7.59276569e-01 4.36675400e-01
-4.40428376e-01 -1.44119310e+00 2.94640847e-02 2.15232655e-01
6.02783680e-01 -1.64653704e-01 1.07028854e+00 -6.52977467e-01
-2.72329539e-01 -5.00219524e-01 -2.30930388e-01 5.85118651e-01
1.85459360e-01 2.85782572e-02 -8.25053334e-01 -2.70752400e-01
-1.41269028e-01 -5.62664747e-01 9.39475119e-01 5.27355790e-01
1.23891020e+00 -8.46721232e-01 -3.07571352e-01 3.45080763e-01
1.08722472e+00 9.43300501e-03 5.57540298e-01 4.05188560e-01
7.02262998e-01 6.99490547e-01 7.96730101e-01 5.40725827e-01
4.11157191e-01 5.98482370e-01 -3.94487604e-02 -2.90503442e-01
-1.49803489e-01 -7.26769090e-01 4.80040133e-01 1.28377998e+00
4.89213914e-01 -7.98005521e-01 -5.35668969e-01 4.56433684e-01
-1.86731827e+00 -8.49359632e-01 1.71674974e-02 2.00158429e+00
1.26851273e+00 7.25525796e-01 1.55523643e-01 -2.28644416e-01
8.73289883e-01 3.85026783e-01 -4.57046479e-01 -5.32975972e-01
-1.69135228e-01 -6.40077665e-02 5.35256088e-01 7.43194282e-01
-9.37189698e-01 1.33081686e+00 7.25996351e+00 1.07090354e+00
-1.04172456e+00 -2.85109401e-01 8.16355944e-01 -1.27622068e-01
-3.78732920e-01 -2.62066782e-01 -1.45914114e+00 4.91213799e-01
1.03424048e+00 -5.87021232e-01 -7.78522938e-02 8.09899449e-01
4.42818761e-01 3.10379654e-01 -9.79183912e-01 5.80712378e-01
1.73403263e-01 -1.42465425e+00 4.67502981e-01 -1.25186950e-01
9.31248903e-01 -2.20682412e-01 1.95715800e-01 5.44154167e-01
3.07016224e-01 -1.08965993e+00 6.40146494e-01 6.49260461e-01
4.91809338e-01 -1.01174068e+00 8.18519473e-01 5.97779751e-01
-6.80400431e-01 2.43682891e-01 -5.55047691e-01 9.73822325e-02
6.90288544e-01 5.97670138e-01 -1.45574355e+00 3.26766431e-01
1.48607641e-01 7.06790268e-01 -5.97050846e-01 9.13973689e-01
-4.86085564e-01 9.56562996e-01 3.91319506e-02 -5.76452672e-01
5.11273682e-01 -6.80507198e-02 4.35060471e-01 1.52249599e+00
2.42916852e-01 -5.05851388e-01 1.39021114e-01 9.05638158e-01
-1.55920252e-01 2.94902295e-01 -4.34611887e-01 -2.69575477e-01
5.57208240e-01 1.12397063e+00 -2.69103557e-01 -3.71596634e-01
-1.60800427e-01 8.55247378e-01 5.53314865e-01 4.40350026e-01
-6.55799866e-01 -8.63891304e-01 9.93635356e-02 -1.51071951e-01
1.79102406e-01 -1.48486540e-01 -3.06736559e-01 -9.85221386e-01
-3.53582464e-02 -6.86273158e-01 2.30825141e-01 -4.72901285e-01
-1.15423036e+00 9.27682400e-01 6.66513816e-02 -1.22202814e+00
-9.47165310e-01 -4.43528235e-01 -8.14956427e-01 9.62607980e-01
-1.70499551e+00 -1.09785819e+00 2.47100666e-01 2.26665348e-01
9.99883175e-01 -4.66755480e-01 7.60707617e-01 -3.13764960e-02
-9.07171905e-01 1.17477846e+00 1.86929584e-01 1.86814502e-01
7.20010996e-01 -1.30890262e+00 8.87988508e-01 7.27482200e-01
5.21913841e-02 7.37878382e-01 7.82232881e-01 -5.55454195e-01
-5.41960180e-01 -1.17836547e+00 1.51253140e+00 -5.31086028e-02
7.81349063e-01 -2.17181191e-01 -7.13384628e-01 3.88949335e-01
6.49259448e-01 -6.99690938e-01 7.24723756e-01 2.52941459e-01
2.35421378e-02 1.10118784e-01 -5.81792057e-01 9.69068825e-01
5.94724536e-01 -5.26139259e-01 -5.25512040e-01 6.14715815e-01
1.22837520e+00 -2.54714757e-01 -8.50516915e-01 4.26349193e-01
3.16033334e-01 -7.27712691e-01 6.14213884e-01 -8.63047421e-01
8.16942334e-01 8.85342881e-02 1.85605854e-01 -1.52775228e+00
-3.35318714e-01 -7.32835650e-01 -4.81949419e-01 1.65611947e+00
1.03850746e+00 -4.32124346e-01 8.12987506e-01 4.81358647e-01
-2.94769347e-01 -1.01310039e+00 -4.71704662e-01 -7.98720896e-01
3.23368162e-01 -3.96965779e-02 4.92338777e-01 5.73897362e-01
3.56057920e-02 8.36299419e-01 -8.08116853e-01 -2.98656017e-01
3.26903313e-01 -4.33005355e-02 6.81694448e-01 -7.70982325e-01
-4.19550806e-01 -7.56615043e-01 2.71707058e-01 -1.71191740e+00
5.99589884e-01 -9.10391927e-01 2.80245095e-01 -1.57892323e+00
1.09377960e-02 -1.85548440e-01 -2.96941221e-01 1.30588248e-01
-6.11474812e-01 -1.05915539e-01 1.61287025e-01 8.95404592e-02
-5.87096810e-01 9.44552660e-01 1.26571262e+00 -1.52514562e-01
-1.61717311e-01 3.82183075e-01 -7.40276277e-01 6.11430407e-01
1.05552077e+00 -2.84170747e-01 -5.82917213e-01 -6.22796118e-01
2.20561683e-01 1.53096586e-01 -2.50604540e-01 -8.01590085e-01
1.96708530e-01 -1.09229930e-01 2.66681880e-01 -1.03796661e+00
2.81919450e-01 -2.46737555e-01 -5.80487490e-01 3.61673504e-01
-1.22600949e+00 2.82951713e-01 -1.27655447e-01 5.23418903e-01
-4.14864063e-01 -8.09159160e-01 6.17197454e-01 -7.04118982e-02
-2.10774973e-01 1.82627439e-01 -3.58533382e-01 1.38244823e-01
3.89869869e-01 -5.30602895e-02 -4.26308841e-01 -7.96559513e-01
-1.44203588e-01 3.76868576e-01 2.33519059e-02 7.00618088e-01
7.12912440e-01 -1.40142119e+00 -1.06096041e+00 -1.21351726e-01
-2.65419297e-02 -4.34426963e-02 -2.84015890e-02 4.09357607e-01
-3.89527321e-01 1.05107164e+00 2.77653962e-01 -2.49249414e-01
-1.06140399e+00 2.75417984e-01 1.09058350e-01 -9.22875106e-01
-3.28986019e-01 1.09106827e+00 2.19391137e-01 -4.32261825e-01
5.36851585e-01 -2.42953319e-02 -5.68437934e-01 1.35543913e-01
6.57862306e-01 3.51894766e-01 5.33257015e-02 -3.74658376e-01
3.12151819e-01 8.67745727e-02 -8.60521555e-01 -3.52952003e-01
9.98734891e-01 -3.03411692e-01 8.45511854e-02 4.49401170e-01
1.43872690e+00 -1.47905350e-01 -1.12227345e+00 -4.04647797e-01
2.69237578e-01 -1.39324352e-01 2.12327898e-01 -8.31129670e-01
-8.42926919e-01 1.00049102e+00 8.28331634e-02 1.95576251e-01
8.51525247e-01 -4.05276090e-01 1.26669729e+00 7.21969545e-01
-2.95737386e-02 -1.55292618e+00 2.89836168e-01 9.18782413e-01
9.99367774e-01 -1.13223398e+00 -1.19927049e-01 2.49581244e-02
-7.73382485e-01 1.13598585e+00 9.48562860e-01 -7.82763809e-02
3.79548818e-01 7.56753311e-02 1.82989687e-01 3.25756460e-01
-1.11109984e+00 3.12358141e-01 5.72996676e-01 4.67432976e-01
9.30902958e-01 -1.66098773e-01 -4.50133085e-01 4.48539853e-01
-7.51382828e-01 -1.79246143e-01 5.22480726e-01 5.17141342e-01
-5.66484928e-01 -1.09394884e+00 -4.69836406e-02 5.92470050e-01
-5.90664744e-01 -6.61678493e-01 -3.85193199e-01 4.82226074e-01
-6.02685034e-01 1.00660932e+00 -1.88985486e-02 -3.13111305e-01
1.91326246e-01 -7.55129978e-02 5.98541237e-02 -5.71797073e-01
-6.43305182e-01 3.38136740e-02 3.96894306e-01 1.20370708e-01
-3.35688256e-02 -3.63293648e-01 -1.00366509e+00 -2.32998058e-02
-8.21875274e-01 4.11965549e-01 5.18080056e-01 7.78443933e-01
1.73863575e-01 6.31698251e-01 9.69280899e-01 -5.58811486e-01
-9.94216025e-01 -1.47285497e+00 -3.81890208e-01 6.97139502e-02
2.84718931e-01 3.34315598e-02 -1.51925698e-01 -2.30584294e-02] | [12.034600257873535, 9.157788276672363] |
e881f7c7-f752-43c0-9792-f10e1f1b8261 | deep-neural-network-or-dermatologist | 1908.06612 | null | https://arxiv.org/abs/1908.06612v1 | https://arxiv.org/pdf/1908.06612v1.pdf | Deep neural network or dermatologist? | Deep learning techniques have proven high accuracy for identifying melanoma in digitised dermoscopic images. A strength is that these methods are not constrained by features that are pre-defined by human semantics. A down-side is that it is difficult to understand the rationale of the model predictions and to identify potential failure modes. This is a major barrier to adoption of deep learning in clinical practice. In this paper we ask if two existing local interpretability methods, Grad-CAM and Kernel SHAP, can shed light on convolutional neural networks trained in the context of melanoma detection. Our contributions are (i) we first explore the domain space via a reproducible, end-to-end learning framework that creates a suite of 30 models, all trained on a publicly available data set (HAM10000), (ii) we next explore the reliability of GradCAM and Kernel SHAP in this context via some basic sanity check experiments (iii) finally, we investigate a random selection of models from our suite using GradCAM and Kernel SHAP. We show that despite high accuracy, the models will occasionally assign importance to features that are not relevant to the diagnostic task. We also show that models of similar accuracy will produce different explanations as measured by these methods. This work represents first steps in bridging the gap between model accuracy and interpretability in the domain of skin cancer classification. | ['Sally Shrapnel', 'Becks Simpson', 'Reuben Dutton', 'Kyle Young', 'Gareth Booth'] | 2019-08-19 | null | null | null | null | ['skin-cancer-classification'] | ['medical'] | [ 6.52674019e-01 5.70910633e-01 -3.24936897e-01 -4.05121475e-01
-7.86272883e-01 -5.14854312e-01 5.60537994e-01 3.74654800e-01
-3.91735017e-01 6.98381662e-01 3.12196463e-01 -7.33836591e-01
-7.52687514e-01 -5.57738483e-01 -3.80325019e-01 -6.24531686e-01
1.18826665e-01 4.16963995e-01 1.31621838e-01 -2.41210029e-01
1.66962832e-01 4.26904768e-01 -1.18081546e+00 8.52530658e-01
8.33354414e-01 9.43096101e-01 -3.56595159e-01 9.62910235e-01
1.53737932e-01 9.93069768e-01 -4.02424097e-01 -5.71599245e-01
2.24884488e-02 -4.91258353e-01 -1.37646484e+00 7.90657848e-02
6.25465512e-01 -3.92349064e-01 9.39324498e-02 6.57473981e-01
2.99538463e-01 -6.17275357e-01 8.08236659e-01 -9.54698861e-01
-6.72189057e-01 3.73937875e-01 -3.41220081e-01 2.23098636e-01
1.69050023e-01 2.88326144e-01 8.98721278e-01 -2.09630996e-01
9.21770155e-01 8.35387886e-01 1.20113480e+00 7.81318307e-01
-1.10307908e+00 -7.74230585e-02 9.04160663e-02 1.58143833e-01
-1.10270619e+00 -1.85248658e-01 1.08392432e-01 -4.71754432e-01
9.58815873e-01 8.02791059e-01 6.66193008e-01 1.28541017e+00
4.61270541e-01 4.20060307e-01 1.50785851e+00 -6.40509963e-01
1.37292266e-01 2.51180857e-01 2.92401314e-01 8.52073014e-01
2.43901089e-01 2.45014191e-01 -3.88340622e-01 -2.33551130e-01
6.16997123e-01 -1.94606319e-01 -1.25211909e-01 9.38931480e-04
-8.85449290e-01 8.84840429e-01 5.55152595e-01 2.36219838e-01
-1.17765822e-01 2.49623775e-01 4.26296145e-01 3.23099315e-01
5.00160158e-01 6.44692004e-01 -5.15334547e-01 4.86478396e-03
-8.02113235e-01 6.72161654e-02 8.37796807e-01 2.31455639e-01
2.35626444e-01 -6.28695011e-01 3.60081978e-02 5.03485620e-01
2.79988229e-01 -1.17411360e-01 3.82573366e-01 -7.47893631e-01
-1.62115902e-01 6.90075815e-01 -8.71372372e-02 -5.99473715e-01
-6.92242980e-01 -4.03172255e-01 -5.71333885e-01 5.67032635e-01
7.57874370e-01 -6.59766719e-02 -1.24854159e+00 1.34069633e+00
1.00893816e-02 -1.43486317e-02 -1.07421860e-01 7.98281312e-01
8.31104219e-01 -4.03037280e-01 4.06033695e-01 3.41612130e-01
1.43979502e+00 -8.24879348e-01 -3.69293928e-01 -1.22599602e-01
1.14916468e+00 -6.90174103e-01 9.99514699e-01 6.30197406e-01
-8.43076110e-01 -1.82865962e-01 -9.77829576e-01 -2.39143491e-01
-6.33801877e-01 1.29106313e-01 9.94526088e-01 8.13233614e-01
-1.30676866e+00 6.22226536e-01 -1.00526774e+00 -9.20534730e-01
5.74940026e-01 6.23522043e-01 -5.38372695e-01 2.11145937e-01
-1.09953153e+00 1.34568679e+00 1.48925304e-01 3.76361310e-02
-7.03627050e-01 -5.98466873e-01 -5.68928421e-01 -3.93654168e-01
8.24717879e-02 -8.43994260e-01 1.22066414e+00 -1.52025342e+00
-1.09111691e+00 1.36638546e+00 -2.41941303e-01 -7.41605341e-01
7.47300625e-01 5.82427569e-02 -5.38414061e-01 2.29612842e-01
-8.53266120e-02 6.34943008e-01 5.24083614e-01 -1.07223082e+00
-7.42254078e-01 -2.09058389e-01 3.39274287e-01 -5.90047538e-02
-2.65939385e-01 -4.02375907e-02 -1.00065954e-01 -2.89367288e-01
-2.14951858e-01 -9.17694390e-01 -3.01568896e-01 3.88032317e-01
-5.70235431e-01 -4.15649451e-02 4.87160921e-01 -6.88241422e-01
1.17463815e+00 -1.82618582e+00 -1.97315529e-01 3.02649140e-01
6.33213997e-01 3.32630664e-01 -1.19311988e-01 4.59056020e-01
-4.84587431e-01 6.95104003e-01 -1.87759563e-01 -1.17930539e-01
-1.90864563e-01 2.88239241e-01 -2.13432610e-01 5.55976629e-01
7.66279399e-01 9.45853591e-01 -8.69864762e-01 -5.78242719e-01
4.77573395e-01 4.09757137e-01 -4.08089846e-01 -3.60674739e-01
-1.03781275e-01 9.93561819e-02 -1.74398765e-01 7.73505569e-01
5.10817349e-01 -6.02019131e-01 3.06064278e-01 -1.14684626e-01
1.09108210e-01 2.69890994e-01 -6.73136234e-01 1.37368751e+00
-3.61956239e-01 8.81057680e-01 -2.07189903e-01 -6.43348515e-01
2.85471946e-01 2.19749779e-01 2.71555007e-01 -2.99397916e-01
-1.34699102e-02 2.90472567e-01 3.63347113e-01 -7.37781227e-01
1.73268378e-01 -3.96409661e-01 3.26174676e-01 3.46987665e-01
-4.40411717e-02 1.48193231e-02 -1.61795281e-02 5.27394079e-02
1.46732044e+00 7.48872161e-02 6.56921864e-01 -4.02300596e-01
3.02928865e-01 4.87343073e-01 1.19851187e-01 7.82960653e-01
-3.74961555e-01 8.90120089e-01 1.04613900e+00 -8.52604866e-01
-9.35039699e-01 -1.05104637e+00 -7.36722291e-01 7.82748640e-01
-4.07978557e-02 -4.07526493e-01 -7.34843850e-01 -1.11617267e+00
5.08021899e-02 5.39928257e-01 -1.51895320e+00 -2.70717349e-02
-6.39493167e-02 -1.07048297e+00 8.03218603e-01 7.18961716e-01
3.10364980e-02 -6.53513372e-01 -6.64687574e-01 -2.00856805e-01
1.96477875e-01 -8.18147480e-01 1.80086419e-01 3.15094978e-01
-7.08644807e-01 -1.56087136e+00 -4.05188262e-01 -5.53215683e-01
9.17557240e-01 -1.59630165e-01 1.27447307e+00 6.64836943e-01
-6.30521417e-01 3.57126743e-01 -2.71074891e-01 -7.25680232e-01
-8.23478043e-01 2.56000847e-01 -3.55862439e-01 -4.00626510e-01
7.70254135e-01 5.99319525e-02 -6.77872598e-01 3.03836942e-01
-1.19945538e+00 2.61223465e-01 8.58183324e-01 1.10072243e+00
5.02900898e-01 -1.70998022e-01 3.37041885e-01 -1.40787899e+00
7.28430092e-01 -4.67000008e-01 1.14048533e-01 3.82916063e-01
-7.22530961e-01 5.15871448e-03 4.10005957e-01 -6.02198802e-02
-7.43708193e-01 7.37406835e-02 -3.01741689e-01 1.57057583e-01
-4.53835249e-01 5.70363224e-01 5.47554493e-01 -3.33184928e-01
1.29322720e+00 -2.42217794e-01 4.27821279e-01 -1.23343512e-01
6.64049089e-02 6.88187659e-01 1.59369588e-01 -1.97937906e-01
4.94921267e-01 8.72002959e-01 2.05954731e-01 -7.01600492e-01
-9.70901012e-01 -2.62701988e-01 -6.88346863e-01 4.35729958e-02
7.65434623e-01 -6.78334355e-01 -4.60942686e-01 3.23496342e-01
-7.94927776e-01 -6.20333970e-01 -3.50056499e-01 7.33033419e-02
-4.62333202e-01 4.89681542e-01 -6.50841415e-01 -6.89187586e-01
-1.53643563e-01 -1.10694373e+00 1.39718866e+00 2.27145389e-01
-9.31967795e-01 -1.68972647e+00 -8.44276398e-02 4.85756874e-01
3.42008442e-01 5.95600843e-01 9.67363477e-01 -7.85440683e-01
-1.31276518e-01 -3.32618833e-01 -4.22491282e-01 2.29292125e-01
3.67136896e-01 4.61435378e-01 -1.26520562e+00 -1.76549584e-01
-3.32594872e-01 -5.63297689e-01 1.07741058e+00 4.96157736e-01
1.45743132e+00 -8.87789354e-02 -5.12302458e-01 6.83468580e-01
1.52248561e+00 -1.89061537e-01 7.65597999e-01 4.51502442e-01
3.25848073e-01 7.44571507e-01 3.56273323e-01 -1.41701654e-01
1.97321698e-01 3.83081406e-01 6.24635816e-01 -6.89963698e-01
-3.56059968e-01 -1.13887347e-01 -4.00404558e-02 -3.68052088e-02
-3.77286524e-01 -1.68740124e-01 -1.10355568e+00 7.60614932e-01
-1.74151266e+00 -6.03349209e-01 -3.92764241e-01 2.01855063e+00
7.38108039e-01 3.95818561e-01 1.42365828e-01 2.45852936e-02
2.83210903e-01 -2.15014294e-01 -3.28469336e-01 -7.37196863e-01
-1.19378448e-01 5.18670976e-01 6.30598783e-01 6.67585135e-01
-1.03056407e+00 5.80110252e-01 7.52964783e+00 7.47721910e-01
-1.21390820e+00 -3.77971940e-02 9.81061101e-01 2.52560377e-02
-3.67305279e-01 2.97133736e-02 -5.29195130e-01 2.94444144e-01
9.66222644e-01 3.56553555e-01 3.06888130e-02 5.50510943e-01
1.56375095e-01 -2.69844800e-01 -1.42991555e+00 3.89302641e-01
7.57569969e-02 -1.52053618e+00 -4.76067513e-02 3.31465483e-01
6.24198914e-01 3.97825502e-02 3.31427723e-01 -1.78428560e-01
3.59605074e-01 -1.76150811e+00 2.47278780e-01 6.57010913e-01
1.04376841e+00 -3.33633989e-01 1.04751766e+00 -4.68002893e-02
-4.05446231e-01 4.40210812e-02 -1.95985690e-01 -8.26761723e-02
-4.34175014e-01 3.32980782e-01 -1.40699852e+00 5.10984719e-01
4.71022189e-01 5.26498079e-01 -9.55268621e-01 9.20953274e-01
-2.83695549e-01 8.21668148e-01 -2.51873612e-01 -7.20428526e-02
4.72404748e-01 4.38723028e-01 2.41428837e-01 1.36364865e+00
-1.93607226e-01 -3.58229727e-01 -5.24368048e-01 7.80855656e-01
4.98889595e-01 -2.11947858e-01 -4.59273428e-01 -7.59000480e-02
-1.22704566e-01 1.32622468e+00 -7.05461204e-01 1.38273221e-02
-2.84328908e-01 8.79570484e-01 1.90827608e-01 2.08258554e-01
-8.05842996e-01 -2.30340868e-01 5.75858653e-01 4.29385722e-01
-1.56043917e-01 3.31383049e-01 -7.01905668e-01 -7.90691316e-01
-1.99262276e-01 -9.62260544e-01 6.74865127e-01 -7.78201401e-01
-1.50010276e+00 5.14389277e-01 -3.35464269e-01 -9.23316240e-01
-4.72079009e-01 -1.31498134e+00 -6.07869983e-01 1.00827408e+00
-1.64542258e+00 -1.47448301e+00 -6.35370314e-01 2.07947627e-01
2.51217812e-01 1.17967792e-01 1.23058844e+00 -2.53907889e-01
-3.54029357e-01 6.95162356e-01 -2.37825308e-02 1.46774665e-01
8.98404717e-01 -1.63563526e+00 3.74070883e-01 3.95397723e-01
-2.22372618e-02 6.69410884e-01 5.53725958e-01 -3.83800805e-01
-6.94508255e-01 -9.41689551e-01 7.84006774e-01 -1.21880102e+00
7.77696967e-01 5.10772914e-02 -7.34708071e-01 7.03728557e-01
1.87771350e-01 2.79529858e-02 1.19992650e+00 5.71556509e-01
-1.69743061e-01 1.54444739e-01 -1.27164245e+00 5.72054744e-01
8.81403208e-01 -5.53535104e-01 -4.23717648e-01 6.47231400e-01
3.80637199e-01 -5.04583359e-01 -9.38689709e-01 3.53488833e-01
8.01988184e-01 -1.17694092e+00 7.02238381e-01 -1.24424112e+00
1.02337611e+00 5.13304537e-03 2.63984591e-01 -1.22143972e+00
-2.67997414e-01 -2.64070481e-01 2.48125106e-01 6.17709398e-01
7.59369671e-01 -7.09980905e-01 1.00901389e+00 7.91877627e-01
6.24845996e-02 -1.38959372e+00 -9.03457820e-01 -3.36883485e-01
4.28503245e-01 -4.07638192e-01 2.93487728e-01 9.46325243e-01
6.83153048e-02 -9.63337570e-02 5.46054505e-02 -3.70837301e-02
3.48734558e-01 -2.51942813e-01 4.50803906e-01 -1.02757287e+00
-4.16049600e-01 -5.06669104e-01 -7.89938927e-01 -3.51189107e-01
-2.70190120e-01 -6.98522091e-01 -4.75909114e-01 -1.65095067e+00
2.75931567e-01 -4.09705967e-01 -4.44938958e-01 7.68766284e-01
-2.09507719e-01 3.76620233e-01 -2.19978318e-01 1.70484915e-01
-3.93692434e-01 -4.43772912e-01 1.05027843e+00 -5.19288853e-02
1.36297792e-01 -1.23415902e-01 -1.12432969e+00 1.02112722e+00
8.48077536e-01 -1.80402786e-01 -3.17637473e-01 -5.62359512e-01
4.28322643e-01 -3.81723464e-01 8.67910922e-01 -7.93438077e-01
1.97861586e-02 -3.33889276e-01 6.72738016e-01 -1.53923603e-02
9.46591645e-02 -7.01361656e-01 9.95941646e-03 7.51980782e-01
-6.00256264e-01 -3.88492525e-01 3.68011892e-01 5.35601139e-01
-2.94145150e-03 -3.39064598e-01 7.70728588e-01 -2.64760941e-01
-8.43809187e-01 -1.13184631e-01 -2.77760863e-01 -1.20841570e-01
1.10667551e+00 -5.31479955e-01 -6.98608994e-01 -1.98150739e-01
-9.45976436e-01 7.50040933e-02 7.78119743e-01 2.56686419e-01
3.14291984e-01 -8.28467548e-01 -7.52647638e-01 1.79582052e-02
4.66787308e-01 -1.29644603e-01 2.32206196e-01 1.13732135e+00
-1.02821732e+00 6.26705229e-01 -1.74721852e-01 -6.75997317e-01
-1.37314296e+00 1.93376094e-01 8.96341324e-01 -4.16861296e-01
-2.60985583e-01 1.04955530e+00 2.21207067e-01 -3.31456453e-01
-6.89854547e-02 -5.72773278e-01 5.84142618e-02 -2.72204995e-01
4.58498955e-01 4.58184928e-02 2.56680787e-01 -1.80435151e-01
-5.10748684e-01 3.12402308e-01 -4.56316054e-01 5.07850088e-02
1.18108761e+00 2.47382089e-01 -9.77976099e-02 9.21950191e-02
9.14097428e-01 -1.50370849e-02 -9.44772840e-01 3.54841441e-01
-2.10815575e-02 -4.22896653e-01 -2.38685254e-02 -1.50048661e+00
-7.55376875e-01 9.24178839e-01 9.07526612e-01 3.65069151e-01
1.26133907e+00 1.46070691e-02 3.59613746e-01 1.09985270e-01
-1.25407115e-01 -1.15480065e+00 -2.11157396e-01 -9.22679901e-02
6.58695877e-01 -1.39304709e+00 2.24941269e-01 -6.82646632e-01
-6.50188148e-01 1.28675151e+00 6.73890889e-01 -1.42187312e-01
3.75309080e-01 2.79461563e-01 4.18062747e-01 -4.27746773e-01
-9.30141151e-01 -2.31903702e-01 3.65723491e-01 8.36615145e-01
7.02025354e-01 7.55644143e-02 -5.23093522e-01 4.55719113e-01
-8.70332941e-02 3.53357911e-01 6.17507458e-01 6.92052722e-01
-3.97538915e-02 -1.24529028e+00 -1.38655782e-01 8.26690555e-01
-7.02148438e-01 -8.86592865e-02 -1.13051867e+00 1.19967771e+00
4.76123422e-01 7.69389391e-01 -7.40779489e-02 -4.85872298e-01
-3.49064171e-02 -1.05667837e-01 4.58561689e-01 -6.11124516e-01
-7.08509862e-01 -3.71115386e-01 5.15632391e-01 -3.23834777e-01
-3.83729041e-01 -3.68048310e-01 -9.80597317e-01 -2.00822666e-01
-3.56563210e-01 -2.09627718e-01 5.55430055e-01 1.00135946e+00
3.13679546e-01 6.89638138e-01 2.69539328e-03 5.59088923e-02
-6.91487551e-01 -7.69309521e-01 -5.14386058e-01 5.39570272e-01
4.41545635e-01 -3.60064209e-01 -4.71758246e-01 1.11990362e-01] | [15.52209758758545, -2.765129566192627] |
75d366d2-647b-462a-95cc-dae6aacbc3f9 | exploring-example-selection-for-few-shot-text | null | null | https://openreview.net/forum?id=tnHT06ijUPZ | https://openreview.net/pdf?id=tnHT06ijUPZ | Exploring Example Selection for Few-shot Text-to-SQL Semantic Parsing | We study example selection methods for few-shot text-to-SQL tasks with unseen databases. Annotating natural language questions with corresponding SQL queries is expensive, but we can use abundant unlabeled questions to efficiently select examples to annotate and then use them to adapt models. Many previous works only randomly sample a few instances for few-shot learning, but this random selection is not sufficient to select representative and informative examples that provide specific domain knowledge. We thus explore methods to efficiently choose annotation examples. We identify two important factors: the diversity of selected instances and the dissimilarity to the source training data if any. A diverse training set contains more domain knowledge, while dissimilar examples are selected to fill in the domain gap between the source and target. We show that our best example selection approach substantially improves few-shot text-to-SQL performance in both finetuning using T5 and in-context learning with Codex: average execution accuracy gains of 8.7% and 4.3% over random selection. Our extensive analysis demonstrates the importance of the similarity metric and the embedding method for example representations. We also find that effective example selection reduces syntax errors on the target domains. Our results encourage future work to further explore example selection for efficient adaptation of text-to-SQL models. | ['Anonymous'] | 2022-01-16 | null | null | null | null | ['text-to-sql'] | ['computer-code'] | [ 4.20442969e-02 -1.31745264e-01 -7.78540611e-01 -8.26523840e-01
-1.20252919e+00 -5.34315705e-01 3.34163308e-01 4.57134604e-01
-5.04912198e-01 6.97966754e-01 2.53938109e-01 -5.89959994e-02
-6.87309802e-02 -9.57581997e-01 -6.63444638e-01 -8.29954147e-02
1.11540131e-01 9.15802956e-01 5.23761153e-01 -4.52944338e-01
3.88008267e-01 1.73490882e-01 -1.86614168e+00 5.76245427e-01
1.02096367e+00 8.76908422e-01 1.60320103e-01 6.24692261e-01
-9.03173685e-01 1.03618252e+00 -8.68542075e-01 -3.41864169e-01
1.45860523e-01 -2.66460001e-01 -7.82315373e-01 -6.89824391e-03
5.18557191e-01 -2.31728330e-01 6.40919358e-02 6.67222261e-01
7.57795453e-01 6.08461142e-01 4.11981851e-01 -1.07164967e+00
-7.27025509e-01 7.34303951e-01 -1.83568552e-01 3.84836018e-01
4.14143831e-01 2.65485674e-01 1.09874988e+00 -1.21840072e+00
8.03351164e-01 8.92399430e-01 5.31405270e-01 7.71093130e-01
-1.25722528e+00 -3.93313706e-01 -1.38192624e-01 3.16172868e-01
-1.34187758e+00 -6.86437488e-01 5.01285732e-01 -2.36446619e-01
1.28290904e+00 4.41383988e-01 3.67003024e-01 9.35204983e-01
-3.41643512e-01 7.55694807e-01 5.30526161e-01 -8.25283647e-01
6.86293602e-01 7.09473014e-01 5.57971358e-01 5.50887108e-01
2.69006342e-01 -1.43085852e-01 -7.15172470e-01 -5.63500643e-01
1.70586288e-01 3.32127392e-01 -1.49030685e-01 -4.69806194e-01
-9.36292410e-01 1.11843336e+00 7.17430860e-02 2.29505301e-01
-2.99199045e-01 2.31438596e-02 6.71723187e-01 5.00473976e-01
2.73935199e-01 1.19544339e+00 -7.47722745e-01 -4.26666558e-01
-7.56325901e-01 2.81738818e-01 1.12340879e+00 1.58387887e+00
1.23380053e+00 6.13049865e-02 -3.81692201e-01 1.27744401e+00
-2.32214749e-01 2.38050625e-01 8.89677525e-01 -1.08091235e+00
4.98592317e-01 8.57998013e-01 1.41826525e-01 -3.36087465e-01
1.23060718e-01 -6.12500831e-02 6.81844503e-02 -1.53394863e-01
2.94037133e-01 -2.89187610e-01 -8.06690514e-01 1.46883512e+00
1.89774185e-01 -1.79801047e-01 2.22740591e-01 5.02138734e-01
6.94413245e-01 6.57570362e-01 2.30421454e-01 -1.90430835e-01
1.32901764e+00 -9.59380448e-01 -6.70481205e-01 -5.87365329e-01
1.10785222e+00 -6.38785422e-01 1.86918151e+00 -1.48312151e-01
-8.47537398e-01 -5.07592976e-01 -9.71394837e-01 1.28815070e-01
-6.74665570e-01 -8.15510452e-02 6.48481905e-01 5.27673781e-01
-4.32226121e-01 5.98859608e-01 -6.28601432e-01 -6.83865309e-01
3.25408041e-01 2.22874820e-01 5.06547466e-02 -3.48492473e-01
-1.26118803e+00 7.13842034e-01 4.80214447e-01 -1.05398512e+00
-5.27992725e-01 -1.08035135e+00 -8.86261761e-01 2.12612674e-01
8.38136375e-01 -4.63850319e-01 1.36337924e+00 -7.80393958e-01
-1.02243257e+00 5.81487536e-01 -3.52075189e-01 -5.27482748e-01
-4.40306365e-02 -2.24452585e-01 -4.59133089e-01 3.50564830e-02
1.76008940e-01 4.23942029e-01 4.89544898e-01 -9.10644114e-01
-6.63940191e-01 -3.04173708e-01 -1.01121485e-01 1.18549913e-01
-8.33974242e-01 -5.91403339e-03 -5.95745385e-01 -5.74894309e-01
-3.32602978e-01 -6.34006858e-01 -2.50174254e-01 -1.38202667e-01
-1.12449806e-02 -3.71058017e-01 7.60281682e-01 -1.82960108e-01
1.64383972e+00 -2.07625151e+00 -5.36953032e-01 -2.04087347e-02
-8.94805789e-02 3.11894596e-01 -1.48488253e-01 5.47783375e-01
5.85647300e-02 2.15949982e-01 -1.40233403e-02 2.09229603e-01
4.34882492e-02 2.79368311e-01 -5.53839564e-01 -2.95449495e-01
2.58425534e-01 1.03376353e+00 -9.57790911e-01 -7.38185942e-01
6.88660145e-02 -2.20369287e-02 -7.51116693e-01 5.04277229e-01
-7.81813860e-01 -3.41957539e-01 -6.13818169e-01 8.13321352e-01
5.25086671e-02 -4.77599233e-01 -5.14276475e-02 -1.19757138e-01
3.77621174e-01 4.26869512e-01 -1.13378561e+00 1.54373312e+00
-7.19743073e-01 3.86768341e-01 -7.03950882e-01 -7.22075820e-01
1.25675678e+00 1.93110794e-01 4.19006795e-01 -6.44537210e-01
-4.19672132e-01 3.97212267e-01 -1.48587525e-01 -9.64021981e-01
6.13170385e-01 -1.78873971e-01 -2.53766596e-01 5.45945227e-01
2.31513843e-01 -1.61887228e-01 4.19619262e-01 2.76269972e-01
1.31063819e+00 -3.64880823e-02 6.89270854e-01 -4.55582775e-02
1.61634218e-02 5.92407227e-01 5.33075690e-01 1.06399441e+00
-1.59250081e-01 5.45364857e-01 2.90249407e-01 -6.08949780e-01
-1.16608262e+00 -8.08799863e-01 -1.37972683e-01 1.86867070e+00
-1.60672188e-01 -5.68490267e-01 -3.78646463e-01 -1.10614562e+00
2.47448921e-01 1.54424381e+00 -3.98604602e-01 -3.87629062e-01
-6.02357149e-01 -5.39246559e-01 3.10299397e-01 8.16894889e-01
4.44817543e-02 -1.04445243e+00 -6.62297845e-01 4.12589937e-01
2.08138395e-02 -6.33037388e-01 -6.67043209e-01 7.59378910e-01
-8.82363677e-01 -9.93831992e-01 -2.28210866e-01 -6.03920102e-01
4.01726544e-01 6.60579205e-02 1.47484136e+00 -8.55112262e-03
-4.94130254e-01 6.22503638e-01 -5.52254081e-01 -5.72873712e-01
-4.97694671e-01 1.97520658e-01 -2.65103489e-01 -4.34343398e-01
1.32295275e+00 -1.85910851e-01 -3.02213967e-01 3.75769615e-01
-9.08170223e-01 -6.55921936e-01 3.44998717e-01 1.11752033e+00
6.64850354e-01 -1.83600113e-01 8.77767146e-01 -1.39650226e+00
7.15032458e-01 -6.38856411e-01 -3.28527808e-01 7.47859120e-01
-9.54548597e-01 3.79019797e-01 7.64997602e-01 -7.46218920e-01
-1.04725969e+00 1.05972677e-01 1.36587054e-01 -5.36985636e-01
-1.92837298e-01 5.45144916e-01 2.91468073e-02 1.28883496e-01
1.55613708e+00 1.08858958e-01 -7.63971508e-02 -2.34404132e-01
4.48249727e-01 7.96171904e-01 7.58201182e-02 -8.05397093e-01
3.80222321e-01 4.78912666e-02 -6.64413512e-01 -8.58534575e-01
-7.53747046e-01 -6.65642738e-01 -4.10882771e-01 2.34083816e-01
4.49960887e-01 -6.95072711e-01 7.32508488e-04 -4.83375728e-01
-8.00449550e-01 -3.49548459e-01 -1.04153895e+00 3.50530118e-01
-5.69074810e-01 6.51096404e-02 -4.15006101e-01 -7.41811633e-01
-4.00959790e-01 -9.58085895e-01 9.38182473e-01 1.25069499e-01
-7.25321889e-01 -9.69478548e-01 3.28156911e-02 1.75077766e-01
6.25739872e-01 -1.69054180e-01 1.37751675e+00 -1.55124748e+00
-5.79230249e-01 -6.92621768e-01 2.28463650e-01 1.43921703e-01
1.94197536e-01 -9.46849138e-02 -7.80242264e-01 -1.39696211e-01
1.39951795e-01 -7.69581318e-01 6.90858662e-01 1.37471063e-02
1.29760313e+00 -3.78118336e-01 -4.22402382e-01 4.01442021e-01
1.59583616e+00 4.34486687e-01 4.10758764e-01 2.28215694e-01
4.41230148e-01 5.24175704e-01 1.28082776e+00 8.14614058e-01
-1.15339942e-02 6.55880332e-01 -9.44729298e-02 5.04260898e-01
-8.95039290e-02 -2.54679739e-01 1.18330278e-01 5.67982435e-01
5.66490173e-01 -1.85045809e-01 -1.21153021e+00 8.07170928e-01
-1.69384646e+00 -1.04341745e+00 4.74509001e-01 2.55732393e+00
1.29273200e+00 2.11163327e-01 8.82535875e-02 -3.02723527e-01
6.52489901e-01 -1.22743007e-03 -7.93217421e-01 -1.62800372e-01
1.99700966e-01 5.41126788e-01 4.07941163e-01 3.42820615e-01
-7.91607022e-01 9.92798269e-01 6.01753092e+00 9.27745104e-01
-8.37193727e-01 4.46608588e-02 6.14273190e-01 -5.21528184e-01
-5.26350796e-01 1.40507460e-01 -1.30060482e+00 4.63539898e-01
1.24979949e+00 -7.34307945e-01 2.70432293e-01 1.62917709e+00
-2.23447561e-01 1.59400553e-01 -1.41871762e+00 5.82548857e-01
2.45241702e-01 -1.49838173e+00 2.45074645e-01 -4.77197289e-01
7.41418481e-01 -7.03895614e-02 -2.19701231e-01 8.44275832e-01
4.06936526e-01 -8.78204823e-01 2.07312331e-02 3.49747181e-01
7.52910852e-01 -6.43450141e-01 4.81413394e-01 2.86216319e-01
-1.08759820e+00 -2.60009766e-01 -7.68729210e-01 3.38236481e-01
-2.45011359e-01 3.26209813e-01 -1.33593500e+00 -6.52680695e-02
6.16915822e-01 4.37029302e-01 -7.53441691e-01 8.57169390e-01
2.56546468e-01 6.17455125e-01 -2.48120874e-01 -7.18247414e-01
-4.05052975e-02 2.61550725e-01 3.23856473e-01 1.12459159e+00
4.29289222e-01 3.09123062e-02 2.92671204e-01 9.88133311e-01
-4.97432724e-02 4.09393251e-01 -9.20029640e-01 -2.80521810e-01
1.15986502e+00 7.99136221e-01 -2.12609187e-01 -8.85629952e-01
-7.20487356e-01 6.34539962e-01 3.93153846e-01 4.84380513e-01
-5.69656014e-01 -1.03172636e+00 6.17559433e-01 3.23014677e-01
4.96356219e-01 2.41620004e-01 -3.11323255e-01 -1.17722154e+00
1.94330111e-01 -1.19963360e+00 7.46636152e-01 -5.03917456e-01
-1.50565541e+00 4.11585689e-01 8.80455971e-03 -1.22135210e+00
-6.58840656e-01 -2.82462656e-01 -6.37552857e-01 7.57532358e-01
-1.03672171e+00 -4.84633029e-01 -3.84954572e-01 3.54325324e-01
1.15082037e+00 -6.36253834e-01 1.08027744e+00 1.75974920e-01
-3.50147903e-01 8.11368883e-01 2.21097201e-01 8.61073583e-02
1.07350039e+00 -1.49726987e+00 4.18495238e-01 5.45408428e-01
1.89032957e-01 1.18888068e+00 5.96409857e-01 -7.27326393e-01
-1.59089684e+00 -1.30401921e+00 9.18527961e-01 -7.01517880e-01
5.87938786e-01 -2.06048816e-01 -1.41018951e+00 8.44535232e-01
5.99988997e-02 4.11721379e-01 1.05290222e+00 1.54879317e-01
-5.42989969e-01 -3.94423932e-01 -1.14011264e+00 8.02253425e-01
6.67152107e-01 -5.39436460e-01 -8.15013766e-01 5.14329135e-01
1.23598862e+00 1.02729462e-02 -1.11928713e+00 3.74588937e-01
1.92464873e-01 -8.56591463e-01 9.25887346e-01 -9.95507479e-01
3.72806549e-01 3.96512374e-02 -5.32345533e-01 -1.05863118e+00
-3.45266499e-02 -3.89688790e-01 -4.19661433e-01 1.27605319e+00
7.24061608e-01 -4.39180911e-01 1.14161396e+00 1.21983409e+00
6.42427057e-02 -9.46467757e-01 -4.89861041e-01 -9.21267152e-01
-5.71388677e-02 -3.89598459e-01 8.55282545e-01 1.06205702e+00
2.33805865e-01 5.86988926e-01 -4.61181067e-02 -3.38308424e-01
3.51686835e-01 3.17111194e-01 9.83651638e-01 -9.73000407e-01
-5.28071582e-01 -1.10006697e-01 -8.10613632e-02 -9.67585802e-01
-5.64131141e-02 -7.62861431e-01 5.38820736e-02 -1.07152832e+00
1.51369557e-01 -6.03524804e-01 -1.55243725e-01 5.54487765e-01
-5.64978004e-01 -5.26591301e-01 -9.56145078e-02 1.92752093e-01
-8.02259803e-01 4.09403563e-01 5.91250896e-01 -1.90588474e-01
-5.61699688e-01 1.32999986e-01 -6.54068232e-01 2.45274499e-01
7.43585110e-01 -5.09893179e-01 -7.72112727e-01 -1.64400592e-01
-2.93243546e-02 1.26783878e-01 -1.53733149e-01 -9.66841936e-01
2.31990874e-01 -4.01564151e-01 3.14988345e-01 -2.61317164e-01
2.14228645e-01 -8.09702873e-01 -3.99526834e-01 2.95803577e-01
-9.75925684e-01 -1.73315983e-02 1.49813935e-01 7.67163455e-01
-2.61765420e-01 -9.60656464e-01 8.64982128e-01 -5.49251318e-01
-1.22113335e+00 2.03690127e-01 -1.58844829e-01 8.10731232e-01
1.03759909e+00 -1.45491168e-01 -3.90381396e-01 -3.60080034e-01
-5.80807090e-01 1.81282446e-01 6.06293499e-01 4.02190924e-01
6.61930561e-01 -1.38877153e+00 -4.49161321e-01 1.98493034e-01
8.93234015e-01 -2.62201011e-01 -2.56769598e-01 3.36643517e-01
-2.99440354e-01 2.97168165e-01 1.77332506e-01 -4.87678468e-01
-1.16791904e+00 8.46754551e-01 1.26740456e-01 -6.04215115e-02
-3.12166095e-01 9.27897334e-01 -3.40137541e-01 -5.81208885e-01
5.79850793e-01 -8.15136209e-02 3.42023820e-02 -1.96967125e-02
6.58912599e-01 4.89241004e-01 1.31585360e-01 3.33889723e-01
-2.30823621e-01 1.09529622e-01 -4.42657828e-01 2.05313470e-02
1.30781543e+00 4.51917835e-02 1.28630698e-01 9.27248716e-01
1.43478274e+00 -1.02482803e-01 -8.44244003e-01 -6.77670419e-01
5.54582894e-01 -6.53268099e-01 -4.16384816e-01 -7.98452497e-01
-7.15030134e-01 8.19963098e-01 4.80145812e-01 2.54348755e-01
9.48734283e-01 1.00915775e-01 6.87744081e-01 1.08691716e+00
4.09249008e-01 -1.38861799e+00 6.18298292e-01 4.89189476e-01
5.52699208e-01 -1.53220677e+00 -2.86885221e-02 -1.85993627e-01
-9.49468613e-01 1.30640125e+00 1.23916841e+00 1.11112878e-01
3.79346818e-01 2.44995072e-01 9.30898171e-03 -3.20437044e-01
-1.17215931e+00 -1.05595559e-01 1.13788128e-01 5.88455319e-01
7.63829231e-01 -3.08356404e-01 -4.88881208e-03 5.63330889e-01
-2.59426795e-03 -3.61915119e-02 3.38434815e-01 1.22633612e+00
-9.07513261e-01 -1.24585438e+00 -1.94121599e-01 1.11855865e+00
-1.45407170e-01 -3.33264053e-01 -3.35207552e-01 8.22475433e-01
-1.39584631e-01 7.95213282e-01 1.23964354e-01 -4.12296712e-01
4.96468186e-01 8.70568991e-01 3.76034621e-03 -1.39518738e+00
-7.00229764e-01 -1.69070452e-01 3.00807774e-01 -4.68687028e-01
2.69892573e-01 -4.06624675e-01 -1.19565785e+00 -8.48690271e-02
-5.42757154e-01 4.54776704e-01 3.14041048e-01 4.64176387e-01
7.38649905e-01 3.42763603e-01 5.40484071e-01 -6.15428202e-03
-1.16710401e+00 -8.87320876e-01 -3.31123948e-01 6.23544574e-01
9.09843761e-03 -3.55648726e-01 -3.79607290e-01 6.01478573e-03] | [10.106372833251953, 3.639873504638672] |
7202b0ce-9a15-4913-b6ee-41fba8037546 | fighting-uncertainty-with-gradients-offline | 2306.14079 | null | https://arxiv.org/abs/2306.14079v1 | https://arxiv.org/pdf/2306.14079v1.pdf | Fighting Uncertainty with Gradients: Offline Reinforcement Learning via Diffusion Score Matching | Offline optimization paradigms such as offline Reinforcement Learning (RL) or Imitation Learning (IL) allow policy search algorithms to make use of offline data, but require careful incorporation of uncertainty in order to circumvent the challenges of distribution shift. Gradient-based policy search methods are a promising direction due to their effectiveness in high dimensions; however, we require a more careful consideration of how these methods interplay with uncertainty estimation. We claim that in order for an uncertainty metric to be amenable for gradient-based optimization, it must be (i) stably convergent to data when uncertainty is minimized with gradients, and (ii) not prone to underestimation of true uncertainty. We investigate smoothed distance to data as a metric, and show that it not only stably converges to data, but also allows us to analyze model bias with Lipschitz constants. Moreover, we establish an equivalence between smoothed distance to data and data likelihood, which allows us to use score-matching techniques to learn gradients of distance to data. Importantly, we show that offline model-based policy search problems that maximize data likelihood do not require values of likelihood; but rather only the gradient of the log likelihood (the score function). Using this insight, we propose Score-Guided Planning (SGP), a planning algorithm for offline RL that utilizes score-matching to enable first-order planning in high-dimensional problems, where zeroth-order methods were unable to scale, and ensembles were unable to overcome local minima. Website: https://sites.google.com/view/score-guided-planning/home | ['Russ Tedrake', 'Abhishek Gupta', 'Lujie Yang', 'Hongkai Dai', 'Glen Chou', 'H. J. Terry Suh'] | 2023-06-24 | null | null | null | null | ['imitation-learning', 'offline-rl'] | ['methodology', 'playing-games'] | [-1.44961298e-01 2.89324045e-01 -4.50805277e-01 -3.71635482e-02
-1.28001595e+00 -9.03889716e-01 6.92885876e-01 2.84104228e-01
-6.96115971e-01 1.12447739e+00 2.60593593e-01 -5.62511742e-01
-4.61540014e-01 -7.37592876e-01 -9.92487192e-01 -6.79344893e-01
-3.18248212e-01 4.97585267e-01 1.18784308e-01 -6.10576756e-02
4.08789128e-01 5.11368811e-01 -1.10277319e+00 -5.60836434e-01
9.59929049e-01 9.99695599e-01 2.46064961e-01 6.05411232e-01
1.91221714e-01 4.01896656e-01 -2.11761206e-01 1.68958575e-01
6.17096603e-01 -3.68148863e-01 -5.84526479e-01 -2.94744909e-01
1.72687173e-01 -6.63446188e-01 -2.19277516e-01 1.11842632e+00
4.34288561e-01 4.80715990e-01 7.90811658e-01 -9.58404958e-01
-2.77580768e-01 1.33516774e-01 -1.11669712e-01 9.99343029e-05
3.51733804e-01 3.62672389e-01 8.73771012e-01 -6.13035023e-01
6.83192909e-01 9.88487601e-01 5.79612732e-01 5.95103681e-01
-1.18592513e+00 -2.44251713e-01 2.36187547e-01 -3.12904745e-01
-9.98114586e-01 -3.50731373e-01 4.33279306e-01 -5.42388082e-01
6.98769748e-01 1.36981800e-01 8.25768113e-01 8.22517037e-01
4.47191238e-01 7.29581058e-01 1.15773201e+00 -3.07110637e-01
7.10908473e-01 7.90994382e-04 -3.57380688e-01 9.20762122e-01
1.02809377e-01 7.24217594e-01 -3.64265382e-01 -2.21256450e-01
1.03104091e+00 -2.76765287e-01 -3.91692698e-01 -8.59060347e-01
-1.16033518e+00 1.07245779e+00 3.97014201e-01 7.97033384e-02
-2.92834222e-01 4.51712281e-01 1.38196364e-01 2.53203571e-01
2.99801320e-01 9.02190864e-01 -4.27070379e-01 -6.61810875e-01
-6.46989465e-01 5.56858778e-01 8.10074329e-01 7.98861682e-01
6.10833347e-01 -5.85452169e-02 -9.05099809e-02 3.77944589e-01
3.95693660e-01 6.12446666e-01 4.06265467e-01 -1.40984893e+00
6.19581282e-01 2.26157501e-01 7.22190797e-01 -7.61210680e-01
-5.24588704e-01 -2.92173594e-01 -3.57015014e-01 6.83471680e-01
8.93262029e-01 -3.54298353e-01 -6.78266108e-01 2.16510177e+00
4.60907489e-01 -2.12457672e-01 8.02003965e-02 9.83475983e-01
-2.97333300e-01 3.74043286e-01 -3.13418031e-01 -5.36547542e-01
6.80700779e-01 -4.34745938e-01 -3.96530569e-01 -7.23137856e-02
6.97886646e-01 -2.77304977e-01 1.41398704e+00 3.64955276e-01
-1.20115638e+00 -2.59560607e-02 -1.13327074e+00 2.40169212e-01
-1.22178331e-01 -3.54389012e-01 4.84274477e-01 4.71425116e-01
-1.00821137e+00 1.09654093e+00 -1.13954699e+00 -3.41287166e-01
2.50815988e-01 3.18241835e-01 -1.38110235e-01 4.97944683e-01
-9.98795033e-01 1.27443886e+00 3.48681152e-01 -8.56329426e-02
-1.09171081e+00 -7.00871646e-01 -8.21703017e-01 -1.47166818e-01
6.17664754e-01 -4.07980174e-01 1.48082614e+00 -6.13473058e-01
-2.01618838e+00 6.63151070e-02 -4.76022251e-02 -6.15983069e-01
9.34639096e-01 -2.64116257e-01 2.41820410e-01 1.03717223e-01
1.91407576e-02 5.26131213e-01 7.98113823e-01 -1.12836087e+00
-5.47201216e-01 -3.42300206e-01 1.90865263e-01 4.99862462e-01
-5.42165339e-02 -4.91605997e-01 6.52264878e-02 -1.62264064e-01
1.61401808e-01 -8.96064341e-01 -4.36877877e-01 9.80823636e-02
-1.20323308e-01 4.10002805e-02 3.74143809e-01 -5.31586587e-01
1.04475725e+00 -1.73693752e+00 1.24928333e-01 5.53192437e-01
-1.15135662e-01 -9.85865444e-02 -7.83554539e-02 3.95488948e-01
4.66952801e-01 2.22516850e-01 -4.60242093e-01 -1.46847755e-01
3.08444977e-01 2.49816820e-01 -5.83503783e-01 7.77665198e-01
3.09572741e-02 9.09611046e-01 -1.25023043e+00 -3.35222900e-01
3.17992806e-01 4.64859195e-02 -8.24320078e-01 -1.34506226e-02
-6.01276159e-01 8.02968502e-01 -6.71590686e-01 4.09895539e-01
2.61102319e-01 5.65839075e-02 1.53635308e-01 1.97238743e-01
-4.60838854e-01 2.59408087e-01 -1.25058031e+00 1.63006115e+00
-5.49283147e-01 2.28031203e-01 3.60300779e-01 -8.33414316e-01
6.63607776e-01 2.14356482e-02 7.50628769e-01 -7.11303711e-01
5.36759906e-02 5.57745159e-01 -2.11874098e-01 -1.89016327e-01
3.45241874e-01 -3.66664946e-01 -9.67371464e-02 5.57964385e-01
-1.32992476e-01 -5.85934818e-01 5.38402572e-02 1.85712595e-02
1.08710897e+00 4.24618334e-01 2.33364597e-01 -4.57331300e-01
1.10445231e-01 8.03093687e-02 5.47593176e-01 8.18414450e-01
-3.63783807e-01 3.39848429e-01 5.89297593e-01 -1.70946822e-01
-1.09294951e+00 -1.31068230e+00 -3.17777604e-01 6.39384627e-01
2.07008928e-01 -9.77170542e-02 -5.76775312e-01 -8.37534070e-01
4.24858302e-01 7.87645698e-01 -4.76012051e-01 -6.03213273e-02
-5.14229059e-01 -4.10226971e-01 2.92281806e-01 3.73595506e-01
1.44721866e-01 -7.09541380e-01 -8.49056780e-01 4.17361945e-01
2.63549656e-01 -4.65765566e-01 -6.95471466e-01 3.77758771e-01
-9.42644298e-01 -1.09958649e+00 -7.40323842e-01 -1.08070694e-01
6.55961156e-01 -2.88003772e-01 6.26544237e-01 -3.16220045e-01
2.17635289e-01 9.58360612e-01 -3.80484089e-02 -3.13345104e-01
-3.53854328e-01 -1.03158122e-02 5.50715387e-01 -4.59611744e-01
-3.56121987e-01 -7.52681613e-01 -6.31551147e-01 3.36401969e-01
-6.81702256e-01 -4.00445700e-01 3.95535022e-01 9.48150218e-01
7.38302708e-01 -2.85648644e-01 4.67896879e-01 -2.75336534e-01
9.08349514e-01 -3.46294284e-01 -1.31489754e+00 2.12495968e-01
-8.96006346e-01 4.74313170e-01 6.53831065e-01 -4.22819674e-01
-7.94283211e-01 6.79327324e-02 5.82774207e-02 -3.94563019e-01
2.85258323e-01 4.54601109e-01 2.36649588e-01 -2.70757705e-01
8.06627333e-01 1.32863835e-01 4.39174324e-01 -2.73800403e-01
5.79134881e-01 3.69745910e-01 2.49627337e-01 -1.16185701e+00
5.73992550e-01 5.55931807e-01 1.59733728e-01 -4.85662878e-01
-7.25047708e-01 -1.07822686e-01 -6.18549049e-01 -2.25012079e-01
6.49577379e-01 -5.05363047e-01 -1.01480544e+00 -2.52149142e-02
-7.85286665e-01 -9.45742428e-01 -7.39553094e-01 7.44170249e-01
-1.29441607e+00 2.96453178e-01 -2.98517257e-01 -1.22689843e+00
5.23364656e-02 -1.06349492e+00 7.59495378e-01 2.53633946e-01
-1.05269894e-01 -1.13678312e+00 2.44729295e-01 -1.76845580e-01
5.17485499e-01 3.78655046e-01 6.95901394e-01 -4.74462360e-01
-5.35700023e-01 -1.40064592e-02 1.47563472e-01 2.30964065e-01
1.34562686e-01 -1.69132262e-01 -5.16957760e-01 -5.69917560e-01
1.96858734e-01 -4.83706474e-01 5.26959300e-01 7.18697488e-01
8.94941151e-01 -5.62383056e-01 -9.66541693e-02 5.56431353e-01
1.34085202e+00 2.41069034e-01 2.24255696e-01 5.92486680e-01
2.91429430e-01 5.21062613e-01 7.78031290e-01 7.36148417e-01
3.16031814e-01 6.94535375e-01 5.02310693e-01 4.26221937e-01
3.08208108e-01 -7.48761654e-01 4.71947759e-01 5.52690089e-01
1.51580364e-01 2.18539238e-01 -9.10218000e-01 4.63118970e-01
-2.03470778e+00 -8.15891206e-01 4.15436715e-01 2.77913713e+00
1.09117591e+00 2.29629695e-01 3.00243527e-01 -4.26398844e-01
2.48418912e-01 -1.36992859e-03 -1.06700122e+00 -5.33912182e-01
2.22905532e-01 8.82610083e-02 9.38356042e-01 1.09409297e+00
-9.37868059e-01 7.49583602e-01 6.37342882e+00 8.38502526e-01
-9.24511731e-01 8.63111094e-02 4.31095690e-01 -2.23203868e-01
-5.49128652e-01 2.01962918e-01 -5.64422607e-01 5.67817748e-01
7.26347327e-01 -2.55548984e-01 8.32956731e-01 1.02251017e+00
4.60237235e-01 -5.78535020e-01 -1.00913143e+00 5.35807788e-01
-6.34907246e-01 -1.14828444e+00 -4.23534721e-01 3.25667411e-01
8.51767063e-01 8.03883374e-02 9.70539823e-02 2.92815953e-01
6.26694024e-01 -1.04333949e+00 8.10639322e-01 6.63088679e-01
6.01595163e-01 -8.20935965e-01 3.95713329e-01 5.85396171e-01
-8.89868438e-01 -2.60243177e-01 -4.45070863e-01 -8.53660554e-02
3.79854500e-01 6.46138847e-01 -8.07282329e-01 1.72026724e-01
1.83756068e-01 3.37921828e-01 1.71403348e-01 1.01827276e+00
-2.60215253e-01 2.52502143e-01 -8.58725250e-01 -3.59412313e-01
4.25186157e-01 -4.78456259e-01 7.52587974e-01 6.48813367e-01
5.05555809e-01 -3.79035063e-02 3.61891508e-01 1.06278813e+00
3.07094932e-01 -6.85611814e-02 -6.66113794e-01 -1.36637703e-01
5.04476130e-01 7.03137219e-01 -3.77495736e-01 1.07974991e-01
6.49592199e-04 6.19680882e-01 4.93222356e-01 3.95150930e-01
-5.65895319e-01 -1.59726545e-01 8.02591264e-01 -7.46756494e-02
1.61206082e-01 -8.53588641e-01 -3.74383271e-01 -9.18542087e-01
2.43111640e-01 -6.89858913e-01 1.82138026e-01 -2.92190999e-01
-9.57488000e-01 -8.84896666e-02 9.18491185e-02 -1.19047201e+00
-6.86455190e-01 -5.45703351e-01 -3.53252321e-01 6.90854847e-01
-1.27863038e+00 -5.67217052e-01 4.05061156e-01 5.20116627e-01
2.92597026e-01 1.02015391e-01 6.24325156e-01 -2.25006700e-01
-1.81531399e-01 4.27011847e-01 5.59173703e-01 -3.14490974e-01
4.99703079e-01 -1.53884554e+00 7.22141042e-02 6.03668094e-01
7.56091997e-02 5.81543922e-01 7.82467544e-01 -8.58622313e-01
-1.72458518e+00 -6.23114109e-01 2.15843156e-01 -7.14571893e-01
8.59983385e-01 -1.83583707e-01 -5.29161215e-01 5.36235094e-01
-4.62324977e-01 -2.26099454e-02 1.80461642e-03 8.38917643e-02
4.01827954e-02 -1.92226749e-02 -1.34265947e+00 7.30982780e-01
9.13365543e-01 -4.80155438e-01 -3.47516418e-01 3.27629805e-01
6.74004018e-01 -7.18310833e-01 -9.11094427e-01 4.18775439e-01
6.97463691e-01 -8.45480621e-01 6.57159507e-01 -5.44891775e-01
1.68225504e-02 -2.63803482e-01 -3.21409255e-01 -1.43562174e+00
7.65622109e-02 -1.04960763e+00 -3.85317981e-01 8.29041719e-01
5.47927201e-01 -1.02497816e+00 7.61079550e-01 9.00773764e-01
-1.15571141e-01 -1.17059457e+00 -1.28724980e+00 -1.30693710e+00
5.72678268e-01 -6.43036366e-01 3.45296651e-01 5.17526925e-01
4.06209260e-01 -3.49205494e-01 -2.04975024e-01 1.59632355e-01
6.55075133e-01 -1.35330260e-02 5.85447609e-01 -6.85813069e-01
-5.57339966e-01 -6.24156296e-01 -8.27402398e-02 -1.28368247e+00
3.40480171e-02 -5.75700343e-01 3.37993622e-01 -1.54126883e+00
-1.91135749e-01 -8.66200924e-01 -1.26106367e-01 3.06431651e-01
1.65328711e-01 -4.40149575e-01 2.83453077e-01 3.29626948e-01
-4.23734218e-01 9.80503976e-01 1.40701663e+00 2.41784841e-01
-6.28203213e-01 1.78661153e-01 -4.44696724e-01 6.42588854e-01
9.22873616e-01 -3.78747642e-01 -5.14609933e-01 -1.47412106e-01
3.62576753e-01 3.25711846e-01 2.79372334e-01 -9.01622176e-01
2.00994015e-01 -6.12583995e-01 1.52023926e-01 -8.95049572e-02
2.91581094e-01 -5.65316677e-01 -2.80789912e-01 5.97887456e-01
-4.75391507e-01 -8.41069594e-02 1.82077169e-01 7.08709061e-01
1.49660572e-01 -3.89036715e-01 7.40803421e-01 -1.16094232e-01
-3.97767216e-01 3.13398212e-01 -3.42519313e-01 3.17160636e-01
8.73853207e-01 9.87032950e-02 -1.57713249e-01 -7.03388453e-01
-5.53565145e-01 4.45626646e-01 7.65022099e-01 -9.02850330e-02
4.85436827e-01 -1.30855060e+00 -3.22620392e-01 -3.36799249e-02
-3.03019464e-01 8.31684992e-02 -2.51982212e-01 9.79091525e-01
-3.04079711e-01 4.79177654e-01 2.10035756e-01 -4.80480909e-01
-4.55818802e-01 4.46411729e-01 5.95508039e-01 -3.39247495e-01
-4.61457282e-01 8.06790829e-01 -2.08492354e-01 -8.34996402e-01
4.25048470e-01 -6.54428065e-01 4.46718127e-01 -1.39081970e-01
4.23900127e-01 3.08075726e-01 -1.72758654e-01 4.02478389e-02
-2.78661668e-01 6.39321923e-01 3.85386080e-01 -8.34051728e-01
1.14602709e+00 -1.38322011e-01 3.44789594e-01 3.62240165e-01
1.03212893e+00 1.28801083e-02 -2.12893772e+00 4.65212651e-02
2.42424995e-01 -5.08831918e-01 4.11087871e-02 -9.02886569e-01
-5.64110100e-01 6.60168290e-01 5.22492766e-01 1.52829170e-01
6.45626783e-01 -2.05856770e-01 5.49295545e-01 4.92720723e-01
7.91084647e-01 -1.44333768e+00 5.68423606e-02 7.61394620e-01
1.03545201e+00 -1.23096693e+00 7.21484050e-02 2.89460301e-01
-5.44245243e-01 1.09904373e+00 3.20568740e-01 -2.68120855e-01
6.24887824e-01 2.70480335e-01 -1.58803225e-01 1.44416898e-01
-6.57374799e-01 -4.39141095e-01 2.37976670e-01 5.87726295e-01
-4.60388027e-02 8.89528543e-02 -4.01670396e-01 2.48460382e-01
-2.91868657e-01 -3.71736400e-02 2.50261724e-01 1.12915325e+00
-7.44402945e-01 -1.17373860e+00 -2.88543999e-01 3.98763448e-01
-8.67794156e-02 1.16136096e-01 -8.17621872e-02 8.60999107e-01
-4.37910259e-01 7.69066274e-01 -1.22396469e-01 -1.26659468e-01
2.72073388e-01 3.43123786e-02 7.86758363e-01 -3.03951204e-01
-7.85713792e-02 -1.26708105e-01 3.64671350e-02 -9.79997396e-01
1.79071585e-03 -7.80578732e-01 -1.42549777e+00 -2.27189809e-01
-2.41065860e-01 2.97677994e-01 1.08770347e+00 1.02187777e+00
2.76796609e-01 -9.66662616e-02 7.15400457e-01 -9.73889649e-01
-1.38935983e+00 -5.86497962e-01 -5.90519845e-01 -1.07458152e-01
5.71814179e-01 -7.77167201e-01 -6.92605674e-01 -6.74635947e-01] | [4.163730144500732, 2.3679394721984863] |
17867915-b0f7-4805-9904-d38de380aa3e | spectral-graph-clustering-for-intentional | 2203.06579 | null | https://arxiv.org/abs/2203.06579v1 | https://arxiv.org/pdf/2203.06579v1.pdf | Spectral Graph Clustering for Intentional Islanding Operations in Resilient Hybrid Energy Systems | Establishing cleaner energy generation therefore improving the sustainability of the power system is a crucial task in this century, and one of the key strategies being pursued is to shift the dependence on fossil fuel to renewable technologies such as wind, solar, and nuclear. However, with the increasing number of heterogeneous components included, the complexity of the hybrid energy system becomes more significant. And the complex system imposes a more stringent requirement of the contingency plan to enhance the overall system resilience. Among different strategies to ensure a reliable system, intentional islanding is commonly applied in practical applications for power systems and attracts abundant interest in the literature. In this study, we propose a hierarchical spectral clustering-based intentional islanding strategy at the transmission level with renewable generations. To incorporate the renewable generation that relies on the inverter technology, the frequency measurements are considered to represent the transient response. And it has been further used as embedded information in the clustering algorithm along with other important electrical information from the system to enrich the modeling capability of the proposed framework. To demonstrate the effectiveness of the islanding strategy, the modified IEEE-9 bus and IEEE-118 bus systems coupled with wind farms are considered as the test cases. | ['Pingfeng Wang', 'Jie Zhang', 'Sobhan Badakhshan', 'Xin Chen', 'Jiaxin Wu'] | 2022-03-13 | null | null | null | null | ['graph-clustering', 'spectral-graph-clustering'] | ['graphs', 'graphs'] | [-7.77908489e-02 -2.78534710e-01 1.11449910e-02 3.51463348e-01
-6.44102693e-03 -7.35182881e-01 3.68588060e-01 1.44327283e-01
3.40280324e-01 1.12965357e+00 -1.73524275e-01 -8.43488574e-02
-5.98461390e-01 -9.68368649e-01 -7.80865252e-02 -1.33357859e+00
-1.43944934e-01 -1.84442595e-01 -1.78850368e-02 -5.14771163e-01
1.55359477e-01 3.85544807e-01 -1.69648576e+00 -3.60264421e-01
1.45805848e+00 7.32314527e-01 2.54348069e-01 8.51527080e-02
2.49852344e-01 5.60925975e-02 -8.62640321e-01 3.93087506e-01
1.91355810e-01 -8.07980478e-01 3.21531966e-02 -8.92783329e-02
-8.94999921e-01 -1.65995568e-01 1.03917368e-01 1.31762969e+00
6.99991286e-01 3.97573620e-01 5.26637316e-01 -1.34718347e+00
-2.03113141e-03 7.16511428e-01 -6.06716335e-01 2.90138751e-01
-1.85963772e-02 3.24660772e-03 7.74677575e-01 -6.23658478e-01
2.06825152e-01 5.79693913e-01 9.81486812e-02 -1.83208808e-01
-8.37089002e-01 -6.06063306e-01 -1.34352431e-01 6.39502466e-01
-1.50324762e+00 -2.22714599e-02 1.20299017e+00 -2.43169934e-01
7.16384590e-01 3.87765050e-01 1.26296651e+00 3.71880621e-01
2.41216779e-01 2.01606229e-01 1.27693725e+00 -4.84241217e-01
5.09800553e-01 -1.35750044e-02 1.45614907e-01 -4.03763680e-03
8.18973362e-01 2.50249263e-02 1.30432963e-01 8.43058676e-02
8.09698403e-02 -1.33304343e-01 -5.43490291e-01 -7.24930391e-02
-6.01741672e-01 5.25369704e-01 3.83807689e-01 9.29071546e-01
-1.23314179e-01 -3.32275480e-01 2.17079028e-01 -1.30096748e-01
2.71142364e-01 2.57281572e-01 1.32475570e-01 -2.59684455e-02
-9.83756483e-01 -2.47159585e-01 4.14480418e-01 6.41078353e-01
3.94976914e-01 8.97639692e-01 4.99220878e-01 4.66693372e-01
1.45680919e-01 7.53430486e-01 6.60040140e-01 -4.74887341e-01
1.11345284e-01 9.80098069e-01 9.96603146e-02 -9.08780515e-01
-4.60557252e-01 -1.13719773e+00 -1.10517859e+00 3.59218150e-01
-2.38668755e-01 -4.81400728e-01 -3.02299410e-01 1.37416208e+00
4.63099897e-01 6.55584261e-02 -9.10415023e-04 7.66376317e-01
1.48225412e-01 1.00938106e+00 -3.09618562e-01 -8.67699444e-01
1.05251288e+00 -3.73465925e-01 -1.01543295e+00 3.98069918e-01
3.35821450e-01 -5.50818920e-01 2.77678341e-01 2.65379399e-01
-8.48571360e-01 -5.31755030e-01 -1.58785295e+00 7.74163544e-01
-6.78852201e-01 2.38159746e-01 -6.56765625e-02 7.89064467e-01
-8.04175079e-01 4.46249515e-01 -6.56157553e-01 -5.15433431e-01
-1.22363575e-01 -1.22794230e-02 4.35855128e-02 1.88834101e-01
-1.50591397e+00 1.18015528e+00 9.13707197e-01 5.60118437e-01
-2.88359284e-01 -3.15584153e-01 -3.20778251e-01 3.92766297e-01
3.48768324e-01 -4.53997076e-01 3.85647446e-01 -7.02393591e-01
-1.44390047e+00 -4.55008119e-01 1.44161880e-01 -3.08625191e-01
2.42156461e-01 4.22901392e-01 -7.62232482e-01 3.23959947e-01
-1.23058781e-01 -4.80051726e-01 4.86773491e-01 -1.09495437e+00
-8.28718066e-01 -2.60662377e-01 -1.80618316e-01 5.02241731e-01
-6.31276727e-01 -4.28263724e-01 4.36987281e-01 -5.11719406e-01
-2.01155409e-01 -8.68620992e-01 -8.26174989e-02 -9.84571159e-01
-5.78628063e-01 -2.75776923e-01 1.11919999e+00 -8.46510708e-01
1.55216813e+00 -2.14889002e+00 5.23053885e-01 5.70108354e-01
-3.76786977e-01 2.59945780e-01 5.71136713e-01 1.13520825e+00
-1.07983284e-01 1.43602565e-01 -4.78347003e-01 3.34116936e-01
1.21904410e-01 2.28037879e-01 9.97330099e-02 3.19718599e-01
-9.86246616e-02 2.03359231e-01 -7.41865158e-01 -2.55691379e-01
6.43600762e-01 2.66618043e-01 -5.57557791e-02 -1.08607389e-01
2.57643819e-01 4.58538592e-01 -3.56293112e-01 4.15074199e-01
6.20817423e-01 2.81599090e-02 3.18670720e-01 -3.44295323e-01
-6.17346346e-01 -4.46792990e-01 -1.53817129e+00 1.26368976e+00
-3.56243819e-01 1.65847450e-01 3.68969053e-01 -1.53492773e+00
6.36344790e-01 5.31419218e-01 7.84738898e-01 -6.46570802e-01
-9.27975327e-02 3.60547334e-01 4.28474009e-01 -4.96361494e-01
7.07037896e-02 7.35509396e-02 1.41422123e-01 6.31417111e-02
-7.98386037e-02 -1.74618065e-01 8.13968480e-01 -1.04260541e-01
5.82540154e-01 1.46366447e-01 5.50521791e-01 -7.31441557e-01
9.30381596e-01 -2.55911648e-02 9.20721054e-01 -3.21531773e-01
2.06684813e-01 -1.30966857e-01 6.03717901e-02 2.33153746e-01
-7.39095151e-01 -8.54843080e-01 -3.83434743e-01 -2.95677427e-02
3.43306571e-01 4.47251387e-02 -5.97776115e-01 -2.90682912e-01
-1.08614199e-01 1.20004106e+00 -4.00663912e-02 -3.81382942e-01
-3.79677117e-01 -1.27537191e+00 -2.41150465e-02 1.09283067e-01
9.11849856e-01 -4.25074100e-01 -7.21926093e-01 3.67448390e-01
-1.67346045e-01 -5.56888163e-01 9.05522779e-02 2.64090776e-01
-5.97800016e-01 -1.18026507e+00 -6.54754400e-01 -5.94775140e-01
7.05753028e-01 6.41284883e-02 5.22009373e-01 2.87805140e-01
-3.02835345e-01 4.43366636e-03 -5.39701641e-01 -8.94156769e-02
-3.17138225e-01 1.58536091e-01 3.78666729e-01 -5.66683821e-02
-4.19712991e-01 -8.65269542e-01 -6.99410677e-01 1.69681832e-01
-8.39754224e-01 1.04881950e-01 5.25589943e-01 9.29781020e-01
1.14674874e-01 1.39178431e+00 1.45962703e+00 -2.71629602e-01
5.17181754e-01 -8.81925941e-01 -9.00646210e-01 2.48859003e-01
-8.29052866e-01 -3.63643706e-01 1.27663863e+00 6.07536957e-02
-1.27855945e+00 -2.38247782e-01 -2.84369253e-02 1.77666575e-01
-4.81460057e-02 8.38733315e-01 -5.92563927e-01 6.33336306e-02
-2.45739907e-01 4.28941220e-01 -2.58695424e-01 -3.05768400e-01
1.58051580e-01 5.83422899e-01 8.12433586e-02 1.65329389e-02
1.31080091e+00 1.76904976e-01 5.56532264e-01 -1.15518773e+00
-6.00495422e-03 -1.83610931e-01 -2.35204935e-01 -5.17473519e-01
6.63492799e-01 -5.90184629e-01 -6.82878733e-01 4.87663895e-01
-7.79503703e-01 2.23263651e-01 -1.89194411e-01 7.47097313e-01
9.09861624e-02 4.94832009e-01 -5.15025854e-02 -1.04170942e+00
-4.05393720e-01 -1.22591317e+00 5.11340462e-02 5.31594694e-01
2.38220155e-01 -1.14055181e+00 -1.75532158e-02 -8.73125419e-02
4.59111691e-01 7.64987767e-01 1.20782936e+00 -3.46145153e-01
-6.29632115e-01 -1.95497647e-03 3.50889742e-01 6.13401592e-01
6.15997970e-01 3.62137228e-01 -4.53233570e-01 -5.14828861e-01
2.72410542e-01 4.39362019e-01 -3.34201567e-02 1.68327913e-01
5.52672446e-01 -2.73266494e-01 -3.35710734e-01 2.71970868e-01
2.03674960e+00 8.98093760e-01 3.30637902e-01 2.12657720e-01
3.11829209e-01 5.01311541e-01 6.08740032e-01 5.75193107e-01
3.26454282e-01 4.61988658e-01 6.43624842e-01 -7.51315802e-02
3.11069071e-01 2.82982022e-01 3.62003386e-01 1.64356780e+00
-2.13306934e-01 -4.01355803e-01 -4.02397960e-01 5.02836883e-01
-1.72460568e+00 -1.16729701e+00 -1.44602939e-01 2.14342999e+00
3.88553381e-01 -7.53375217e-02 -2.13175923e-01 9.20302749e-01
9.99026239e-01 5.45249134e-02 -4.97226566e-01 -1.68313116e-01
-3.30776781e-01 1.19154807e-02 1.70220792e-01 1.43123552e-01
-5.60516357e-01 -2.96794683e-01 4.28563547e+00 1.10355806e+00
-1.05793726e+00 -7.90818706e-02 3.20421398e-01 1.12652630e-01
-2.00306475e-01 -2.42898823e-03 -4.43254679e-01 1.22095490e+00
1.01379132e+00 -9.34291303e-01 5.53357899e-01 4.88567322e-01
8.67165565e-01 -6.28267825e-01 -4.95198220e-01 3.92023563e-01
-1.09279230e-01 -6.86259389e-01 -1.93623319e-01 4.77340259e-02
1.06977963e+00 -2.95362502e-01 -3.76610279e-01 2.01740358e-02
-1.34787396e-01 -3.11674833e-01 4.73291725e-01 4.36746806e-01
3.23955804e-01 -1.19304216e+00 1.14737368e+00 6.22423172e-01
-1.55614793e+00 -3.34633678e-01 1.58878565e-02 -6.13416024e-02
5.77202618e-01 9.39738512e-01 -3.16226184e-01 1.53636444e+00
7.82541096e-01 6.67838216e-01 -3.98473620e-01 1.14843905e+00
-3.80127728e-01 7.52171040e-01 -5.96304119e-01 -1.34534717e-01
-1.99700311e-01 -8.29869032e-01 5.33049762e-01 4.27419990e-01
8.41776252e-01 1.54085487e-01 2.24183097e-01 4.69976753e-01
2.99762487e-01 2.02296630e-01 -5.85967541e-01 5.43597043e-02
9.36119914e-01 1.61404979e+00 -8.49230587e-01 -4.18684751e-01
-3.92102331e-01 3.09484273e-01 -5.22385180e-01 5.10154188e-01
-9.68726993e-01 -8.05557489e-01 1.85307473e-01 -1.89165398e-01
-6.53159171e-02 -1.90183967e-01 -2.39819765e-01 -9.67272401e-01
2.33297721e-02 -5.68683624e-01 1.46423742e-01 -6.09195709e-01
-1.16967487e+00 2.80485094e-01 1.78548381e-01 -1.65057778e+00
-5.21418810e-01 -1.37278782e-02 -1.00099337e+00 1.20295334e+00
-1.59600556e+00 -7.03429401e-01 -4.15068567e-01 4.36224073e-01
5.22745132e-01 1.17941955e-02 6.11435473e-01 6.15100622e-01
-1.18148088e+00 -1.12546019e-01 9.09610450e-01 -3.19974929e-01
-1.13859084e-02 -1.15539968e+00 -5.32995582e-01 1.40807927e+00
-4.21823114e-01 3.22355032e-01 6.30211473e-01 -5.49052238e-01
-1.54236746e+00 -6.26665175e-01 2.32170060e-01 6.68551803e-01
7.27938533e-01 -2.55345870e-02 -7.09583580e-01 1.50876209e-01
1.22134364e+00 -4.03150767e-01 4.64864790e-01 -7.36659467e-01
7.07247913e-01 -4.14972723e-01 -1.23387206e+00 5.84263086e-01
3.43286037e-01 -7.80061781e-02 -4.48944002e-01 6.59144148e-02
2.27970660e-01 1.04637742e-02 -1.17011285e+00 5.59977293e-01
2.62676477e-01 -7.73871303e-01 5.09859145e-01 3.33401531e-01
-2.02280372e-01 -9.43764269e-01 2.17914879e-01 -2.18340707e+00
-2.26899579e-01 -5.11637926e-01 6.41732337e-03 1.56623471e+00
1.97117284e-01 -1.28531516e+00 2.26966262e-01 -6.36465028e-02
-2.45453462e-01 -6.14355206e-01 -1.31751943e+00 -8.01761150e-01
-1.37643397e-01 5.72405934e-01 8.44174922e-01 1.06783116e+00
6.72717333e-01 2.10316226e-01 1.55695587e-01 6.72161579e-01
9.10125494e-01 2.92846799e-01 2.40643948e-01 -1.16600811e+00
3.91853116e-02 -4.63523090e-01 -5.81875406e-02 -1.47804126e-01
-2.43061796e-01 -6.49205387e-01 -2.48407856e-01 -2.03030992e+00
-4.22281176e-01 -1.43222377e-01 -3.72896105e-01 6.16701618e-02
-2.12776542e-01 4.07178812e-02 1.69126913e-01 1.45765081e-01
2.12099209e-01 9.91571963e-01 8.34945261e-01 -7.33682737e-02
9.64283869e-02 1.33974016e-01 1.36963263e-01 5.98112643e-01
1.14855146e+00 4.44123670e-02 -8.28251362e-01 2.96783689e-02
3.36719640e-02 4.56065714e-01 -2.44864404e-01 -1.46756625e+00
2.89782792e-01 -1.38154134e-01 1.74784124e-01 -8.12876463e-01
-2.31082924e-02 -1.59749544e+00 8.08615565e-01 8.94257963e-01
4.68484312e-01 3.67007315e-01 -8.01969599e-03 4.20957595e-01
-4.20478433e-01 -3.14070165e-01 6.13880932e-01 1.39451787e-01
-5.02070248e-01 -2.72973955e-01 -5.97240150e-01 -3.34973454e-01
1.61503410e+00 -1.74479812e-01 -7.75363624e-01 -6.08581305e-02
-5.28701603e-01 7.20787168e-01 3.99801493e-01 2.59316117e-01
1.70082867e-01 -1.08634567e+00 -5.24256587e-01 -1.95033140e-02
-2.48396531e-01 -2.99884349e-01 4.72647816e-01 8.83750141e-01
-4.21749413e-01 3.30186903e-01 -3.10917556e-01 -5.26035905e-01
-8.84054244e-01 3.71042937e-01 2.66378969e-01 -9.92095768e-02
-2.41727307e-01 -3.46593529e-01 -4.83293444e-01 2.83804357e-01
-3.75545233e-01 -1.89808786e-01 -5.73203921e-01 5.55141807e-01
-3.25993150e-02 1.02669787e+00 3.18518192e-01 -4.17542994e-01
-3.02675605e-01 5.36376595e-01 8.16063762e-01 -2.04661787e-02
1.09998178e+00 -5.15326619e-01 -4.26957995e-01 6.08362854e-01
7.55805492e-01 2.29469880e-01 -8.02846074e-01 5.60855150e-01
-1.37187898e-01 7.70736635e-02 7.00568780e-02 -8.34926844e-01
-1.08052838e+00 4.75408614e-01 4.81548280e-01 1.11926162e+00
1.35112298e+00 -1.01491261e+00 3.17196786e-01 -1.13811698e-02
9.44843948e-01 -1.40903544e+00 -4.61654216e-01 -1.14604510e-01
4.47383940e-01 -6.21946037e-01 2.16830790e-01 -2.97233999e-01
5.40738413e-03 1.07008195e+00 3.12429130e-01 -1.80382922e-01
7.39112616e-01 3.75263691e-01 -3.85279536e-01 4.46732670e-01
-3.61200839e-01 -2.00372607e-01 -2.16319129e-01 2.77644455e-01
2.06535831e-01 -8.61635245e-03 -1.08132315e+00 5.25300801e-01
7.03815371e-02 -5.89043312e-02 7.62937844e-01 9.30070460e-01
-3.76064330e-01 -1.00039446e+00 -5.44868529e-01 4.08947587e-01
-4.04485881e-01 2.81364739e-01 4.73867446e-01 9.58843350e-01
5.84144235e-01 1.36075854e+00 -2.48913497e-01 -4.96889651e-02
3.91803294e-01 1.96181893e-01 -2.27350816e-02 -3.08650225e-01
-5.75179636e-01 4.44346815e-02 -7.50005171e-02 1.31324068e-01
-5.91578424e-01 -5.97615361e-01 -1.56750178e+00 -1.72874808e-01
-9.70491648e-01 1.04544795e+00 9.44840789e-01 8.93854320e-01
2.46953115e-01 9.88524854e-01 1.29446638e+00 -7.13330925e-01
-5.39940000e-01 -1.02786076e+00 -9.45228219e-01 -1.13133624e-01
-1.62459567e-01 -7.58145452e-01 -9.48195696e-01 -3.07293236e-01] | [5.7233195304870605, 2.5720808506011963] |
88e8744d-0e74-429e-b033-07466e80293d | reliable-natural-language-understanding-with | 2302.03780 | null | https://arxiv.org/abs/2302.03780v2 | https://arxiv.org/pdf/2302.03780v2.pdf | Reliable Natural Language Understanding with Large Language Models and Answer Set Programming | Humans understand language by extracting information (meaning) from sentences, combining it with existing commonsense knowledge, and then performing reasoning to draw conclusions. While large language models (LLMs) such as GPT-3 and ChatGPT are able to leverage patterns in the text to solve a variety of NLP tasks, they fall short in problems that require reasoning. They also cannot reliably explain the answers generated for a given question. In order to emulate humans better, we propose STAR, a framework that combines LLMs with Answer Set Programming (ASP). We show how LLMs can be used to effectively extract knowledge -- represented as predicates -- from language. Goal-directed ASP is then employed to reliably reason over this knowledge. We apply the STAR framework to three different NLU tasks requiring reasoning: qualitative reasoning, mathematical reasoning, and goal-directed conversation. Our experiments reveal that STAR is able to bridge the gap of reasoning in NLU tasks, leading to significant performance improvements, especially for smaller LLMs, i.e., LLMs with a smaller number of parameters. NLU applications developed using the STAR framework are also explainable: along with the predicates generated, a justification in the form of a proof tree can be produced for a given output. | ['Gopal Gupta', 'Parth Padalkar', 'Yankai Zeng', 'Abhiramon Rajasekharan'] | 2023-02-07 | null | null | null | null | ['mathematical-reasoning'] | ['natural-language-processing'] | [ 2.75595635e-01 9.26639795e-01 -2.21900821e-01 -3.69349569e-01
-7.87822902e-01 -7.42171228e-01 6.31845772e-01 5.20152390e-01
2.63287693e-01 8.00003111e-01 2.55756140e-01 -9.25130546e-01
-2.34047294e-01 -1.29781544e+00 -8.98971498e-01 2.17267826e-01
2.28041053e-01 6.25954211e-01 2.49479696e-01 -4.32943970e-01
3.70841622e-01 6.32202774e-02 -1.48738778e+00 6.67812586e-01
1.26670957e+00 7.41243362e-01 1.11894093e-01 5.96811950e-01
-7.75924742e-01 1.71724641e+00 -5.24502993e-01 -6.54187262e-01
-1.45278603e-01 -5.69585562e-01 -1.41671443e+00 -3.16941798e-01
7.46325701e-02 -3.23518574e-01 3.54922488e-02 8.49355042e-01
-1.93775207e-01 3.38117518e-02 5.71082950e-01 -1.51308179e+00
-6.42264307e-01 1.26929879e+00 1.42449260e-01 -1.60012648e-01
1.03032362e+00 3.16084206e-01 1.49071527e+00 -4.92629439e-01
6.72727227e-01 1.81186926e+00 3.88166070e-01 8.92842829e-01
-1.19882905e+00 -2.38509789e-01 2.02725027e-02 4.05185729e-01
-7.57379413e-01 -2.73141116e-01 6.20289683e-01 -2.21127555e-01
1.47550046e+00 5.45662165e-01 5.12218535e-01 8.02338243e-01
2.55228609e-01 9.36227679e-01 1.08642745e+00 -6.79505229e-01
3.63635659e-01 2.42330268e-01 5.86202085e-01 1.01246428e+00
3.25506032e-01 -2.83322752e-01 -7.41518497e-01 -2.40520522e-01
4.13588405e-01 -2.83156037e-01 -2.60123461e-01 -2.50931680e-02
-1.18640101e+00 9.92766440e-01 2.18834460e-01 2.68785208e-01
-2.54808187e-01 3.42704743e-01 3.26495290e-01 4.58026588e-01
1.78738192e-01 8.06363940e-01 -5.42182148e-01 -1.97622731e-01
-5.34275293e-01 5.64952433e-01 1.69725466e+00 1.08060479e+00
6.68616891e-01 -3.41822058e-01 -1.98347732e-01 4.62283909e-01
5.15622735e-01 5.78111351e-01 2.38064289e-01 -1.43247712e+00
7.42368758e-01 1.09683430e+00 2.05568671e-01 -9.64700878e-01
-2.17537254e-01 4.37042303e-02 -1.20692849e-01 -6.23855218e-02
4.36997533e-01 -2.10820705e-01 -4.47082341e-01 1.82551551e+00
2.41666391e-01 -7.02382401e-02 6.13919914e-01 6.56859279e-01
1.00926447e+00 8.19058776e-01 8.93872306e-02 -5.85980713e-02
1.49250245e+00 -7.64169633e-01 -6.28080130e-01 -6.55669451e-01
1.07062078e+00 -1.35600520e-02 1.22332931e+00 3.27158988e-01
-1.17928910e+00 2.22435128e-02 -9.90341723e-01 -5.39932072e-01
-2.49228463e-01 -4.39170241e-01 8.98226142e-01 1.85897142e-01
-9.53310847e-01 5.76204419e-01 -4.51044768e-01 -2.06473202e-01
4.56423372e-01 1.84949473e-01 -7.54278749e-02 -4.04946595e-01
-1.35970294e+00 1.10694265e+00 7.44823873e-01 -2.08174825e-01
-5.00863016e-01 -8.82808924e-01 -1.28859735e+00 4.26127523e-01
8.78404081e-01 -1.12184370e+00 1.54342520e+00 -5.32095313e-01
-1.42610049e+00 7.77224779e-01 -3.21575820e-01 -9.16147590e-01
3.66075635e-01 -1.32154822e-01 3.26238796e-02 1.87626377e-01
2.77687877e-01 6.87556505e-01 2.44944304e-01 -1.08386683e+00
-4.43006217e-01 -2.79035360e-01 8.23838115e-01 -6.20509163e-02
1.15014695e-01 1.24122716e-01 -1.53314751e-02 -5.00099994e-02
1.30389154e-01 -7.27586687e-01 -9.19403955e-02 -9.36750397e-02
-5.83782256e-01 -6.60406590e-01 5.27392924e-01 -5.59567451e-01
1.24248421e+00 -1.58001351e+00 3.12260538e-01 1.99126765e-01
4.66964483e-01 2.24886555e-02 8.91208500e-02 5.18337131e-01
3.00034523e-01 4.71612245e-01 -3.61572504e-01 -2.22603399e-02
5.45552075e-01 7.85467863e-01 -6.83705628e-01 -4.90431696e-01
4.11269248e-01 1.37153780e+00 -1.08502030e+00 -7.03880727e-01
1.00614764e-01 -2.49363080e-01 -8.49119008e-01 2.60963768e-01
-1.25405002e+00 -5.14377616e-02 -6.92961395e-01 5.08337379e-01
2.40717977e-01 -5.48314691e-01 3.47642452e-01 4.05868202e-01
2.63374090e-01 7.79465854e-01 -9.63297844e-01 1.53216493e+00
-7.65285611e-01 6.75422847e-01 -1.85231626e-01 -8.37499499e-01
8.73176754e-01 2.27936000e-01 -2.81539410e-01 -2.56636560e-01
4.02458161e-02 2.82196313e-01 1.34235118e-02 -8.61914277e-01
1.13719478e-01 -4.01116818e-01 -1.18499495e-01 8.63311589e-01
-6.40274435e-02 -7.30435371e-01 5.61034739e-01 5.47611475e-01
1.26555860e+00 1.26034722e-01 5.88059604e-01 -1.95743397e-01
6.93945467e-01 6.12070024e-01 3.63359779e-01 9.53788698e-01
2.47132704e-01 -2.07859185e-02 1.08339560e+00 -5.84164619e-01
-7.33299851e-01 -8.34150791e-01 2.64696002e-01 8.64528775e-01
3.50825265e-02 -8.35887253e-01 -8.34574044e-01 -7.06307113e-01
1.29113242e-01 1.51635695e+00 -1.40451819e-01 1.27870892e-03
-5.67888260e-01 -1.68722123e-01 6.74184144e-01 4.91394907e-01
4.75588441e-01 -1.56822538e+00 -9.76665914e-01 2.35966012e-01
-6.68169677e-01 -1.35633349e+00 4.21237171e-01 1.97345480e-01
-9.00051117e-01 -1.33522260e+00 1.89557046e-01 -5.38585722e-01
5.94407082e-01 -5.56409359e-02 1.39224017e+00 4.12360102e-01
1.07333720e-01 4.80552375e-01 -3.89144748e-01 -7.74191618e-01
-9.84217465e-01 -2.58454949e-01 -2.95818210e-01 -6.58858180e-01
6.24668479e-01 -6.24300957e-01 4.93178591e-02 -1.44084051e-01
-9.55832720e-01 4.62668687e-01 2.69009650e-01 5.17974913e-01
1.41512424e-01 3.28790024e-02 3.98747534e-01 -1.10246682e+00
9.71633613e-01 -5.87901890e-01 -3.68624032e-01 6.09274626e-01
-4.32759374e-01 6.53597355e-01 8.12768519e-01 -1.55418247e-01
-9.83578205e-01 -6.03383005e-01 7.08819404e-02 4.35152687e-02
-1.21634446e-01 9.27843511e-01 -1.46609262e-01 3.23755413e-01
8.60957205e-01 1.09400138e-01 6.28634468e-02 -7.72588179e-02
5.99288464e-01 5.16458690e-01 5.51084697e-01 -1.28678465e+00
7.59840190e-01 2.25107327e-01 8.61860290e-02 -4.05368567e-01
-1.23444057e+00 -4.28789183e-02 -1.72674388e-01 1.63054496e-01
6.77733481e-01 -5.66443741e-01 -9.11685348e-01 -1.99797601e-01
-1.52030778e+00 -5.83770990e-01 -3.39558870e-01 1.73443392e-01
-8.62460494e-01 4.13673252e-01 -6.52147353e-01 -9.95068133e-01
-5.34998298e-01 -8.27585757e-01 8.23669732e-01 1.28705040e-01
-8.01969409e-01 -1.06073439e+00 -2.77821094e-01 6.56420827e-01
1.93207249e-01 2.12809920e-01 1.70591664e+00 -9.96835291e-01
-6.75363779e-01 2.35481858e-02 -2.45946184e-01 2.25020587e-01
-2.09139660e-01 -1.26006722e-01 -8.02307129e-01 5.31083941e-01
8.55225101e-02 -5.13578534e-01 3.58950436e-01 8.59484971e-02
1.02390873e+00 -7.93253481e-01 -2.06981733e-01 5.87248020e-02
1.02794600e+00 -6.65821061e-02 6.69636190e-01 3.72415185e-01
2.28343219e-01 8.76737475e-01 4.86635506e-01 2.19707116e-01
7.16305971e-01 1.73266932e-01 2.67632931e-01 5.29217124e-01
1.43862382e-01 -6.49233222e-01 3.34340364e-01 3.55506867e-01
-7.46796792e-03 -5.06419167e-02 -1.27759194e+00 3.54896098e-01
-2.22146726e+00 -9.99904633e-01 -2.73019671e-01 1.54765725e+00
1.22934091e+00 3.06195170e-01 -2.44903326e-01 2.46541142e-01
2.11833492e-01 -1.00239664e-01 -4.79887038e-01 -8.30912769e-01
-2.54331920e-02 3.20077807e-01 -2.09000930e-01 9.07741606e-01
-4.12935376e-01 1.12457466e+00 6.03276825e+00 5.96986353e-01
-5.39920688e-01 -2.42146045e-01 1.63659066e-01 2.05309853e-01
-8.74687672e-01 5.27566433e-01 -4.51776415e-01 -7.90197030e-02
9.99239504e-01 -4.30036485e-01 6.23276949e-01 9.49008465e-01
1.25370353e-01 -2.26099953e-01 -1.60424197e+00 7.38132358e-01
1.80838462e-02 -1.63558757e+00 3.86502385e-01 -2.19404235e-01
3.73951018e-01 -4.12375242e-01 -5.17799437e-01 5.71635187e-01
5.69637001e-01 -1.10536277e+00 8.09570432e-01 4.82704550e-01
2.42670700e-01 -4.57310945e-01 7.74009824e-01 1.04896760e+00
-6.06956601e-01 -3.76068383e-01 -2.78158963e-01 -7.16782868e-01
-6.31961301e-02 5.55875540e-01 -1.03787816e+00 5.91686785e-01
2.01400369e-01 5.06183743e-01 -3.01477045e-01 4.92486745e-01
-1.06981242e+00 4.75410968e-01 -3.18796337e-01 -5.35305858e-01
1.13766618e-01 -1.57748401e-01 5.59721828e-01 1.05702412e+00
1.53091103e-01 5.36754429e-01 5.46623878e-02 1.65193474e+00
-4.46094386e-02 -1.15274705e-01 -6.93692803e-01 -2.86592185e-01
4.54765797e-01 9.55455124e-01 -3.54006082e-01 -7.30881929e-01
-2.39558384e-01 6.57149136e-01 4.19975758e-01 3.11209738e-01
-6.86060548e-01 -2.97125489e-01 2.82254159e-01 -1.06676303e-01
-1.69782534e-01 -1.82999000e-01 -4.96461868e-01 -1.25154102e+00
2.12106869e-01 -1.17839742e+00 3.79052937e-01 -1.40976274e+00
-1.04978859e+00 2.78663665e-01 4.38287377e-01 -4.96743530e-01
-7.91756928e-01 -8.51087868e-01 -7.04721272e-01 8.72301042e-01
-1.58854723e+00 -1.10697496e+00 -1.42479256e-01 2.90514916e-01
6.23874187e-01 3.68351251e-01 9.47447598e-01 -5.94017565e-01
-1.79540291e-01 -1.47259325e-01 -8.72904360e-01 9.09479335e-02
1.64060012e-01 -1.48540199e+00 3.52727145e-01 6.91602886e-01
2.27438256e-01 1.10211921e+00 8.85809183e-01 -6.23735487e-01
-1.79412794e+00 -6.70622647e-01 1.43426859e+00 -8.53091478e-01
1.00794125e+00 -2.56113380e-01 -9.34034705e-01 9.58223104e-01
2.09886841e-02 -2.45827287e-01 6.40842021e-01 3.01962405e-01
-6.91904604e-01 3.42209786e-01 -1.17591095e+00 8.42300713e-01
1.05199933e+00 -7.08301544e-01 -1.47949541e+00 4.60424393e-01
1.13495827e+00 -5.35833597e-01 -7.12468088e-01 4.62018214e-02
4.31493402e-01 -8.05710495e-01 6.06098652e-01 -1.07872486e+00
1.12965965e+00 -3.36887866e-01 -3.20200354e-01 -1.15464699e+00
1.68495059e-01 -7.75918841e-01 -5.64851165e-01 9.82144654e-01
8.16531658e-01 -6.91035748e-01 3.76219034e-01 1.30564845e+00
3.78695317e-02 -8.78136039e-01 -6.24891818e-01 -5.15800178e-01
9.71453264e-02 -1.04138160e+00 7.31059730e-01 5.85895121e-01
8.86736929e-01 7.18163013e-01 1.61561459e-01 8.73904675e-02
5.23258209e-01 6.01632774e-01 8.80173504e-01 -1.39137280e+00
-4.10132259e-01 -3.03521007e-01 1.06051251e-01 -9.53298867e-01
6.63237929e-01 -1.14812160e+00 6.46542683e-02 -2.11237121e+00
3.09517413e-01 -2.88445741e-01 3.59144509e-01 9.84731793e-01
-1.49346769e-01 -5.26292980e-01 4.61531967e-01 9.64976773e-02
-5.41804075e-01 2.69014537e-02 1.22953653e+00 -2.54381299e-01
-1.57683760e-01 -2.15716779e-01 -1.05108786e+00 1.02949142e+00
7.29981363e-01 -3.27152520e-01 -4.88977551e-01 -2.56444722e-01
8.24958205e-01 3.70534152e-01 6.55616581e-01 -7.45667458e-01
2.90810347e-01 -5.76874316e-01 -2.33198568e-01 -7.57939816e-02
6.06281625e-04 -6.18131042e-01 -3.09332520e-01 5.24656534e-01
-6.63974524e-01 -2.17117146e-01 3.27483356e-01 2.89540738e-01
-1.09803431e-01 -6.20387435e-01 5.90404831e-02 -3.83626640e-01
-7.34526873e-01 -3.35685521e-01 -2.99414426e-01 3.36364567e-01
7.73584962e-01 1.41100558e-02 -6.55259430e-01 -5.72125018e-01
-5.69774210e-01 6.30793571e-01 1.62840426e-01 1.39060497e-01
7.22384632e-01 -8.59122813e-01 -6.51126981e-01 -1.92137897e-01
1.29523993e-01 2.54481852e-01 -1.50949135e-01 7.10867941e-01
-5.96974671e-01 7.15195417e-01 8.97201970e-02 -3.02083015e-01
-1.07567680e+00 5.35626709e-01 3.55208874e-01 -4.92353976e-01
-7.45386004e-01 5.23749948e-01 8.78334939e-02 -6.77903593e-01
2.60434859e-02 -1.03483891e+00 -5.26398458e-02 -3.48611683e-01
6.22139513e-01 3.40582803e-02 -1.77116632e-01 1.17393121e-01
-3.79208386e-01 2.00042695e-01 2.95895129e-01 -2.24268630e-01
1.33100951e+00 1.30795941e-01 -5.55974722e-01 3.57923031e-01
5.18679202e-01 3.55851725e-02 -6.17546499e-01 -3.78710717e-01
3.45348924e-01 -7.36457407e-02 -3.85210842e-01 -1.15999734e+00
-9.16912258e-02 8.73032272e-01 -7.59363890e-01 5.42792201e-01
7.56627679e-01 2.97924280e-01 9.06710804e-01 1.01155889e+00
6.43560648e-01 -7.17576802e-01 -2.37757806e-02 7.24708140e-01
9.95188594e-01 -1.14964819e+00 -1.39268816e-01 -7.15418637e-01
-6.26079202e-01 1.43702245e+00 4.28990453e-01 1.75735265e-01
-9.77744758e-02 4.32305008e-01 -1.50231913e-01 -3.98382574e-01
-1.14017928e+00 -1.15689702e-01 8.43837485e-02 5.05029142e-01
2.10597172e-01 1.22062474e-01 -2.07209185e-01 9.05300140e-01
-5.96193671e-01 3.15975279e-01 6.43595755e-01 1.06328416e+00
-7.83801675e-01 -1.09606123e+00 -5.05539596e-01 3.64108264e-01
-2.57433176e-01 -3.02927226e-01 -8.62831295e-01 7.75930524e-01
-7.36788809e-02 1.41857171e+00 -4.55624431e-01 -1.81788430e-01
9.52107981e-02 4.50211138e-01 6.74408793e-01 -9.67052698e-01
-4.77079570e-01 -6.91183388e-01 7.02141225e-01 -6.51320517e-01
-3.46293271e-01 -2.33655587e-01 -1.88194656e+00 -5.62368274e-01
-2.26996075e-02 4.42093223e-01 2.79780746e-01 1.50098276e+00
1.60313085e-01 3.67048889e-01 1.63860470e-02 -4.79815453e-02
-7.71486163e-01 -6.66378379e-01 -1.01139434e-01 2.98011750e-01
7.51924366e-02 -1.75854191e-01 -4.68552500e-01 1.85921080e-02] | [9.406176567077637, 7.29893684387207] |
dcbf959b-831b-49be-b32f-8075d90e1c38 | unsupervised-term-extraction-for-highly | 2210.13118 | null | https://arxiv.org/abs/2210.13118v1 | https://arxiv.org/pdf/2210.13118v1.pdf | Unsupervised Term Extraction for Highly Technical Domains | Term extraction is an information extraction task at the root of knowledge discovery platforms. Developing term extractors that are able to generalize across very diverse and potentially highly technical domains is challenging, as annotations for domains requiring in-depth expertise are scarce and expensive to obtain. In this paper, we describe the term extraction subsystem of a commercial knowledge discovery platform that targets highly technical fields such as pharma, medical, and material science. To be able to generalize across domains, we introduce a fully unsupervised annotator (UA). It extracts terms by combining novel morphological signals from sub-word tokenization with term-to-topic and intra-term similarity metrics, computed using general-domain pre-trained sentence-encoders. The annotator is used to implement a weakly-supervised setup, where transformer-models are fine-tuned (or pre-trained) over the training data generated by running the UA over large unlabeled corpora. Our experiments demonstrate that our setup can improve the predictive performance while decreasing the inference latency on both CPUs and GPUs. Our annotators provide a very competitive baseline for all the cases where annotations are not available. | ['Diego Antognini', 'Peter Staar', 'Francesco Fusco'] | 2022-10-24 | null | null | null | null | ['term-extraction'] | ['natural-language-processing'] | [ 4.82157290e-01 -1.97186563e-02 -3.96569520e-01 -3.56618345e-01
-1.35310960e+00 -1.01989615e+00 4.26233947e-01 6.62084520e-01
-6.13114357e-01 7.49410152e-01 -8.14320743e-02 -6.92875743e-01
1.06352665e-01 -5.66980898e-01 -7.63879299e-01 -5.04577100e-01
-7.48624876e-02 7.17791617e-01 1.38161346e-01 -2.78802104e-02
8.40246230e-02 3.67463946e-01 -1.18478310e+00 5.08843839e-01
9.57146704e-01 8.17988455e-01 3.11497509e-01 6.22956216e-01
-4.04004514e-01 3.17979842e-01 -5.49251080e-01 -5.18558264e-01
1.04164712e-01 2.81301979e-02 -1.01934600e+00 -2.76201487e-01
2.51670986e-01 9.44808200e-02 1.29691511e-01 1.02210200e+00
5.23437440e-01 -1.06168702e-01 5.62034607e-01 -5.97874999e-01
-4.22952265e-01 6.88624144e-01 -4.28595930e-01 2.01439038e-01
2.47273117e-01 4.27533686e-02 1.14968252e+00 -8.92437398e-01
9.92288351e-01 8.14171731e-01 5.97166359e-01 5.01625955e-01
-1.18639922e+00 -5.65257192e-01 2.66150355e-01 -5.05478214e-03
-1.47289932e+00 -3.73690784e-01 3.54025632e-01 -5.10138750e-01
1.37035191e+00 2.68288374e-01 1.04873121e-01 1.25891030e+00
1.36483938e-01 6.61657751e-01 7.33245134e-01 -5.23053706e-01
3.65203679e-01 4.84399468e-01 1.14629880e-01 7.55422831e-01
3.10177505e-01 -4.62791860e-01 -5.97059846e-01 -5.02656162e-01
3.24328125e-01 -2.15091869e-01 -2.66078174e-01 -1.01364389e-01
-1.20651412e+00 7.26043165e-01 1.89242698e-02 5.57861030e-01
-4.69180793e-01 -9.42760259e-02 8.42421114e-01 2.66861200e-01
8.61516118e-01 8.07283938e-01 -1.01908016e+00 -1.29945889e-01
-1.19322097e+00 -2.83329543e-02 9.92533386e-01 1.11134207e+00
4.59801465e-01 -5.53078890e-01 -4.47958201e-01 6.78803980e-01
-1.48703948e-01 3.07542235e-01 7.08421826e-01 -4.04869825e-01
3.98814708e-01 4.42057908e-01 -3.42346877e-02 -4.56141502e-01
-4.52073365e-01 -8.22456181e-01 -5.52792788e-01 -3.09184849e-01
1.74844250e-01 -3.78050983e-01 -1.19233227e+00 1.50777113e+00
4.08170849e-01 8.42854083e-02 3.04697175e-03 5.64368367e-01
8.23886216e-01 3.73805016e-01 5.31812727e-01 -2.51826257e-01
1.96234512e+00 -9.59663212e-01 -7.22752273e-01 -3.79851431e-01
1.03649986e+00 -9.15711880e-01 7.25203753e-01 4.04829204e-01
-8.85523319e-01 -1.87952280e-01 -8.04476202e-01 -4.18651402e-01
-8.34402144e-01 3.62359762e-01 6.98441386e-01 4.01190490e-01
-7.63258398e-01 5.76295435e-01 -1.14479101e+00 -4.42306459e-01
5.85714340e-01 3.94858837e-01 -4.00785685e-01 -2.25447044e-02
-1.26447570e+00 1.09421539e+00 5.56535780e-01 -2.86770910e-01
-9.32996035e-01 -1.01594758e+00 -8.43301773e-01 3.69466454e-01
4.71872032e-01 -8.47555578e-01 1.28111911e+00 -5.88678956e-01
-1.20933151e+00 1.04127872e+00 -3.69573504e-01 -5.53776503e-01
-8.85252375e-03 -2.28008375e-01 -2.85807312e-01 1.20716475e-01
3.06927770e-01 2.98015505e-01 5.47965705e-01 -7.47149587e-01
-6.29751682e-01 -3.24667275e-01 3.36308195e-03 6.39277846e-02
-6.86969697e-01 2.03281641e-01 -6.69856906e-01 -7.08119690e-01
-3.64206642e-01 -7.80085444e-01 -4.35703546e-01 -2.25261033e-01
-5.22717237e-01 -4.10000861e-01 4.11623180e-01 -8.79743874e-01
1.24227500e+00 -1.89744532e+00 -1.39412850e-01 2.22364426e-01
6.81185424e-02 5.24025083e-01 -2.71430854e-02 4.33064520e-01
-3.51573527e-02 1.00319609e-01 -2.46429175e-01 -3.26536506e-01
-2.06274986e-02 7.04993680e-02 -2.32628599e-01 7.95293674e-02
4.75860327e-01 8.33602071e-01 -1.14210236e+00 -6.25726879e-01
-1.40856668e-01 4.42604065e-01 -3.83749574e-01 1.52049288e-01
-7.10632026e-01 2.93007046e-01 -7.58400381e-01 6.64125264e-01
4.87538993e-01 -6.37336552e-01 4.79373395e-01 -7.45303333e-02
-3.04245353e-02 7.92004049e-01 -7.40938723e-01 2.39819002e+00
-8.63592029e-01 3.01167220e-01 9.84352198e-04 -1.21117151e+00
6.44886494e-01 6.08613253e-01 3.57069224e-01 -2.67237931e-01
1.03639357e-01 5.45493543e-01 -3.03896397e-01 -6.60133421e-01
3.86240810e-01 -1.06862195e-01 -1.12593591e-01 3.59099865e-01
5.85300386e-01 1.93490848e-01 4.19099659e-01 1.21529572e-01
1.55049145e+00 1.39270559e-01 4.08001930e-01 -2.75399506e-01
5.56074798e-01 3.29649329e-01 4.29369897e-01 7.35414803e-01
2.57821918e-01 2.20816761e-01 3.94908100e-01 -3.76325279e-01
-8.74647975e-01 -7.62530506e-01 -4.43707585e-01 1.50912738e+00
-3.31897527e-01 -5.12107909e-01 -6.13727391e-01 -9.13467228e-01
-9.23227146e-02 5.53967774e-01 -3.61082971e-01 7.29725405e-04
-3.52637976e-01 -6.82511270e-01 6.42144382e-01 4.49349612e-01
6.55745938e-02 -7.81944454e-01 -2.71782398e-01 5.03673732e-01
-1.21924013e-01 -1.47053003e+00 -4.65134412e-01 7.88137376e-01
-6.60576761e-01 -6.52853251e-01 -8.02593112e-01 -1.09039843e+00
6.54067695e-01 -1.79201156e-01 1.20352948e+00 -2.93344975e-01
-5.39907217e-01 -1.34829566e-01 -2.43480965e-01 -6.81519985e-01
-1.20351151e-01 7.51064062e-01 -2.13100001e-01 -2.60058433e-01
8.86039436e-01 -4.69443113e-01 -5.09523451e-01 -2.02984646e-01
-8.91044796e-01 -1.23384394e-01 8.31507564e-01 9.78601873e-01
6.93304539e-01 -1.96918458e-01 5.25476873e-01 -1.17835665e+00
6.09251976e-01 -4.55875456e-01 -6.16778910e-01 6.33404911e-01
-6.19414866e-01 5.85745335e-01 6.54704392e-01 -6.36320293e-01
-1.04771662e+00 3.33066881e-01 -3.84007543e-01 -7.65713453e-02
-3.12768847e-01 8.15054953e-01 -7.48686865e-02 2.04357412e-02
8.19084525e-01 2.90860534e-01 -4.92272258e-01 -7.39104629e-01
5.54660976e-01 1.04504228e+00 5.54176509e-01 -6.73811495e-01
6.14722729e-01 3.90978992e-01 -1.55988589e-01 -5.03901601e-01
-1.21910644e+00 -8.72899950e-01 -3.69462073e-01 4.63354588e-01
8.50652397e-01 -1.15272236e+00 -5.58131099e-01 -1.66278705e-01
-1.38930738e+00 -6.52773902e-02 -1.55869156e-01 6.34068191e-01
1.34880096e-02 1.55052856e-01 -7.38802075e-01 -4.33645636e-01
-1.02402711e+00 -1.01922882e+00 1.44358218e+00 1.92342997e-01
-3.17382067e-01 -1.11072910e+00 2.51861930e-01 2.38309681e-01
4.37973082e-01 6.25981688e-02 8.56382668e-01 -1.32184017e+00
-2.18025684e-01 -2.72103071e-01 -1.27876744e-01 2.04108447e-01
1.26348287e-01 -3.63838196e-01 -1.19100523e+00 -1.27684921e-01
-1.81299612e-01 -3.84931862e-01 9.96455252e-01 9.18152556e-02
1.47373378e+00 -1.84059620e-01 -8.87112916e-01 4.87155586e-01
1.27244902e+00 -1.04350718e-02 1.39186651e-01 3.33577514e-01
6.12162828e-01 3.82969826e-01 5.09775996e-01 3.13061208e-01
1.75506860e-01 6.16346598e-01 -2.19313323e-01 -2.48410925e-01
2.18411982e-01 -3.69753130e-02 2.30117291e-01 6.26159012e-01
3.34846713e-02 -1.92568451e-01 -9.20438766e-01 8.71386647e-01
-1.56857646e+00 -7.70533144e-01 8.44386369e-02 2.10393929e+00
1.76757538e+00 2.86495149e-01 -1.80974185e-01 -3.11888963e-01
4.83808607e-01 -3.84506583e-01 -4.81004953e-01 -6.09345436e-01
3.08311671e-01 1.03966439e+00 7.18175054e-01 4.14712459e-01
-1.28650570e+00 1.08600545e+00 5.45175409e+00 1.25762951e+00
-1.30663049e+00 3.05226952e-01 5.03709078e-01 -1.97226182e-01
-1.84409931e-01 -3.51036154e-02 -9.92969096e-01 4.51488733e-01
1.31028676e+00 -1.18663140e-01 8.60632583e-02 1.07385135e+00
-1.44609481e-01 3.09958339e-01 -1.35810018e+00 6.48290515e-01
-4.84726392e-02 -1.33055341e+00 -1.46683484e-01 -1.59954622e-01
6.54781103e-01 2.80028313e-01 -1.10144928e-01 4.27049100e-01
4.10430580e-01 -8.97070825e-01 1.15247987e-01 2.83847034e-01
9.15349364e-01 -5.43639481e-01 8.97045255e-01 3.80992085e-01
-9.36566114e-01 3.58914763e-01 -3.42125028e-01 4.18367207e-01
1.56262591e-01 1.20680118e+00 -1.26658607e+00 7.72781253e-01
4.45996374e-01 3.60580921e-01 -2.56417871e-01 8.60281467e-01
-2.99786031e-01 5.52500546e-01 -6.04531646e-01 -9.76116583e-02
1.65626496e-01 2.26205781e-01 2.86506653e-01 1.84996951e+00
3.11348468e-01 -3.10444534e-02 2.71527708e-01 7.86969900e-01
-4.14164960e-01 2.95822680e-01 -4.17181045e-01 -4.27377790e-01
3.76575559e-01 1.63416827e+00 -7.07867622e-01 -5.97943485e-01
-4.55748409e-01 1.08824956e+00 3.59188586e-01 1.97940528e-01
-6.43412709e-01 -8.37771118e-01 6.21197879e-01 -1.29186481e-01
6.21187747e-01 -2.83086835e-03 -4.42968518e-01 -1.22368562e+00
1.53488696e-01 -8.94341648e-01 5.68414330e-01 -3.01325202e-01
-1.40170074e+00 7.30491459e-01 -2.12659717e-01 -1.00081027e+00
-2.40305275e-01 -8.34187865e-01 -2.65934050e-01 9.99277711e-01
-1.70502794e+00 -1.24178505e+00 8.84795412e-02 2.75491983e-01
4.10219312e-01 1.15182064e-01 1.27478218e+00 5.99021971e-01
-3.75324786e-01 8.73080254e-01 1.39243811e-01 2.97412038e-01
9.09967065e-01 -1.28646004e+00 5.01514316e-01 4.24925059e-01
2.20221817e-01 1.18855453e+00 5.24558246e-01 -5.40866375e-01
-1.24939096e+00 -1.24945998e+00 1.48994386e+00 -5.06559670e-01
7.71667778e-01 -6.75287843e-01 -1.04533732e+00 5.72561502e-01
1.28860638e-01 1.66291937e-01 1.21545756e+00 5.41794717e-01
-4.68815029e-01 1.43542662e-01 -9.55196559e-01 3.31604213e-01
9.20927465e-01 -8.29318702e-01 -5.51611304e-01 9.53369439e-01
7.46718526e-01 -6.78125679e-01 -1.12307239e+00 2.53185809e-01
3.81581038e-01 6.26371503e-02 9.92974997e-01 -9.17874336e-01
4.07368273e-01 -1.55441791e-01 3.73945594e-01 -1.10410690e+00
-3.07649951e-02 -6.49863303e-01 -1.91661790e-01 1.24091280e+00
9.73901808e-01 -3.63924265e-01 5.24724662e-01 4.42182958e-01
-7.83464685e-02 -8.43483806e-01 -7.63282955e-01 -7.80962706e-01
1.06490053e-01 -3.15146655e-01 4.94200110e-01 1.10156417e+00
5.46768308e-01 8.38847220e-01 -1.43516995e-02 1.37884557e-01
3.12132299e-01 2.96089143e-01 2.70687968e-01 -1.11872578e+00
-5.78345835e-01 -1.98800996e-01 -2.00429305e-01 -1.03183126e+00
1.60546601e-01 -1.26278710e+00 2.98109353e-01 -1.51814091e+00
3.24204564e-01 -2.80536860e-01 -4.80281472e-01 9.06805038e-01
-3.67729336e-01 1.53848901e-01 -6.42262280e-01 -4.92950194e-02
-7.87431955e-01 1.26633316e-01 9.63406265e-01 -3.30708504e-01
-1.30795226e-01 -1.10630348e-01 -7.63797641e-01 6.52559459e-01
5.56112707e-01 -8.62400532e-01 -1.97356880e-01 -6.03726804e-01
2.13932067e-01 -2.19174981e-01 -1.17941432e-01 -6.90177321e-01
3.97229671e-01 1.16533279e-01 2.12891102e-01 -2.67630190e-01
-2.26252526e-03 -6.58374071e-01 -2.29934111e-01 2.31657386e-01
-6.48131490e-01 -3.44178468e-01 3.35000515e-01 3.43678355e-01
-6.10033460e-02 -4.62255627e-01 3.80745888e-01 -2.20490023e-01
-3.07741135e-01 2.74973959e-01 -2.26594135e-01 2.15763643e-01
8.37727785e-01 5.10804594e-01 -2.75344491e-01 1.19375497e-01
-8.04810643e-01 3.03618580e-01 1.05264448e-01 1.35910138e-01
1.21272199e-01 -1.06261599e+00 -6.61547661e-01 -1.20593645e-01
4.01687980e-01 1.21510632e-01 -1.20962247e-01 6.58216894e-01
-4.26865429e-01 8.26181591e-01 2.37363130e-01 -3.48668426e-01
-1.26066732e+00 8.01518798e-01 -1.31206615e-02 -9.94605303e-01
-3.34857672e-01 1.10229480e+00 -3.21852118e-02 -5.01063049e-01
3.14907014e-01 -4.85602856e-01 -1.10825092e-01 -1.54662803e-02
7.18855917e-01 -1.66375145e-01 6.50949538e-01 9.42752957e-02
-6.21412098e-01 2.23964170e-01 -6.09698832e-01 -2.38227211e-02
1.45922637e+00 3.72269899e-01 -5.21714203e-02 1.73331480e-02
1.35995054e+00 -1.70810416e-01 -4.31071520e-01 -6.25934720e-01
4.45803225e-01 1.13897860e-01 2.09627062e-01 -9.20279741e-01
-6.98250413e-01 8.08368206e-01 4.14025635e-01 -4.77451012e-02
1.09003997e+00 2.21848935e-01 1.04760706e+00 7.14969039e-01
3.14201832e-01 -1.29459941e+00 -3.31374556e-01 4.67766553e-01
3.95133704e-01 -1.15242100e+00 1.42764807e-01 -5.37718177e-01
-3.36959273e-01 1.02617598e+00 2.09235415e-01 1.89982012e-01
6.01600289e-01 6.00884378e-01 -1.21845016e-02 -4.15122181e-01
-8.87639046e-01 -4.85785156e-01 4.63481754e-01 4.74811584e-01
7.82216072e-01 -1.30985022e-01 -7.53611207e-01 9.19296801e-01
2.75192852e-03 2.78311431e-01 -3.85735296e-02 1.10656071e+00
-2.19372809e-01 -1.59518170e+00 -8.23486149e-02 5.29984057e-01
-1.13142824e+00 -5.56869686e-01 -2.24351883e-01 2.01168180e-01
3.02543640e-01 8.42668772e-01 -7.36778453e-02 -1.18287444e-01
7.45804086e-02 4.67647463e-01 3.07596385e-01 -1.13175106e+00
-8.37382436e-01 1.05077542e-01 3.78737837e-01 -3.50508153e-01
-2.27566779e-01 -4.16117311e-01 -1.29406464e+00 1.44755542e-01
-5.60051382e-01 4.67811048e-01 8.81173253e-01 1.16288316e+00
7.16819763e-01 7.05691218e-01 2.66418666e-01 -3.43744457e-01
-5.74066639e-01 -1.19075346e+00 -1.51794925e-01 2.92380542e-01
1.31941199e-01 -3.77424806e-01 -5.18957153e-02 4.47344363e-01] | [8.703512191772461, 8.749719619750977] |
371b2a22-0dbc-467a-b3ba-2dce0abdd6ff | simplify-the-usage-of-lexicon-in-chinese-ner | 1908.05969 | null | https://arxiv.org/abs/1908.05969v2 | https://arxiv.org/pdf/1908.05969v2.pdf | Simplify the Usage of Lexicon in Chinese NER | Recently, many works have tried to augment the performance of Chinese named entity recognition (NER) using word lexicons. As a representative, Lattice-LSTM (Zhang and Yang, 2018) has achieved new benchmark results on several public Chinese NER datasets. However, Lattice-LSTM has a complex model architecture. This limits its application in many industrial areas where real-time NER responses are needed. In this work, we propose a simple but effective method for incorporating the word lexicon into the character representations. This method avoids designing a complicated sequence modeling architecture, and for any neural NER model, it requires only subtle adjustment of the character representation layer to introduce the lexicon information. Experimental studies on four benchmark Chinese NER datasets show that our method achieves an inference speed up to 6.15 times faster than those of state-ofthe-art methods, along with a better performance. The experimental results also show that the proposed method can be easily incorporated with pre-trained models like BERT. | ['Ruotian Ma', 'Xuanjing Huang', 'Minlong Peng', 'Qi Zhang'] | 2019-08-16 | simplify-the-usage-of-lexicon-in-chinese-ner-1 | https://aclanthology.org/2020.acl-main.528 | https://aclanthology.org/2020.acl-main.528.pdf | acl-2020-6 | ['chinese-named-entity-recognition'] | ['natural-language-processing'] | [-1.43872663e-01 -2.68911391e-01 1.31972313e-01 -3.07076007e-01
-6.43956959e-01 -5.10429740e-01 2.41928488e-01 -7.74666071e-02
-1.04721105e+00 8.17728460e-01 2.36928031e-01 -5.13667405e-01
4.68461037e-01 -9.73835468e-01 -3.27310950e-01 -4.67717409e-01
2.92393506e-01 2.06920817e-01 1.84126928e-01 -2.46715605e-01
1.00649357e-01 4.27118152e-01 -8.45109046e-01 2.59133801e-02
1.08415794e+00 6.72921062e-01 4.85943615e-01 3.70267034e-01
-6.94830656e-01 7.26473927e-01 -6.63649797e-01 -5.54756224e-01
-1.07274488e-01 -3.15315574e-01 -7.06418455e-01 -4.88274992e-01
-2.41093561e-01 -2.44171411e-01 -2.28062660e-01 1.14181805e+00
9.70673263e-01 3.10288012e-01 3.24283719e-01 -3.07767302e-01
-9.17801619e-01 1.12790310e+00 -1.65931195e-01 1.27237409e-01
5.68022486e-03 -2.24563047e-01 6.93550766e-01 -8.18876982e-01
4.63240623e-01 9.33496952e-01 8.43688428e-01 8.43572199e-01
-6.20591760e-01 -9.32056367e-01 1.94997251e-01 2.63154954e-01
-1.44279742e+00 -1.95848078e-01 4.75460112e-01 2.30503216e-01
1.36786079e+00 -5.12475297e-02 2.99993664e-01 1.12822425e+00
2.26734411e-02 8.38453531e-01 9.27699804e-01 -4.43801910e-01
1.54270217e-01 2.07584389e-02 2.56500691e-01 4.84918416e-01
3.07748735e-01 -3.10011864e-01 9.10346135e-02 -7.80429468e-02
8.35787594e-01 6.21455647e-02 3.22356783e-02 4.12168622e-01
-1.20250082e+00 6.96073711e-01 4.16682541e-01 8.56646180e-01
-5.28083503e-01 7.62935802e-02 5.81760705e-01 -6.75590709e-02
4.64428544e-01 4.68224406e-01 -7.87086308e-01 -3.76902759e-01
-6.32136881e-01 -3.80208582e-01 9.53761160e-01 1.09570062e+00
4.89053667e-01 4.85410094e-01 -1.51322708e-01 9.89713550e-01
-4.64679999e-03 4.92765218e-01 9.16891158e-01 -2.71004826e-01
5.65352559e-01 3.63586038e-01 3.54607329e-02 -8.25654626e-01
-5.44303834e-01 -4.30763990e-01 -1.15294492e+00 -6.01197541e-01
2.47012764e-01 -7.40144372e-01 -9.29717839e-01 1.44760680e+00
4.34585437e-02 3.66174459e-01 2.52555102e-01 6.72933400e-01
8.29781711e-01 9.23710108e-01 4.72921133e-01 -1.02232069e-01
1.38574719e+00 -1.21302760e+00 -1.09557390e+00 -2.31617346e-01
8.87651205e-01 -8.52678299e-01 1.03088486e+00 2.92615652e-01
-6.96487367e-01 -5.28146744e-01 -9.45431769e-01 -5.48273250e-02
-6.81789756e-01 6.87506258e-01 9.14159596e-01 9.53076065e-01
-8.39034081e-01 5.70862770e-01 -9.91648376e-01 -5.13087988e-01
-1.48589417e-01 3.20685506e-01 -2.15035483e-01 1.85398366e-02
-1.68034875e+00 1.06038237e+00 8.32229614e-01 5.65819025e-01
-4.50187325e-01 -1.12996399e-01 -8.62171531e-01 2.93839008e-01
1.73881516e-01 -1.92310467e-01 1.25280452e+00 -4.14912015e-01
-1.90325463e+00 1.69220358e-01 -2.43869603e-01 -3.14291179e-01
8.08460861e-02 -4.89942551e-01 -8.52259815e-01 -3.64093602e-01
-2.13945866e-01 4.75417942e-01 2.68124118e-02 -5.88045180e-01
-5.32975912e-01 6.92582801e-02 -6.89101070e-02 7.73482583e-03
-8.86594594e-01 3.22424799e-01 -5.12064636e-01 -8.12666774e-01
-5.05601838e-02 -6.25255644e-01 -8.08032513e-01 -8.41612041e-01
-4.00332749e-01 -5.73482811e-01 4.65346396e-01 -8.10516357e-01
1.71151876e+00 -1.89131927e+00 -5.93295693e-01 -1.31021664e-01
-2.59466797e-01 7.90430129e-01 -4.56428051e-01 6.84947550e-01
1.50536224e-01 6.04530394e-01 -1.26275882e-01 -2.83455878e-01
4.32938896e-02 5.84605634e-02 -3.31356764e-01 3.67750488e-02
4.04235005e-01 1.02113342e+00 -9.18673575e-01 -3.90264899e-01
1.05472513e-01 6.80339456e-01 -2.59038925e-01 2.54905939e-01
-4.19985652e-02 2.25733981e-01 -7.08742261e-01 3.91402096e-01
7.61273026e-01 -1.96091056e-01 4.63392794e-01 1.14079587e-01
-4.25860584e-01 7.37143815e-01 -1.21280098e+00 1.61843109e+00
-6.02690339e-01 3.83183241e-01 -5.04517674e-01 -6.78123415e-01
1.32120502e+00 6.23442888e-01 -1.37018666e-01 -7.63589621e-01
3.38241190e-01 3.63685966e-01 -9.92945582e-02 -2.92482108e-01
7.17317700e-01 -4.38689142e-02 -2.55194217e-01 4.82248187e-01
8.88009295e-02 3.63552123e-01 3.39646637e-01 -3.12059700e-01
9.68952596e-01 1.64001986e-01 4.87757117e-01 -9.86028537e-02
6.82194352e-01 -4.06341627e-02 1.00487065e+00 7.17695832e-01
-8.41880962e-02 3.27450752e-01 1.61932502e-02 -4.32054222e-01
-1.03660154e+00 -6.15591168e-01 -2.37993747e-02 1.02267551e+00
-1.03240989e-01 -3.94509137e-01 -1.03327453e+00 -7.09655523e-01
-7.71530688e-01 7.69050479e-01 -2.17548549e-01 3.29990499e-02
-1.04479563e+00 -7.42372811e-01 1.33758831e+00 8.92000079e-01
1.03787100e+00 -1.38935542e+00 -1.42813087e-01 7.12678730e-01
-4.00504827e-01 -1.37076855e+00 -6.14966393e-01 1.45583376e-01
-1.14829731e+00 -4.68475103e-01 -9.51116800e-01 -1.36550641e+00
5.97560763e-01 4.16350812e-02 7.48527586e-01 1.00371867e-01
9.44938287e-02 -2.40917936e-01 -6.60482526e-01 -3.44943106e-01
-1.92847148e-01 6.76309884e-01 3.82291265e-02 -3.82788837e-01
8.29534590e-01 -3.49160522e-01 -1.61773458e-01 1.63678229e-01
-9.11991775e-01 -1.29430011e-01 9.93441403e-01 9.92863894e-01
4.35638905e-01 4.26833406e-02 7.49920666e-01 -8.49910975e-01
7.97970295e-01 -2.95672357e-01 -5.77289343e-01 5.77180028e-01
-5.39002180e-01 3.00023347e-01 1.09184897e+00 -6.69509709e-01
-1.42717957e+00 1.16267651e-01 -5.92444837e-01 1.85625568e-01
-3.72597873e-01 8.81325305e-01 -3.34561288e-01 1.17705807e-01
3.32257032e-01 4.35086876e-01 -7.85609245e-01 -9.94698167e-01
3.00371557e-01 9.45008099e-01 2.69291878e-01 -5.81961870e-01
5.44357955e-01 -5.49715757e-02 -5.67189991e-01 -8.29239547e-01
-6.83422327e-01 -3.56371284e-01 -7.70885289e-01 2.65093952e-01
9.59886432e-01 -8.86876285e-01 -6.35644794e-01 6.36619031e-01
-1.64443672e+00 -2.60559231e-01 3.30073744e-01 8.57922852e-01
-1.40811624e-02 5.31096876e-01 -1.21589077e+00 -9.22937393e-01
-6.75863385e-01 -8.60182166e-01 4.83298272e-01 4.87645298e-01
1.33917004e-01 -1.16412890e+00 2.18692236e-02 -1.89899817e-01
6.95438504e-01 -2.81405717e-01 8.25254261e-01 -9.45988953e-01
-2.19785482e-01 -2.76036978e-01 -2.39437282e-01 4.54172224e-01
7.50598535e-02 -2.88364410e-01 -9.31912005e-01 3.24315089e-03
9.09535736e-02 -2.40348071e-01 7.23347366e-01 -5.80946021e-02
1.07607770e+00 -2.91473716e-01 -2.57213205e-01 4.92223382e-01
1.41291976e+00 5.20657122e-01 8.53747725e-01 5.34587741e-01
7.10753918e-01 2.05188140e-01 5.87262332e-01 3.20935875e-01
4.89820540e-01 2.17834741e-01 -1.02847748e-01 -3.29493314e-01
1.05638653e-01 -4.84280348e-01 5.86579859e-01 1.71065390e+00
-2.15090394e-01 -4.50074345e-01 -1.02546406e+00 4.91735727e-01
-1.66846359e+00 -7.84034610e-01 -1.26067936e-01 1.95249629e+00
1.12993813e+00 6.00091778e-02 -4.76232946e-01 -2.67289370e-01
1.05018234e+00 5.92015088e-02 -4.38529819e-01 -5.61965585e-01
-1.46880686e-01 5.35478771e-01 6.69519365e-01 1.72623008e-01
-1.25933683e+00 1.61649394e+00 6.38341618e+00 1.08049428e+00
-1.22056115e+00 2.43365049e-01 4.41417366e-01 5.13090014e-01
-1.76849052e-01 -9.66008678e-02 -1.21034658e+00 2.72179693e-01
1.27394533e+00 -1.79577321e-01 2.22770572e-01 8.27678800e-01
-1.87502638e-03 3.95989627e-01 -6.41244292e-01 7.84040868e-01
-7.93243200e-02 -1.04869711e+00 8.73809233e-02 -1.14601485e-01
7.08851755e-01 2.80327797e-01 -3.87257904e-01 7.49520779e-01
4.75888014e-01 -9.57380295e-01 3.68667066e-01 4.57876980e-01
7.33358502e-01 -9.41185713e-01 1.24976945e+00 5.31800270e-01
-1.48972666e+00 1.45282716e-01 -9.41514313e-01 -3.07980597e-01
4.02498752e-01 4.71041411e-01 -7.36809373e-01 6.22011364e-01
4.96781409e-01 4.97360468e-01 -4.25888568e-01 1.12037086e+00
-5.97985685e-01 9.22109127e-01 -2.13309810e-01 -6.93780303e-01
4.55474824e-01 -1.31960094e-01 -1.66642759e-02 1.69030857e+00
5.01543820e-01 2.87420303e-01 1.08636938e-01 6.91584170e-01
-1.20732486e-01 5.05808830e-01 -4.27100241e-01 -6.10100865e-01
6.45644426e-01 1.38336897e+00 -8.53026927e-01 -3.73732537e-01
-3.49312186e-01 1.11595845e+00 5.82856357e-01 4.41002220e-01
-9.07457769e-01 -1.04409325e+00 3.37006778e-01 -7.51516640e-01
5.56143463e-01 -6.66941941e-01 -3.98600966e-01 -1.35578299e+00
-1.46110266e-01 -6.71340287e-01 2.44401127e-01 -5.13785124e-01
-1.40572822e+00 1.09536386e+00 -6.16168737e-01 -1.13698077e+00
-1.93396971e-01 -7.86284566e-01 -7.22132742e-01 9.10014331e-01
-1.57013786e+00 -1.03305411e+00 4.73805107e-02 3.00418496e-01
4.90119785e-01 -2.13424116e-02 1.33422542e+00 6.27119660e-01
-9.08616543e-01 8.74901175e-01 2.05516115e-01 8.68886292e-01
6.82586491e-01 -1.00899398e+00 1.06880760e+00 1.02189779e+00
1.16447739e-01 1.08401108e+00 1.52507991e-01 -7.74881065e-01
-1.17621517e+00 -1.03985870e+00 1.55191696e+00 -1.10442594e-01
6.39079154e-01 -4.89036351e-01 -8.27645779e-01 6.90645814e-01
4.75929230e-01 -3.05808961e-01 8.97425711e-01 9.43453833e-02
-1.43166199e-01 1.20599695e-01 -8.17578197e-01 7.77807474e-01
9.45664525e-01 -5.20230412e-01 -7.00887203e-01 -8.86362791e-02
1.03035319e+00 -2.91119099e-01 -9.33887184e-01 3.92850310e-01
3.95495653e-01 -5.42723596e-01 5.59063375e-01 -7.62767673e-01
4.02123220e-02 -3.57459605e-01 5.77056147e-02 -1.18590856e+00
-5.55705130e-01 -4.23277527e-01 1.36964589e-01 1.55541444e+00
5.61425328e-01 -7.54586220e-01 4.55143929e-01 4.96739954e-01
-2.09280506e-01 -5.28695941e-01 -6.40291274e-01 -9.91406500e-01
2.49299526e-01 -5.05286336e-01 8.91052008e-01 1.11264336e+00
3.22075933e-02 4.07612443e-01 -4.45003778e-01 1.29051223e-01
1.55995682e-01 -3.48237246e-01 2.93401957e-01 -1.00314307e+00
1.11091301e-01 -3.59689891e-01 -1.38793781e-01 -1.38077188e+00
3.19660127e-01 -7.76699960e-01 3.01821321e-01 -1.74869132e+00
1.38127983e-01 -5.08946300e-01 -6.99984014e-01 7.38394856e-01
-4.10129696e-01 8.22291672e-02 1.92238405e-01 -1.19795099e-01
-6.24444723e-01 6.68506980e-01 1.12222207e+00 -2.45722607e-02
-1.67417377e-01 -2.65680015e-01 -5.76056838e-01 7.43442476e-01
1.14357424e+00 -7.08560586e-01 2.17761826e-02 -9.26366746e-01
2.16018811e-01 -2.97208190e-01 -3.24261546e-01 -1.01732826e+00
6.61616027e-01 2.51547638e-02 4.96853560e-01 -7.12336123e-01
-1.13220990e-01 -6.03498399e-01 -1.45053446e-01 5.31757593e-01
-1.80174008e-01 3.14797312e-01 3.61653179e-01 3.12454820e-01
-2.64609605e-01 -5.83815217e-01 3.01289558e-01 -2.32223466e-01
-9.99133468e-01 2.70387888e-01 -6.40593767e-01 5.94247654e-02
4.79153246e-01 6.49835635e-03 -2.18545511e-01 -1.44018784e-01
-4.08234268e-01 1.56169549e-01 -1.77249070e-02 5.80772698e-01
3.85857463e-01 -1.43674552e+00 -5.63311279e-01 1.28095478e-01
-1.43211573e-01 -2.51047194e-01 3.48792523e-01 5.54064155e-01
-5.19222140e-01 7.89277375e-01 -1.98022928e-02 6.12604693e-02
-6.54086590e-01 4.05885577e-01 1.82826549e-01 -6.37505651e-01
-4.77168560e-01 6.70121729e-01 -7.33194798e-02 -8.32137108e-01
3.32003087e-01 -4.28616554e-01 -6.46741331e-01 -2.00635001e-01
7.93669164e-01 4.50623721e-01 1.87357347e-02 -3.32864732e-01
-3.41417491e-01 4.20380354e-01 -1.22319236e-01 -6.50344715e-02
1.34296870e+00 -4.68295626e-02 -2.66882211e-01 4.32402223e-01
9.77414012e-01 7.65194148e-02 -6.48608565e-01 -2.99408436e-01
6.05153263e-01 2.48749465e-01 -1.06042556e-01 -8.24130237e-01
-8.99047673e-01 1.18484604e+00 4.14111882e-01 -8.93364623e-02
1.04157639e+00 -5.78646898e-01 1.46420252e+00 1.12047076e+00
7.10158587e-01 -1.23617494e+00 -4.52026278e-01 1.35930657e+00
2.99185127e-01 -1.01301551e+00 -5.03435969e-01 -2.08907813e-01
-5.40222585e-01 1.49634838e+00 8.48797619e-01 -5.57202175e-02
5.82381546e-01 3.75225484e-01 4.44647610e-01 3.29855531e-01
-5.74175179e-01 -4.00535375e-01 1.08412877e-01 3.63183916e-01
8.77877831e-01 5.42399585e-02 -9.65413868e-01 1.27401602e+00
-8.28504711e-02 4.76930067e-02 4.93425220e-01 6.64490402e-01
-4.57246155e-01 -1.55409920e+00 1.93666164e-02 9.57074910e-02
-7.99665809e-01 -7.18411982e-01 -1.88004896e-01 6.58941627e-01
-9.83326063e-02 1.06933105e+00 -1.75987501e-02 -5.58237612e-01
3.31154078e-01 2.40217224e-01 -8.10762048e-02 -7.01087534e-01
-8.34756017e-01 1.35182217e-01 2.48333097e-01 -2.71182746e-01
-2.81029135e-01 -1.84440970e-01 -1.68606365e+00 -9.10012498e-02
-8.05873275e-01 6.36732042e-01 7.14382946e-01 1.01605761e+00
4.33561027e-01 4.76789892e-01 5.06760299e-01 -5.05192816e-01
-6.70799851e-01 -1.37845731e+00 -5.17923295e-01 -1.78417847e-01
-2.53836602e-01 -1.40494928e-01 -4.48018722e-02 -1.94530055e-01] | [9.842507362365723, 9.76589584350586] |
8575fb81-6834-4357-98bb-dfe7517fcaf3 | randomized-greedy-learning-for-non-monotone | 2302.01324 | null | https://arxiv.org/abs/2302.01324v1 | https://arxiv.org/pdf/2302.01324v1.pdf | Randomized Greedy Learning for Non-monotone Stochastic Submodular Maximization Under Full-bandit Feedback | We investigate the problem of unconstrained combinatorial multi-armed bandits with full-bandit feedback and stochastic rewards for submodular maximization. Previous works investigate the same problem assuming a submodular and monotone reward function. In this work, we study a more general problem, i.e., when the reward function is not necessarily monotone, and the submodularity is assumed only in expectation. We propose Randomized Greedy Learning (RGL) algorithm and theoretically prove that it achieves a $\frac{1}{2}$-regret upper bound of $\tilde{\mathcal{O}}(n T^{\frac{2}{3}})$ for horizon $T$ and number of arms $n$. We also show in experiments that RGL empirically outperforms other full-bandit variants in submodular and non-submodular settings. | ['Mohamed-Slim Alouini', 'Christopher John Quinn', 'Vaneet Aggarwal', 'Fares Fourati'] | 2023-02-02 | null | null | null | null | ['multi-armed-bandits'] | ['miscellaneous'] | [ 2.19095852e-02 4.48382288e-01 -9.97401536e-01 -2.32758000e-01
-1.15766251e+00 -9.37716126e-01 -3.46479505e-01 -1.72571898e-01
-4.26866859e-01 1.50080121e+00 6.73356652e-02 -7.87073314e-01
-9.81366634e-01 -6.94230855e-01 -1.14839935e+00 -9.12908852e-01
-3.59025210e-01 7.04255641e-01 -2.08783433e-01 -4.31529656e-02
2.49721169e-01 3.11517984e-01 -8.04941237e-01 1.00273959e-01
7.64843941e-01 1.56442475e+00 1.13388654e-02 6.78524196e-01
-8.95235911e-02 9.05147433e-01 -3.33406270e-01 -4.35135096e-01
8.25017333e-01 -5.14123797e-01 -8.99292827e-01 3.48652184e-01
4.55425441e-01 -6.23996973e-01 -2.65139699e-01 1.15369391e+00
2.11747780e-01 -4.62203883e-02 3.16378653e-01 -1.23592103e+00
-4.05715048e-01 1.15329099e+00 -1.08575070e+00 3.64782065e-01
-7.12451786e-02 -7.88569301e-02 1.64147186e+00 5.47883101e-02
2.74167627e-01 1.32391214e+00 1.12126820e-01 3.97562146e-01
-1.21036899e+00 -7.55314171e-01 8.31257939e-01 -1.80136219e-01
-7.84209549e-01 3.19793373e-02 5.41117668e-01 -2.12584868e-01
5.49611926e-01 8.38806152e-01 6.72321975e-01 2.43859619e-01
-2.38115832e-01 1.18547630e+00 1.30870473e+00 -4.71348852e-01
2.62402147e-01 -1.08178981e-01 2.28837252e-01 5.71298063e-01
8.66575181e-01 2.09916875e-01 -1.97440565e-01 -4.21417579e-02
8.53277266e-01 1.88501716e-01 -2.42008537e-01 -4.37215149e-01
-7.18260586e-01 1.15987539e+00 2.48562977e-01 -1.29238635e-01
-5.54347873e-01 8.99645329e-01 -4.87342328e-02 7.17376709e-01
3.99086863e-01 2.74380267e-01 -5.79099357e-01 -6.79884478e-02
-1.04529011e+00 4.43390965e-01 8.29041660e-01 1.39619410e+00
5.23065329e-01 1.51908085e-01 -5.07884443e-01 5.05731285e-01
-1.96313504e-02 8.97201955e-01 -2.83091545e-01 -1.31256962e+00
1.23122144e+00 3.76892179e-01 9.05361712e-01 -4.78633493e-01
-5.60539126e-01 -6.30308986e-01 -3.83837640e-01 -6.46531628e-03
7.03918099e-01 -7.36088753e-01 -7.27133155e-01 1.59279358e+00
1.75537318e-01 -5.92551291e-01 -2.82667279e-01 9.88189340e-01
1.22374743e-01 5.37081659e-01 -5.15007734e-01 -9.44495261e-01
6.08351231e-01 -8.59360337e-01 -6.20604634e-01 -3.05085123e-01
5.05628109e-01 -5.37457764e-01 6.16008341e-01 5.85753560e-01
-1.51601493e+00 4.15065855e-01 -7.24713147e-01 4.42030460e-01
3.59693378e-01 -2.09884629e-01 1.10251033e+00 1.16780210e+00
-7.27933228e-01 1.83314100e-01 -4.47494179e-01 3.72323513e-01
6.32696807e-01 7.48008132e-01 3.58547390e-01 -3.23383778e-01
-4.30585325e-01 2.30223373e-01 3.34001929e-01 -2.33577914e-03
-1.12702024e+00 -4.91292506e-01 -2.48124883e-01 1.61189184e-01
1.22397661e+00 -2.63274699e-01 1.50610054e+00 -6.69238806e-01
-1.10101128e+00 5.67320406e-01 4.73447330e-02 -6.08347058e-01
7.69539356e-01 -2.21625090e-01 1.98843688e-01 7.27624148e-02
-6.84802327e-03 1.87597096e-01 5.43651342e-01 -1.17424262e+00
-1.01298964e+00 -7.90819347e-01 8.19837093e-01 3.68745662e-02
-3.77737522e-01 -9.87588018e-02 7.65614063e-02 -3.79434615e-01
4.41523194e-01 -9.34275210e-01 -4.53056216e-01 -5.34171879e-01
-5.42152882e-01 4.70314408e-03 5.66533469e-02 -2.72170693e-01
1.47633636e+00 -1.64589548e+00 2.11627737e-01 4.51046467e-01
-1.07566908e-01 -3.22189659e-01 4.54341434e-02 4.11774665e-01
2.52577454e-01 3.68130356e-01 2.21012548e-01 8.02307874e-02
2.65043646e-01 2.89645016e-01 -3.33478332e-01 6.53081119e-01
-6.84635818e-01 8.82815957e-01 -6.50262058e-01 -1.95138708e-01
-3.30125183e-01 -7.36269236e-01 -7.31531501e-01 -3.01868230e-01
-8.64449143e-01 7.48135000e-02 -8.53489161e-01 1.03504717e+00
7.83887625e-01 -4.65612739e-01 4.53401476e-01 5.95274568e-01
-3.28532279e-01 1.01066992e-01 -1.74061275e+00 1.11076820e+00
-3.70491058e-01 -5.30071510e-03 5.99704385e-01 -1.35932255e+00
5.34099817e-01 7.65841082e-02 8.03141773e-01 -7.17573106e-01
3.50668550e-01 4.95117366e-01 -1.74489677e-01 -2.54601032e-01
2.62886763e-01 -5.62268734e-01 -1.19274944e-01 9.46874201e-01
-3.42679590e-01 3.64961885e-02 3.78721535e-01 1.64468288e-02
1.09486830e+00 -3.61904740e-01 7.52592236e-02 -3.67448717e-01
8.06972086e-02 1.57419384e-01 6.94206953e-01 1.40628052e+00
3.99531871e-02 2.23233074e-01 1.15541577e+00 -4.58808661e-01
-9.82980788e-01 -7.74459779e-01 1.37169912e-01 1.57663989e+00
1.99684083e-01 3.76416087e-01 -3.36199760e-01 -6.11118436e-01
6.52550697e-01 8.78885567e-01 -6.77428126e-01 4.63879794e-01
-3.47115010e-01 -8.92466605e-01 -4.56113257e-02 3.88650358e-01
3.05907160e-01 -6.29493952e-01 -4.80642855e-01 3.50689530e-01
2.94759125e-02 -8.70454848e-01 -7.80578434e-01 4.87982929e-01
-1.14053893e+00 -8.80758166e-01 -8.64363194e-01 -3.17979217e-01
6.44689739e-01 5.88678539e-01 7.42417812e-01 -3.95309627e-01
1.32398918e-01 5.00823081e-01 -4.15975690e-01 -8.07911813e-01
3.73328865e-01 8.08496727e-04 -2.02637628e-01 -6.29518703e-02
1.47099897e-01 -3.55227649e-01 -7.46522427e-01 3.33993167e-01
-7.79372454e-01 -2.94108033e-01 5.61158657e-01 6.72476947e-01
8.74957800e-01 -1.85887869e-02 7.02431440e-01 -1.03015637e+00
4.48539197e-01 -3.61370176e-01 -1.56173110e+00 4.45596397e-01
-5.42455375e-01 1.09397478e-01 4.66612190e-01 -3.29921037e-01
-7.69426823e-01 -1.86305691e-03 5.79253674e-01 -2.30291069e-01
5.95383763e-01 4.75089997e-01 1.36089280e-01 1.36909308e-02
4.78463918e-01 3.23226571e-01 -3.58317196e-01 -4.75016177e-01
1.72020584e-01 5.54363012e-01 -1.45210937e-01 -8.79720807e-01
5.55456638e-01 6.76822245e-01 4.99885380e-01 -2.73702294e-01
-1.56113219e+00 -3.45253706e-01 2.05855787e-01 -8.78904983e-02
1.73153833e-01 -7.20466256e-01 -1.58997309e+00 -3.13061744e-01
-7.45650530e-01 -4.42804724e-01 -5.12814760e-01 7.16652870e-01
-1.17440748e+00 -2.47953972e-03 -1.23306870e-01 -1.77653360e+00
-2.22586468e-01 -6.89264834e-01 5.78732789e-01 1.57157242e-01
4.66025770e-01 -4.61719900e-01 -1.11336969e-01 9.37771201e-01
2.66604930e-01 1.24595210e-01 7.68739700e-01 -2.72821128e-01
-1.24639368e+00 -2.78187156e-01 -3.79767060e-01 3.21560621e-01
2.51561087e-02 -8.35196912e-01 -3.27790171e-01 -6.15478098e-01
-8.74972567e-02 -6.17124498e-01 9.56273496e-01 1.10394406e+00
1.31782949e+00 -1.04428351e+00 -8.45724270e-02 5.18558025e-01
1.56030655e+00 5.07618785e-01 2.75196284e-01 4.72481966e-01
6.24544807e-02 3.10776114e-01 9.12105918e-01 1.14168406e+00
6.48845658e-02 1.61467373e-01 1.02842081e+00 4.31140423e-01
5.68387032e-01 -4.40522693e-02 9.01438743e-02 -2.54914284e-01
-2.75988519e-01 -5.07806003e-01 -4.73965704e-01 8.15712690e-01
-2.18118739e+00 -1.12897718e+00 1.87838788e-03 2.43598628e+00
9.97424304e-01 3.04506987e-01 7.36803472e-01 -7.24478886e-02
8.20304394e-01 1.16961166e-01 -9.04201388e-01 -7.10130513e-01
-3.04707378e-01 2.23680273e-01 1.35046744e+00 6.15300179e-01
-7.03448653e-01 8.13031256e-01 5.61770344e+00 7.55252302e-01
-6.94348574e-01 1.83155373e-01 8.33962858e-01 -1.20919728e+00
-5.00827134e-01 2.61815097e-02 -1.11820352e+00 3.80762786e-01
4.20096934e-01 -2.26396322e-01 8.75465512e-01 9.68955398e-01
3.61659080e-01 -3.22085649e-01 -9.01343942e-01 8.81791234e-01
-3.47339451e-01 -1.67035306e+00 -4.67339933e-01 4.30409789e-01
1.40978909e+00 -1.10142455e-01 2.97503382e-01 1.30901650e-01
8.64220023e-01 -1.04566419e+00 7.29824603e-01 2.92282719e-02
7.38604069e-01 -1.00854278e+00 4.28801209e-01 6.49340093e-01
-6.83117628e-01 -9.64081645e-01 -4.13920462e-01 -1.83189452e-01
1.83145881e-01 6.46189153e-01 -4.39187735e-01 5.49114108e-01
6.64905965e-01 -5.65488003e-02 4.81481194e-01 1.33810282e+00
-1.06793851e-01 5.69684744e-01 -6.76144183e-01 -4.88069922e-01
7.23864973e-01 -2.19062284e-01 4.53256607e-01 8.06679368e-01
2.20308170e-01 5.00084996e-01 3.53530675e-01 6.06910884e-01
-3.65321726e-01 2.52350539e-01 -5.71445860e-02 -2.39764959e-01
3.25881720e-01 7.71381259e-01 -6.38716340e-01 9.46863517e-02
-2.46332094e-01 3.01007330e-01 1.62689805e-01 3.33160967e-01
-8.04012477e-01 -9.56024900e-02 4.14416343e-01 2.80539334e-01
1.04130960e+00 -1.46924019e-01 -5.70615828e-01 -6.72760904e-01
4.42232937e-01 -3.02456707e-01 7.83038497e-01 -1.10707924e-01
-1.05116367e+00 -1.19121060e-01 1.31015152e-01 -8.41919541e-01
-8.26503057e-03 -5.58196127e-01 -1.24535516e-01 5.62512457e-01
-1.71620572e+00 -8.70146036e-01 2.64525115e-01 6.69684768e-01
2.21327335e-01 -1.71453178e-01 5.11339605e-02 3.26161496e-02
-3.78554463e-01 6.81600928e-01 7.60117710e-01 -3.35591614e-01
-1.52114585e-01 -1.23142838e+00 -6.43033028e-01 4.44253117e-01
-3.11572403e-01 2.26843074e-01 7.92854607e-01 -5.20043254e-01
-1.79424953e+00 -9.34527755e-01 3.60963553e-01 1.36650622e-01
9.23606932e-01 -1.26423880e-01 2.34254956e-01 1.02627218e+00
-1.27817497e-01 4.97815274e-02 5.53938806e-01 2.01358855e-01
-1.84751168e-01 -5.37576377e-01 -1.35460794e+00 3.31236482e-01
1.37836611e+00 2.74545461e-01 2.66743656e-02 8.86777997e-01
8.87135923e-01 -5.02675653e-01 -7.89443552e-01 1.99067131e-01
5.94499171e-01 -8.15032542e-01 6.39436066e-01 -1.04054070e+00
1.62251800e-01 3.68031591e-01 -7.18665600e-01 -6.72274232e-01
-2.21340030e-01 -1.31098807e+00 -4.12709326e-01 6.23476744e-01
7.84151316e-01 -6.25759363e-01 1.42461801e+00 4.64104325e-01
8.56402814e-02 -9.30238307e-01 -1.40995479e+00 -1.39916074e+00
5.22245765e-01 -3.71871442e-01 5.61230242e-01 5.71090639e-01
2.60545649e-02 -1.46463603e-01 -8.87955964e-01 2.63948292e-02
6.69182181e-01 8.15602422e-01 5.70881605e-01 -8.71358812e-01
-8.10501814e-01 -3.06578934e-01 3.05488944e-01 -1.49320626e+00
-3.52260202e-01 -5.77120483e-01 -3.23853582e-01 -1.47918665e+00
5.12859881e-01 -7.36849010e-01 -5.02541900e-01 5.81866801e-01
4.26440805e-01 -3.93961012e-01 4.04132456e-01 -1.41802728e-01
-1.09068692e+00 2.59706289e-01 1.65375197e+00 -2.87797660e-01
-4.92630035e-01 7.31512129e-01 -1.26823103e+00 1.78568244e-01
8.07538807e-01 -7.14490712e-01 -3.57869744e-01 -6.00886524e-01
7.10147202e-01 8.89045894e-01 -5.54786390e-03 -1.54078618e-01
-5.28725907e-02 -1.13715947e+00 -2.10059583e-01 -8.40322316e-01
2.60275632e-01 -9.70830381e-01 -1.48579597e-01 6.43097281e-01
-3.97819430e-01 -2.56230861e-01 -5.34671284e-02 6.74342215e-01
3.57242316e-01 -6.63590074e-01 5.82567573e-01 -4.26660001e-01
2.15134576e-01 4.17642713e-01 6.69414625e-02 2.89207548e-01
1.33226550e+00 -1.39109671e-01 -5.72021544e-01 -8.73803735e-01
-6.94154978e-01 7.27406919e-01 -9.20225605e-02 -1.85037270e-01
2.09274411e-01 -1.12321186e+00 -7.89194465e-01 -5.25749326e-01
-2.40782306e-01 5.21886274e-02 3.05194020e-01 1.01220119e+00
-2.62097925e-01 1.04309773e+00 1.53131112e-01 -2.90245950e-01
-9.26053584e-01 6.31655037e-01 2.91889817e-01 -6.67787731e-01
8.80672783e-02 1.10698307e+00 -3.18894684e-01 -2.67934818e-02
5.38693964e-01 -2.52072901e-01 4.20046329e-01 1.80140048e-01
2.33313233e-01 8.59639108e-01 -2.27474436e-01 2.02276856e-01
-3.31734233e-02 1.03319511e-01 -4.59095806e-01 -3.75778705e-01
1.68192852e+00 -6.16962239e-02 -1.75624460e-01 1.28105372e-01
8.89645815e-01 -1.13414548e-01 -1.32831228e+00 -4.25171316e-01
-2.47027893e-02 -8.11174691e-01 6.80400357e-02 -8.77276123e-01
-1.20948625e+00 3.79687488e-01 4.19214636e-01 7.52864540e-01
9.66173828e-01 2.48875141e-01 6.10964119e-01 6.25507891e-01
7.57102609e-01 -1.52550709e+00 1.77922487e-01 3.13223541e-01
8.48282456e-01 -1.07974803e+00 4.36072767e-01 -1.84186488e-01
-4.05211896e-01 8.63727570e-01 2.61337638e-01 -2.01102629e-01
4.30733442e-01 -1.30836383e-01 -7.21677780e-01 -9.67940539e-02
-5.61781704e-01 -4.57182318e-01 -1.10672429e-01 3.62030901e-02
3.64577584e-02 5.00559688e-01 -7.94045925e-01 7.28727043e-01
-2.91363627e-01 1.46507427e-01 5.33977926e-01 1.21100521e+00
-9.77981210e-01 -9.46698964e-01 -7.19150007e-01 8.80440652e-01
-8.24878514e-01 1.40199006e-01 -3.17471474e-01 7.21927285e-01
-2.61203915e-01 1.12425733e+00 -2.13797390e-01 3.07494134e-01
2.22556382e-01 -4.33585495e-01 1.00630379e+00 -3.88075054e-01
-2.73495287e-01 5.74434876e-01 1.21746980e-01 -2.73913234e-01
-3.06738645e-01 -7.06236303e-01 -1.06191874e+00 -4.50434178e-01
-4.89747405e-01 1.40654370e-01 5.92642367e-01 8.28144789e-01
-1.87769949e-01 1.63170785e-01 1.17546391e+00 -1.08117245e-01
-1.29864144e+00 -7.66592503e-01 -1.07266462e+00 -1.33506149e-01
5.15350521e-01 -5.25142908e-01 -2.09521279e-01 -7.58786857e-01] | [4.559925556182861, 3.3553757667541504] |
6728e29d-2520-4007-ace0-58edd02a2c63 | uni-em-an-environment-for-deep-neural-network | null | null | https://doi.org/10.1101/607366 | https://www.biorxiv.org/content/biorxiv/early/2019/04/12/607366.full-text.pdf | UNI-EM: An Environment for Deep Neural Network-Based Automated Segmentation of Neuronal Electron Microscopic Images | Recently, there has been a rapid expansion in the field of micro-connectomics, which targets the three-dimensional (3D) reconstruction of neuronal networks from a stack of two-dimensional (2D) electron microscopic (EM) images. The spatial scale of the 3D reconstruction grows rapidly owing to deep neural networks (DNNs) that enable automated image segmentation. Several research teams have developed their own software pipelines for DNN-based segmentation. However, the complexity of such pipelines makes their use difficult even for computer experts and impossible for non-experts. In this study, we developed a new software program, called UNI-EM, that enables 2D- and 3D-DNN-based segmentation for non-computer experts. UNI-EM is a software collection for DNN-based EM image segmentation, including ground truth generation, training, inference, postprocessing, proofreading, and visualization. UNI-EM comes with a set of 2D DNNs, i.e., U-Net, ResNet, HighwayNet, and DenseNet. We further wrapped flood-filling networks (FFNs) as a representative 3D DNN-based neuron segmentation algorithm. The 2D- and 3D-DNNs are known to show state-of-the-art level segmentation performance. We then provided two-example workflows: mitochondria segmentation using a 2D DNN as well as neuron segmentation using FFNs. Following these example workflows, users can benefit from DNN-based segmentation without any knowledge of Python programming or DNN frameworks. | ['Hidetoshi Urakubo', 'Yoshiyuki Kubota', 'Torsten Bullmann', 'Shin Ishii', 'Shigeyuki Oba'] | 2019-04-12 | null | null | null | biorxiv-neuroscience-2019-4 | ['electron-microscopy-image-segmentation'] | ['computer-vision'] | [-2.65012197e-02 -2.66593099e-02 4.77047920e-01 -1.62604034e-01
-1.83774993e-01 -5.57590067e-01 1.15970418e-01 8.62865001e-02
-9.02314484e-01 8.16537499e-01 -5.54614425e-01 -6.38041794e-01
2.21826985e-01 -7.66278148e-01 -7.38969326e-01 -7.16296554e-01
1.04001187e-01 7.18925834e-01 7.13405311e-01 1.33537725e-01
2.75494874e-01 8.84571671e-01 -1.22416568e+00 -8.06389526e-02
7.06159294e-01 6.60961032e-01 8.38659644e-01 7.67109513e-01
-4.72776473e-01 -1.88319082e-03 -4.09039736e-01 -2.06902981e-01
2.78608263e-01 -2.93060720e-01 -9.25810814e-01 -3.83983940e-01
1.53535726e-02 -3.60957891e-01 -8.96959603e-02 1.22733784e+00
7.94496894e-01 -2.72622168e-01 3.76195222e-01 -1.13459158e+00
-4.12365288e-01 4.60609496e-01 -2.41910845e-01 5.15983164e-01
-1.26461729e-01 4.58030462e-01 7.88704380e-02 -5.39531350e-01
1.08788931e+00 1.23805237e+00 7.08071351e-01 8.51984978e-01
-1.49855375e+00 -6.45573139e-01 -2.96734244e-01 -1.42274499e-01
-1.06557953e+00 6.93092942e-02 3.10975552e-01 -7.46301770e-01
1.08873296e+00 -3.26837689e-01 1.19431686e+00 8.29893708e-01
4.05566543e-01 5.59219718e-01 1.34785390e+00 -3.28253396e-02
5.49736500e-01 -3.93199295e-01 1.43026263e-01 3.48582804e-01
3.06383699e-01 -2.91693270e-01 1.50289536e-01 1.23634838e-01
1.51185250e+00 1.49952754e-01 5.53797930e-04 -6.97343722e-02
-1.27087426e+00 4.85986233e-01 3.29692781e-01 4.99586880e-01
-2.83790886e-01 3.18940962e-03 4.80890691e-01 -9.33863297e-02
2.41759747e-01 1.03449389e-01 -6.34449780e-01 -1.42929271e-01
-1.13006556e+00 3.28009665e-01 4.84747976e-01 7.47351766e-01
9.26549077e-01 -4.53624316e-02 3.07009816e-01 5.28074384e-01
3.94897819e-01 1.54950649e-01 5.73386788e-01 -1.30088735e+00
-1.84424251e-01 9.39061463e-01 -2.95197338e-01 -5.30903995e-01
-9.10073102e-01 -1.28150880e-02 -1.09033644e+00 6.76358104e-01
6.40142620e-01 -2.69820213e-01 -1.22577548e+00 1.70035744e+00
6.13658547e-01 1.03211333e-03 -1.70472458e-01 1.07836819e+00
1.03100204e+00 1.30617768e-01 8.39888379e-02 2.23143362e-02
1.18992209e+00 -5.42324185e-01 -4.63634789e-01 1.04221895e-01
7.96889424e-01 -3.66983950e-01 7.05219567e-01 1.88004002e-01
-1.19371080e+00 -3.00499380e-01 -8.75752449e-01 -3.40341508e-01
-8.04581523e-01 -3.23649555e-01 4.43535030e-01 1.74847245e-01
-1.48636675e+00 8.40928257e-01 -1.35236359e+00 -9.25734162e-01
1.00058138e+00 5.54511428e-01 -6.30606413e-01 2.58617222e-01
-7.61817276e-01 1.09746563e+00 5.22214711e-01 -1.01000763e-01
-1.13318729e+00 -8.75284970e-01 -5.08927345e-01 -1.45546362e-01
-3.03844094e-01 -1.00527716e+00 1.08854532e+00 -3.80148590e-01
-1.22704864e+00 1.49500024e+00 1.36070224e-02 -6.03502631e-01
2.58864582e-01 2.58642495e-01 2.17161365e-02 5.63973784e-01
1.62611887e-01 1.35832059e+00 -6.19487874e-02 -1.06855321e+00
-2.02983618e-01 -7.30403781e-01 1.32336738e-02 -2.19001368e-01
2.81308472e-01 8.97804275e-02 -1.79388091e-01 -2.18714669e-01
3.01374018e-01 -4.37440306e-01 -4.76706982e-01 5.77561200e-01
-5.89731991e-01 3.50664253e-03 1.20815492e+00 -6.59765601e-01
5.67698359e-01 -1.90339124e+00 9.53606069e-02 -1.55858293e-01
7.07134843e-01 5.01493812e-01 3.44028547e-02 -5.36588207e-02
-1.71362042e-01 5.48056424e-01 -4.84706968e-01 -3.49262983e-01
-1.64624393e-01 1.99541494e-01 5.55472493e-01 5.22445083e-01
-2.80525889e-02 9.18482780e-01 -7.47497439e-01 -7.27873504e-01
3.83500189e-01 6.13718987e-01 -2.79143512e-01 3.44477221e-02
-2.44395941e-01 8.23043525e-01 -1.57688975e-01 6.60034359e-01
9.87077713e-01 -3.54516178e-01 2.86656201e-01 -2.03014344e-01
-5.05966604e-01 5.33852763e-02 -7.17031300e-01 1.92036128e+00
3.43303904e-02 6.89539492e-01 7.41287708e-01 -1.13568759e+00
8.37653756e-01 2.59170920e-01 4.51562494e-01 -6.21409714e-01
2.82456458e-01 3.09735835e-01 5.71975000e-02 -2.72729337e-01
-1.66057616e-01 -1.52607873e-01 5.57717144e-01 5.36785424e-01
4.55209196e-01 -2.36482427e-01 6.37638450e-01 2.39785701e-01
1.11027026e+00 3.74855965e-01 5.05660586e-02 -5.85337460e-01
1.63681209e-02 1.15020886e-01 6.86758161e-01 2.51192749e-01
-5.01126885e-01 7.32139289e-01 7.63504326e-01 -5.30713916e-01
-1.59525657e+00 -1.12486291e+00 -4.96163845e-01 3.73421341e-01
1.37902573e-01 2.14468036e-02 -1.38565254e+00 -2.80157804e-01
-2.10548550e-01 2.10796282e-01 -4.10881996e-01 4.97085392e-01
-3.05442631e-01 -8.19331765e-01 8.53041053e-01 2.78376967e-01
7.69503474e-01 -1.23951697e+00 -9.24411952e-01 2.11998627e-01
1.94100052e-01 -1.21537721e+00 1.68791607e-01 4.70131099e-01
-1.17023826e+00 -1.17645240e+00 -1.13813341e+00 -1.03432250e+00
9.95056331e-01 1.18569553e-01 9.49052036e-01 2.75355190e-01
-5.32103658e-01 1.03905484e-01 1.05830841e-02 -2.25453645e-01
-2.20857337e-01 -6.81588799e-02 1.75480336e-01 -8.28697801e-01
3.32130849e-01 -1.33827543e+00 -8.03161621e-01 3.06064785e-01
-1.16958845e+00 4.64955628e-01 4.25543875e-01 3.03788304e-01
1.32945514e+00 -1.25067562e-01 3.85140240e-01 -6.24130428e-01
3.87283266e-01 -3.50087970e-01 -7.95760632e-01 2.34333463e-02
-2.99386442e-01 -3.29603851e-01 6.94709599e-01 -3.11286151e-01
-4.76500362e-01 2.32702456e-02 -5.99240422e-01 -3.58772188e-01
-7.94380128e-01 2.58442372e-01 -2.47405797e-01 -1.39724180e-01
4.51897144e-01 1.55727074e-01 3.77362311e-01 -5.48100114e-01
-1.29662111e-01 4.57368761e-01 5.13821244e-01 -4.46039110e-01
2.85947382e-01 5.85032880e-01 1.91799209e-01 -7.35352576e-01
-4.98553187e-01 -1.69737533e-01 -1.03290176e+00 -2.63547659e-01
1.47980499e+00 -4.75055069e-01 -9.16917682e-01 9.01815057e-01
-1.24368930e+00 -9.77774441e-01 -1.32039815e-01 2.81270921e-01
-5.22096992e-01 3.63406748e-01 -1.01995385e+00 -4.50942695e-01
-6.70790851e-01 -1.44271731e+00 7.64259279e-01 6.81458056e-01
-9.00463313e-02 -1.03817582e+00 2.58373380e-01 1.45319505e-02
3.95727754e-01 3.14027518e-01 1.09300160e+00 -7.67651916e-01
-4.62053955e-01 2.18127802e-01 -4.05930042e-01 2.56900162e-01
-3.07166815e-01 2.73813397e-01 -8.23108852e-01 2.42935941e-01
-2.61441797e-01 -2.30824798e-01 5.77697694e-01 9.21459973e-01
1.32195723e+00 8.88230428e-02 -5.84040225e-01 8.56455207e-01
1.27180326e+00 1.32886022e-01 8.69324505e-01 5.88258922e-01
5.15572309e-01 4.02231991e-01 -2.24504471e-01 -6.91434592e-02
4.33339685e-01 3.00851375e-01 6.89554811e-01 -5.03302097e-01
1.20165767e-02 1.55285060e-01 -1.65996194e-01 6.39171600e-01
-1.24666020e-01 4.76124473e-02 -1.00530148e+00 4.62471664e-01
-1.72442651e+00 -8.16643238e-01 -5.77824712e-01 1.71214747e+00
8.67819488e-01 9.34618562e-02 4.16610330e-01 -2.70204190e-02
9.93007958e-01 -5.17179608e-01 -1.04326665e+00 -2.81889975e-01
-4.25645828e-01 1.55879810e-01 2.49148935e-01 9.08841118e-02
-9.05972183e-01 9.46963310e-01 6.55216312e+00 6.67205691e-01
-1.07591200e+00 1.68780223e-01 9.25792336e-01 4.07226011e-02
-1.47318214e-01 7.62575716e-02 -8.14933598e-01 6.45338178e-01
9.48084056e-01 1.13225363e-01 5.20409942e-01 6.18543863e-01
4.77195323e-01 -4.96470630e-01 -7.90155828e-01 1.04027867e+00
-7.19767153e-01 -1.91491640e+00 -1.52426735e-01 3.17643225e-01
4.96092409e-01 6.45785928e-01 -6.32358134e-01 -4.02360074e-02
4.39564615e-01 -9.54817176e-01 5.33593297e-01 6.06616974e-01
8.63233387e-01 -6.99528992e-01 8.73209774e-01 5.28640628e-01
-8.69260192e-01 5.83032787e-01 -6.97703004e-01 9.77362245e-02
6.11906648e-01 1.07992232e+00 -3.23110431e-01 1.73369363e-01
1.11344814e+00 5.60363889e-01 -3.09296995e-01 1.31732821e+00
6.41299635e-02 3.44855934e-01 -5.41908383e-01 2.83628821e-01
-5.58254719e-02 -5.44985592e-01 3.10183465e-01 1.29067206e+00
2.85938799e-01 1.38214380e-01 -9.19793993e-02 1.58091128e+00
-1.59095004e-01 -3.40093315e-01 -3.91202539e-01 -4.25187677e-01
4.04099941e-01 1.99018550e+00 -1.68362498e+00 -3.40938658e-01
-6.44232854e-02 6.82067633e-01 3.35230410e-01 3.52807701e-01
-4.85573381e-01 -4.94720638e-01 8.49236608e-01 2.15680793e-01
1.91393256e-01 -4.11591232e-01 -5.45256317e-01 -7.18354225e-01
-5.20423710e-01 -3.99569750e-01 -7.62193576e-02 -1.09696507e+00
-1.25583494e+00 3.52590382e-01 -2.22361892e-01 -3.94673049e-01
2.74623692e-01 -9.37540531e-01 -7.99791038e-01 7.75726914e-01
-1.01950574e+00 -7.38234520e-01 -4.28627506e-02 3.45145702e-01
7.96058998e-02 2.71051794e-01 7.81232059e-01 4.58412260e-01
-9.20973122e-01 -1.99395224e-01 -3.16589314e-04 3.41987133e-01
1.34018630e-01 -1.29742265e+00 4.64222908e-01 6.07979953e-01
-4.97948468e-01 7.36697435e-01 3.57833326e-01 -7.63355613e-01
-9.42759693e-01 -9.70597565e-01 6.33254945e-01 -1.15021490e-01
6.08167052e-01 -4.37214434e-01 -8.28903079e-01 7.37704396e-01
-3.01306006e-02 1.08267151e-01 7.57148623e-01 -5.49189687e-01
2.75812089e-01 3.47932667e-01 -1.75729632e+00 7.08140373e-01
1.04856479e+00 -3.43761832e-01 -2.37570316e-01 2.50302404e-01
7.40124404e-01 -3.25611204e-01 -1.12592280e+00 5.45887128e-02
5.12185395e-01 -1.19173491e+00 8.91726673e-01 -2.45906979e-01
4.46083963e-01 -6.53457165e-01 6.37279525e-02 -1.00612617e+00
-1.37455627e-01 -3.79002869e-01 7.92691763e-03 1.12401414e+00
3.20024669e-01 -5.61764359e-01 6.75472438e-01 6.07279122e-01
-6.31805122e-01 -8.56218696e-01 -9.71832156e-01 -5.66811442e-01
3.71793866e-01 -3.82073581e-01 4.77583319e-01 6.55701101e-01
8.26482028e-02 8.62240791e-02 6.14197552e-01 -2.54906654e-01
6.81375861e-01 -2.19216034e-01 5.33239245e-01 -1.44770551e+00
3.00359637e-01 -6.44273043e-01 -7.18465924e-01 -1.03632665e+00
1.30669951e-01 -1.10495639e+00 -9.50316265e-02 -2.21972346e+00
5.25208950e-01 1.19061612e-01 2.40287110e-01 6.37806892e-01
3.24926764e-01 3.84275526e-01 -3.12337838e-02 2.49488875e-01
-6.82749808e-01 1.03095010e-01 1.62835920e+00 2.93558508e-01
-5.77356331e-02 -3.96302164e-01 -5.02389491e-01 9.81638908e-01
1.02845573e+00 -4.57916260e-01 -5.87925985e-02 -5.03160834e-01
6.13674037e-02 -2.40736157e-01 7.63095200e-01 -1.21649623e+00
4.13596243e-01 1.89564317e-01 7.91487098e-01 -1.08708465e+00
-4.31456305e-02 -5.69724441e-01 2.63206482e-01 3.80994856e-01
1.03025936e-01 1.24100208e-01 3.49202752e-01 -1.58698410e-01
1.64075404e-01 -2.25819737e-01 1.06679606e+00 -7.83926427e-01
-3.51636708e-01 4.35338676e-01 -9.43816841e-01 8.19823742e-02
9.32556510e-01 -7.45573759e-01 -6.94463372e-01 1.79884527e-02
-1.00103056e+00 2.04177663e-01 1.04066575e+00 -7.21291423e-01
5.82250714e-01 -1.02589607e+00 -1.68500934e-02 1.64726134e-02
-6.64862573e-01 6.94175541e-01 4.99448121e-01 1.34440839e+00
-1.01247275e+00 2.94546664e-01 -7.84086287e-01 -7.07952797e-01
-7.73138642e-01 3.37925971e-01 6.07417464e-01 -2.28341490e-01
-8.28548551e-01 7.10196376e-01 2.31488273e-01 -1.06136799e+00
-7.26390257e-02 -4.56505954e-01 -1.30048186e-01 -2.86194503e-01
5.39223850e-01 5.99174917e-01 -3.39152105e-02 -4.22741890e-01
-3.43586326e-01 5.38631260e-01 1.69627666e-01 6.41073007e-03
1.63435864e+00 -3.19725037e-01 -6.18112683e-01 4.59218413e-01
8.78836036e-01 -9.86680806e-01 -1.63991201e+00 4.60614115e-01
-1.42271042e-01 2.84780473e-01 1.22617379e-01 -8.47534657e-01
-1.34387839e+00 1.25920916e+00 5.95397770e-01 1.96326226e-01
9.46308911e-01 3.69236665e-03 1.18402827e+00 1.16407752e-01
4.66675520e-01 -1.01189160e+00 -2.72595227e-01 7.28506863e-01
2.09283382e-01 -7.78437853e-01 -2.22666308e-01 1.43649906e-01
6.02329895e-02 1.28055406e+00 8.79031181e-01 -4.77189235e-02
7.55183399e-01 7.22698569e-01 1.01635061e-01 -5.24075091e-01
-4.06727940e-01 -2.34618038e-01 -2.95488864e-01 1.03565764e+00
3.28809321e-01 -2.02738166e-01 -5.88145852e-02 8.41660023e-01
-1.36618108e-01 3.56496125e-01 4.59627330e-01 9.44434762e-01
-6.50084376e-01 -8.88819516e-01 -2.16526419e-01 5.66630721e-01
-5.25315166e-01 -6.83596656e-02 -4.65524286e-01 7.56669700e-01
3.00889879e-01 4.56238180e-01 4.45108235e-01 -1.89975098e-01
-1.14608750e-01 2.85116434e-02 5.33169627e-01 -5.34230113e-01
-5.34374535e-01 1.73517376e-01 -4.14181530e-01 -5.20635068e-01
-5.46234071e-01 -3.38564247e-01 -2.20748496e+00 -7.39238679e-01
1.01491138e-02 -1.67689636e-01 9.90718663e-01 1.12764001e+00
6.78881407e-01 4.25856769e-01 -2.95560658e-01 -1.44511986e+00
5.65424323e-01 -9.43254173e-01 -9.28240001e-01 -6.14033826e-02
-9.85186696e-02 -4.08761740e-01 -3.53516608e-01 1.94618583e-01] | [14.265244483947754, -3.10966420173645] |
d90cc550-89f8-47aa-ae3c-9996de6286d8 | chatgpt-is-more-likely-to-be-perceived-as | 2305.12564 | null | https://arxiv.org/abs/2305.12564v1 | https://arxiv.org/pdf/2305.12564v1.pdf | ChatGPT Is More Likely to Be Perceived as Male Than Female | We investigate how people perceive ChatGPT, and, in particular, how they assign human-like attributes such as gender to the chatbot. Across five pre-registered studies (N = 1,552), we find that people are more likely to perceive ChatGPT to be male than female. Specifically, people perceive male gender identity (1) following demonstrations of ChatGPT's core abilities (e.g., providing information or summarizing text), (2) in the absence of such demonstrations, and (3) across different methods of eliciting perceived gender (using various scales and asking to name ChatGPT). Moreover, we find that this seemingly default perception of ChatGPT as male can reverse when ChatGPT's feminine-coded abilities are highlighted (e.g., providing emotional support for a user). | ['Jin Kim', 'Jared Wong'] | 2023-05-21 | null | null | null | null | ['chatbot', 'chatbot'] | ['methodology', 'natural-language-processing'] | [-2.71330714e-01 6.80199802e-01 3.44024897e-02 -4.09386635e-01
-2.75803894e-01 -8.88914645e-01 7.53888607e-01 1.40344635e-01
-3.84541899e-01 5.34440100e-01 5.14978886e-01 -2.55340189e-01
3.25006694e-01 -5.71558177e-01 3.01625639e-01 -5.23525953e-01
7.55250573e-01 5.36302984e-01 -3.86028767e-01 -2.02053756e-01
8.64074707e-01 -5.29153980e-02 -1.17220902e+00 1.11351442e-02
7.33989775e-01 3.44972044e-01 6.42539412e-02 5.00014544e-01
-7.12293386e-02 9.38656747e-01 -1.13793409e+00 -8.82886052e-01
-2.70822167e-01 -6.33584440e-01 -1.00563943e+00 -4.26152945e-02
4.85142916e-01 -7.28803992e-01 -1.08804509e-01 8.31020594e-01
3.36392701e-01 7.41516575e-02 6.68316126e-01 -1.58618331e+00
-9.47732985e-01 6.36057198e-01 -1.91765562e-01 -2.15223268e-01
8.23321223e-01 5.69207609e-01 1.04465270e+00 -4.91673410e-01
9.25676167e-01 1.41490984e+00 7.80868709e-01 9.31882024e-01
-1.57707167e+00 -8.41594577e-01 -2.41054013e-01 -2.89566606e-01
-1.02036417e+00 -6.24485731e-01 3.23935896e-01 -7.10491657e-01
7.41814435e-01 4.52426255e-01 7.39028573e-01 1.40597618e+00
-6.58808276e-02 -6.57988945e-03 1.62438583e+00 5.13438098e-02
1.84640199e-01 6.08657122e-01 2.58727763e-02 1.14659041e-01
3.56912374e-01 -4.15012270e-01 -5.96796095e-01 -5.54106712e-01
6.48283899e-01 -3.15877438e-01 2.02844635e-01 4.73714888e-01
-1.30081689e+00 8.10965896e-01 1.69682860e-01 5.17125309e-01
-4.48811978e-01 1.40900001e-01 6.70133173e-01 4.42005545e-01
2.91441828e-01 8.72613966e-01 -1.16203330e-01 -1.26908708e+00
-3.17032337e-01 3.06227028e-01 1.19955254e+00 8.90373468e-01
4.54231620e-01 -5.46363890e-02 -3.04453045e-01 8.27704608e-01
-4.05444764e-02 3.59495997e-01 2.28543356e-01 -1.59022474e+00
4.65252191e-01 5.97367644e-01 5.79973519e-01 -9.53293681e-01
-4.11422640e-01 2.59100378e-01 -5.28825760e-01 3.82335275e-01
1.02979279e+00 -4.35462087e-01 1.00347511e-01 2.08620358e+00
1.81898788e-01 -7.66285002e-01 -1.27096415e-01 9.10267889e-01
8.88047099e-01 2.09048003e-01 4.99004662e-01 -1.68390512e-01
1.62237275e+00 -1.93507761e-01 -7.27783203e-01 -4.69501466e-01
6.94775224e-01 -8.25533330e-01 1.40822041e+00 1.47180736e-01
-9.31128144e-01 -5.39844871e-01 -5.88075876e-01 -1.68446824e-01
-2.25566119e-01 -1.65574644e-02 6.82047367e-01 1.37370539e+00
-1.20512974e+00 4.96078074e-01 -3.58010471e-01 -9.88992751e-01
4.60363664e-02 6.30962253e-02 -5.75280607e-01 1.98032171e-01
-8.63416672e-01 1.15058064e+00 -1.36943564e-01 -3.71843250e-03
-4.66660321e-01 -4.13381636e-01 -7.72329271e-01 -3.39947245e-03
-1.05983140e-02 -7.04255462e-01 1.52579176e+00 -9.68031287e-01
-1.35852134e+00 1.22504056e+00 -1.29894495e-01 9.46652144e-02
5.56023896e-01 -6.24523275e-02 1.03562736e-04 2.62037218e-01
6.57716632e-01 8.87466967e-01 7.33968675e-01 -1.11020434e+00
-4.34511423e-01 -4.47340816e-01 7.11722493e-01 5.29282987e-01
-3.93794119e-01 5.46629071e-01 5.78220725e-01 -2.41647184e-01
1.93219736e-01 -8.20510864e-01 2.07014278e-01 3.29050310e-02
-8.25890005e-01 -5.80249548e-01 9.17469740e-01 -1.00455272e+00
8.53054404e-01 -2.29251194e+00 -2.88793325e-01 -1.64402947e-02
9.84006047e-01 -3.98564637e-01 5.09621382e-01 9.91696835e-01
2.45230049e-01 6.62216485e-01 3.21266294e-01 -3.71172518e-01
6.63900912e-01 -4.64632688e-03 3.52017522e-01 6.18790269e-01
-2.68525425e-02 7.20245957e-01 -8.82731915e-01 -6.11545324e-01
-1.63798958e-01 3.73113483e-01 -2.79761463e-01 1.00681283e-01
4.01089132e-01 6.45429552e-01 5.44581451e-02 3.90651733e-01
5.80458224e-01 -9.03505906e-02 5.47577918e-01 3.53058487e-01
-5.32093525e-01 8.07893932e-01 -6.12240493e-01 6.14087224e-01
-5.72527528e-01 9.83616948e-01 7.39272118e-01 -3.57000083e-01
1.02290654e+00 5.44659615e-01 -1.12315744e-01 -2.95854419e-01
2.71014869e-01 1.68563530e-01 6.09734595e-01 -5.82700491e-01
6.23930514e-01 -3.40401053e-01 -4.94826972e-01 1.11821020e+00
-5.49133494e-02 -3.30318928e-01 2.34346405e-01 3.60229522e-01
8.97831261e-01 -1.69922516e-01 3.81790638e-01 -4.12951380e-01
-5.60208224e-02 -1.84486300e-01 3.72555286e-01 8.40845585e-01
-5.34428537e-01 7.51903355e-02 1.25113904e+00 4.63919900e-02
-1.12507892e+00 -6.81405723e-01 2.81192005e-01 1.65200424e+00
1.04396860e-03 -2.80832171e-01 -8.47275913e-01 -4.93614197e-01
-1.85665116e-02 1.10950112e+00 -6.66833103e-01 -2.02739581e-01
-2.46415794e-01 -4.86655943e-02 8.14217567e-01 5.00284076e-01
8.07579458e-01 -1.19289052e+00 -1.04556942e+00 -1.33866325e-01
-7.68779695e-01 -9.19095159e-01 -6.69522166e-01 -1.70787215e-01
-7.33154476e-01 -6.48192644e-01 -4.35250729e-01 -5.95316529e-01
5.33532560e-01 2.91013390e-01 1.13093281e+00 8.74719024e-01
9.70317721e-02 5.85963786e-01 -3.70036393e-01 6.54890910e-02
-5.82935452e-01 2.23857686e-02 -1.30382970e-01 -4.45170671e-01
3.82897466e-01 -8.49936306e-01 -4.52553183e-01 4.88905907e-01
-2.87252218e-02 9.25986022e-02 3.75923514e-01 4.65290457e-01
-7.51109600e-01 -3.77185732e-01 6.42699420e-01 -9.75536287e-01
1.02884066e+00 -5.55159211e-01 2.93373644e-01 -2.97468811e-01
-2.82897681e-01 -8.30130160e-01 4.99504000e-01 -6.09489679e-01
-1.16841674e+00 -7.45544970e-01 5.87309338e-03 4.61255461e-01
-6.16881728e-01 1.47838285e-02 2.48218924e-02 -8.04345608e-02
7.39111841e-01 -8.63153636e-02 3.81557554e-01 3.65596614e-03
9.31927189e-02 8.95206213e-01 3.80436718e-01 -8.69975686e-01
8.00124288e-01 1.74199015e-01 -4.65565145e-01 -7.69107282e-01
-3.54733407e-01 -3.00500751e-01 -3.82795185e-01 -7.01443553e-01
1.08617938e+00 -6.47865832e-01 -1.90366042e+00 3.88426542e-01
-9.04810190e-01 -8.19633067e-01 2.91006535e-01 5.48117399e-01
-5.53889215e-01 2.28802323e-01 -1.07929981e+00 -1.34026909e+00
-1.03585847e-01 -6.24098003e-01 8.54643643e-01 3.95429969e-01
-1.39735091e+00 -8.69985878e-01 -4.08785373e-01 6.12353623e-01
5.06730139e-01 4.20577675e-02 1.04670966e+00 -8.77110183e-01
1.36717230e-01 3.83173227e-02 -4.57195163e-01 -1.31543249e-01
3.13578159e-01 1.51452526e-01 -8.07263494e-01 6.71835020e-02
2.26085752e-01 -7.03190386e-01 -9.32767540e-02 2.40713023e-04
4.23293889e-01 -8.47588718e-01 -2.12327942e-01 -1.02711700e-01
8.30427110e-01 3.66417885e-01 6.84952021e-01 2.16919631e-01
3.15242350e-01 1.00298321e+00 7.23874569e-01 4.18184608e-01
7.73726404e-01 5.87523878e-01 3.83111686e-01 6.46266937e-02
3.03711355e-01 -2.92845398e-01 4.21462327e-01 9.11809430e-02
-5.10340154e-01 2.21386939e-01 -6.77848816e-01 2.33854622e-01
-1.44626594e+00 -1.13992584e+00 -4.32852626e-01 1.86141622e+00
7.36547828e-01 2.94033140e-01 6.92772448e-01 -7.15158740e-03
9.93252099e-01 6.76965639e-02 -5.14180958e-01 -8.05740356e-01
2.03578904e-01 -1.53930128e-01 1.33206714e-02 2.96164006e-01
-5.88604033e-01 7.30707586e-01 6.48911572e+00 2.21244901e-01
-7.89301515e-01 3.16599250e-01 1.03900766e+00 3.31456214e-01
-9.11567360e-02 1.41038030e-01 -2.88780212e-01 3.21042776e-01
4.64272827e-01 -2.95674115e-01 4.94285613e-01 8.07068586e-01
1.57303199e-01 -4.56364393e-01 -1.23312640e+00 9.14583206e-01
-2.61747986e-01 -5.72660983e-01 -8.02057862e-01 7.03546852e-02
4.22369726e-02 -1.11182690e+00 7.99859092e-02 6.88945055e-01
3.11549425e-01 -1.09587216e+00 1.06115341e+00 -7.70376176e-02
9.36238468e-01 -2.97594398e-01 7.93305874e-01 4.50680584e-01
-7.27311373e-01 -1.00285508e-01 -1.22866802e-01 -1.08367908e+00
1.56160295e-01 -1.24684624e-01 -1.16220796e+00 -9.60888565e-02
7.94817805e-01 2.64457285e-01 -5.39536536e-01 3.27296108e-01
-4.77211088e-01 6.91822171e-01 -1.08359195e-01 -6.32906854e-01
-2.12758869e-01 -4.84373599e-01 4.41245854e-01 9.25681055e-01
3.13286453e-01 1.99468508e-01 -3.24926525e-01 1.25575554e+00
8.03246126e-02 -1.48600042e-01 -4.96725351e-01 -4.46656078e-01
1.07787418e+00 1.62381244e+00 -1.11364758e+00 -4.10419434e-01
-2.47842431e-01 8.72102380e-01 1.76529754e-02 4.73589480e-01
-5.45279860e-01 -4.18141305e-01 9.44420099e-01 2.09101364e-01
-3.17207664e-01 -2.86249667e-01 -9.09867287e-01 -5.14754772e-01
-2.53485054e-01 -6.99712932e-01 -7.15909302e-02 -1.16234052e+00
-1.33133960e+00 1.33263811e-01 -1.33719355e-01 -8.36499751e-01
-2.00767130e-01 -1.96813032e-01 -8.94552946e-01 9.37774658e-01
-3.93516533e-02 -9.26644921e-01 -5.57938218e-01 2.15778828e-01
3.58253270e-02 3.55221838e-01 8.06222320e-01 2.67576259e-02
-6.15869053e-02 5.89007974e-01 -6.99783087e-01 3.53061646e-01
1.12117255e+00 -1.43647122e+00 4.00001407e-01 1.99510530e-02
-6.38694406e-01 1.20596206e+00 9.71497178e-01 -7.70439923e-01
-1.03741002e+00 -1.72979102e-01 1.30158746e+00 -6.26983106e-01
6.18866444e-01 -8.27343285e-01 -4.17159081e-01 8.16801012e-01
4.89901721e-01 -9.41364765e-01 6.43751800e-01 6.05282068e-01
-5.38771152e-01 4.24083620e-01 -1.83035719e+00 8.72604728e-01
1.47532153e+00 -7.86748827e-01 -7.44946003e-01 3.23237002e-01
5.37012875e-01 -3.36525321e-01 -8.25725794e-01 -5.57385743e-01
9.01078403e-01 -1.46013963e+00 5.58563590e-01 -4.03262377e-01
6.34581685e-01 3.24079782e-01 1.78927466e-01 -1.25838375e+00
-3.52947593e-01 -9.13351059e-01 7.06501305e-01 1.63093090e+00
8.16262886e-02 -1.14366663e+00 5.27987421e-01 1.11054277e+00
-2.30308145e-01 -3.02648306e-01 -1.07644892e+00 -3.51153523e-01
-2.60683522e-02 1.93513006e-01 3.76304895e-01 1.25419188e+00
7.36175776e-01 4.11285758e-01 -1.81160808e-01 -4.76080000e-01
1.16438672e-01 -2.17602417e-01 8.52533400e-01 -1.43968558e+00
-1.43412445e-02 -7.97549486e-01 -3.29338670e-01 -6.70947969e-01
-1.04796439e-01 -3.01740825e-01 9.91153270e-02 -1.66134024e+00
5.01902640e-01 -3.78969520e-01 9.07995820e-01 5.53013325e-01
-1.51934013e-01 2.52215952e-01 4.20252770e-01 8.91043916e-02
-3.37236971e-01 1.31685019e-01 1.15007389e+00 3.74415785e-01
-1.57060791e-02 8.82407799e-02 -1.49913013e+00 6.61177456e-01
1.07563972e+00 -4.05227929e-01 -1.43907025e-01 3.34388107e-01
5.50718486e-01 3.99053156e-01 8.09342623e-01 -5.46173155e-01
-4.27793153e-02 -4.14720565e-01 2.10342884e-01 5.16626984e-03
6.64354801e-01 -3.67147565e-01 8.58679265e-02 5.38288355e-01
-1.23088129e-01 2.82963544e-01 -1.59263745e-01 -1.66449085e-01
6.54053614e-02 -3.11459839e-01 4.70057666e-01 -2.58737653e-01
1.07290238e-01 -6.23671293e-01 -1.04741621e+00 2.68851012e-01
7.83068180e-01 -6.01910353e-01 -8.90995502e-01 -1.31252086e+00
-6.13158643e-01 -3.06506157e-02 9.46306050e-01 3.51580009e-02
1.73428521e-01 -1.09865665e+00 -3.74681294e-01 -3.51627529e-01
-4.77854572e-02 -6.87848270e-01 2.58379310e-01 1.37516785e+00
-2.33626738e-01 -7.63856806e-03 -4.34488267e-01 -1.64464667e-01
-1.33377469e+00 -2.32537761e-02 -2.18170807e-02 3.82808030e-01
-4.00935531e-01 9.77394283e-01 3.56343120e-01 -5.60572088e-01
-9.77122933e-02 5.10618016e-02 -2.39664778e-01 3.83913875e-01
2.16187000e-01 5.64583123e-01 -5.21611273e-01 -7.17918158e-01
-3.73185575e-01 8.16788599e-02 2.40878776e-01 -5.25306940e-01
9.43003595e-01 -1.89430043e-01 -5.31814754e-01 7.47708261e-01
6.80356920e-01 2.54575491e-01 -9.60105658e-01 5.66880465e-01
-1.97374359e-01 -8.54574144e-01 -1.10098529e+00 -5.38303733e-01
-3.55718464e-01 4.94453162e-01 -2.19087258e-01 8.82561445e-01
5.50511122e-01 2.94164985e-01 3.32715124e-01 3.43442947e-01
6.83791757e-01 -1.31786561e+00 4.22805429e-01 3.43329906e-01
8.07991803e-01 -9.14553940e-01 -2.07785040e-01 -6.47037566e-01
-1.07680559e+00 1.06402004e+00 1.17471814e+00 1.03604466e-01
2.44502544e-01 -1.54398337e-01 1.34034187e-01 -3.92563462e-01
-9.65276122e-01 2.60132998e-01 -5.22117376e-01 1.02518082e+00
9.60532904e-01 3.38847697e-01 -8.96143436e-01 7.98160017e-01
-1.05828023e+00 -5.13529122e-01 9.39070702e-01 6.18575096e-01
-2.81005740e-01 -6.54010177e-01 -4.98100489e-01 7.57392704e-01
-3.44168365e-01 7.17670321e-02 -1.07411504e+00 1.10747421e+00
1.59338173e-02 1.47098470e+00 1.93396837e-01 -4.33454752e-01
6.72263578e-02 2.36146837e-01 3.54903013e-01 -8.60380173e-01
-1.20537972e+00 -2.87735850e-01 9.64059353e-01 -2.35996589e-01
-3.84314924e-01 -1.05838573e+00 -8.49629760e-01 -9.39514339e-01
3.50623466e-02 1.91256195e-01 1.87028244e-01 9.99737084e-01
4.20359969e-02 -1.30565196e-01 2.52722323e-01 -8.19578171e-01
-3.43830317e-01 -1.71472216e+00 -7.22767115e-01 5.57232141e-01
-1.56808600e-01 -7.51271605e-01 -6.41113698e-01 -1.57324120e-01] | [12.470973014831543, 7.773597717285156] |
58cb80f1-6b91-417f-9d20-67e3b449127d | masking-of-quantum-information-into | 1910.00938 | null | https://arxiv.org/abs/1910.00938v4 | https://arxiv.org/pdf/1910.00938v4.pdf | Masking of Quantum Information into Restricted Set of states | Masking of data is a method to protect information by shielding it from a third party, however keeping it usable for further usages like application development, building program extensions to name a few. Whereas it is possible for classical information encoded in composite quantum states to be completely masked from reduced sub-systems, it has to be checked if quantum information can also be masked when the future possibilities of a quantum computer are increasing day by day. Newly proposed no-masking theorem [Phys. Rev. Lett. 120, 230501 (2018)], one of the no-go theorems, demands that except for some restricted sets of non-orthogonal states, it's impossible to mask arbitrary quantum states. Here, we explore the possibility of masking in the IBM quantum experience platform by designing the quantum circuits and running them on the 5-qubit quantum computer. We choose two particular states considering both the orthogonal and non-orthogonal basis states and illustrate their masking through both the theoretical calculation as well as verification in the quantum computer. By quantum state tomography, it is concluded that the experimental results are collected with high fidelity and hence the possibility of masking is realized. | ['Prasanta K. Panigrahi', 'Bikash K. Behera', 'Soumya Sarkar', 'Tamal Ghosh'] | 2019-10-01 | null | null | null | null | ['quantum-state-tomography'] | ['medical'] | [ 6.14122391e-01 3.77944469e-01 1.65154472e-01 -1.63068935e-01
-6.35253370e-01 -6.90391302e-01 4.43741411e-01 -7.92216435e-02
-4.06698763e-01 9.99545395e-01 -1.60416767e-01 -8.32791030e-01
1.14252470e-01 -1.04554701e+00 -5.27254403e-01 -1.00299418e+00
-2.23298177e-01 1.63710322e-02 2.63331085e-01 -5.17395020e-01
5.22452652e-01 4.86236334e-01 -1.75576174e+00 4.33111697e-01
8.03272307e-01 3.87773246e-01 -2.42920846e-01 6.52645290e-01
2.71977812e-01 5.08320689e-01 -4.28412914e-01 -2.92982966e-01
7.03703821e-01 -7.00565815e-01 -9.75684524e-01 -2.84008801e-01
-1.00697197e-01 -2.88211197e-01 -5.54133594e-01 1.72481024e+00
1.88242644e-01 -3.34243715e-01 1.70312032e-01 -1.00256705e+00
-4.73968804e-01 6.91580296e-01 8.28485878e-04 -7.78918415e-02
4.93015051e-01 4.80571419e-01 7.22245395e-01 -1.41150296e-01
6.04424894e-01 7.12377727e-01 4.48242091e-02 5.72723448e-01
-1.39732635e+00 -8.03920090e-01 -8.41170847e-01 2.69991308e-01
-1.75965631e+00 -4.56737518e-01 2.99912691e-01 -1.93830937e-01
1.19701457e+00 6.72278762e-01 3.39792937e-01 5.10648549e-01
5.96164346e-01 -1.81167200e-01 1.84225178e+00 -8.74522388e-01
3.17736208e-01 6.69050813e-01 4.54497486e-01 4.98986244e-01
3.38136882e-01 8.62513483e-01 -3.45369428e-01 -3.14375967e-01
1.74112141e-01 -4.78812903e-01 -2.64264435e-01 -1.93100765e-01
-1.39115834e+00 5.45154214e-01 2.39727482e-01 5.11180043e-01
-1.39603153e-01 -3.04786973e-02 7.14484677e-02 6.93462133e-01
-3.51257205e-01 3.23743016e-01 -1.00006260e-01 -1.25303892e-02
-5.83894432e-01 1.19363934e-01 8.32433760e-01 9.83412981e-01
1.23774922e+00 -3.69695485e-01 -1.90203831e-01 -3.68399918e-01
1.10307872e-01 6.30217552e-01 3.54640447e-02 -5.09496868e-01
1.00313984e-01 5.05556881e-01 2.97781378e-01 -3.23400825e-01
-4.91708606e-01 -1.22868707e-02 -8.77404511e-01 4.90082622e-01
2.99755394e-01 -4.98966128e-02 -8.53277147e-01 1.79577005e+00
3.05688232e-01 8.39261264e-02 4.50821191e-01 8.73706281e-01
6.08404875e-01 5.81061959e-01 -3.12053770e-01 -3.95553112e-01
1.60924923e+00 -2.43556738e-01 -9.73608136e-01 2.01594174e-01
1.03140450e+00 -8.67613018e-01 3.37172568e-01 5.17179847e-01
-7.45086312e-01 -4.67500567e-01 -1.62376559e+00 1.49615824e-01
-5.14681756e-01 -3.47914577e-01 7.16466844e-01 1.67914248e+00
-1.01390076e+00 8.68377686e-01 -6.57513380e-01 7.00818524e-02
-8.53099003e-02 7.73929358e-01 -6.94106936e-01 7.19810138e-03
-1.73410535e+00 1.11176777e+00 8.30063939e-01 1.05785765e-01
-6.06755257e-01 1.33015797e-01 -4.61864799e-01 8.80462080e-02
6.92990571e-02 -3.73288572e-01 6.24252141e-01 -3.53683591e-01
-1.60356259e+00 1.08531582e+00 -1.19771250e-01 -3.64325494e-01
-1.72215533e-02 5.27885199e-01 -1.01221097e+00 5.68163916e-02
-1.22892305e-01 1.00335330e-01 6.70037329e-01 -7.22808897e-01
-3.55359405e-01 -3.57075036e-01 4.77983356e-01 -1.68606535e-01
1.31620392e-01 1.56168163e-01 2.83184439e-01 5.87249994e-01
4.77230042e-01 -1.35858321e+00 -3.58456165e-01 -8.36124361e-01
-9.26025927e-01 2.42033929e-01 4.90321279e-01 -2.11321026e-01
1.14069867e+00 -2.34797764e+00 -1.66472107e-01 5.95153987e-01
-2.09386498e-02 4.28126782e-01 1.87156916e-01 8.33303273e-01
-5.37655652e-01 2.50761569e-01 -3.36817145e-01 1.18238181e-01
2.08637476e-01 -2.38735937e-02 -2.74249166e-01 1.08256555e+00
1.78047970e-01 6.42596066e-01 -5.12891233e-01 -1.70951605e-01
3.08443159e-01 1.74822539e-01 -6.44765854e-01 -1.96464881e-01
2.68463910e-01 8.29620242e-01 -1.07268579e-01 3.69815737e-01
1.33300805e+00 9.72448662e-03 3.92656207e-01 -4.78722379e-02
-6.24904513e-01 6.80639446e-01 -1.61462557e+00 1.46408224e+00
2.05886632e-01 2.36225963e-01 1.32434675e-02 -7.17046916e-01
5.86942673e-01 6.48345470e-01 -9.39905494e-02 -8.97733092e-01
3.54311466e-01 3.31144661e-01 6.97546124e-01 -3.45523000e-01
9.08799946e-01 -6.98164284e-01 -3.63853931e-01 8.12791526e-01
-1.13201641e-01 -3.31107497e-01 2.10490614e-01 2.64356047e-01
1.33744740e+00 -1.62108779e-01 3.54004502e-01 -5.43089867e-01
8.44240129e-01 6.13676570e-02 3.59855175e-01 8.44699085e-01
-3.24352294e-01 2.81532019e-01 4.84213710e-01 -5.54595739e-02
-1.22355783e+00 -1.02823436e+00 -7.08716631e-01 2.68152028e-01
6.08430147e-01 -6.98352754e-01 -8.05277526e-01 -7.11473748e-02
-3.61141145e-01 8.83389652e-01 -2.38675877e-01 -5.23648262e-01
-2.52891779e-01 -9.35050249e-01 6.70793712e-01 -3.76279414e-01
5.60327828e-01 -9.98215497e-01 -5.71346104e-01 -6.47322685e-02
1.63284659e-01 -1.09691894e+00 2.49742985e-01 5.55508673e-01
-6.39702857e-01 -8.32016408e-01 3.55577141e-01 -2.75834531e-01
6.64892972e-01 2.98104405e-01 6.41769290e-01 4.57253605e-01
-1.00021280e-01 -1.76577881e-01 -3.24836195e-01 -2.84437016e-02
-1.02584577e+00 -3.14011157e-01 4.21873063e-01 -1.43573657e-01
6.73983157e-01 -5.31896055e-01 -4.60192531e-01 1.29604444e-01
-1.11957812e+00 -3.15325707e-02 5.91118336e-01 6.47613645e-01
2.16627672e-01 4.43995267e-01 -1.33283073e-02 -9.24857020e-01
1.70438439e-01 -1.86024606e-01 -7.58390367e-01 2.48068683e-02
-3.30752164e-01 3.86764258e-01 6.24724507e-01 2.33982831e-01
-7.17723012e-01 9.68394801e-02 -1.93242580e-01 3.63312870e-01
-4.56660062e-01 2.25824699e-01 -3.55808944e-01 -3.27451676e-01
8.00495923e-01 3.72052014e-01 -2.88744181e-01 4.45592366e-02
5.43250978e-01 8.39014888e-01 3.32606912e-01 -2.73703068e-01
1.28011310e+00 7.11382508e-01 5.88908613e-01 -8.49749863e-01
-3.36216062e-01 -3.28411013e-01 -8.81170809e-01 1.94572508e-01
9.45956528e-01 -6.93160474e-01 -1.22345102e+00 2.35768184e-01
-1.12927914e+00 4.22703177e-01 -4.75079473e-03 5.93323112e-01
-2.74167925e-01 7.90043533e-01 -4.13552791e-01 -1.16419709e+00
8.58861059e-02 -1.46797562e+00 7.64752984e-01 1.23182349e-01
-5.19881137e-02 -4.23494697e-01 7.06591383e-02 3.20044607e-01
1.70327231e-01 4.22910973e-02 7.25722015e-01 -5.12923837e-01
-1.24723983e+00 -5.71783721e-01 -8.03092495e-02 3.68669271e-01
4.47398238e-02 -1.60890087e-01 -1.37571514e+00 -5.30219078e-01
4.08504725e-01 -1.65402800e-01 6.60704255e-01 -2.84430593e-01
6.33322835e-01 8.74087289e-02 -1.64045066e-01 4.32870626e-01
1.50576389e+00 2.32119843e-01 1.40374756e+00 1.39330879e-01
2.91370153e-01 3.27883244e-01 4.53391910e-01 1.18854024e-01
-2.20412210e-01 6.33774459e-01 5.72050691e-01 2.84853518e-01
4.62618142e-01 4.14596349e-02 4.38573629e-01 1.02264881e+00
-4.45002690e-02 1.45193800e-01 -5.34924924e-01 5.11729233e-02
-1.25843823e+00 -1.22782660e+00 -8.32067549e-01 2.75131249e+00
6.74872935e-01 4.24753696e-01 -4.16141152e-01 4.52176362e-01
8.62973928e-01 -1.40541568e-01 3.04399058e-02 -9.12114680e-01
-2.95103043e-01 7.76545227e-01 8.77705157e-01 7.14352667e-01
-1.07938957e+00 6.74283624e-01 5.97975874e+00 5.47736228e-01
-9.91303027e-01 4.63447899e-01 -5.19796386e-02 2.53145039e-01
-5.73382020e-01 1.04521155e+00 -7.04332530e-01 4.85460043e-01
1.36569238e+00 -2.36273229e-01 5.80425739e-01 2.27423862e-01
-1.77171156e-01 -6.39223039e-01 -1.04981911e+00 8.27872217e-01
-3.29373181e-01 -1.16465616e+00 -4.62033868e-01 4.82787162e-01
7.35552073e-01 -1.79504111e-01 -1.08146332e-01 3.44676197e-01
-1.47396877e-01 -1.13985062e+00 6.17305458e-01 3.10853302e-01
8.03803205e-01 -7.57760763e-01 8.05577695e-01 6.77739203e-01
-7.18085229e-01 6.69106022e-02 -4.67716128e-01 -5.80614150e-01
1.84793562e-01 5.19252300e-01 -6.36860967e-01 9.24584985e-01
2.07632974e-01 -1.69691205e-01 -3.72095525e-01 7.01000750e-01
-3.34759235e-01 4.26099539e-01 -3.53862435e-01 1.95000097e-02
2.43369862e-01 -7.03113496e-01 5.29276133e-01 7.49320567e-01
4.64761645e-01 5.66993117e-01 -2.69799054e-01 1.22265422e+00
3.26541007e-01 -1.07109800e-01 -8.73700559e-01 -1.65028498e-01
4.45048064e-01 1.10270786e+00 -8.39348197e-01 -3.41449380e-01
-3.97033900e-01 9.82482851e-01 -1.13085896e-01 -7.17846081e-02
-6.41555727e-01 -5.14434576e-01 3.69092077e-01 -1.86517864e-01
2.06150919e-01 -3.44595790e-01 -3.02812099e-01 -1.33298051e+00
-1.19569683e-02 -9.05679464e-01 -1.33676261e-01 -3.90746534e-01
-7.21697450e-01 4.75009620e-01 -2.66724914e-01 -1.56697226e+00
6.83256164e-02 -6.27936602e-01 -2.90659636e-01 1.29796529e+00
-1.22209966e+00 -6.78613663e-01 3.21615160e-01 5.40043056e-01
-8.96555960e-01 1.31406516e-01 1.36436212e+00 2.74698138e-01
-3.81522089e-01 3.65844816e-01 3.07012200e-01 -1.86813220e-01
4.69018608e-01 -1.02043402e+00 3.60791802e-01 1.17678595e+00
1.34058863e-01 1.25768673e+00 1.03526175e+00 -5.28593838e-01
-1.95620406e+00 -3.45833480e-01 1.04987812e+00 -4.89305854e-01
5.75012743e-01 -6.70516491e-01 -9.58652437e-01 5.61231673e-01
3.60947162e-01 -1.24948092e-01 8.12206209e-01 -5.15073240e-02
-3.27670187e-01 2.59709954e-01 -1.35590029e+00 4.82674718e-01
7.37745583e-01 -1.16436756e+00 -4.97258723e-01 5.62999964e-01
5.51869810e-01 -4.26362425e-01 -7.93268025e-01 2.43493021e-01
3.05178940e-01 -1.62400353e+00 6.39757574e-01 -5.06037891e-01
7.21185580e-02 -8.30400944e-01 -4.80108440e-01 -8.43978226e-01
-2.29136795e-01 -1.11414886e+00 4.97077316e-01 8.40171337e-01
3.71671528e-01 -9.75448012e-01 6.74723089e-01 7.69927323e-01
-5.15395030e-02 2.65216798e-01 -1.49163604e+00 -8.84001136e-01
9.52885523e-02 -6.02603555e-01 8.47632110e-01 1.01418900e+00
7.01706171e-01 3.29048425e-01 -5.98297238e-01 8.60012949e-01
7.27079511e-01 1.89985394e-01 8.54398131e-01 -8.27888131e-01
-4.32282060e-01 1.16543658e-02 -8.52725327e-01 -7.11461484e-01
-9.19367895e-02 -1.23155355e+00 -4.41854075e-02 -7.76992679e-01
2.08650246e-01 -3.22179765e-01 -4.11851078e-01 1.13284402e-01
4.41737846e-02 4.00766581e-01 1.26570731e-01 4.42103520e-02
-3.62499416e-01 2.72563964e-01 1.18681109e+00 1.05602905e-01
8.29065070e-02 -3.24051380e-02 -5.30087888e-01 1.42526641e-01
6.12025857e-01 -7.91009247e-01 5.00993468e-02 3.56487542e-01
5.69084108e-01 3.65641296e-01 4.10889715e-01 -1.19567549e+00
2.19847903e-01 1.17053635e-01 -3.38559091e-01 -5.89774251e-01
4.75767702e-01 -8.51934254e-01 6.49056196e-01 9.27487314e-01
1.63163722e-01 -3.63853067e-01 1.30519405e-01 3.01440269e-01
-9.94629934e-02 -7.58984506e-01 8.40489149e-01 -1.91247806e-01
-6.27452135e-01 -1.30534261e-01 -5.92017353e-01 -6.25353098e-01
1.21410871e+00 -3.55269313e-01 -6.08180165e-01 -2.28938018e-03
-7.89516687e-01 -2.01843128e-01 8.21761012e-01 -1.39526904e-01
1.94685295e-01 -1.24603426e+00 -5.05497217e-01 6.36561275e-01
2.69283712e-01 -7.18857229e-01 7.14938879e-01 9.55393493e-01
-7.17527866e-01 8.79511833e-01 -3.61615062e-01 -6.31268680e-01
-1.16473544e+00 1.01831126e+00 1.73550919e-01 -8.18274450e-03
-4.07911748e-01 3.66515309e-01 -3.59409079e-02 -5.57489157e-01
-5.67941904e-01 -1.99899197e-01 2.92480469e-01 -4.87947911e-01
7.32731521e-01 4.19239961e-02 2.83438116e-01 -9.08601701e-01
-3.98138434e-01 2.55412906e-01 1.18006214e-01 -2.74469465e-01
7.85189092e-01 -1.72416449e-01 -7.40567982e-01 2.97486007e-01
1.09389174e+00 4.13890213e-01 -3.17836106e-01 -1.56091601e-01
-5.04034497e-02 -5.21726906e-01 -1.72419757e-01 -3.24974149e-01
-3.45221043e-01 1.03009605e+00 8.86809230e-01 8.47754836e-01
9.30946112e-01 -2.73547560e-01 4.23352838e-01 5.98588347e-01
1.12432301e+00 -8.14230204e-01 -7.10147977e-01 5.53349614e-01
4.36190218e-02 -1.07078707e+00 1.97815344e-01 -5.30654788e-01
-2.40507618e-01 1.00511158e+00 3.58446990e-03 -6.94259629e-02
6.85509980e-01 2.05401838e-01 -8.93605426e-02 -2.15324163e-01
-6.44289672e-01 -3.81358445e-01 -1.31252920e-02 4.20485795e-01
3.17021579e-01 5.76330304e-01 -6.57154858e-01 7.48890862e-02
-5.20460129e-01 -2.16514334e-01 1.18638122e+00 1.02768862e+00
-4.48073477e-01 -1.68278670e+00 -8.49685311e-01 1.09467015e-01
-3.95742476e-01 -3.02874148e-01 -1.09444723e-01 6.69254720e-01
4.35149372e-01 1.27377856e+00 -2.72155613e-01 -5.89529395e-01
1.23512648e-01 2.18804955e-01 7.01460004e-01 -8.51020932e-01
-5.71275413e-01 -4.99385267e-01 1.85986638e-01 -5.28825581e-01
-5.61910570e-01 -5.41903377e-01 -1.38730943e+00 -7.51569927e-01
-8.35766971e-01 3.48106414e-01 7.57342219e-01 1.14911520e+00
2.00622454e-01 3.40357184e-01 6.78900659e-01 -4.82290745e-01
-6.95829690e-01 -7.99447954e-01 -1.03903890e+00 2.97128469e-01
2.67388731e-01 -4.49093044e-01 -4.73308325e-01 -3.95401299e-01] | [5.599791049957275, 4.943629741668701] |
d33b1f4b-e554-4b04-951d-a7b72cdbac0c | leveraging-gpt-4-for-food-effect | 2306.16275 | null | https://arxiv.org/abs/2306.16275v1 | https://arxiv.org/pdf/2306.16275v1.pdf | Leveraging GPT-4 for Food Effect Summarization to Enhance Product-Specific Guidance Development via Iterative Prompting | Food effect summarization from New Drug Application (NDA) is an essential component of product-specific guidance (PSG) development and assessment. However, manual summarization of food effect from extensive drug application review documents is time-consuming, which arouses a need to develop automated methods. Recent advances in large language models (LLMs) such as ChatGPT and GPT-4, have demonstrated great potential in improving the effectiveness of automated text summarization, but its ability regarding the accuracy in summarizing food effect for PSG assessment remains unclear. In this study, we introduce a simple yet effective approach, iterative prompting, which allows one to interact with ChatGPT or GPT-4 more effectively and efficiently through multi-turn interaction. Specifically, we propose a three-turn iterative prompting approach to food effect summarization in which the keyword-focused and length-controlled prompts are respectively provided in consecutive turns to refine the quality of the generated summary. We conduct a series of extensive evaluations, ranging from automated metrics to FDA professionals and even evaluation by GPT-4, on 100 NDA review documents selected over the past five years. We observe that the summary quality is progressively improved throughout the process. Moreover, we find that GPT-4 performs better than ChatGPT, as evaluated by FDA professionals (43% vs. 12%) and GPT-4 (64% vs. 35%). Importantly, all the FDA professionals unanimously rated that 85% of the summaries generated by GPT-4 are factually consistent with the golden reference summary, a finding further supported by GPT-4 rating of 72% consistency. These results strongly suggest a great potential for GPT-4 to draft food effect summaries that could be reviewed by FDA professionals, thereby improving the efficiency of PSG assessment cycle and promoting the generic drug product development. | ['Hualou Liang', 'Liang Zhao', 'Meng Hu', 'Yi Zhang', 'Felix Agbavor', 'Taha ValizadehAslani', 'Biao Han', 'Jing Wang', 'Ping Ren', 'Yiwen Shi'] | 2023-06-28 | null | null | null | null | ['text-summarization'] | ['natural-language-processing'] | [ 3.25115919e-01 1.52859211e-01 -7.43625522e-01 -1.19233690e-01
-1.41279364e+00 -1.08833778e+00 4.38985080e-01 1.09088862e+00
-9.84867066e-02 7.31672585e-01 7.42502928e-01 -6.66908145e-01
-4.36632186e-01 -3.01168889e-01 -4.74079400e-01 -2.64934003e-01
2.33792156e-01 2.67841011e-01 -2.02815488e-01 -1.00562789e-01
4.72460687e-01 1.72859833e-01 -8.31428945e-01 6.55601084e-01
1.50046599e+00 3.25390249e-01 1.63284495e-01 5.87374806e-01
-1.21932425e-01 6.03764832e-01 -7.67324328e-01 -4.38736677e-01
-1.95695713e-01 -5.81499755e-01 -7.46452749e-01 -1.41746223e-01
2.68080980e-01 -4.52437222e-01 3.65298778e-01 8.38184774e-01
7.20875323e-01 -2.52923340e-01 6.37713492e-01 -6.75363064e-01
-5.46931744e-01 9.04636562e-01 -5.52843034e-01 8.66474435e-02
9.58174884e-01 4.09613907e-01 9.91679132e-01 -5.36196828e-01
7.51274824e-01 1.19137609e+00 5.67667484e-01 4.14686382e-01
-1.07410252e+00 -5.92192113e-01 2.65156716e-01 -1.79460540e-01
-8.64070714e-01 -4.26442564e-01 3.50481063e-01 -6.89302683e-01
1.21925688e+00 4.53902066e-01 8.21201921e-01 6.41305327e-01
7.91432261e-01 5.65702856e-01 8.83672476e-01 -2.89567620e-01
4.22209561e-01 -2.71418951e-02 4.37804371e-01 9.82850119e-02
7.39630044e-01 -2.75746077e-01 -3.11135441e-01 -7.72188425e-01
2.46944547e-01 -2.55328238e-01 -2.70667315e-01 3.83289546e-01
-8.79573286e-01 9.90102828e-01 -1.43385783e-01 2.63233930e-01
-8.28181863e-01 -1.60467193e-01 5.35778880e-01 1.55352324e-01
7.63716102e-01 7.68053532e-01 -6.12875879e-01 -3.38552058e-01
-1.06747532e+00 6.02303028e-01 9.47807014e-01 8.06734264e-01
-1.16032973e-01 -3.13524127e-01 -6.37426496e-01 8.59811425e-01
4.31740403e-01 7.56187499e-01 3.03919852e-01 -7.91849554e-01
4.55513328e-01 6.88526273e-01 3.16362381e-01 -9.60163176e-01
-4.95070517e-01 -4.80471790e-01 -4.78948176e-01 -1.54794678e-01
-1.36752508e-03 -3.54453802e-01 -7.21653283e-01 1.54174864e+00
2.06475675e-01 -5.77232420e-01 2.23193273e-01 3.28197271e-01
1.25835347e+00 8.10952604e-01 6.66655064e-01 -7.70419657e-01
1.61533284e+00 -5.97471297e-01 -1.25129294e+00 5.18157892e-02
1.15604889e+00 -1.18999612e+00 7.64631271e-01 7.41212547e-01
-1.31299341e+00 -3.04053962e-01 -1.23479569e+00 1.86417252e-01
-8.31454694e-02 2.32331127e-01 4.54086959e-01 6.89702392e-01
-7.40284145e-01 7.71026075e-01 -6.54533684e-01 -2.77522236e-01
5.84769487e-01 2.98310459e-01 -1.92945957e-01 -1.94734648e-01
-1.16536736e+00 1.03533566e+00 1.71063915e-01 -1.11808226e-01
-6.18119836e-01 -1.05587673e+00 -8.80539119e-01 -1.16920210e-01
2.77954638e-01 -1.03572142e+00 1.69547248e+00 -1.23342253e-01
-1.48059487e+00 2.98878372e-01 -3.40490729e-01 -3.56989294e-01
2.26445287e-01 -3.44642878e-01 -4.76452321e-01 2.07920030e-01
3.91563267e-01 3.78057688e-01 1.48456976e-01 -6.72038376e-01
-6.81499839e-01 -3.37506652e-01 -1.39484159e-03 2.68619686e-01
6.45946488e-02 3.13141018e-01 -1.17017120e-01 -5.07619381e-01
-3.42817605e-01 -7.74900436e-01 -4.73338157e-01 -6.91778064e-01
-5.72635770e-01 -6.75797284e-01 3.27886313e-01 -9.73083735e-01
1.94905519e+00 -1.80583107e+00 -3.16270500e-01 3.54823954e-02
3.89580965e-01 5.65330565e-01 -3.88118505e-01 9.23217595e-01
-3.69587541e-01 5.79363704e-01 -1.31175026e-01 5.77102676e-02
-9.17723700e-02 -2.71734059e-01 -1.74772486e-01 2.65474170e-01
2.13288039e-01 1.04327476e+00 -1.27070236e+00 -1.94825426e-01
1.56897441e-01 1.32432312e-01 -5.86199284e-01 -2.25406706e-01
-4.44500655e-01 4.56732243e-01 -7.62400389e-01 5.75912416e-01
6.35737896e-01 -4.55637813e-01 3.22516799e-01 -2.54986703e-01
-4.43855941e-01 7.70839989e-01 -6.87485516e-01 1.54154730e+00
-1.86859533e-01 1.64145917e-01 -4.57787365e-01 -3.17672879e-01
6.00857735e-01 5.34557402e-01 6.65284574e-01 -8.15022349e-01
7.75894448e-02 3.21958482e-01 2.26960674e-01 -6.85086191e-01
4.99046236e-01 -7.95181915e-02 -1.76537320e-01 4.60970759e-01
-2.75099069e-01 -3.61879289e-01 6.20907485e-01 5.09210825e-01
1.11539793e+00 -1.67788953e-01 7.33481169e-01 -2.63879389e-01
3.44789833e-01 4.01509345e-01 3.09683472e-01 6.96456194e-01
9.44945291e-02 4.42116827e-01 6.12498045e-01 -4.52443324e-02
-7.70618618e-01 -3.73077869e-01 -2.30221853e-01 3.46732020e-01
-4.59300458e-01 -1.18481815e+00 -9.86055493e-01 -6.60428703e-01
-1.69516206e-01 1.13472557e+00 -4.52765524e-01 -9.88987386e-02
-3.89104903e-01 -8.21517467e-01 3.67935151e-01 1.03858605e-01
1.71838015e-01 -7.39927411e-01 -4.78375673e-01 7.84415662e-01
-3.44680071e-01 -9.07995582e-01 -7.31216609e-01 -1.75610244e-01
-7.99222946e-01 -1.26276779e+00 -9.98025656e-01 -3.57500732e-01
2.83752471e-01 1.31155461e-01 8.07148635e-01 -2.58085698e-01
9.57206562e-02 3.54365855e-01 -4.37310725e-01 -8.79532635e-01
-1.00015962e+00 -8.69219527e-02 -2.55093157e-01 -7.65715122e-01
3.81668746e-01 -1.09672785e-01 -8.00504744e-01 -1.60213113e-02
-9.99342799e-01 -6.41613826e-02 7.39922464e-01 2.41871476e-01
5.36801219e-01 -1.42281190e-01 1.02813137e+00 -1.07698786e+00
1.31030214e+00 -4.04949158e-01 -3.19753796e-01 2.73308009e-01
-8.58294010e-01 -1.85865209e-01 3.87879223e-01 -3.92073512e-01
-7.24831581e-01 -2.64226615e-01 -5.61074853e-01 4.03879911e-01
1.21531293e-01 1.23903108e+00 -2.37154029e-02 1.97807267e-01
8.93171549e-01 -2.47884691e-01 1.26181692e-01 -3.35885853e-01
3.71083885e-01 8.43657494e-01 2.49445159e-02 -2.35283956e-01
2.34769285e-01 -1.43572018e-01 -3.72353762e-01 -5.19411445e-01
-6.40049398e-01 -5.64367950e-01 1.83311462e-01 2.52675526e-02
7.68397450e-01 -9.44037855e-01 -6.13866091e-01 2.51337588e-01
-1.45241678e+00 -1.35913044e-01 -1.47536546e-01 6.82954669e-01
-2.25143507e-01 7.13086367e-01 -6.38480723e-01 -5.42275786e-01
-9.22428012e-01 -1.29151261e+00 9.69151855e-01 2.04593197e-01
-1.18504429e+00 -7.00050592e-01 3.78992349e-01 4.06842113e-01
2.23628998e-01 3.95276129e-01 1.25328672e+00 -9.40611362e-01
1.33427782e-02 -4.05273885e-01 3.58115807e-02 1.43629059e-01
7.57247627e-01 1.27433091e-01 -5.35701334e-01 -1.02418378e-01
1.59189031e-01 2.06788674e-01 3.07040930e-01 1.23268771e+00
5.82681537e-01 -6.12571418e-01 -5.62808394e-01 -3.90547544e-01
1.07593453e+00 9.41220403e-01 6.41958714e-01 1.29678860e-01
2.95476079e-01 6.81806207e-01 9.80057657e-01 5.00541151e-01
2.98315942e-01 6.26400530e-01 -1.12548709e-01 -1.29460141e-01
-3.05820942e-01 -3.19492072e-01 4.90332365e-01 9.25243199e-01
1.78943753e-01 -3.57447982e-01 -6.34543002e-01 3.76072824e-01
-1.76564801e+00 -6.77843988e-01 -5.33091068e-01 2.13970184e+00
1.04468191e+00 1.44811332e-01 2.02048659e-01 -6.73085377e-02
4.72484946e-01 -1.26895517e-01 -5.67891419e-01 -7.16042638e-01
5.32372221e-02 2.34210968e-01 3.85588467e-01 3.95496577e-01
-6.69881701e-01 5.17518878e-01 6.81193447e+00 9.82262790e-01
-8.91216576e-01 -9.71194804e-02 7.38171518e-01 6.97333692e-03
-6.35918975e-01 -1.33799657e-01 -7.53856778e-01 5.86522818e-01
1.28238821e+00 -8.70958269e-01 -2.39900410e-01 4.53924954e-01
1.14967489e+00 -2.79413521e-01 -1.15523541e+00 7.90023565e-01
-1.58628523e-01 -1.68544090e+00 3.85628134e-01 1.98326916e-01
8.41296077e-01 -3.66949797e-01 -9.31848044e-05 1.62055820e-01
3.55385154e-01 -8.12009633e-01 4.40871358e-01 5.81549630e-02
8.43167901e-01 -6.05311573e-01 8.63866210e-01 1.74582556e-01
-9.74030614e-01 9.53799188e-02 -1.93170443e-01 1.67227849e-01
5.99591851e-01 9.36997890e-01 -1.02779150e+00 1.22120476e+00
2.65891701e-01 1.01186717e+00 -3.56072187e-01 1.18877959e+00
-2.00035691e-01 6.10591590e-01 2.57402025e-02 -2.33056277e-01
2.56076306e-01 -2.67522167e-02 5.82306623e-01 1.23693943e+00
4.19908792e-01 4.95456487e-01 -3.13483253e-02 5.16200602e-01
3.86125371e-02 6.34367704e-01 -5.90495646e-01 -5.81685305e-01
3.13350588e-01 9.45920765e-01 -6.74497247e-01 -6.04642510e-01
-2.39680827e-01 2.40471080e-01 -5.41022778e-01 1.98828772e-01
-7.61508346e-01 -2.83934236e-01 3.07864487e-01 1.57277718e-01
9.02914554e-02 2.72149652e-01 -3.98508489e-01 -4.10129279e-01
-1.38707325e-01 -1.36986434e+00 5.09206653e-01 -7.71088839e-01
-9.41960275e-01 6.74834907e-01 2.37373635e-01 -1.53729165e+00
-1.85063407e-01 -8.33771974e-02 -3.99609894e-01 8.54720294e-01
-1.18207061e+00 -7.10130095e-01 3.18300396e-01 -5.58544463e-03
8.68895352e-01 3.29592764e-01 6.95161045e-01 3.92797619e-01
-3.44738156e-01 4.45068747e-01 -2.09826156e-01 -5.19833446e-01
8.36711705e-01 -9.03377652e-01 3.64043742e-01 3.88586104e-01
-3.47254932e-01 1.08146942e+00 9.70722079e-01 -1.24736869e+00
-1.05705535e+00 -9.52829480e-01 1.20348990e+00 -5.84592700e-01
5.43569088e-01 3.07907254e-01 -7.71806419e-01 7.30326027e-02
4.97701824e-01 -1.18000519e+00 1.04798281e+00 2.50844359e-02
1.49303451e-01 1.30328029e-01 -1.09272003e+00 6.51854455e-01
4.98305917e-01 -2.83808291e-01 -6.99515164e-01 7.51823187e-01
1.07582426e+00 -5.22368252e-01 -1.22049904e+00 5.92724025e-01
5.96292853e-01 -1.80542126e-01 6.22949541e-01 -5.43260992e-01
7.16480136e-01 -4.16368037e-01 3.12397659e-01 -1.41770983e+00
-3.43670100e-01 -8.24409902e-01 3.32448989e-01 1.08973086e+00
9.42527533e-01 -6.34224594e-01 3.36217284e-01 8.31419110e-01
-5.86269557e-01 -8.66356969e-01 -5.40937901e-01 -4.14638638e-01
1.04485877e-01 -4.22781289e-01 3.97577614e-01 7.16850221e-01
5.60284555e-01 6.71399832e-01 -1.02760337e-01 -4.04128358e-02
2.86163054e-02 -2.23998427e-01 4.83328372e-01 -1.05646133e+00
-1.24074653e-01 -6.18787110e-01 1.82277352e-01 -1.07327485e+00
-5.23556590e-01 -6.76201761e-01 -2.10524723e-01 -2.23936439e+00
5.91685951e-01 1.10559098e-01 -1.00212999e-01 5.28408229e-01
-1.84534937e-01 -3.44028711e-01 -4.73214947e-02 2.05383599e-01
-4.38866943e-01 1.81069463e-01 1.51235354e+00 -5.14176548e-01
-8.03227127e-01 8.19470435e-02 -1.44769287e+00 3.45283836e-01
6.32781923e-01 -5.79430044e-01 -6.74895287e-01 4.93336320e-02
5.13675392e-01 3.07454288e-01 -2.08995625e-01 -3.41995686e-01
3.15817505e-01 -1.02487132e-01 6.96201921e-02 -8.94935668e-01
-3.99980396e-01 -2.98120767e-01 6.28940463e-01 8.30111623e-01
-3.84340227e-01 1.10828720e-01 7.45158076e-01 3.75311852e-01
-2.49266416e-01 -2.91195512e-01 7.15128854e-02 -9.73637998e-02
-1.68480724e-02 1.28340811e-01 -6.80363357e-01 -3.23913217e-01
5.90213358e-01 -1.56798035e-01 -5.58661342e-01 -4.91253912e-01
-6.93217039e-01 3.72296780e-01 2.99967565e-02 3.77544314e-01
4.06455219e-01 -1.00686383e+00 -9.47543681e-01 -3.73285800e-01
1.97163478e-01 -1.41706422e-01 7.90879726e-01 1.17421412e+00
-2.48141155e-01 1.06615841e+00 3.48352969e-01 -5.32009006e-01
-1.46563041e+00 3.37678641e-01 -1.34530038e-01 -8.05447221e-01
-4.84023005e-01 1.85967341e-01 4.24252957e-01 2.83080004e-02
-5.69627844e-02 -9.32560027e-01 -5.97271621e-01 2.90405899e-01
9.52497125e-01 2.79690236e-01 5.31859517e-01 -1.50436193e-01
-3.40893477e-01 6.96302533e-01 -6.45299971e-01 -1.00824408e-01
1.22432351e+00 -4.95403595e-02 -2.77740881e-02 1.01108387e-01
9.37030375e-01 2.51902878e-01 -6.13254964e-01 3.45564753e-01
1.06013089e-01 1.07679889e-01 -1.44601747e-01 -1.55445707e+00
-3.93611014e-01 2.32926339e-01 2.90086001e-01 2.61081249e-01
9.86731946e-01 5.02118804e-02 7.22588658e-01 -3.94240990e-02
1.22561954e-01 -9.16712403e-01 -4.04676013e-02 1.00610591e-01
1.20200729e+00 -7.76833475e-01 3.06063086e-01 -6.30105078e-01
-6.49552763e-01 8.46037507e-01 -1.10815592e-01 3.87333184e-01
3.52743655e-01 -9.60172564e-02 -9.48724616e-03 -6.48546457e-01
-8.31659973e-01 3.99056137e-01 5.17641544e-01 3.55365783e-01
6.76757753e-01 5.50561212e-02 -1.25089598e+00 9.72313404e-01
-2.22111749e-03 3.98422033e-01 6.46435559e-01 9.12831128e-01
-1.98085055e-01 -1.44804013e+00 -1.14427805e-01 5.22078514e-01
-6.90600276e-01 -3.27310711e-01 -3.89178574e-01 7.16597080e-01
-2.29279041e-01 1.58313298e+00 -7.35115469e-01 -4.52937931e-02
7.23629296e-01 -1.58702970e-01 4.03243870e-01 -8.90422702e-01
-9.31073546e-01 7.82364309e-01 6.54135466e-01 -4.63913292e-01
-6.37542725e-01 -6.38091326e-01 -1.11292791e+00 -1.67055011e-01
-4.63956505e-01 6.50082707e-01 7.64884710e-01 1.00186706e+00
6.40824258e-01 7.68069148e-01 4.70281690e-01 -1.35158256e-01
-5.07402301e-01 -1.10474837e+00 -2.23632410e-01 -1.14376955e-01
3.36445868e-01 -2.58069336e-01 -1.11202322e-01 -2.73926239e-02] | [12.241127014160156, 9.640447616577148] |
d8d548ff-d0b8-4df2-9077-b7d1403ec07c | neural-boxer-at-the-iwcs-shared-task-on-drs | null | null | https://aclanthology.org/W19-1204 | https://aclanthology.org/W19-1204.pdf | Neural Boxer at the IWCS Shared Task on DRS Parsing | This paper describes our participation in the shared task of Discourse Representation Structure parsing. It follows the work of Van Noord et al. (2018), who employed a neural sequence-to-sequence model to produce DRSs, also exploiting linguistic information with multiple encoders. We provide a detailed look in the performance of this model and show that (i) the benefit of the linguistic features is evident across a number of experiments which vary the amount of training data and (ii) the model can be improved by applying a number of postprocessing methods to fix ill-formed output. Our model ended up in second place in the competition, with an F-score of 84.5. | ['Rik van Noord'] | 2019-05-01 | null | null | null | ws-2019-5 | ['drs-parsing'] | ['natural-language-processing'] | [ 4.25757021e-01 9.40974474e-01 3.81934121e-02 -4.51776683e-01
-1.03273165e+00 -8.90698731e-01 8.45742702e-01 3.19246858e-01
-4.63213772e-01 8.56540501e-01 8.74376774e-01 -5.79908609e-01
1.17005527e-01 -4.60210621e-01 -6.90887213e-01 -9.24893022e-02
-9.94632915e-02 3.43360513e-01 3.43575686e-01 -6.54619157e-01
3.78586560e-01 1.20489173e-01 -1.28620887e+00 8.37356746e-01
5.43176949e-01 4.27632868e-01 3.71252239e-01 9.84712601e-01
1.41519867e-02 1.35377872e+00 -9.21485007e-01 -5.43170631e-01
-2.96365209e-02 -4.81476486e-01 -1.40722954e+00 -2.61763304e-01
2.49419585e-01 -2.83364683e-01 -2.47866347e-01 5.44172406e-01
4.63137805e-01 2.33217016e-01 4.83534038e-01 -4.65963393e-01
-7.35747516e-01 1.17566979e+00 -2.30506882e-01 3.59206885e-01
6.42022967e-01 1.04067720e-01 1.39312124e+00 -6.96233451e-01
1.05110049e+00 1.42551494e+00 5.28941035e-01 5.94144821e-01
-1.14735579e+00 -7.05572516e-02 1.91345215e-01 -3.32008936e-02
-8.88600886e-01 -8.61124635e-01 3.41028631e-01 -3.01742285e-01
1.73341477e+00 1.08241551e-01 2.35397562e-01 9.61873233e-01
-1.83009014e-01 7.76721358e-01 9.89051044e-01 -8.46581936e-01
-3.85610946e-02 -1.57524332e-01 1.77773684e-01 7.08775461e-01
-1.39561668e-01 8.17749873e-02 -4.97432530e-01 9.33504105e-02
4.40284789e-01 -9.18848693e-01 -2.09844559e-01 1.65969685e-01
-9.77472067e-01 1.09136832e+00 3.77568334e-01 8.57477486e-01
-7.28768781e-02 -3.00059188e-02 7.11201787e-01 3.57126892e-01
2.82670438e-01 7.24296987e-01 -5.16378343e-01 -3.44710618e-01
-6.96398556e-01 3.84249806e-01 9.43167329e-01 5.60425341e-01
2.06390694e-01 5.12839407e-02 -4.05834556e-01 9.22321796e-01
2.22807527e-01 -2.28637606e-01 6.45236969e-01 -1.30653071e+00
7.80055165e-01 3.78507763e-01 5.36903776e-02 -6.30822420e-01
-4.60712790e-01 -3.38881575e-02 -1.02192432e-01 1.54197335e-01
7.90492356e-01 -4.76105303e-01 -6.32173061e-01 2.14748931e+00
-7.58525729e-02 -2.30025783e-01 2.81969875e-01 6.75914347e-01
9.58522439e-01 6.57472610e-01 3.18855643e-01 -1.49466604e-01
1.40891075e+00 -1.03337276e+00 -7.86365330e-01 -4.69546437e-01
1.00443912e+00 -8.24046731e-01 9.39640224e-01 2.33425528e-01
-1.55457211e+00 -4.55470532e-01 -1.11890686e+00 -5.27484000e-01
3.96530032e-02 1.41169280e-01 8.61979485e-01 5.63242555e-01
-1.35765088e+00 7.16312349e-01 -7.91320801e-01 -3.78767550e-01
2.11860552e-01 1.14307888e-01 -3.47490340e-01 8.96067396e-02
-1.27815247e+00 1.32133901e+00 7.06051707e-01 -1.96401700e-01
-3.75562996e-01 -3.94360304e-01 -1.07937241e+00 8.87248944e-03
3.63420635e-01 -6.98583603e-01 1.65472794e+00 -1.06059718e+00
-1.76781893e+00 1.14737356e+00 -3.07920516e-01 -6.32866323e-01
3.65390718e-01 -3.98320764e-01 7.30732549e-03 2.77617306e-01
-2.77233366e-02 7.07872510e-01 1.88394532e-01 -9.21121418e-01
-6.35459363e-01 -1.56144187e-01 4.22987133e-01 1.90481514e-01
3.38089436e-01 5.18084705e-01 1.25007749e-01 -6.51418328e-01
-9.29938704e-02 -6.62630796e-01 -1.77758917e-01 -6.27886772e-01
-2.57134974e-01 -5.32439947e-01 1.69527024e-01 -1.12000883e+00
1.23567712e+00 -1.82989311e+00 5.20944357e-01 -3.50767285e-01
-9.05947201e-03 3.78063023e-01 -1.41059771e-01 8.61613095e-01
-3.01479340e-01 4.87896681e-01 -5.17880559e-01 -4.84552026e-01
6.57508790e-04 7.89161921e-02 -3.65326554e-01 1.15564600e-01
6.87600613e-01 1.16798198e+00 -8.02286685e-01 -2.20858276e-01
1.23798465e-02 2.13409141e-01 -4.72213179e-01 3.00107002e-01
-3.15644920e-01 3.83284062e-01 -9.32019725e-02 6.97365403e-02
7.60527179e-02 -1.08483106e-01 5.55839181e-01 2.16277838e-01
-4.03028339e-01 1.26679063e+00 -9.54892337e-01 2.01686788e+00
-3.63200814e-01 7.71547735e-01 2.26165324e-01 -1.05449581e+00
5.50047517e-01 6.54147446e-01 -1.70841962e-01 -4.18999583e-01
1.91033378e-01 2.08095208e-01 5.27463555e-01 -5.08585036e-01
8.10893714e-01 -2.32797593e-01 -2.87063837e-01 6.74450099e-01
2.28703529e-01 6.92530572e-02 4.73431736e-01 4.44015384e-01
1.30934942e+00 4.29457128e-01 6.58173144e-01 -1.28733084e-01
4.73641336e-01 9.73509774e-02 2.88356990e-01 6.99133992e-01
-1.29594192e-01 5.52057028e-01 7.58604884e-01 -6.27019703e-02
-1.16760695e+00 -6.58638120e-01 -1.50621161e-01 1.21597707e+00
-5.29309273e-01 -5.50445139e-01 -7.90606439e-01 -5.37616014e-01
-3.99360448e-01 1.03670049e+00 -6.55468702e-01 1.35390282e-01
-1.28054452e+00 -3.38650793e-01 9.70820904e-01 6.52924001e-01
7.68209174e-02 -1.54421294e+00 -8.09502363e-01 3.40453833e-01
-2.50288486e-01 -1.09010351e+00 6.07821271e-02 2.66294897e-01
-6.68133378e-01 -9.66998875e-01 -3.34943473e-01 -8.93148363e-01
1.53362900e-01 -5.40034398e-02 1.15287209e+00 4.47370559e-01
6.79932386e-02 1.02548063e-01 -5.36795259e-01 -2.68489242e-01
-9.08656478e-01 3.07704657e-01 -6.05122328e-01 -7.28925169e-01
6.25822693e-02 -3.62367928e-01 4.62134294e-02 -4.26488459e-01
-7.59183586e-01 2.39272028e-01 3.50817591e-01 8.81939828e-01
-1.90952551e-02 -5.08568585e-01 9.25255179e-01 -1.21931505e+00
8.80394459e-01 -4.16164279e-01 -3.88003916e-01 6.91221878e-02
-7.26449788e-02 2.83672154e-01 4.57186550e-01 8.63920301e-02
-1.23932266e+00 -1.38758570e-01 -6.38319016e-01 3.82034183e-01
-3.04050058e-01 6.97482228e-01 -5.35096489e-02 5.64146459e-01
5.90800941e-01 -3.67548555e-01 1.88703015e-02 -5.30579388e-01
6.36753976e-01 5.36710143e-01 5.44076025e-01 -6.32983267e-01
4.03929174e-01 -3.44536081e-02 -1.99345186e-01 -6.04375720e-01
-1.05937016e+00 -5.89826815e-02 -6.87222004e-01 2.76824236e-01
1.01094079e+00 -9.83064950e-01 -4.68397170e-01 2.40701035e-01
-1.57680237e+00 -7.80899823e-01 -2.54711598e-01 2.01664031e-01
-5.22017300e-01 4.26846951e-01 -1.07812142e+00 -8.12039196e-01
-1.38669312e-01 -9.84504163e-01 8.21773469e-01 4.17417213e-02
-8.05600464e-01 -1.08427191e+00 1.10872701e-01 4.42344129e-01
2.76777357e-01 3.23327839e-01 1.15268290e+00 -9.46200192e-01
-3.00021529e-01 3.34340572e-01 -1.57039434e-01 1.81001961e-01
-1.06962010e-01 -5.48334680e-02 -1.05506706e+00 -6.56951666e-02
-1.44767109e-02 -7.73781955e-01 9.21471655e-01 2.98988998e-01
7.13999093e-01 -2.60040134e-01 -4.31201383e-02 1.93404704e-01
1.15082288e+00 4.60763946e-02 7.73557007e-01 5.91827154e-01
3.89355421e-01 1.04419851e+00 4.46400762e-01 2.80396864e-02
6.80680454e-01 7.20249355e-01 3.00150394e-01 2.37163052e-01
-3.00459474e-01 -3.06044966e-01 4.40951347e-01 6.85968280e-01
8.67239386e-02 -4.38249171e-01 -7.99316287e-01 7.74369180e-01
-2.00996542e+00 -1.07097793e+00 -3.23944956e-01 1.76001358e+00
9.92426336e-01 2.55515248e-01 3.38438392e-01 9.98344123e-02
6.36815012e-01 3.98310632e-01 -1.33691682e-02 -1.12734687e+00
-2.53825903e-01 7.14177310e-01 2.30945647e-01 9.20112073e-01
-9.86036241e-01 1.25247657e+00 7.17678690e+00 3.11463386e-01
-7.31280982e-01 5.91746606e-02 4.38687205e-01 8.86031687e-02
-3.35160196e-01 1.28699273e-01 -5.85968077e-01 2.38922015e-01
1.55990529e+00 -1.01337470e-01 4.43566799e-01 4.35920149e-01
-1.35563135e-01 -1.71431124e-01 -1.15143180e+00 2.08426952e-01
1.21738255e-01 -1.36072373e+00 -3.26115221e-01 -3.09509993e-01
3.40865910e-01 1.53137833e-01 -3.43003333e-01 5.58481991e-01
8.28251064e-01 -1.42510438e+00 9.38046575e-01 1.69994757e-01
3.97502184e-01 -5.96002460e-01 5.72812080e-01 5.22371292e-01
-7.26089835e-01 -1.31745443e-01 -7.74312541e-02 -6.16876185e-01
4.84149814e-01 -7.35179707e-02 -8.71924043e-01 7.27467120e-01
3.36001754e-01 4.69932437e-01 -4.83853877e-01 5.17428100e-01
-8.96555722e-01 7.75697589e-01 -1.84651449e-01 -9.04083923e-02
2.24913046e-01 1.12353355e-01 5.33508003e-01 1.51392305e+00
-6.43733889e-02 3.46308529e-01 2.34308131e-02 6.31643236e-01
-1.37761354e-01 1.68204010e-01 -4.23488677e-01 -2.33399540e-01
5.57695448e-01 1.09543002e+00 -4.66110498e-01 -3.26102644e-01
-3.47868443e-01 8.45559120e-01 9.38131332e-01 -3.70581448e-02
-4.78678048e-01 -3.56640816e-01 3.03321004e-01 -1.68107882e-01
7.03661323e-01 -3.38106483e-01 -4.18415964e-01 -8.01110506e-01
-4.98603657e-02 -9.80156481e-01 5.03017247e-01 -7.72341728e-01
-1.02181017e+00 6.38809443e-01 8.44072476e-02 -4.02417928e-01
-8.28661323e-01 -6.22111201e-01 -6.75212502e-01 1.10229433e+00
-1.43574059e+00 -1.09946072e+00 3.75282019e-01 -3.74769829e-02
5.66655338e-01 -2.17514634e-02 1.06305766e+00 8.69824514e-02
-4.96867299e-01 4.13386077e-01 -3.60794663e-01 3.37028980e-01
6.98774338e-01 -1.41780508e+00 7.04630077e-01 1.01169360e+00
2.12734833e-01 8.68885219e-01 7.82834232e-01 -4.30850476e-01
-9.86586392e-01 -7.25027204e-01 1.45128405e+00 -5.64342380e-01
7.56666780e-01 -2.71780223e-01 -1.13338637e+00 1.16671491e+00
9.45826411e-01 -4.58604932e-01 6.85110092e-01 3.63727152e-01
-3.41911346e-01 7.22258389e-01 -1.07137167e+00 4.12353516e-01
1.08874774e+00 -4.56834882e-01 -1.37337041e+00 1.06966920e-01
9.56212997e-01 -8.12575936e-01 -8.10796142e-01 2.18566909e-01
3.69719416e-01 -1.13354897e+00 6.51149988e-01 -8.10702920e-01
9.36959207e-01 -5.44309355e-02 -2.09178254e-01 -1.40513813e+00
-3.88497055e-01 -5.79979777e-01 -7.65404254e-02 1.47665584e+00
5.32042027e-01 -3.74004602e-01 1.50045097e-01 5.26337028e-01
-5.83310604e-01 -6.30094171e-01 -9.00684059e-01 -4.83039409e-01
5.55865169e-01 -2.92502254e-01 2.67896682e-01 6.45768762e-01
2.87480503e-01 9.84890044e-01 2.03846879e-02 -2.87282020e-01
-8.63912702e-02 -5.92817068e-02 5.86292446e-01 -1.03917837e+00
-7.45828211e-01 -5.05937636e-01 1.49063215e-01 -1.00389659e+00
4.91278380e-01 -1.17999220e+00 2.39342600e-02 -1.81952035e+00
-1.32838160e-01 -1.19149402e-01 -6.69930205e-02 5.97397387e-01
-3.54160577e-01 -4.66082431e-02 6.65565610e-01 1.33352712e-01
-5.03112674e-01 1.89251453e-01 8.14375043e-01 3.14235359e-01
-2.88843483e-01 -5.11201262e-01 -1.22893524e+00 6.86770022e-01
9.22183216e-01 -3.70269179e-01 -7.48594925e-02 -7.36479819e-01
2.06257373e-01 2.24820450e-01 4.34841029e-02 -5.72159648e-01
-2.16275170e-01 6.40226454e-02 2.06261814e-01 -3.54426771e-01
3.74003142e-01 -1.47340789e-01 -2.19633460e-01 4.86381859e-01
-7.22559512e-01 2.60703087e-01 5.22988856e-01 6.40241355e-02
-6.88928217e-02 -6.93835378e-01 6.36484802e-01 -3.55318815e-01
-7.14001715e-01 -6.78935707e-01 -5.80225945e-01 3.38385940e-01
8.43813896e-01 -7.89042190e-02 -6.43281460e-01 -3.48157436e-01
-5.53042352e-01 2.34455448e-02 3.69888246e-01 3.82361978e-01
2.73017101e-02 -9.11312759e-01 -9.43096995e-01 -1.70546353e-01
-1.67514488e-01 7.37308152e-03 -2.05052838e-01 5.56368709e-01
-3.51257026e-01 6.14266455e-01 -5.80356084e-02 -2.59782225e-01
-1.32297635e+00 1.89883903e-01 1.38687432e-01 -7.35185266e-01
-5.98614514e-01 8.34157109e-01 -2.35312998e-01 -3.76935571e-01
-1.49564520e-01 -1.66721549e-02 -5.32232285e-01 1.83018953e-01
6.73868537e-01 2.00146630e-01 9.79297981e-02 -8.40516031e-01
-3.30363989e-01 8.85175616e-02 -2.74780631e-01 -5.44561863e-01
1.68185711e+00 -1.60228014e-01 -1.34908825e-01 7.07516849e-01
1.02310085e+00 3.07520390e-01 -1.10166860e+00 -1.23410240e-01
5.21707952e-01 -6.61453158e-02 -2.27900729e-01 -1.09560251e+00
-4.81206715e-01 7.75813580e-01 -3.09730954e-02 3.64525259e-01
6.63591444e-01 1.35702431e-01 7.68334091e-01 3.43424946e-01
2.08418936e-01 -9.73116994e-01 -1.28684625e-01 1.14846051e+00
9.89708722e-01 -9.41261172e-01 -4.29371409e-02 -3.22998673e-01
-8.66698503e-01 1.18333709e+00 3.76005381e-01 -4.65182096e-01
9.54314426e-04 3.42118353e-01 6.22054264e-02 -1.90320835e-01
-1.09859979e+00 -3.27258289e-01 -2.63626669e-02 5.75979352e-01
1.11033630e+00 -8.07283744e-02 -8.14580858e-01 7.57036388e-01
-6.26412332e-01 -1.72361642e-01 7.00445652e-01 1.06329632e+00
-6.06261134e-01 -1.49183726e+00 -1.01680256e-01 5.13440073e-02
-8.26757431e-01 -1.59759685e-01 -7.09916949e-01 9.05509830e-01
-2.39063129e-01 1.23893130e+00 4.37544063e-02 7.11870044e-02
4.20020193e-01 4.75804418e-01 7.34421611e-01 -1.06818461e+00
-9.97844100e-01 -1.93642348e-01 9.42579627e-01 -3.67232770e-01
-5.53372025e-01 -9.47399199e-01 -1.64912832e+00 -1.00804053e-01
-1.62475005e-01 -2.86391713e-02 5.21300733e-01 1.13824821e+00
2.51730889e-01 9.11640763e-01 2.99854219e-01 -7.01003551e-01
-6.76453948e-01 -1.36172092e+00 -1.55603483e-01 4.92025226e-01
3.44292611e-01 -3.79248202e-01 -2.41008729e-01 9.94324163e-02] | [10.74677848815918, 9.304732322692871] |
afab943d-dc8e-4fca-a617-0cbb8116b0e4 | evaluating-the-robustness-of-self-supervised | 2105.06986 | null | https://arxiv.org/abs/2105.06986v1 | https://arxiv.org/pdf/2105.06986v1.pdf | Evaluating the Robustness of Self-Supervised Learning in Medical Imaging | Self-supervision has demonstrated to be an effective learning strategy when training target tasks on small annotated data-sets. While current research focuses on creating novel pretext tasks to learn meaningful and reusable representations for the target task, these efforts obtain marginal performance gains compared to fully-supervised learning. Meanwhile, little attention has been given to study the robustness of networks trained in a self-supervised manner. In this work, we demonstrate that networks trained via self-supervised learning have superior robustness and generalizability compared to fully-supervised learning in the context of medical imaging. Our experiments on pneumonia detection in X-rays and multi-organ segmentation in CT yield consistent results exposing the hidden benefits of self-supervision for learning robust feature representations. | ['Bjoern H. Menze', 'Stephanie E. Combs', 'Jan C. Peeken', 'Anjany Sekuboyina', 'Suprosanna Shit', 'Christopher Watanabe', 'Fernando Navarro'] | 2021-05-14 | null | null | null | null | ['pneumonia-detection'] | ['medical'] | [ 6.65853858e-01 5.91804326e-01 -4.26909626e-01 -7.42655277e-01
-8.83580089e-01 -3.07080243e-02 3.09953570e-01 3.05112839e-01
-3.85563731e-01 4.87875164e-01 3.75805080e-01 -1.67250067e-01
-1.65280804e-01 -3.90766472e-01 -6.98410928e-01 -5.90896368e-01
-3.99033725e-01 4.62051451e-01 5.45048751e-02 1.06990188e-01
-2.49253899e-01 4.03796822e-01 -1.06595373e+00 4.44738984e-01
7.15583861e-01 4.38918889e-01 1.29844084e-01 6.90894365e-01
3.70170623e-01 9.80123997e-01 -5.24520338e-01 1.46156969e-02
1.68944135e-01 -3.58089417e-01 -1.22322679e+00 3.99570346e-01
7.69332230e-01 -3.73268962e-01 -4.36156005e-01 7.53914177e-01
6.44770861e-01 2.61399522e-02 7.18284130e-01 -8.78022850e-01
-6.06032252e-01 7.40192354e-01 -1.18314415e-01 7.54579127e-01
7.31305182e-02 1.09054625e-01 1.04815674e+00 -4.44957018e-01
5.23796082e-01 9.30384576e-01 9.83627617e-01 8.54378879e-01
-1.09379005e+00 -6.21396601e-01 -1.51339963e-01 -3.28077108e-01
-8.94998670e-01 -4.93261814e-01 4.07367110e-01 -3.64449322e-01
1.10685635e+00 1.13112547e-01 1.63068101e-01 1.41444206e+00
2.79825777e-01 1.08509338e+00 9.40442085e-01 -3.62301052e-01
-6.84841424e-02 3.57392162e-01 5.41639209e-01 1.07005501e+00
2.99167901e-01 3.52798820e-01 -2.47081995e-01 -2.50515372e-01
6.94505334e-01 2.33183950e-01 -2.43523479e-01 -4.86187965e-01
-1.35870600e+00 7.79824078e-01 7.38081217e-01 6.05087340e-01
-3.61463726e-01 1.46266103e-01 6.77555799e-01 3.53773713e-01
7.34535873e-01 8.27072024e-01 -5.61924458e-01 9.30775329e-02
-7.66866148e-01 -2.41667435e-01 5.35004616e-01 8.89682114e-01
4.01386172e-01 3.11560780e-01 -1.72857851e-01 7.01204836e-01
1.35104328e-01 2.55852580e-01 8.74069631e-01 -7.34701216e-01
5.54156601e-01 6.18268788e-01 -3.61026436e-01 -5.11105835e-01
-8.75439584e-01 -8.27063143e-01 -8.58335376e-01 1.24888927e-01
2.95186371e-01 -3.82732749e-01 -1.01721191e+00 1.53622329e+00
8.82438645e-02 6.82501495e-02 4.36394840e-01 6.08742356e-01
9.64397669e-01 1.90121725e-01 2.51933724e-01 -7.88654983e-02
1.00587738e+00 -1.16458118e+00 -6.34217083e-01 -2.92330891e-01
8.01788270e-01 -3.79539490e-01 1.01798332e+00 1.33854300e-01
-1.02754045e+00 -5.73838294e-01 -1.00734675e+00 2.34804556e-01
-7.99992532e-02 4.77086902e-02 6.28325462e-01 7.49799192e-01
-8.10144067e-01 9.00871515e-01 -1.14261556e+00 -5.18686533e-01
9.03919399e-01 4.86092836e-01 -4.71345305e-01 -6.49940744e-02
-7.86491454e-01 1.10010660e+00 4.05834019e-01 -3.16848338e-01
-1.12226701e+00 -8.36640477e-01 -8.94132435e-01 8.28504562e-04
2.65692770e-01 -7.99600065e-01 1.35671842e+00 -1.08711195e+00
-1.16879272e+00 9.85229492e-01 2.65633047e-01 -8.11576068e-01
4.35367167e-01 -2.70994097e-01 -3.01374018e-01 5.54251015e-01
2.57599652e-01 6.60659075e-01 1.05160677e+00 -1.17082036e+00
-4.46775347e-01 -2.79803693e-01 -8.87127295e-02 2.46623650e-01
-6.50948405e-01 1.86462134e-01 1.09610200e-01 -9.17485714e-01
5.91144376e-02 -1.08555901e+00 -4.88772362e-01 -2.55647212e-01
-3.96586031e-01 -3.60043913e-01 7.97494650e-01 -4.81425911e-01
8.31184447e-01 -2.29477739e+00 -2.76153684e-01 5.74010536e-02
3.51078123e-01 3.42970371e-01 -2.56988890e-02 1.58483490e-01
-5.78903615e-01 9.62883234e-02 -3.99943531e-01 -2.05957472e-01
-3.99712086e-01 4.55471426e-01 -1.56794518e-01 5.71950376e-01
4.40950185e-01 9.33865368e-01 -1.06743443e+00 -6.30772114e-01
7.71551505e-02 4.33743358e-01 -4.08678740e-01 4.88539815e-01
4.11792286e-02 5.22193253e-01 -6.21448517e-01 6.64049268e-01
-1.97534133e-02 -7.89799750e-01 1.31303877e-01 -7.51929954e-02
4.18287814e-01 4.00106221e-01 -5.76165378e-01 1.69688463e+00
-6.08913422e-01 4.69642639e-01 -1.39950752e-01 -1.13004088e+00
5.46144247e-01 5.96857548e-01 8.34119499e-01 -2.97243029e-01
1.73133492e-01 -9.82167348e-02 1.17526561e-01 -8.09432566e-01
2.89204046e-02 -3.33307087e-01 2.10006103e-01 8.17485511e-01
2.97128618e-01 -1.09745502e-01 -6.86384290e-02 1.74350336e-01
1.42364585e+00 7.00120442e-03 3.27332616e-01 -3.46829951e-01
1.36476949e-01 2.97444880e-01 2.02952147e-01 7.97967970e-01
-3.14746290e-01 9.12858427e-01 -1.17981747e-01 -4.58463162e-01
-8.05938601e-01 -9.92613494e-01 -4.20161217e-01 1.38348305e+00
-3.29848707e-01 -2.49377862e-01 -7.43138492e-01 -1.37580633e+00
4.12577130e-02 4.76350456e-01 -8.44269872e-01 -2.64889628e-01
-6.70893908e-01 -9.08299923e-01 7.77820349e-01 9.72002566e-01
4.24770087e-01 -1.28670859e+00 -1.06042838e+00 1.36103436e-01
1.62423253e-01 -1.07450807e+00 -4.88752604e-01 6.22537494e-01
-1.53047168e+00 -1.38279641e+00 -6.73482418e-01 -9.80812430e-01
1.11109889e+00 4.05925423e-01 1.09941757e+00 2.79817611e-01
-6.76212430e-01 9.08216178e-01 -2.65991569e-01 -3.81558031e-01
-7.40834355e-01 2.91808486e-01 -6.35321485e-03 -3.74721199e-01
6.30584508e-02 -2.22336292e-01 -4.68573958e-01 3.55253905e-01
-1.10443091e+00 -2.59827942e-01 7.55796075e-01 1.17470098e+00
3.41521204e-01 1.42235145e-01 7.50015378e-01 -1.49369407e+00
6.28379464e-01 -6.15254045e-01 1.60299450e-01 2.29407579e-01
-9.13148105e-01 9.57992673e-02 6.11828566e-01 -3.35796177e-01
-1.19060671e+00 2.66571879e-01 -2.94377431e-02 -3.42815936e-01
-3.41161847e-01 3.41423154e-01 4.07820553e-01 -9.75915715e-02
1.18319499e+00 -8.33943039e-02 2.93213964e-01 -3.91200215e-01
1.72973394e-01 5.48571885e-01 4.20792282e-01 -4.58165228e-01
7.43457913e-01 6.09371305e-01 -2.26761922e-01 -6.51047170e-01
-1.26423156e+00 -6.31312370e-01 -7.99409091e-01 2.21246555e-01
9.24468279e-01 -9.98638928e-01 -3.98670956e-02 4.77919206e-02
-5.28061569e-01 -5.38888156e-01 -7.77879298e-01 6.13249660e-01
-5.21366954e-01 4.40045118e-01 -6.69533134e-01 -3.05280536e-01
-6.17605507e-01 -1.00751114e+00 8.46265137e-01 -4.94519509e-02
-5.19133031e-01 -1.32131100e+00 1.84817106e-01 6.34497404e-01
4.68357742e-01 6.82644695e-02 8.06030393e-01 -1.23507738e+00
-2.63656139e-01 -2.39406958e-01 -1.65687382e-01 6.84017420e-01
7.54910469e-01 -4.33007509e-01 -1.18947756e+00 -6.60930753e-01
1.88825890e-01 -7.67156303e-01 8.23083103e-01 2.61256844e-01
1.17951477e+00 -3.26685399e-01 -3.78188401e-01 6.02079690e-01
1.06363165e+00 -1.71302184e-01 2.30517790e-01 3.79004359e-01
5.97456217e-01 5.88546336e-01 3.76121551e-01 6.13201223e-02
1.71536312e-01 1.47567526e-01 1.89369857e-01 -3.99179995e-01
-3.98214608e-01 1.17248334e-02 1.60232782e-01 6.10688388e-01
-3.94440740e-02 1.12570442e-01 -1.27502096e+00 5.23487926e-01
-1.64936388e+00 -8.33970726e-01 8.87262076e-02 1.78026962e+00
9.88478482e-01 2.63302952e-01 6.53703213e-02 4.44402471e-02
5.12973428e-01 5.28725609e-02 -5.84054470e-01 5.92177771e-02
4.81891930e-02 3.68800819e-01 5.00193238e-01 1.46270525e-02
-1.47562194e+00 5.52541196e-01 8.08827496e+00 2.41082892e-01
-1.07772136e+00 3.87312114e-01 6.52145684e-01 2.77069714e-02
-1.60528440e-02 -4.22409087e-01 -4.41915900e-01 -7.19214324e-03
1.03508794e+00 2.52856791e-01 -1.83587730e-01 1.12025607e+00
-2.98977885e-02 2.64198184e-01 -1.34610629e+00 6.80007219e-01
3.08139294e-01 -1.18331110e+00 -2.00858906e-01 -1.91957429e-01
9.30490613e-01 4.26353693e-01 2.26771146e-01 2.84029245e-01
4.56887305e-01 -1.16796541e+00 9.18658972e-02 4.50010896e-02
6.29691780e-01 -3.14751238e-01 8.03564012e-01 2.35873654e-01
-6.56215906e-01 -2.83071518e-01 -2.84271568e-01 1.50430515e-01
-2.96025336e-01 4.26067770e-01 -1.60842538e+00 4.45067108e-01
8.19719851e-01 9.07788277e-01 -8.07167351e-01 9.31817830e-01
-9.28465873e-02 1.06686890e+00 -9.08045992e-02 3.18437874e-01
3.26053917e-01 2.91245073e-01 2.98517257e-01 1.65134656e+00
-3.37734044e-01 -8.25941283e-03 4.06241208e-01 4.90833789e-01
-3.64440046e-02 2.42219329e-01 -8.07325721e-01 -4.65343073e-02
-2.84360677e-01 1.01767898e+00 -7.78855026e-01 -5.42379677e-01
-4.53602344e-01 7.24166751e-01 3.46860737e-01 1.52193651e-01
-4.64225441e-01 5.98222017e-03 1.27251804e-01 1.66682646e-01
2.47410327e-01 5.82606606e-02 -5.25033236e-01 -1.04747653e+00
-3.09815735e-01 -1.03803885e+00 8.65622282e-01 -5.43046117e-01
-1.62620711e+00 6.54336810e-01 1.32139847e-01 -1.11796367e+00
-2.98484951e-01 -5.46141744e-01 -7.18635321e-01 3.37358832e-01
-1.53598154e+00 -1.18306017e+00 -2.31378809e-01 7.42785871e-01
7.09358454e-01 -4.99069065e-01 1.14269638e+00 -3.09798270e-02
-6.94189787e-01 5.62834322e-01 1.47050142e-01 3.00178796e-01
7.77046382e-01 -1.48116028e+00 1.52235284e-01 6.62016213e-01
2.10985675e-01 8.63564730e-01 4.79066461e-01 -6.79458082e-01
-1.21011448e+00 -1.20706487e+00 4.72954482e-01 -6.11477792e-01
5.24009645e-01 8.25830773e-02 -1.04917157e+00 9.73201752e-01
3.56281161e-01 3.77105832e-01 9.21968222e-01 1.64497316e-01
-4.29710716e-01 1.26242861e-01 -1.21217096e+00 1.77879989e-01
1.11747801e+00 -6.93089247e-01 -1.10518634e+00 8.02748144e-01
6.50805712e-01 -4.04595256e-01 -1.15225804e+00 6.84023917e-01
2.07883716e-01 -6.57077014e-01 1.06513536e+00 -9.57399368e-01
4.45941895e-01 2.21192509e-01 2.69470960e-01 -1.23395872e+00
-3.37314278e-01 -3.86646360e-01 4.50911112e-02 8.15884948e-01
4.53597486e-01 -4.87596333e-01 1.05586338e+00 4.84560937e-01
-4.38250750e-01 -6.67071760e-01 -5.81244886e-01 -6.86126053e-01
1.45832792e-01 -1.92108974e-01 3.19455937e-02 1.28423762e+00
1.27934918e-01 3.62012506e-01 -3.79392579e-02 2.26775393e-01
6.61262691e-01 -2.13682149e-02 5.94243526e-01 -1.08739686e+00
-2.38269672e-01 -1.80305928e-01 -2.80772775e-01 -6.22469425e-01
3.68287861e-01 -1.29300451e+00 1.77629516e-01 -1.52553093e+00
4.66995448e-01 -6.07819736e-01 -6.34512305e-01 8.51675332e-01
-3.83007675e-01 1.30750984e-01 -3.81989665e-02 4.09083039e-01
-7.46515453e-01 2.16456547e-01 1.23877668e+00 -3.40367824e-01
-3.58530469e-02 3.51698935e-01 -7.61276782e-01 7.37083673e-01
1.02310443e+00 -7.44558215e-01 -6.89030528e-01 -3.73234093e-01
-3.44367653e-01 -9.59590897e-02 2.11422876e-01 -9.52561557e-01
5.31005599e-02 7.19391108e-02 3.95484507e-01 -2.83384949e-01
-8.07863399e-02 -9.44524109e-01 -3.68398637e-01 6.54694080e-01
-7.16539621e-01 1.50744636e-02 2.70327121e-01 6.57695591e-01
-1.38356686e-01 -5.79021931e-01 7.60846555e-01 -4.36106026e-01
-4.03385729e-01 1.86480269e-01 -4.18301702e-01 3.89832973e-01
8.42304289e-01 -1.44059226e-01 -1.24182232e-01 -1.52798906e-01
-1.03464687e+00 7.67968968e-02 1.24827892e-01 3.29222858e-01
7.98697114e-01 -9.39929485e-01 -7.44432330e-01 1.49217919e-01
1.35087535e-01 2.13469729e-01 -2.77683660e-02 5.82731605e-01
-3.24101627e-01 4.64112639e-01 -2.90114045e-01 -9.56612408e-01
-1.34348142e+00 5.73558748e-01 4.18956846e-01 -5.25022924e-01
-1.00009227e+00 7.99692273e-01 2.36478761e-01 -6.33885562e-01
5.48487246e-01 -5.08912444e-01 -1.30321890e-01 -2.02429548e-01
3.33483785e-01 2.46536642e-01 2.75651246e-01 -2.63049543e-01
-2.07982659e-01 1.00388221e-01 -3.82392764e-01 4.05237556e-01
1.64892447e+00 1.52774781e-01 2.60487169e-01 3.63538027e-01
1.28946733e+00 -3.63729954e-01 -1.16327703e+00 -4.41255957e-01
4.32525903e-01 -1.42911553e-01 1.02037624e-01 -7.70676613e-01
-1.12359178e+00 7.73237109e-01 8.12961340e-01 2.91237002e-03
8.71412635e-01 1.13104641e-01 5.33283949e-01 8.95757020e-01
1.18255422e-01 -9.87648189e-01 7.17584968e-01 3.13727468e-01
6.56829834e-01 -1.66515994e+00 2.72258729e-01 -3.13342094e-01
-8.63800526e-01 1.21983325e+00 6.96776390e-01 -2.26116568e-01
8.05577874e-01 1.90339804e-01 2.40133300e-01 -4.52201486e-01
-6.74336493e-01 -1.37860522e-01 4.53202724e-01 6.91784918e-01
6.37132525e-01 -1.54140204e-01 4.30171847e-01 2.73696959e-01
-1.05852280e-02 1.22333989e-01 3.28219503e-01 1.32145822e+00
-3.34284037e-01 -7.95731366e-01 -3.48937839e-01 7.42960155e-01
-7.67162025e-01 -8.91152248e-02 -1.38330743e-01 8.31062973e-01
-1.80423155e-01 7.57988453e-01 -1.47464752e-01 -6.73525706e-02
2.29802415e-01 2.55227804e-01 5.36003411e-01 -1.34032869e+00
-1.18508041e+00 -9.96101350e-02 3.24330419e-01 -3.99477631e-01
-8.62361789e-01 -7.16132522e-01 -1.44173622e+00 4.67607021e-01
-5.06966412e-01 -1.23741562e-02 3.78554970e-01 1.02762175e+00
4.85957563e-02 8.81103754e-01 7.15360403e-01 -4.75646555e-01
-9.91793275e-01 -1.00869966e+00 -2.19483182e-01 7.98004031e-01
4.65515912e-01 -4.04705733e-01 -2.76829958e-01 1.94819584e-01] | [14.878756523132324, -2.2396321296691895] |
d304d07d-de4d-4968-b1e7-9b93c73f24f0 | calibrated-data-dependent-constraints-with | 2301.06195 | null | https://arxiv.org/abs/2301.06195v1 | https://arxiv.org/pdf/2301.06195v1.pdf | Calibrated Data-Dependent Constraints with Exact Satisfaction Guarantees | We consider the task of training machine learning models with data-dependent constraints. Such constraints often arise as empirical versions of expected value constraints that enforce fairness or stability goals. We reformulate data-dependent constraints so that they are calibrated: enforcing the reformulated constraints guarantees that their expected value counterparts are satisfied with a user-prescribed probability. The resulting optimization problem is amendable to standard stochastic optimization algorithms, and we demonstrate the efficacy of our method on a fairness-sensitive classification task where we wish to guarantee the classifier's fairness (at test time). | ['Mikhail Yurochkin', 'Yuekai Sun', 'Songkai Xue'] | 2023-01-15 | null | null | null | null | ['stochastic-optimization'] | ['methodology'] | [ 2.36737952e-01 4.68660980e-01 -8.66169393e-01 -1.11159968e+00
-8.82769823e-01 -5.81097305e-01 4.85832423e-01 8.66269544e-02
-9.61471677e-01 1.29514253e+00 -2.20646486e-01 -4.87039655e-01
-3.44249457e-01 -5.33925712e-01 -3.50989282e-01 -6.61844552e-01
1.22131750e-01 5.28611839e-01 -3.47309887e-01 9.85149965e-02
5.27698994e-01 5.85271001e-01 -1.23988414e+00 8.62959679e-03
1.07307637e+00 1.24272358e+00 -7.25338280e-01 5.80472291e-01
3.77338916e-01 8.64286840e-01 -5.63020945e-01 -9.07390356e-01
5.73092580e-01 -2.81077117e-01 -1.00327718e+00 1.65430814e-01
2.96802104e-01 -2.49452770e-01 1.21650591e-01 1.25572920e+00
1.92500651e-01 3.46148193e-01 8.68761063e-01 -1.93291306e+00
-5.25611401e-01 2.51736343e-01 -3.27920318e-01 6.44982904e-02
2.51019597e-02 1.26829624e-01 1.35316348e+00 -4.73560870e-01
2.58806080e-01 1.19623387e+00 4.03398365e-01 1.11159158e+00
-1.39849579e+00 -4.46714282e-01 2.47274309e-01 -2.33795211e-01
-1.01461554e+00 -8.69392455e-01 3.31275940e-01 -3.86511743e-01
5.16143084e-01 5.30315042e-01 9.46228132e-02 9.45949614e-01
2.24115714e-01 7.45891094e-01 1.04024923e+00 -3.70166332e-01
7.71624744e-01 3.81750792e-01 2.69613355e-01 4.54941124e-01
4.91712332e-01 4.60812271e-01 -4.77162778e-01 -5.74784517e-01
2.81366646e-01 -2.97895044e-01 -2.10387647e-01 -5.40771306e-01
-6.00831270e-01 1.36808658e+00 2.55538169e-02 -3.13176006e-01
-2.06114143e-01 8.93569812e-02 4.28968042e-01 4.86272007e-01
7.49920845e-01 5.47448277e-01 -5.81905782e-01 2.01504394e-01
-8.99703503e-01 5.18510461e-01 9.61743474e-01 1.03454649e+00
1.65072680e-01 1.62827760e-01 -6.30301535e-01 6.58986330e-01
3.85800272e-01 3.71406108e-01 1.63509294e-01 -1.31795883e+00
4.52504516e-01 2.73777489e-02 7.00064898e-01 -4.91647780e-01
-1.48911700e-01 -4.77571875e-01 -4.86143589e-01 5.19108653e-01
6.70179963e-01 -2.10674495e-01 -4.97683972e-01 2.15891933e+00
2.97301680e-01 -2.33843088e-01 2.39874013e-02 9.17060912e-01
7.82999545e-02 1.92399964e-01 3.05043757e-01 -8.69031250e-01
7.45809495e-01 -4.52325791e-01 -8.17655146e-01 -7.94343743e-03
3.76382142e-01 -2.96568751e-01 1.22381556e+00 4.89902735e-01
-1.27619159e+00 1.00667648e-01 -9.84826267e-01 4.84271087e-02
-2.95472884e-04 -4.34370339e-01 6.00519359e-01 1.27787662e+00
-6.62274837e-01 7.83185720e-01 -4.34079677e-01 2.78494507e-01
7.82457888e-01 4.50623155e-01 -3.08062017e-01 2.41009131e-01
-1.20539856e+00 9.70273912e-01 1.14855491e-01 2.44850233e-01
-9.96049166e-01 -7.82148302e-01 -8.22217703e-01 9.55461338e-02
5.66701293e-01 -6.13048851e-01 1.55001199e+00 -1.05496383e+00
-1.60255027e+00 1.48409367e+00 1.23851448e-02 -5.63140452e-01
1.21239865e+00 8.09531808e-02 -2.18892157e-01 -4.34430689e-01
1.82789281e-01 9.68792811e-02 8.32435429e-01 -1.12364495e+00
-7.76475608e-01 -2.55211204e-01 2.72253066e-01 1.76836506e-01
-3.76934767e-01 3.14228028e-01 7.49785081e-03 -5.46496749e-01
-5.67808568e-01 -7.41581678e-01 -4.91860181e-01 4.73768115e-01
-5.16222477e-01 -1.33426532e-01 3.60580236e-01 -8.73016268e-02
1.11086166e+00 -2.00547194e+00 -3.01657259e-01 4.49723035e-01
1.99634627e-01 3.22685763e-02 -8.66437505e-04 -2.30345547e-01
-1.04927674e-01 5.01314580e-01 -7.14120924e-01 -5.85503519e-01
4.77754772e-01 4.48306471e-01 -2.97329217e-01 9.06866908e-01
1.43012345e-01 4.44730729e-01 -7.60094643e-01 -4.38217223e-01
-9.12999827e-03 -2.28613868e-01 -8.43580961e-01 4.56925809e-01
-2.29995757e-01 2.13112757e-01 -5.57470262e-01 3.19475085e-01
6.89142525e-01 -1.57010015e-02 9.44385901e-02 4.12094563e-01
1.95350572e-01 1.43512204e-01 -1.12820876e+00 9.00489330e-01
-4.04670447e-01 1.16338149e-01 3.42721909e-01 -1.25901425e+00
5.80354214e-01 1.23720050e-01 4.81051356e-01 -1.90674841e-01
3.79204273e-01 -3.45520256e-03 -1.71633109e-01 -2.99925476e-01
3.71039331e-01 -8.92921388e-01 -3.65271926e-01 5.55207253e-01
-1.55671746e-01 -3.26481074e-01 -1.98469460e-01 5.28612398e-02
5.69470704e-01 -4.20000404e-01 6.20534360e-01 -8.33663404e-01
3.77893031e-01 -2.14099795e-01 1.01383936e+00 1.13482940e+00
-7.03892946e-01 3.86727124e-01 1.02830184e+00 -2.24950522e-01
-9.47254002e-01 -1.04354012e+00 -5.22172213e-01 1.16352320e+00
-3.03918451e-01 2.06128851e-01 -5.11915684e-01 -1.17139339e+00
6.28019512e-01 1.09249246e+00 -9.29642558e-01 -2.75447577e-01
2.78609172e-02 -1.05811965e+00 2.77319133e-01 2.29789570e-01
8.94065201e-02 -3.59139919e-01 -6.02919877e-01 -1.34008393e-01
2.31450960e-01 -8.57484400e-01 -7.06829011e-01 2.14465454e-01
-6.35730147e-01 -1.22761941e+00 -4.22299862e-01 -1.78616613e-01
8.21551502e-01 -4.64831054e-01 1.25485611e+00 1.16872363e-01
-8.16437379e-02 2.52686322e-01 1.03562482e-01 -6.42068088e-01
-4.70255971e-01 -3.79079908e-01 2.70753950e-01 1.91743314e-01
1.44579962e-01 -7.55667612e-02 -2.86085576e-01 4.02687818e-01
-7.85235584e-01 -4.62068141e-01 -3.30609977e-01 1.19173896e+00
5.27263582e-01 -2.73548309e-02 8.53502095e-01 -1.52561069e+00
1.04646373e+00 -3.47235411e-01 -1.11117077e+00 4.65318888e-01
-1.28480864e+00 1.74970642e-01 7.57578611e-01 -2.55481362e-01
-1.14226973e+00 -1.42450750e-01 2.85545588e-01 -3.60399097e-01
2.73096144e-01 3.37503344e-01 -4.92143750e-01 -1.69725746e-01
6.72009647e-01 -5.17628133e-01 -1.67384204e-02 -1.95696950e-01
2.93894142e-01 7.35894799e-01 5.10404170e-01 -1.16833723e+00
4.93758947e-01 7.62448609e-02 2.65035391e-01 1.04195550e-01
-1.24687541e+00 2.13218719e-01 -2.73504704e-01 -1.86891973e-01
4.61794972e-01 -4.64518666e-01 -1.24618053e+00 -3.65300439e-02
-7.05708921e-01 -3.14197779e-01 -6.37963057e-01 3.86216283e-01
-8.38495016e-01 3.93575877e-02 -1.20559946e-01 -1.46229029e+00
-4.59402353e-02 -9.95366454e-01 6.71539485e-01 1.72804650e-02
-1.30365536e-01 -1.16349924e+00 -1.77271649e-01 1.31750584e-01
2.76247621e-01 5.55202961e-01 1.01505542e+00 -8.32700968e-01
3.98446620e-03 -4.26720738e-01 -4.48304787e-02 4.78118569e-01
-7.47790188e-02 2.15874463e-01 -1.08879054e+00 -3.57857555e-01
3.89699906e-01 -5.82706511e-01 6.75551355e-01 5.72161376e-01
1.83671713e+00 -8.22500706e-01 1.19113937e-01 6.19839430e-01
1.40162802e+00 -7.34285777e-03 -1.35218605e-01 2.50036027e-02
5.40176779e-03 8.80853176e-01 7.14563489e-01 1.09573019e+00
9.37137380e-02 4.92353588e-01 4.48021382e-01 1.25833005e-01
7.67989576e-01 -9.90835112e-03 -7.91680738e-02 -1.84272394e-01
1.16668381e-01 -3.75090569e-01 -7.70382881e-01 3.28116328e-01
-2.07669663e+00 -9.24402714e-01 1.74969845e-02 2.81706285e+00
1.24603367e+00 2.21478641e-01 1.98079437e-01 -1.13219637e-02
7.81052172e-01 1.19790815e-01 -9.81164694e-01 -9.78487730e-01
5.07518537e-02 2.23419905e-01 7.88372934e-01 1.08980823e+00
-1.06444263e+00 4.81505364e-01 8.49180508e+00 8.08713675e-01
-4.90898818e-01 2.16175690e-02 1.31225419e+00 -6.72690451e-01
-6.74770772e-01 -9.60097536e-02 -5.25651991e-01 4.45902944e-01
9.22826111e-01 -9.10293102e-01 5.44704556e-01 1.05484962e+00
3.72841477e-01 -1.11471429e-01 -1.59868371e+00 7.26601660e-01
-4.02592450e-01 -9.57506835e-01 -2.32105255e-01 8.81349742e-02
7.46217191e-01 -4.71275568e-01 4.55733985e-01 2.82366008e-01
9.69373047e-01 -1.41266096e+00 8.47814798e-01 4.78639483e-01
1.03596509e+00 -1.11492884e+00 6.88384295e-01 3.07309657e-01
-2.34993637e-01 -1.69172645e-01 -3.95706356e-01 -1.48470938e-01
5.82063459e-02 1.17763197e+00 -3.10042650e-01 2.68585503e-01
8.75932500e-02 4.22174603e-01 1.36528267e-02 7.70964682e-01
-2.73561001e-01 4.62646425e-01 -1.76982269e-01 6.03710674e-02
-4.09247400e-03 -2.20021784e-01 3.14911753e-01 1.05280864e+00
-2.42960438e-01 2.30997175e-01 4.75050658e-02 1.06312323e+00
-4.60171282e-01 4.87988070e-02 -4.63571101e-01 1.60555139e-01
5.32673776e-01 9.44095135e-01 -2.34440207e-01 -1.80357575e-01
-1.71910211e-01 6.77628398e-01 4.90348548e-01 2.72692919e-01
-9.24396098e-01 -1.96166575e-01 1.09653008e+00 -4.60287392e-01
-4.54499751e-01 3.03787380e-01 -8.27776611e-01 -1.25996840e+00
-1.52014417e-03 -9.25816774e-01 9.29292679e-01 -3.41032684e-01
-1.88682032e+00 2.12154090e-01 -5.21043502e-02 -9.02509749e-01
-2.34161198e-01 -8.32220137e-01 -5.84703982e-01 1.14259648e+00
-1.52318561e+00 -4.11550611e-01 3.57811391e-01 8.15198898e-01
2.07254360e-03 -3.22919935e-01 9.55214262e-01 -1.81683153e-02
-8.68028820e-01 1.05644763e+00 8.29324946e-02 -1.84141204e-01
5.94544411e-01 -1.49149716e+00 -1.33122817e-01 9.71640408e-01
-3.40727150e-01 5.34823954e-01 9.65716600e-01 -2.62547851e-01
-8.88897359e-01 -1.05473793e+00 9.65209663e-01 -5.76445699e-01
5.15465438e-01 -3.25102270e-01 -5.66432595e-01 6.77653909e-01
-2.38156900e-01 5.88468730e-01 8.77857447e-01 4.05873060e-01
-3.70316982e-01 -1.71552032e-01 -1.88083363e+00 3.44150007e-01
9.35591280e-01 -5.52295506e-01 -4.55025554e-01 6.13053262e-01
4.05328721e-01 -4.16767031e-01 -8.45298707e-01 3.70363563e-01
6.03257298e-01 -7.61080503e-01 5.46497941e-01 -1.64986086e+00
4.43529427e-01 2.73673743e-01 -3.53888988e-01 -1.27269924e+00
-9.90659669e-02 -9.12201941e-01 6.96617216e-02 9.64902878e-01
6.29387081e-01 -8.70840013e-01 9.16403532e-01 1.86704302e+00
3.12152714e-01 -6.98090553e-01 -1.32750607e+00 -9.81977642e-01
5.32249153e-01 -4.61492956e-01 6.29239082e-01 1.26800776e+00
2.83509463e-01 -2.02938750e-01 -6.89666390e-01 -3.41582932e-02
9.34387684e-01 3.79786119e-02 2.75197387e-01 -1.09584379e+00
-3.74755174e-01 -8.53142262e-01 -3.75753641e-02 -2.80166775e-01
6.94888711e-01 -9.33140457e-01 1.43464282e-01 -7.06231833e-01
3.92275304e-01 -7.14867651e-01 -6.38512135e-01 6.06947005e-01
-5.05387664e-01 -2.07292661e-01 1.05127729e-01 -1.66708589e-01
-5.29596031e-01 6.69378877e-01 8.65462124e-01 7.77163953e-02
1.47151813e-01 6.73006833e-01 -1.20084774e+00 3.90385628e-01
7.68094301e-01 -7.97020674e-01 -5.47521770e-01 -1.38340071e-01
1.90996245e-01 1.93057761e-01 2.56724834e-01 -2.18267724e-01
-1.26848118e-02 -1.02963984e+00 2.05324203e-01 3.07437062e-01
-1.41799018e-01 -7.58304536e-01 -5.83114438e-02 5.27569592e-01
-1.16830909e+00 -1.74697742e-01 -2.15720594e-01 4.72940415e-01
-5.05699404e-02 -4.16498899e-01 1.35805678e+00 1.50304750e-01
2.25651842e-02 5.77758312e-01 -2.85723180e-01 5.47014654e-01
1.15030909e+00 3.02146167e-01 -1.52445346e-01 -4.98677760e-01
-7.13395298e-01 6.32216811e-01 3.54497761e-01 -5.54709546e-02
3.68378758e-01 -1.21150279e+00 -9.04428124e-01 5.79483919e-02
2.59331286e-01 -2.93526709e-01 -3.97460222e-01 6.05849147e-01
4.45424989e-02 2.77883887e-01 -8.26793462e-02 -1.32089213e-01
-9.99239028e-01 7.39830494e-01 7.13600576e-01 -3.01840663e-01
-2.95401514e-02 1.13526034e+00 1.73839927e-01 -4.68128979e-01
4.10664111e-01 -5.20761451e-03 2.98675954e-01 -2.23502859e-01
3.88190061e-01 3.76994163e-01 -7.40053654e-02 -4.51976694e-02
-4.67493117e-01 -1.06982917e-01 2.32222248e-02 -4.08604383e-01
1.14287746e+00 -6.89514428e-02 -1.73978191e-02 2.13494718e-01
1.04548669e+00 6.31576926e-02 -1.35116708e+00 -2.84923047e-01
1.76154211e-01 -9.22613502e-01 1.16779707e-01 -1.11583698e+00
-1.03510869e+00 6.45809770e-01 2.10593164e-01 3.36626880e-02
1.02057672e+00 -5.89992583e-01 3.70036624e-02 2.86514699e-01
4.30690795e-01 -1.19242215e+00 -5.78473628e-01 2.30065405e-01
7.44165540e-01 -1.41701066e+00 1.60801038e-01 -2.74377257e-01
-8.80500674e-01 8.64819527e-01 4.49610680e-01 -1.23053156e-01
7.35506713e-01 4.56401289e-01 -1.06928073e-01 3.89148444e-01
-9.33260202e-01 1.48841769e-01 3.94617200e-01 6.72752023e-01
2.60391325e-01 3.79197776e-01 -7.39519298e-01 8.95587504e-01
-1.83436856e-01 6.24307431e-02 5.05709827e-01 7.19682157e-01
-3.24014187e-01 -1.08355176e+00 -1.11531876e-01 8.31467509e-01
-7.51537502e-01 2.00989656e-02 -3.57492208e-01 4.45236236e-01
-2.44097993e-01 1.20379639e+00 -2.71962211e-02 7.31042027e-02
3.24168652e-01 -3.17764804e-02 4.00969476e-01 -5.71334660e-01
-4.22935337e-01 -3.98827404e-01 3.26969355e-01 -6.96657419e-01
-3.94724637e-01 -6.96709156e-01 -9.19664919e-01 -7.55421579e-01
-3.48477662e-01 4.25674766e-01 3.39248210e-01 9.96660769e-01
-4.08311561e-02 1.92407697e-01 1.25359166e+00 -8.25878233e-02
-1.55793846e+00 -3.72652531e-01 -8.71750414e-01 4.39669490e-01
5.27059495e-01 -5.02131701e-01 -5.80344200e-01 -1.62978709e-01] | [8.841141700744629, 5.309725761413574] |
9fd332e5-4f53-4694-9035-a6aeaa38c69e | learning-to-transfer-role-assignment-across | 2204.12937 | null | https://arxiv.org/abs/2204.12937v1 | https://arxiv.org/pdf/2204.12937v1.pdf | Learning to Transfer Role Assignment Across Team Sizes | Multi-agent reinforcement learning holds the key for solving complex tasks that demand the coordination of learning agents. However, strong coordination often leads to expensive exploration over the exponentially large state-action space. A powerful approach is to decompose team works into roles, which are ideally assigned to agents with the relevant skills. Training agents to adaptively choose and play emerging roles in a team thus allows the team to scale to complex tasks and quickly adapt to changing environments. These promises, however, have not been fully realised by current role-based multi-agent reinforcement learning methods as they assume either a pre-defined role structure or a fixed team size. We propose a framework to learn role assignment and transfer across team sizes. In particular, we train a role assignment network for small teams by demonstration and transfer the network to larger teams, which continue to learn through interaction with the environment. We demonstrate that re-using the role-based credit assignment structure can foster the learning process of larger reinforcement learning teams to achieve tasks requiring different roles. Our proposal outperforms competing techniques in enriched role-enforcing Prey-Predator games and in new scenarios in the StarCraft II Micro-Management benchmark. | ['Truyen Tran', 'Svetha Venkatesh', 'Phuoc Nguyen', 'Dung Nguyen'] | 2022-04-17 | null | null | null | null | ['starcraft-ii'] | ['playing-games'] | [ 7.44423121e-02 6.69236248e-03 9.83354151e-02 8.55522975e-02
-4.03374195e-01 -7.50398457e-01 5.26986361e-01 3.69683713e-01
-9.27486420e-01 1.17931199e+00 -1.88922375e-01 1.10726401e-01
-3.66246372e-01 -7.97068357e-01 -6.10607803e-01 -9.93021011e-01
-5.46126544e-01 1.14691269e+00 5.74122369e-01 -9.57165480e-01
2.50236630e-01 7.92009458e-02 -1.81632841e+00 -3.67421545e-02
7.83753872e-01 5.06698608e-01 4.26250577e-01 8.51849258e-01
4.25026596e-01 1.18651557e+00 -9.83600616e-01 1.77587613e-01
5.83302319e-01 -5.07859111e-01 -6.39878929e-01 1.91501826e-01
-2.30904877e-01 -2.67467976e-01 3.11272889e-01 6.29565299e-01
5.84057629e-01 5.81856966e-01 3.06537330e-01 -1.56846678e+00
8.37495085e-03 9.63485718e-01 -4.80499983e-01 1.64973944e-01
1.67249814e-01 3.59097600e-01 1.09053397e+00 -2.32062832e-01
4.21630144e-01 1.26774180e+00 4.56102639e-01 7.04772413e-01
-1.09557617e+00 -8.21966708e-01 5.00536442e-01 5.14136031e-02
-7.67414689e-01 1.28355324e-01 3.50061744e-01 -2.90538162e-01
9.80465591e-01 -3.85515660e-01 1.05540824e+00 9.44885373e-01
2.14789748e-01 4.91640925e-01 1.28779435e+00 -3.08321863e-01
4.24854308e-01 -1.66311368e-01 -4.47416186e-01 6.82690382e-01
2.38166526e-01 2.60445625e-01 -5.24052978e-01 -2.98440099e-01
9.73019421e-01 -2.43578991e-03 1.20636396e-01 -6.38589859e-01
-1.47434092e+00 9.20272410e-01 2.20429972e-01 1.86565921e-01
-7.05424428e-01 5.06802499e-01 5.83248019e-01 1.10408437e+00
6.75448552e-02 1.23059797e+00 -5.78794181e-01 -2.76681781e-01
-2.29157850e-01 5.18868744e-01 7.93770075e-01 3.04649889e-01
7.38095760e-01 6.46591410e-02 1.76457494e-01 6.05059683e-01
1.25506669e-01 2.46391743e-01 3.48849058e-01 -1.44667888e+00
3.14818978e-01 7.57799923e-01 5.64635038e-01 -2.75658518e-01
-3.99897248e-01 -2.60769516e-01 -6.48128867e-01 1.03182101e+00
5.25699139e-01 -7.05649495e-01 -3.82191747e-01 1.79039586e+00
6.75416112e-01 1.57434270e-01 3.39898229e-01 6.83113158e-01
-6.33700117e-02 5.05605578e-01 -1.17732100e-01 -3.56700838e-01
1.07250416e+00 -1.24524891e+00 -2.96546817e-01 -1.35706946e-01
3.19099009e-01 -1.67240098e-01 8.28719378e-01 6.68627560e-01
-1.17704439e+00 -5.18827140e-01 -1.08797753e+00 9.57873285e-01
-3.32849175e-02 -6.95818007e-01 6.50006592e-01 4.08834428e-01
-1.24162817e+00 7.20775545e-01 -8.86127412e-01 -1.54345542e-01
1.94888189e-01 8.80354464e-01 -2.14399308e-01 3.63565445e-01
-1.24599016e+00 1.03072548e+00 3.84791702e-01 -2.31058493e-01
-1.70625162e+00 -3.82337391e-01 -6.70029819e-01 3.62958498e-02
1.00002682e+00 -7.50118077e-01 1.62221467e+00 -1.24588776e+00
-1.91227400e+00 2.83813328e-01 8.34740698e-01 -5.42568207e-01
4.90288407e-01 1.45988792e-01 3.93870085e-01 2.06910267e-01
1.85677290e-01 6.34357750e-01 1.08954823e+00 -1.41173398e+00
-1.15483129e+00 -7.63649866e-03 7.72156537e-01 9.97152984e-01
-3.65052074e-01 2.90025711e-01 3.66761178e-01 -3.98233056e-01
-4.91401047e-01 -1.06837690e+00 -8.24313581e-01 -3.41539085e-01
4.34664369e-01 -4.82934386e-01 1.80926383e-01 3.42207193e-01
6.31671011e-01 -1.78951395e+00 7.91331410e-01 2.33164608e-01
4.69411403e-01 -6.59659132e-02 -3.49881321e-01 7.43128181e-01
5.93152307e-02 -1.76406145e-01 3.81109305e-02 -3.72071832e-01
2.11771429e-01 6.04932427e-01 1.95370853e-01 3.36539656e-01
-5.73933795e-02 4.19411719e-01 -1.24853694e+00 -2.52241641e-01
-2.02578798e-01 -9.80678126e-02 -8.47166359e-01 5.01621544e-01
-4.72522855e-01 6.42884493e-01 -4.83378589e-01 3.04253668e-01
6.60496503e-02 -2.14091986e-01 6.54879212e-01 1.01934278e+00
-2.25571394e-01 5.37518673e-02 -1.55194509e+00 1.26494443e+00
-3.99040520e-01 8.69233608e-02 7.42339551e-01 -1.11045659e+00
8.17477167e-01 4.30148154e-01 8.77945006e-01 -5.28662980e-01
3.78136560e-02 1.64686009e-01 6.76140368e-01 -5.12626246e-02
4.16565269e-01 -4.97238159e-01 -3.55301470e-01 1.13868487e+00
4.04741541e-02 -5.35647154e-01 5.82735002e-01 2.58203540e-02
1.30507147e+00 6.52864501e-02 2.87786782e-01 -5.54976352e-02
3.59055340e-01 2.40742594e-01 6.16700947e-01 1.25676811e+00
-2.83492178e-01 -9.22516733e-02 6.84967101e-01 -6.27245963e-01
-9.48221862e-01 -7.88392365e-01 4.67931122e-01 2.01630640e+00
2.59175420e-01 -1.66546240e-01 -4.53610390e-01 -7.91518688e-01
2.14652449e-01 6.12202771e-02 -8.17506492e-01 -1.12398431e-01
-9.72516119e-01 -7.30682611e-01 3.23308796e-01 3.33046436e-01
2.52592713e-01 -1.77437818e+00 -1.38411653e+00 6.69964314e-01
1.99104086e-01 -4.99198198e-01 -4.81265485e-01 8.68515015e-01
-6.44619942e-01 -1.29280233e+00 -6.27822042e-01 -9.58820403e-01
6.52420878e-01 1.51759461e-01 1.13036895e+00 4.01589811e-01
-2.51843818e-02 7.49016225e-01 -5.78445494e-01 -5.46811163e-01
-7.37590134e-01 4.01024781e-02 6.98557138e-01 -2.57012516e-01
-3.37720692e-01 -8.39915812e-01 -5.10552883e-01 6.75218940e-01
-7.32290924e-01 -1.17244400e-01 6.09544039e-01 1.29201210e+00
1.68513849e-01 4.03993458e-01 7.76756227e-01 -5.51725268e-01
1.04138207e+00 -3.25226098e-01 -8.34531665e-01 1.95094392e-01
-5.37929833e-01 1.87180340e-01 7.47354031e-01 -8.80558312e-01
-7.54792452e-01 -1.24308899e-01 3.61886740e-01 -1.14848144e-01
1.67155355e-01 3.84133846e-01 2.65544623e-01 -1.15281753e-01
7.17431605e-01 9.43481475e-02 5.15028536e-01 1.32105835e-02
-6.25984669e-02 4.29263301e-02 3.56157832e-02 -1.16472018e+00
8.82562220e-01 1.62462920e-01 -8.67942870e-02 -2.37438917e-01
-4.69709396e-01 -3.41944873e-01 -5.60074925e-01 -4.02085602e-01
6.20180964e-01 -1.02486610e+00 -1.25640738e+00 6.79886699e-01
-8.65745068e-01 -1.06916320e+00 -5.01394331e-01 2.37921163e-01
-8.02729607e-01 -8.80070310e-03 -5.27256429e-01 -6.72450006e-01
-1.56611484e-02 -1.10937595e+00 6.04250908e-01 3.16860348e-01
5.63122295e-02 -9.81667817e-01 6.47308409e-01 1.40525654e-01
5.82948983e-01 1.99391142e-01 5.78065991e-01 -5.84771395e-01
-3.11685532e-01 3.34342837e-01 5.35125911e-01 1.52889013e-01
1.20795824e-01 -3.09783340e-01 -2.82671720e-01 -7.99792230e-01
-1.27786189e-01 -1.15413451e+00 4.99500751e-01 1.31572157e-01
5.32805800e-01 -2.14842051e-01 -9.48665440e-02 -1.87610798e-02
9.86303985e-01 3.60317588e-01 6.37198612e-02 8.45149279e-01
1.91515610e-01 7.62780726e-01 1.02532136e+00 8.53368104e-01
4.76067305e-01 6.36477411e-01 9.43673611e-01 1.73182160e-01
5.22724569e-01 1.08345831e-02 7.47711301e-01 4.36407357e-01
-5.70533991e-01 -6.59113377e-03 -8.72597694e-01 2.97442317e-01
-2.31594682e+00 -1.24850142e+00 4.43388522e-01 2.02961755e+00
1.19817615e+00 5.08762777e-01 7.34070063e-01 -8.88304487e-02
5.48501253e-01 -8.26940909e-02 -7.85582364e-01 -2.06731781e-01
1.44364998e-01 3.31069201e-01 1.62563011e-01 5.80074966e-01
-1.01968896e+00 9.00723696e-01 6.21654177e+00 2.82728076e-01
-7.79665768e-01 2.25767896e-01 2.61074215e-01 -3.69002342e-01
1.67825356e-01 -5.46852015e-02 -5.79721630e-01 1.10945150e-01
7.67873228e-01 -2.77783006e-01 7.79783905e-01 7.29365528e-01
2.06330836e-01 -3.52104545e-01 -1.13040638e+00 6.28532290e-01
-1.97290614e-01 -1.16168439e+00 -5.21479487e-01 2.17006177e-01
8.66242230e-01 5.96096255e-02 -5.23778908e-02 8.54014099e-01
1.31119025e+00 -1.15513790e+00 7.84395158e-01 -8.87935832e-02
3.53175610e-01 -9.10874248e-01 7.17312813e-01 8.03141892e-01
-1.17513144e+00 -6.60302401e-01 -3.78363937e-01 -7.71351814e-01
-1.70594186e-01 -4.81827319e-01 -1.01649296e+00 1.73473343e-01
7.10495234e-01 5.37228048e-01 -3.24852943e-01 9.08388317e-01
-1.82793379e-01 3.63964200e-01 -2.51232594e-01 -3.53217065e-01
5.32623947e-01 -3.35117072e-01 4.38663960e-01 4.04017895e-01
-2.54986100e-02 8.87465551e-02 1.02292836e+00 2.23972365e-01
1.84909716e-01 -1.70513690e-01 -4.24724817e-01 1.14825899e-02
4.97491330e-01 1.19627655e+00 -8.69667530e-01 -2.36442789e-01
-6.23864345e-02 7.74393618e-01 6.57616377e-01 -6.02683704e-03
-8.38006258e-01 -8.97824764e-02 9.13837731e-01 1.02901489e-01
3.77398312e-01 -2.99898326e-01 4.06391412e-01 -8.48666191e-01
-4.17277068e-01 -1.56421781e+00 5.43224275e-01 -3.28374892e-01
-1.24003792e+00 4.96197671e-01 9.18550789e-02 -1.31044555e+00
-4.88697976e-01 -4.77156609e-01 -6.74683988e-01 4.35523897e-01
-1.43641317e+00 -8.47336054e-01 -2.28414580e-01 8.33013415e-01
6.57757282e-01 -6.49073660e-01 8.14575076e-01 -6.89844862e-02
-5.38656652e-01 2.01582000e-01 3.36550698e-02 -1.49113342e-01
8.85680199e-01 -1.79095697e+00 -8.22056159e-02 2.57628798e-01
-2.20779747e-01 4.03011262e-01 7.85238147e-01 -4.68815863e-01
-1.28106821e+00 -6.05657876e-01 1.70589611e-02 -5.12074232e-01
1.07751048e+00 -4.66249913e-01 -4.97039706e-01 4.36451584e-01
5.73653579e-01 -7.41087124e-02 5.75554430e-01 2.30875924e-01
-3.75930406e-02 -1.56845734e-01 -7.98581302e-01 4.67131525e-01
9.25006628e-01 -2.47618798e-02 -6.32335722e-01 2.31889099e-01
6.24947071e-01 -4.07936126e-01 -7.28294790e-01 -4.84433658e-02
3.81152183e-01 -8.71360838e-01 7.86186635e-01 -8.01462591e-01
3.69514257e-01 -3.61223698e-01 3.49542588e-01 -1.92523623e+00
-3.31214458e-01 -9.41331208e-01 1.98399127e-01 6.72842681e-01
3.76503348e-01 -6.69835925e-01 6.79907024e-01 4.23256271e-02
-6.12991154e-02 -5.97592592e-01 -8.63847315e-01 -7.07538843e-01
3.67838115e-01 2.65022814e-01 3.74955148e-01 9.13350642e-01
2.32625827e-01 4.92040753e-01 -5.09160578e-01 1.21048808e-01
6.35878563e-01 1.38896435e-01 9.39429462e-01 -1.53384757e+00
-1.07481039e+00 -4.91723657e-01 1.44891143e-01 -6.77583873e-01
2.64066547e-01 -4.95045513e-01 3.77962291e-01 -1.40399086e+00
1.94856346e-01 -1.05454040e+00 -4.18194503e-01 7.93499589e-01
-2.67968029e-01 -7.88847953e-02 3.54536176e-01 3.28795910e-01
-1.37314403e+00 5.88201284e-01 1.61347640e+00 -7.57327452e-02
-4.80773121e-01 2.86152124e-01 -7.98228502e-01 6.09420776e-01
8.99448276e-01 -7.28016436e-01 -5.52556396e-01 -1.70483261e-01
5.96366584e-01 3.42875868e-01 8.82148743e-02 -9.92957711e-01
4.09700722e-01 -6.90613449e-01 -6.33839220e-02 2.48948425e-01
2.59864867e-01 -8.11323285e-01 -1.04690127e-01 9.42454278e-01
-4.06464219e-01 6.70084536e-01 -1.45392790e-02 7.39840150e-01
-2.25039124e-01 -2.74190873e-01 6.67794049e-01 -5.51728487e-01
-5.86324334e-01 1.01344869e-01 -9.14120495e-01 2.56386012e-01
1.49347377e+00 -3.17208648e-01 -1.47147745e-01 -5.75074494e-01
-7.81641483e-01 1.10431194e+00 5.55081069e-01 3.19612563e-01
4.18928087e-01 -8.78910720e-01 -9.85232532e-01 -1.47156253e-01
-2.10140329e-02 2.83730298e-01 -1.76605918e-02 7.06434190e-01
-3.78904015e-01 -3.82123291e-01 -9.43896770e-01 -3.55179369e-01
-1.44818926e+00 2.79726028e-01 7.22053349e-01 -8.85424852e-01
-3.15512538e-01 8.54728758e-01 1.18720112e-02 -6.82454228e-01
3.92720670e-01 -1.53929129e-01 -5.31886280e-01 3.95550966e-01
4.77905631e-01 5.42653948e-02 -3.58556777e-01 -4.41785865e-02
-1.16401657e-01 1.36447579e-01 -2.08874211e-01 -3.85755807e-01
1.76591587e+00 1.98115110e-01 -2.06745476e-01 2.67074198e-01
-5.90316169e-02 -2.58242369e-01 -1.84816265e+00 -9.35842097e-02
-2.48296335e-02 -7.49441683e-02 -4.77947503e-01 -6.95508778e-01
-6.16475523e-01 3.89131069e-01 2.37470597e-01 4.48001713e-01
8.15276206e-01 -1.52562872e-01 8.05943739e-03 9.40988481e-01
9.41050887e-01 -1.46082747e+00 1.25666749e+00 1.00768626e+00
7.83410668e-01 -1.27129161e+00 -7.07637295e-02 1.68869972e-01
-1.06137049e+00 9.71561670e-01 1.15738821e+00 -4.30795103e-01
3.21858019e-01 4.24057156e-01 5.08127771e-02 -1.99105635e-01
-1.34486890e+00 -3.03339720e-01 -6.13522053e-01 9.01662111e-01
-1.98753387e-01 1.87169164e-02 1.31978571e-01 2.68766463e-01
2.97163446e-02 -4.55538452e-01 9.93306577e-01 1.27966559e+00
-8.19718599e-01 -1.58574414e+00 -7.11549580e-01 2.78469115e-01
-9.97283980e-02 2.05039158e-01 -3.21954727e-01 7.99764335e-01
4.32716519e-01 8.55311930e-01 1.33851841e-01 7.74733350e-02
3.02231371e-01 -1.47636339e-01 7.41209507e-01 -1.18004358e+00
-1.42745805e+00 2.00340040e-02 8.60676020e-02 -3.47085804e-01
-6.60659313e-01 -8.55349183e-01 -1.30850184e+00 -1.43680692e-01
1.46978197e-03 5.27892888e-01 6.13455139e-02 7.11735785e-01
-1.61111951e-02 7.65282273e-01 8.33153844e-01 -1.12861276e+00
-1.28466809e+00 -8.92963588e-01 -6.09414756e-01 3.67597222e-01
5.04035890e-01 -1.14855731e+00 -2.46522829e-01 -3.11280757e-01] | [3.7699625492095947, 1.9702125787734985] |
e7770a8d-1cd3-448c-87b5-755c7fcb4ab2 | mmdag-multimodal-directed-acyclic-graph | null | null | https://aclanthology.org/2022.lrec-1.733 | https://aclanthology.org/2022.lrec-1.733.pdf | MMDAG: Multimodal Directed Acyclic Graph Network for Emotion Recognition in Conversation | Emotion recognition in conversation is important for an empathetic dialogue system to understand the user’s emotion and then generate appropriate emotional responses. However, most previous researches focus on modeling conversational contexts primarily based on the textual modality or simply utilizing multimodal information through feature concatenation. In order to exploit multimodal information and contextual information more effectively, we propose a multimodal directed acyclic graph (MMDAG) network by injecting information flows inside modality and across modalities into the DAG architecture. Experiments on IEMOCAP and MELD show that our model outperforms other state-of-the-art models. Comparative studies validate the effectiveness of the proposed modality fusion method. | ['Hongying Zan', 'Changyong Niu', 'Yuxiang Jia', 'Shuo Xu'] | null | null | null | null | lrec-2022-6 | ['emotion-recognition-in-conversation'] | ['natural-language-processing'] | [-1.86291516e-01 -1.01341106e-01 -1.94925129e-01 -4.99909669e-01
-4.20009404e-01 -4.47984934e-01 7.48139143e-01 1.07110523e-01
-2.32628509e-01 7.99377024e-01 9.37763572e-01 2.30989642e-02
1.80264324e-01 -5.03701687e-01 1.59447134e-01 -4.32315379e-01
2.21051559e-01 -8.64422601e-03 -3.21550280e-01 -6.97474122e-01
9.45634842e-02 2.03048497e-01 -1.13414800e+00 6.40846968e-01
6.86946750e-01 9.34363782e-01 -2.79670954e-01 6.50779188e-01
-8.48743558e-01 1.52494466e+00 -5.82184672e-01 -6.92181349e-01
-3.77325058e-01 -9.30977106e-01 -1.03964972e+00 3.61067243e-02
-4.72109020e-01 -1.08026713e-01 -5.22385359e-01 7.64294267e-01
7.63757467e-01 5.56482375e-01 4.80192155e-01 -1.43372405e+00
-1.38755441e-01 1.02621210e+00 -3.22108209e-01 -1.60264641e-01
9.91271734e-01 -1.31614581e-01 8.98186803e-01 -8.46132398e-01
6.70595288e-01 1.43336439e+00 4.45948184e-01 7.16050923e-01
-8.43513131e-01 -4.32900757e-01 4.30325150e-01 4.22485888e-01
-8.34969699e-01 -5.53183079e-01 1.48764658e+00 -3.91382501e-02
1.06442118e+00 3.68910313e-01 5.49229145e-01 1.58968604e+00
1.99680641e-01 1.06186903e+00 1.18973923e+00 -4.16239560e-01
-2.41886415e-02 1.77600920e-01 3.48295063e-01 6.22931480e-01
-6.17052972e-01 -3.93030703e-01 -1.13295686e+00 -4.35992002e-01
1.91806868e-01 -4.90877777e-02 -1.57058910e-01 1.96136937e-01
-1.16934538e+00 9.34018254e-01 1.60217062e-01 4.70012963e-01
-6.36171818e-01 -1.00583270e-01 9.01377857e-01 4.94732380e-01
4.17349279e-01 3.12162519e-01 -7.61376023e-02 -6.53684497e-01
-2.29048997e-01 -1.71510860e-01 1.15643227e+00 5.72919011e-01
4.91433680e-01 -2.79122908e-02 -1.49012432e-01 1.09806037e+00
4.48480666e-01 8.06192160e-02 3.80597711e-01 -8.44379604e-01
5.53369641e-01 1.04536009e+00 -7.20221326e-02 -1.15886748e+00
-8.48302960e-01 5.69230020e-02 -8.59003425e-01 -3.42522979e-01
1.78687759e-02 -8.32555950e-01 -3.56463432e-01 1.85224223e+00
4.01257038e-01 1.22814886e-01 6.75614953e-01 1.01468420e+00
1.41133726e+00 7.81111479e-01 4.94626880e-01 -3.29464108e-01
1.45452213e+00 -9.66586351e-01 -1.27005529e+00 -2.60907292e-01
6.91780031e-01 -8.99995327e-01 6.95719719e-01 2.27980852e-01
-9.99493480e-01 -1.95727527e-01 -8.94766450e-01 -7.52886087e-02
-4.87079263e-01 -1.18107915e-01 9.65695441e-01 5.30354202e-01
-6.21094644e-01 -1.15579486e-01 -5.22935033e-01 -4.74703908e-01
-1.12671554e-01 2.46328503e-01 -5.85460603e-01 1.45238817e-01
-1.72893596e+00 1.08839440e+00 3.98672879e-01 6.16711855e-01
-2.76772559e-01 -9.98310223e-02 -1.04538512e+00 -1.52823441e-02
5.88637814e-02 -7.32224464e-01 1.10207856e+00 -1.03298783e+00
-2.13107777e+00 3.17030311e-01 -2.21712753e-01 -9.07714963e-02
1.17840216e-01 -9.69921499e-02 -8.81969273e-01 1.99789613e-01
-7.20104218e-01 4.72704053e-01 2.78987944e-01 -1.22967553e+00
-5.57744443e-01 -3.26808274e-01 4.25980449e-01 7.34086394e-01
-7.40298390e-01 2.98429579e-01 -2.85681099e-01 -2.35771626e-01
5.72013259e-02 -8.77282023e-01 -1.60896957e-01 -7.88344383e-01
-4.60572124e-01 -3.19465399e-01 1.02917624e+00 -6.37004614e-01
1.59911990e+00 -1.98870337e+00 5.38545728e-01 2.38064349e-01
2.04250574e-01 -1.29740223e-01 -1.91602647e-01 1.11902463e+00
2.32405394e-01 -1.21012405e-01 5.44810705e-02 -6.00179970e-01
2.24686801e-01 2.59394616e-01 -6.66745603e-02 1.79739773e-01
1.28324237e-02 6.67245984e-01 -7.17182875e-01 -7.24041343e-01
3.41933787e-01 6.71775222e-01 -2.65743762e-01 6.64252758e-01
1.72057860e-02 7.15488613e-01 -6.82286739e-01 5.72447658e-01
4.12178099e-01 2.07535595e-01 6.17345214e-01 -4.24897879e-01
7.91849010e-03 2.25646108e-01 -9.81871128e-01 2.10586691e+00
-9.52020109e-01 5.64965844e-01 1.89373851e-01 -6.42582893e-01
1.07848144e+00 6.86559260e-01 5.46556354e-01 -6.37210310e-01
5.74420691e-01 -1.26293421e-01 -8.33656639e-02 -8.50464225e-01
6.92014813e-01 -2.04279393e-01 -6.89283371e-01 4.97223735e-01
2.76245087e-01 2.97508359e-01 -3.29813585e-02 4.21263397e-01
8.20407271e-01 4.88996506e-02 2.36108214e-01 2.82397270e-01
8.65661621e-01 -3.42325389e-01 5.00555813e-01 3.30746591e-01
-2.68576503e-01 -9.68718007e-02 7.75436401e-01 -2.64814138e-01
-2.36413985e-01 -5.42406082e-01 2.69681394e-01 1.44746923e+00
1.98043644e-01 -5.54281235e-01 -5.66038549e-01 -6.97019458e-01
-5.52053094e-01 6.33173466e-01 -5.73058665e-01 -1.98855251e-01
-2.97113806e-01 -7.09388793e-01 7.68249035e-01 3.98489118e-01
8.32142591e-01 -1.22386491e+00 -4.11299676e-01 4.34797764e-01
-8.99270415e-01 -1.21520698e+00 -1.58449322e-01 -7.98178688e-02
-3.47753674e-01 -6.76813960e-01 -2.04941154e-01 -6.69562101e-01
2.06550032e-01 -9.04893428e-02 9.54089880e-01 -2.76828468e-01
1.46146864e-01 7.04449296e-01 -7.25706637e-01 -1.01750292e-01
-4.94306892e-01 2.75555730e-01 -1.22548804e-01 4.66373265e-01
3.48688930e-01 -7.10159063e-01 -6.06688499e-01 1.91887036e-01
-8.17498565e-01 2.44934171e-01 3.65852416e-01 9.29912329e-01
-1.79567248e-01 -3.43304753e-01 9.62042928e-01 -8.87716293e-01
1.32206798e+00 -8.41023922e-01 2.10497826e-01 4.95096624e-01
-2.79706456e-02 6.87896386e-02 5.01817465e-01 -4.11183208e-01
-1.82439327e+00 -1.05338339e-02 -2.40674436e-01 -8.98204818e-02
-1.13010779e-01 9.37171221e-01 -4.20147419e-01 3.12910266e-02
1.93196267e-01 -7.94306099e-02 -2.14714944e-01 -2.12141052e-01
7.03333139e-01 1.14024639e+00 4.67794478e-01 -8.00979793e-01
-1.14032976e-01 1.92202270e-01 -7.57939294e-02 -6.98520064e-01
-3.70191455e-01 -5.42398214e-01 -3.54550809e-01 -8.62406135e-01
9.84394670e-01 -7.71639824e-01 -1.39045632e+00 4.48754162e-01
-1.37572598e+00 1.76200241e-01 5.69571853e-01 6.74707055e-01
-2.63273388e-01 2.84829557e-01 -9.56707776e-01 -1.24266493e+00
-5.58977008e-01 -9.13028836e-01 6.59837365e-01 5.27158499e-01
-4.90043163e-01 -1.34360385e+00 3.22163701e-01 5.06542265e-01
4.29319024e-01 4.38773602e-01 7.49266326e-01 -8.55315328e-01
3.48783880e-01 -2.77251244e-01 -8.34342912e-02 5.62598929e-02
9.68307108e-02 -7.41543919e-02 -1.08063757e+00 2.74190128e-01
-8.36433619e-02 -7.37367749e-01 6.44389153e-01 -2.40146503e-01
6.62298918e-01 -3.99601221e-01 -9.37099531e-02 9.71293896e-02
1.02343142e+00 3.13613713e-01 5.50983727e-01 4.35208157e-02
7.98998952e-01 9.42587137e-01 4.51804668e-01 8.16425025e-01
8.03280652e-01 4.15519506e-01 3.87497008e-01 -1.45439506e-01
3.08043569e-01 -1.21398062e-01 4.94353056e-01 1.19705594e+00
-2.40654871e-02 -6.30006671e-01 -5.55002153e-01 2.65613824e-01
-2.38146496e+00 -1.11697340e+00 -1.59665763e-01 1.48964071e+00
7.46418297e-01 -2.72071898e-01 8.85867104e-02 -1.54081836e-01
5.16563177e-01 4.29238796e-01 -2.30504483e-01 -1.08477545e+00
-4.07241046e-01 -8.14728513e-02 -2.29433239e-01 6.27625763e-01
-9.37293887e-01 8.19776833e-01 5.75306034e+00 5.58822155e-01
-1.13465393e+00 -1.86636355e-02 3.93547714e-01 1.96529441e-02
-3.88644725e-01 -1.55333551e-02 -3.70453060e-01 2.47215167e-01
9.18010712e-01 -5.18582202e-02 5.74018419e-01 4.22423869e-01
3.06136906e-01 -1.41862571e-01 -9.10861135e-01 1.17036891e+00
2.23948345e-01 -9.49416816e-01 -1.57482654e-01 -3.91720861e-01
5.04361689e-01 -4.17152673e-01 -2.17359424e-01 5.52237749e-01
2.83476084e-01 -7.52415180e-01 1.83453321e-01 1.02444065e+00
2.03265343e-02 -1.04992592e+00 1.04053569e+00 1.07238419e-01
-1.21930742e+00 1.96695793e-02 1.44328192e-01 -1.33745506e-01
4.32272971e-01 4.23535667e-02 -8.13470483e-01 8.36927533e-01
1.98787525e-01 5.39270878e-01 -3.29965830e-01 5.47974586e-01
-1.55928865e-01 5.87591410e-01 -1.78415507e-01 -3.12988907e-01
3.87063473e-01 -1.99813440e-01 4.35671628e-01 1.71232891e+00
9.43213105e-02 2.77310938e-01 1.57276705e-01 3.71179938e-01
-3.18553388e-01 5.61024904e-01 -5.87495029e-01 -3.03860188e-01
4.92348969e-01 1.70713484e+00 -3.22309405e-01 -1.57025993e-01
-6.37543142e-01 1.13250136e+00 2.91336805e-01 3.80829304e-01
-9.64004159e-01 -5.62417746e-01 6.09688938e-01 -9.15205240e-01
-2.82528460e-01 -1.29516363e-01 -1.27092093e-01 -1.22221065e+00
-3.35711151e-01 -8.01148534e-01 7.52586424e-01 -7.62358069e-01
-1.46800017e+00 9.36093450e-01 -2.47479603e-01 -1.08254278e+00
-5.24638534e-01 -1.86119527e-01 -8.14385951e-01 8.52861762e-01
-1.14322042e+00 -1.56706321e+00 -3.37959439e-01 8.97500992e-01
3.24113041e-01 1.39340162e-02 1.26376855e+00 3.06301773e-01
-7.96951413e-01 5.72733760e-01 -3.43008578e-01 1.39884889e-01
8.42603743e-01 -8.23319554e-01 -3.88536572e-01 4.80508864e-01
-1.30140752e-01 6.05950475e-01 7.22053766e-01 -3.91736716e-01
-1.69310451e+00 -5.33404231e-01 9.04824376e-01 -8.04019123e-02
6.77879810e-01 -2.70035416e-01 -7.93251574e-01 4.80835259e-01
1.14276695e+00 -5.35614967e-01 1.24331164e+00 4.39630419e-01
-3.44860494e-01 6.42611682e-02 -1.14182615e+00 7.32711554e-01
8.23645771e-01 -5.90520382e-01 -5.99965692e-01 -7.83022791e-02
5.86018920e-01 -3.32727224e-01 -1.19059992e+00 2.80830145e-01
7.17518687e-01 -8.66284966e-01 4.93052065e-01 -9.83154833e-01
5.06352425e-01 3.38741913e-02 -2.42637724e-01 -1.44431400e+00
2.28660092e-01 -8.35654140e-01 -1.72415853e-01 1.59635723e+00
4.97222453e-01 -6.09147072e-01 3.17608714e-01 9.22479272e-01
-1.00625493e-01 -5.44509530e-01 -7.55822003e-01 1.21335804e-01
-5.04101276e-01 -5.14910579e-01 6.51568890e-01 1.35991693e+00
1.11261916e+00 8.51039529e-01 -9.35854912e-01 -1.49227589e-01
9.23608765e-02 1.57696664e-01 6.81467295e-01 -9.19734359e-01
-6.69805110e-02 -5.64711154e-01 -1.72769383e-01 -4.85867709e-01
6.60230160e-01 -5.93514919e-01 -8.18897039e-02 -1.71808183e+00
1.98417038e-01 8.43927190e-02 -4.62422103e-01 5.64262927e-01
-1.55391037e-01 4.55387309e-03 1.93027318e-01 -2.48799577e-01
-1.02155423e+00 7.61754990e-01 1.02989292e+00 5.24570197e-02
-5.34638524e-01 -3.25494349e-01 -6.93981290e-01 7.01856136e-01
8.13995361e-01 -8.93296152e-02 -3.99730980e-01 -1.82743534e-01
5.12510836e-01 7.66727388e-01 8.00222754e-02 -4.98902678e-01
5.56987226e-01 -3.41379404e-01 8.45018923e-02 -4.54756945e-01
7.99684167e-01 -8.28085542e-01 5.99933118e-02 -1.61870182e-01
-6.90812945e-01 9.60209146e-02 3.62861812e-01 3.45813632e-01
-6.20981693e-01 9.04287398e-02 1.45154908e-01 1.66663080e-01
-8.20969343e-01 -1.13573037e-01 -8.35146368e-01 -1.99368224e-01
7.22414374e-01 2.79507309e-01 -5.54193735e-01 -9.84817505e-01
-9.39030707e-01 4.80497152e-01 -1.03363596e-01 6.49452090e-01
7.74708986e-01 -1.55277777e+00 -5.82440615e-01 -2.90090919e-01
3.41445774e-01 -8.05448890e-01 7.75704801e-01 1.00221109e+00
-4.49548885e-02 2.20899373e-01 -4.07225847e-01 -2.48937532e-02
-1.42509532e+00 5.73055595e-02 4.13627565e-01 -3.58064890e-01
3.75146121e-02 5.64038873e-01 -2.15051979e-01 -5.55797338e-01
2.71520674e-01 2.93132484e-01 -7.35909343e-01 4.75760579e-01
2.91563034e-01 3.79024923e-01 -1.25400513e-01 -8.30543935e-01
-4.92019236e-01 1.00039557e-01 1.36190906e-01 -5.57614744e-01
1.01544249e+00 -4.33414370e-01 -3.07395220e-01 6.56021476e-01
1.29493284e+00 5.11401296e-02 -5.63652933e-01 -1.95893079e-01
-2.43302479e-01 -1.69399619e-01 7.50939474e-02 -1.06649041e+00
-8.74866486e-01 8.42198730e-01 3.46281767e-01 2.25045651e-01
1.26198196e+00 -2.22112313e-01 9.28700686e-01 5.79155326e-01
9.80239436e-02 -1.21507454e+00 1.91674381e-01 5.29918611e-01
8.34841073e-01 -1.11253488e+00 -4.02422309e-01 -1.73464924e-01
-1.36302698e+00 1.40147448e+00 7.76436508e-01 4.46541548e-01
4.91546780e-01 1.47585869e-01 5.63251436e-01 -1.95829466e-01
-9.63293433e-01 -1.13009080e-01 1.86856776e-01 9.44479555e-02
7.51909852e-01 3.11861075e-02 -6.92343235e-01 1.11574972e+00
4.93260846e-02 -1.36930853e-01 4.21331048e-01 1.05819607e+00
-1.09050415e-01 -1.28781414e+00 -1.91009283e-01 -2.30849087e-02
-5.13218284e-01 8.83535668e-03 -9.14785087e-01 5.38953900e-01
-2.38714367e-01 1.56062305e+00 -2.27959380e-01 -8.89795721e-01
4.13007617e-01 5.50679982e-01 3.09546560e-01 1.48057751e-03
-1.10288763e+00 1.13198198e-01 8.10571253e-01 -4.61512208e-01
-8.98998380e-01 -4.61499751e-01 -1.39183569e+00 -4.47287023e-01
-1.95333362e-01 3.25013191e-01 8.79647553e-01 9.94914293e-01
4.45406616e-01 5.82200348e-01 1.03462362e+00 -6.34933531e-01
1.70591146e-01 -1.20878673e+00 -2.57338762e-01 4.19419050e-01
-9.64084119e-02 -4.36015129e-01 -3.65917563e-01 -2.79762477e-01] | [13.057839393615723, 5.881637096405029] |
07ca1187-37c0-45b8-b458-d10f69ed8c0d | simple-yet-effective-synthetic-dataset | 2303.11660 | null | https://arxiv.org/abs/2303.11660v1 | https://arxiv.org/pdf/2303.11660v1.pdf | Simple Yet Effective Synthetic Dataset Construction for Unsupervised Opinion Summarization | Opinion summarization provides an important solution for summarizing opinions expressed among a large number of reviews. However, generating aspect-specific and general summaries is challenging due to the lack of annotated data. In this work, we propose two simple yet effective unsupervised approaches to generate both aspect-specific and general opinion summaries by training on synthetic datasets constructed with aspect-related review contents. Our first approach, Seed Words Based Leave-One-Out (SW-LOO), identifies aspect-related portions of reviews simply by exact-matching aspect seed words and outperforms existing methods by 3.4 ROUGE-L points on SPACE and 0.5 ROUGE-1 point on OPOSUM+ for aspect-specific opinion summarization. Our second approach, Natural Language Inference Based Leave-One-Out (NLI-LOO) identifies aspect-related sentences utilizing an NLI model in a more general setting without using seed words and outperforms existing approaches by 1.2 ROUGE-L points on SPACE for aspect-specific opinion summarization and remains competitive on other metrics. | ['Yassine Benajiba', 'Miguel Ballesteros', 'Kalpit Dixit', 'Yogarshi Vyas', 'Shuai Wang', 'Jie Ma', 'Ming Shen'] | 2023-03-21 | null | null | null | null | ['unsupervised-opinion-summarization'] | ['natural-language-processing'] | [ 2.99756318e-01 4.89136308e-01 -4.23236489e-01 -3.48943740e-01
-1.50019324e+00 -9.21426594e-01 7.73238301e-01 8.82887542e-01
-2.88054168e-01 1.12613285e+00 6.53935134e-01 -1.89228982e-01
6.42491803e-02 -7.98484683e-01 -5.11555314e-01 -3.47202599e-01
3.85736793e-01 6.16867185e-01 1.85753629e-01 -4.15845037e-01
8.61034334e-01 3.01874112e-02 -1.35357320e+00 3.24847013e-01
1.40007007e+00 4.83961433e-01 -2.13995829e-01 1.02225792e+00
-5.78639209e-01 4.63768572e-01 -1.07821441e+00 -5.34953833e-01
-1.10517673e-01 -6.85537636e-01 -6.74236178e-01 1.66100696e-01
4.80384201e-01 1.08390927e-01 5.67138612e-01 9.80995595e-01
6.26254559e-01 2.37119600e-01 9.28315520e-01 -9.71684277e-01
-7.35082507e-01 6.73952520e-01 -6.64628506e-01 -1.22638837e-01
7.31636882e-01 -2.36733928e-01 1.52092898e+00 -9.66846526e-01
8.84625196e-01 1.32186460e+00 5.08446693e-01 3.43542606e-01
-9.29094553e-01 2.46462934e-02 4.40096796e-01 -4.26193625e-01
-7.84442425e-01 -2.86006570e-01 6.29034638e-01 -1.11681506e-01
1.32835877e+00 4.98406529e-01 7.50043988e-01 4.49725300e-01
5.29980659e-01 1.03688967e+00 8.83870423e-01 -4.66128647e-01
6.95611477e-01 3.10698837e-01 5.90344727e-01 2.37147778e-01
9.44609761e-01 -7.67157376e-01 -6.31791294e-01 -5.43268502e-01
-1.41905025e-02 -1.51869148e-01 -5.56153543e-02 4.87427451e-02
-1.06110370e+00 9.26522255e-01 -2.02139512e-01 1.49670973e-01
-6.34170234e-01 6.60815164e-02 5.34820080e-01 3.27024281e-01
1.12058139e+00 1.03282928e+00 -7.35249341e-01 -1.70127794e-01
-1.29705608e+00 6.00804090e-01 1.20799327e+00 1.06130719e+00
8.17065954e-01 6.42446280e-02 -6.39442444e-01 8.47579122e-01
7.78931752e-02 9.83923435e-01 6.24126673e-01 -8.79904032e-01
4.48270679e-01 8.80946755e-01 2.99357086e-01 -8.18172872e-01
-1.60407051e-01 -5.25645256e-01 -5.11848927e-01 -2.80578732e-01
-3.56400251e-01 -4.75569785e-01 -1.05907083e+00 1.20074201e+00
1.79127753e-01 -6.52808070e-01 5.55575967e-01 2.30356216e-01
1.25390983e+00 9.34869170e-01 -1.98808342e-01 -6.20800614e-01
1.59510970e+00 -1.25755477e+00 -8.32911789e-01 -5.83730757e-01
7.35545397e-01 -9.21661317e-01 1.03996992e+00 3.61805707e-01
-1.19249415e+00 -9.25164521e-02 -8.70522797e-01 3.06432862e-02
-5.01091421e-01 3.69966030e-01 5.47392368e-01 4.89238739e-01
-1.29231775e+00 1.65322140e-01 -2.32459694e-01 -6.21605217e-01
2.21707791e-01 1.81760713e-01 -4.24087882e-01 -4.59326915e-02
-7.99764693e-01 6.26104116e-01 2.85703614e-02 -3.97914678e-01
-2.82777727e-01 -6.70494497e-01 -1.14345479e+00 4.57377434e-02
5.77900231e-01 -9.89897072e-01 1.43379092e+00 -6.07831120e-01
-1.28198719e+00 5.84815204e-01 -9.39972281e-01 -3.54210734e-01
-1.23523856e-02 -4.88981783e-01 -2.20635653e-01 3.43819946e-01
8.04514408e-01 6.09209538e-01 6.48539722e-01 -1.62603617e+00
-5.91142058e-01 -2.16048002e-01 2.30899379e-01 4.29786384e-01
-5.77589571e-02 7.24266171e-02 -3.14518541e-01 -7.00672805e-01
-8.31895173e-02 -7.57949233e-01 -6.93061054e-01 -6.52485311e-01
-6.28060758e-01 -4.26474869e-01 4.50202227e-01 -5.45410931e-01
1.35207427e+00 -1.39067078e+00 -1.67820811e-01 -8.80277380e-02
3.57047506e-02 4.05764878e-01 -5.49667358e-01 8.77433836e-01
2.83309817e-01 4.95369852e-01 -5.06622016e-01 -5.11093259e-01
2.96731200e-03 -7.43580237e-02 -4.28973973e-01 -3.10887277e-01
4.55758244e-01 1.11279643e+00 -1.29402769e+00 -6.21956646e-01
-2.35696584e-01 3.37977484e-02 -5.91503203e-01 -7.11087584e-02
-4.16790426e-01 -7.17601478e-02 -5.43931484e-01 6.46279991e-01
5.04643321e-01 -6.91207685e-03 -3.19245271e-02 -1.04519099e-01
-1.82447433e-01 4.84970152e-01 -7.06644297e-01 1.38272214e+00
-7.00073302e-01 6.36639893e-01 -3.65416765e-01 -6.41328931e-01
1.09576881e+00 3.22473556e-01 2.08527938e-01 -2.93205649e-01
-3.79118547e-02 3.98096949e-01 -5.20465255e-01 -2.93906778e-01
1.35663259e+00 -1.89592347e-01 -5.62178910e-01 1.02138937e+00
3.21927033e-02 -9.26219463e-01 9.37787533e-01 6.76147163e-01
1.11691368e+00 -2.24617288e-01 9.23131824e-01 -2.60528952e-01
6.68421745e-01 2.98138708e-01 3.88378888e-01 1.05697238e+00
4.41301167e-02 1.01597726e+00 8.49587083e-01 -1.41456619e-01
-8.37815642e-01 -7.20824897e-01 3.98958445e-01 7.44703770e-01
-1.09956330e-02 -9.71830845e-01 -7.35379934e-01 -9.83398438e-01
-1.96334302e-01 1.57063437e+00 -5.52394927e-01 -3.62829454e-02
-3.28688025e-01 -7.35181272e-01 1.61333337e-01 3.21287841e-01
1.26936868e-01 -1.22067857e+00 -2.19823763e-01 2.04008386e-01
-3.79771888e-01 -9.21964467e-01 -6.48163438e-01 -3.57160484e-03
-8.90632212e-01 -7.25761712e-01 -9.01054084e-01 -4.08056945e-01
1.07923114e+00 5.85218251e-01 1.56586504e+00 -1.43577948e-01
2.17659622e-01 5.70841789e-01 -7.81658113e-01 -8.59822989e-01
-4.18504596e-01 4.07391757e-01 -1.59087896e-01 -2.87138462e-01
5.72106659e-01 -4.13590997e-01 -4.47179317e-01 -1.84690982e-01
-1.04061139e+00 -3.98279876e-01 7.53712654e-01 5.66187024e-01
6.63329363e-01 -3.05245161e-01 1.00279725e+00 -1.32168281e+00
1.43880594e+00 -3.24845791e-01 -6.74195886e-02 3.65896702e-01
-7.53145456e-01 3.24508369e-01 6.40399456e-01 -1.49855822e-01
-8.36042941e-01 -3.57235879e-01 -7.62412250e-02 5.26689589e-01
4.60673086e-02 6.65105402e-01 -5.16620558e-03 6.01811767e-01
6.32168293e-01 4.47486222e-01 -2.34744594e-01 -4.38084640e-02
6.34418726e-01 6.53949499e-01 9.24008489e-02 -1.93255275e-01
7.12610483e-01 4.96462941e-01 -3.97813231e-01 -1.09448302e+00
-1.49411786e+00 -6.27358437e-01 -4.41446126e-01 -2.27268822e-02
7.72253573e-01 -9.07634974e-01 1.18192032e-01 1.80185646e-01
-1.41089416e+00 2.18175694e-01 -8.51171017e-01 2.06713408e-01
-3.69470268e-01 4.89525408e-01 -1.06948204e-01 -1.06256080e+00
-1.28395224e+00 -8.66173506e-01 1.48604453e+00 4.77106959e-01
-9.12036538e-01 -1.00213635e+00 3.71405184e-01 3.74422759e-01
3.99719536e-01 7.13757724e-02 6.63259685e-01 -1.05149674e+00
-1.44222349e-01 -7.24974334e-01 -2.21276581e-02 5.08779764e-01
3.30170512e-01 6.64738268e-02 -5.51065981e-01 -4.71214429e-02
-1.24243632e-01 -1.42025858e-01 1.02715516e+00 6.35829806e-01
2.51471490e-01 -6.34751976e-01 -1.91498637e-01 -3.20499778e-01
1.18481207e+00 4.20605354e-02 4.23115820e-01 1.74404725e-01
4.58432049e-01 6.60757363e-01 9.86382842e-01 3.71434689e-01
6.27166986e-01 7.12427944e-02 -8.01801216e-03 3.47896218e-02
-1.03799552e-02 -2.16023028e-01 4.60087687e-01 1.24314046e+00
2.19161659e-01 -7.37081110e-01 -3.38790745e-01 1.27605832e+00
-1.82865167e+00 -8.21337283e-01 -2.94686735e-01 1.90916729e+00
8.02790344e-01 3.31668317e-01 1.03419032e-02 5.75928437e-03
5.17014921e-01 7.10977674e-01 -4.44969893e-01 -9.78878081e-01
-2.74980009e-01 1.78031459e-01 2.00664327e-01 7.91148663e-01
-6.86339617e-01 1.04199767e+00 5.90487289e+00 8.28541577e-01
-5.73935986e-01 -1.19338438e-01 4.77154672e-01 -6.68151975e-02
-1.13383853e+00 4.21014041e-01 -1.01270032e+00 8.88577476e-02
7.92583227e-01 -6.90012813e-01 -3.16418022e-01 9.40907478e-01
4.38508838e-01 -6.80905998e-01 -6.84962928e-01 5.04678905e-01
7.89949238e-01 -1.32362854e+00 6.17921650e-01 -1.93591088e-01
1.43563867e+00 -3.43756407e-01 -3.22948068e-01 1.13504939e-01
3.27993155e-01 -5.66800296e-01 4.19623107e-01 5.32937169e-01
6.56513274e-01 -8.06263208e-01 1.28704119e+00 1.92446798e-01
-1.15233481e+00 5.26211083e-01 -4.91128117e-01 -3.12270802e-02
7.13620901e-01 1.19092357e+00 -7.05945432e-01 8.24464321e-01
2.62485266e-01 8.96284699e-01 -5.83840907e-01 7.76070058e-01
-6.37331963e-01 7.01006711e-01 -5.06831296e-02 -6.38112128e-01
4.00810242e-01 -2.86779642e-01 9.97027993e-01 1.29302502e+00
4.80652362e-01 -3.21631357e-02 -2.57190429e-02 2.99994260e-01
-1.27195343e-01 4.02015358e-01 -7.63807178e-01 -2.30100647e-01
3.13678801e-01 1.44009519e+00 -8.16101134e-01 -7.76418149e-01
2.71795373e-02 9.45522785e-01 -1.12266205e-01 2.86031246e-01
-1.63415298e-01 -9.18858588e-01 3.16690385e-01 -1.03135049e-01
5.85917294e-01 3.58899273e-02 -4.52253878e-01 -1.25373149e+00
2.26477429e-01 -1.00882161e+00 -4.36140597e-02 -9.00061429e-01
-1.12001443e+00 8.86434019e-01 7.59821236e-02 -1.30310225e+00
-5.14122963e-01 -2.16850370e-01 -1.11592293e+00 5.66654325e-01
-1.52473783e+00 -9.51019883e-01 -7.30608627e-02 -3.64501089e-01
1.14152873e+00 -1.71623662e-01 8.00576568e-01 -4.03512388e-01
-2.73279339e-01 3.69070619e-01 4.33271453e-02 -4.27397519e-01
7.85151362e-01 -1.54057980e+00 7.97595024e-01 9.00645494e-01
2.06968822e-02 9.04405534e-01 1.06131756e+00 -8.71620595e-01
-9.55861509e-01 -1.09888804e+00 1.91546607e+00 -5.50462067e-01
5.00108540e-01 8.77994392e-03 -5.96670508e-01 3.86381656e-01
6.49411738e-01 -7.53249049e-01 9.93587077e-01 6.66407645e-02
-1.53246611e-01 -1.56513140e-01 -1.00100100e+00 8.55260432e-01
5.47221839e-01 -3.11033398e-01 -9.65556681e-01 5.63423932e-01
1.20706320e+00 1.34854436e-01 -5.81396163e-01 3.75741214e-01
3.70321214e-01 -8.54872942e-01 4.56148386e-01 -5.31209350e-01
8.65537584e-01 -5.43491006e-01 2.21990584e-03 -1.69804752e+00
1.97651714e-01 -7.91390896e-01 1.31684750e-01 1.50844026e+00
9.86216784e-01 -6.74731135e-01 5.46722829e-01 1.67446032e-01
-1.35393783e-01 -9.92376208e-01 -2.83264190e-01 -4.03366834e-01
8.74003619e-02 -2.91597605e-01 5.83793998e-01 2.49560371e-01
1.62845512e-03 9.62119043e-01 -2.04501674e-01 -1.74354374e-01
3.97718668e-01 4.88716930e-01 1.04260492e+00 -1.03873575e+00
1.30994052e-01 -4.27946270e-01 -3.14601287e-02 -1.19181883e+00
1.58693492e-01 -5.37912130e-01 2.15445459e-01 -2.46926880e+00
4.02653217e-01 2.26345420e-01 1.60429075e-01 2.07410902e-01
-7.82893479e-01 2.19253346e-01 1.00104466e-01 1.91574953e-02
-1.19295096e+00 7.70972192e-01 1.20574975e+00 -2.26972327e-01
-2.83603549e-01 1.43894389e-01 -1.45160425e+00 7.62929499e-01
8.12275529e-01 -6.34904802e-01 -5.68237066e-01 -8.56863335e-02
7.01753736e-01 -2.24919945e-01 -3.34578067e-01 -6.18831873e-01
1.44798785e-01 -5.06728515e-02 -1.97384849e-01 -1.35546267e+00
9.42172185e-02 -1.33694410e-01 -4.76886809e-01 2.44852528e-01
-4.96744990e-01 3.63433629e-01 1.03852071e-01 5.71670353e-01
-5.33678651e-01 -7.42961586e-01 8.07287619e-02 -4.72549319e-01
-2.40627110e-01 -1.71274871e-01 -7.60402918e-01 5.47587633e-01
7.01025307e-01 -2.20479637e-01 -5.59840918e-01 -9.83993411e-01
-1.30010426e-01 3.91479164e-01 5.24454594e-01 1.89663768e-01
6.90069973e-01 -9.82861519e-01 -1.01335514e+00 -1.61665052e-01
3.37495118e-01 1.29324719e-01 1.22224316e-01 5.12958944e-01
-5.37202358e-01 6.91271722e-01 4.25677419e-01 -3.48735005e-01
-1.17323673e+00 1.19102478e-01 -2.40444466e-01 -8.69737864e-01
-1.67231113e-01 7.55115271e-01 5.35385422e-02 -9.18504834e-01
-4.71374661e-01 -3.84966075e-01 -5.82121134e-01 5.68686187e-01
6.98385298e-01 3.13025147e-01 5.93132116e-02 -5.26845872e-01
-3.20782214e-01 9.11463857e-01 -3.52138817e-01 -4.66607928e-01
1.33472574e+00 -2.70002574e-01 -6.40547574e-01 7.01722443e-01
9.85078931e-01 6.90010130e-01 -5.86617231e-01 1.72319915e-02
1.81834340e-01 -4.77824360e-03 -3.12765330e-01 -7.22403347e-01
-5.51766157e-01 4.09686685e-01 -4.16742712e-01 5.30104578e-01
9.90462959e-01 1.59607023e-01 8.89526486e-01 6.63805246e-01
-2.62595657e-02 -1.09577239e+00 1.25438675e-01 7.08350420e-01
1.03508353e+00 -1.20414996e+00 5.89970350e-01 -4.97635096e-01
-1.04647398e+00 7.42478788e-01 2.98251092e-01 -2.65226185e-01
3.27089995e-01 -2.27195919e-01 3.32406431e-01 -3.87726068e-01
-1.10697591e+00 -2.41281331e-01 5.10321319e-01 5.61330378e-01
6.47971988e-01 2.64206454e-02 -1.02841234e+00 6.01095259e-01
-5.53329229e-01 -3.80485505e-01 9.36818123e-01 1.03381729e+00
-6.79667413e-01 -1.17013848e+00 -2.93968841e-02 9.12846625e-01
-4.84572649e-01 -5.07742882e-01 -8.54225457e-01 4.84072268e-01
-4.76072162e-01 1.49843645e+00 -1.71283811e-01 5.26703894e-02
4.79242891e-01 4.07587886e-02 -1.68088786e-02 -1.16491902e+00
-8.11197102e-01 9.33067948e-02 6.73954487e-01 -1.78346738e-01
-6.48757339e-01 -7.23412991e-01 -1.08010828e+00 1.12238653e-01
-5.65807819e-01 7.49157965e-01 6.47534013e-01 1.01820374e+00
7.60168374e-01 5.63635111e-01 6.21561587e-01 -6.65306211e-01
-5.19581079e-01 -1.17828107e+00 -4.90221590e-01 1.09333128e-01
3.81049603e-01 6.70936191e-03 -6.32654130e-01 -5.94145525e-03] | [12.431170463562012, 9.358367919921875] |
dfa41940-3ce4-4962-9614-1f33a217bd46 | object-proposal-generation-applying-the | 1704.03706 | null | http://arxiv.org/abs/1704.03706v1 | http://arxiv.org/pdf/1704.03706v1.pdf | Object proposal generation applying the distance dependent Chinese restaurant process | In application domains such as robotics, it is useful to represent the
uncertainty related to the robot's belief about the state of its environment.
Algorithms that only yield a single "best guess" as a result are not
sufficient. In this paper, we propose object proposal generation based on
non-parametric Bayesian inference that allows quantification of the likelihood
of the proposals. We apply Markov chain Monte Carlo to draw samples of image
segmentations via the distance dependent Chinese restaurant process. Our method
achieves state-of-the-art performance on an indoor object discovery data set,
while additionally providing a likelihood term for each proposal. We show that
the likelihood term can effectively be used to rank proposals according to
their quality. | ['Mikko Lauri', 'Simone Frintrop'] | 2017-04-12 | null | null | null | null | ['object-proposal-generation'] | ['computer-vision'] | [-4.59155366e-02 1.07396245e-01 -1.21454068e-01 -6.09148324e-01
-1.16257441e+00 -5.34381926e-01 6.13720357e-01 3.72841090e-01
-7.17862785e-01 6.81403875e-01 -2.27435037e-01 -4.62769955e-01
-6.24939762e-02 -9.33478773e-01 -9.02711332e-01 -5.42690456e-01
7.16340840e-02 1.11065781e+00 6.18992627e-01 2.58550495e-01
4.35349703e-01 5.76206028e-01 -1.25155962e+00 -1.51176527e-01
9.87020552e-01 7.02400088e-01 7.43157387e-01 6.41149521e-01
-1.19129211e-01 1.33534208e-01 -5.68167150e-01 -1.93626136e-01
3.15831006e-02 -1.74606204e-01 -7.65091836e-01 2.29517311e-01
-6.01216871e-03 -4.76649433e-01 2.02522770e-01 1.27463114e+00
2.73540560e-02 2.74661422e-01 9.68624055e-01 -8.52197886e-01
2.91985050e-02 5.21026909e-01 -2.97665417e-01 -9.80826691e-02
2.87193120e-01 1.83876425e-01 8.67370069e-01 -7.31676877e-01
5.20205855e-01 1.40753305e+00 4.05593812e-01 1.68313950e-01
-1.35285032e+00 -4.05683488e-01 2.69328654e-01 2.99790949e-02
-1.43114746e+00 -1.93816260e-01 4.71511692e-01 -4.10026163e-01
6.13605678e-01 -1.87033862e-01 6.98951721e-01 5.64155102e-01
2.32947662e-01 1.06115484e+00 1.04848599e+00 -3.67563218e-01
8.69632483e-01 -9.12625194e-02 3.07007879e-02 8.09587479e-01
6.16053402e-01 -6.23000897e-02 -3.12802166e-01 -3.08551073e-01
7.51579940e-01 -3.11323076e-01 1.77808776e-01 -6.21102333e-01
-1.03940749e+00 9.56776023e-01 4.96973723e-01 -1.84547409e-01
-5.58181882e-01 5.28158486e-01 -2.17472583e-01 -5.16706467e-01
2.19999552e-01 5.01797438e-01 -3.17629099e-01 -1.68714628e-01
-9.15256143e-01 6.13934278e-01 9.41716492e-01 9.52932894e-01
8.77087831e-01 -3.43059599e-01 1.18408769e-01 5.66142321e-01
1.04515529e+00 6.23310268e-01 -6.22987226e-02 -1.36684346e+00
-1.22325524e-04 -2.60449182e-02 8.38987648e-01 -4.16709095e-01
-1.42217070e-01 -1.72574759e-01 -1.76619068e-02 3.77604246e-01
6.05631649e-01 -1.49655670e-01 -1.29767942e+00 1.39877641e+00
4.33017403e-01 1.38890952e-01 2.01969258e-02 7.96828270e-01
2.67046958e-01 5.73275387e-01 7.84693211e-02 -6.62262887e-02
1.19033337e+00 -5.75285017e-01 -5.02771437e-01 -4.27548885e-01
2.37692490e-01 -8.86782348e-01 5.49497306e-01 6.23654425e-01
-7.92145610e-01 -3.86425227e-01 -8.64185393e-01 3.55998933e-01
2.80819237e-02 7.91572407e-02 8.19538474e-01 7.34611690e-01
-6.83722138e-01 5.91382623e-01 -1.34079945e+00 -3.28068167e-01
3.35901171e-01 3.48412514e-01 8.50411505e-02 -2.99771577e-01
-4.61830348e-01 1.01782060e+00 6.23597264e-01 2.05357611e-01
-1.17478311e+00 7.58148581e-02 -7.19039798e-01 -2.22775385e-01
4.77813184e-01 -5.19339025e-01 1.72628415e+00 -1.49417400e-01
-1.41082335e+00 3.91314775e-01 -3.23075265e-01 -6.11991882e-01
5.49256325e-01 -3.04536849e-01 1.48113161e-01 1.50763378e-01
4.03149575e-01 1.06564999e+00 4.98943746e-01 -1.64592803e+00
-8.00917089e-01 -2.37228632e-01 -9.95606706e-02 2.34453335e-01
7.02867866e-01 -1.74733683e-01 -7.51622915e-01 -8.34217519e-02
8.14840913e-01 -1.16209495e+00 -9.39558864e-01 -2.54130252e-02
-4.67420340e-01 -3.18329334e-01 4.43958521e-01 -3.75385344e-01
5.40977895e-01 -1.91463506e+00 -3.03126097e-01 3.52136433e-01
-3.25942457e-01 -1.31726712e-01 2.10616365e-01 1.14108205e-01
8.20482254e-01 9.82413664e-02 -3.60843897e-01 -4.00875568e-01
1.43420845e-01 5.68150818e-01 -1.41026378e-01 6.54937506e-01
3.16827774e-01 5.18587768e-01 -1.16915381e+00 -6.77506089e-01
3.90491754e-01 1.80361047e-01 -5.45804620e-01 1.75361872e-01
-5.49667239e-01 3.64188343e-01 -8.61390114e-01 4.85076368e-01
6.76536798e-01 -1.35611683e-01 2.57223159e-01 3.85964476e-02
-1.40014812e-01 4.30919439e-01 -1.50589144e+00 1.72551072e+00
-1.00801229e-01 1.70457676e-01 8.43890384e-02 -4.80348080e-01
9.14163053e-01 -1.18491895e-01 1.35873154e-01 2.61462317e-03
1.99990347e-01 3.57353628e-01 1.65656343e-01 -2.25737691e-01
8.53833377e-01 -3.31641287e-01 -3.07006568e-01 3.45984370e-01
-4.05939715e-03 -8.79435241e-01 2.58556038e-01 1.11467294e-01
1.04761016e+00 5.66582143e-01 4.09019589e-01 -1.75226256e-01
-2.34518558e-01 2.19581798e-01 4.87009943e-01 1.14303648e+00
-3.10736746e-01 5.98694623e-01 2.64087051e-01 -4.37900499e-02
-1.01369619e+00 -1.44226205e+00 -2.47051135e-01 3.80595833e-01
4.54100102e-01 1.30842045e-01 -7.53379524e-01 -6.66119456e-01
-4.09459919e-02 1.13726461e+00 -2.00583041e-01 1.53790951e-01
-3.53789449e-01 -9.35744822e-01 8.32940117e-02 4.33990538e-01
3.46069217e-01 -9.46715653e-01 -9.73630428e-01 5.71359694e-01
-2.98046529e-01 -1.04308581e+00 1.66288137e-01 5.14426529e-01
-1.06261861e+00 -8.67586136e-01 -4.07797307e-01 -3.19028169e-01
8.48142982e-01 2.64115334e-01 8.68484855e-01 -2.84795582e-01
-3.95196602e-02 3.41916054e-01 -3.69047582e-01 -4.05473620e-01
-6.00046396e-01 -8.21668804e-02 5.31153157e-02 -3.90977085e-01
1.89754158e-01 -2.00389996e-01 -4.46979463e-01 3.36251616e-01
-5.52726328e-01 -1.27694920e-01 8.43478680e-01 5.06576478e-01
8.77998590e-01 2.71471977e-01 1.59376383e-01 -5.86071968e-01
3.94392580e-01 -3.84268254e-01 -8.87919486e-01 2.86494810e-02
-3.96764189e-01 4.28279489e-01 -5.18233851e-02 -3.86265367e-01
-1.36417210e+00 5.31294107e-01 -2.18746990e-01 -5.52251898e-02
-5.61968744e-01 4.52051461e-01 -8.56172964e-02 1.10039979e-01
3.68824214e-01 -1.75424084e-01 -2.60679752e-01 -6.71469510e-01
4.76500183e-01 4.94441062e-01 6.77861512e-01 -9.75684047e-01
4.17311341e-01 5.38452506e-01 3.14750522e-02 -6.89506948e-01
-7.78582275e-01 -6.71920240e-01 -6.45960331e-01 -2.65895218e-01
9.39393580e-01 -7.09560692e-01 -6.89108670e-01 3.15734327e-01
-1.33197665e+00 -4.29153770e-01 -1.50939345e-01 1.00331783e+00
-8.23922038e-01 2.57421732e-01 -3.92929345e-01 -1.40818024e+00
2.94423819e-01 -1.32979202e+00 1.09145021e+00 3.57599288e-01
-1.80539474e-01 -5.27234018e-01 -7.25416020e-02 2.26502433e-01
-1.15321577e-01 -1.01129655e-02 5.08957446e-01 -6.12413585e-01
-1.11227930e+00 -3.45011920e-01 -2.03236625e-01 2.88562980e-02
-9.51673687e-02 1.94252118e-01 -8.59001160e-01 2.27627512e-02
-1.50005102e-01 -1.42556846e-01 8.87986958e-01 7.99868405e-01
7.04178691e-01 4.16683294e-02 -5.56195974e-01 -5.02093695e-02
1.42062020e+00 3.63784343e-01 6.03264153e-01 2.63587356e-01
3.07038158e-01 6.61457360e-01 1.18555081e+00 5.33863246e-01
3.99662435e-01 5.48695624e-01 6.70607686e-01 3.92227679e-01
2.60470748e-01 -3.66255164e-01 -3.28469276e-02 2.53202200e-01
1.98180318e-01 -4.44038957e-01 -1.10094380e+00 8.04925621e-01
-1.91864896e+00 -6.35926187e-01 -2.58156747e-01 2.26751399e+00
4.94419903e-01 6.42581105e-01 -3.69519182e-02 -1.72983900e-01
7.50414073e-01 -2.73036599e-01 -5.35678148e-01 -5.84907345e-02
4.13798213e-01 1.98839717e-02 9.15039301e-01 7.20203698e-01
-1.03526962e+00 1.02351248e+00 7.00756216e+00 6.85239255e-01
-4.29690868e-01 -3.00719030e-03 4.60690200e-01 4.75361884e-01
-2.83990294e-01 4.40233141e-01 -1.05529833e+00 3.05812836e-01
7.91688025e-01 3.16723526e-01 1.48829356e-01 9.41678226e-01
2.94394255e-01 -1.03300643e+00 -8.20065916e-01 4.89898562e-01
-3.38330328e-01 -9.79316175e-01 -1.85541883e-01 3.30038697e-01
6.46524012e-01 2.21837938e-01 -1.39518157e-01 2.82715466e-02
1.16100967e+00 -6.59845412e-01 1.15056443e+00 6.25090718e-01
1.08646631e-01 -8.03622901e-01 7.87490964e-01 6.62441194e-01
-9.12143409e-01 2.90039834e-02 -2.15878442e-01 -2.93404181e-02
6.19194090e-01 6.78299904e-01 -1.78374207e+00 3.10788929e-01
6.83455050e-01 -7.49792755e-02 -3.10786009e-01 1.49416625e+00
-5.14422476e-01 7.11273193e-01 -8.97799611e-01 -4.27524745e-01
3.82617474e-01 -3.06769013e-01 4.20999408e-01 1.02603090e+00
3.93054426e-01 1.76150426e-01 6.69265449e-01 9.77778137e-01
3.32651556e-01 -1.86149806e-01 -2.55666912e-01 1.66595265e-01
8.21691573e-01 8.64026248e-01 -1.51532304e+00 -3.27678740e-01
8.50986317e-02 7.28521645e-01 1.94605112e-01 1.87845036e-01
-5.23837209e-01 -1.20024718e-01 3.05238903e-01 -2.85597265e-01
6.27416909e-01 -7.42434859e-01 -3.71726304e-01 -6.77933931e-01
-2.09028080e-01 -3.93551439e-01 -8.48918185e-02 -9.26040113e-01
-9.96005177e-01 1.89845011e-01 4.71939355e-01 -8.36247444e-01
-6.59996450e-01 -6.42279625e-01 -2.24761888e-01 8.38804424e-01
-1.19451678e+00 -8.46296847e-01 1.82570622e-03 -2.89213806e-01
5.86042702e-01 4.17508900e-01 7.07033694e-01 -2.45979577e-01
-8.71371850e-02 -2.99436092e-01 7.57958069e-02 3.02555393e-02
3.40531468e-01 -1.37821078e+00 5.27390659e-01 9.74253356e-01
1.50642335e-01 7.27909386e-01 1.32729483e+00 -1.07810593e+00
-1.05634558e+00 -6.87923372e-01 4.25847530e-01 -3.67969126e-01
4.35373992e-01 -9.41944588e-03 -7.03822672e-01 5.79908609e-01
-4.24119771e-01 1.14543678e-03 2.87018925e-01 7.92593211e-02
8.85868967e-02 1.92574039e-01 -1.34862268e+00 5.21148443e-01
5.57144821e-01 -1.50136352e-01 -6.17083371e-01 4.13042679e-02
6.09910131e-01 -3.50538880e-01 -6.32212639e-01 4.55836058e-01
5.77588260e-01 -7.68717229e-01 7.30118334e-01 -2.27716565e-02
-2.17003599e-01 -7.97793329e-01 -5.31142175e-01 -1.11165893e+00
-1.70365050e-01 -3.53283674e-01 1.86680675e-01 9.07679796e-01
5.10001183e-01 -1.89055622e-01 1.03101361e+00 6.24603271e-01
-2.49249607e-01 -3.44227791e-01 -1.10597646e+00 -7.26016521e-01
-2.72723228e-01 -9.09560919e-01 5.23339510e-01 2.45162666e-01
-5.59263349e-01 1.56952381e-01 5.33601120e-02 5.62198043e-01
1.01210153e+00 1.20257959e-01 8.70286047e-01 -1.32640660e+00
-3.35492224e-01 -1.89623222e-01 -2.87026048e-01 -1.48468161e+00
-1.90941691e-02 -2.40313306e-01 1.15104330e+00 -1.97371101e+00
2.94260740e-01 -8.40814948e-01 8.44181105e-02 8.59576315e-02
-5.48865423e-02 9.73051265e-02 5.63473627e-02 1.17250174e-01
-8.88191640e-01 3.65412593e-01 1.01231468e+00 1.13577940e-01
-1.81638509e-01 4.04915959e-01 -2.05041692e-01 1.08851600e+00
7.39867747e-01 -6.87037885e-01 -5.15867881e-02 -2.49715537e-01
7.35505968e-02 6.46488741e-02 3.82847428e-01 -1.04480493e+00
1.34023443e-01 -4.18939084e-01 2.21899137e-01 -1.07302451e+00
6.51775837e-01 -6.07792735e-01 1.44938156e-01 5.94064295e-01
-1.43446267e-01 -3.42747837e-01 -9.65264142e-02 1.01460373e+00
2.19468236e-01 -9.76002038e-01 7.68338680e-01 -4.61740524e-01
-7.98563242e-01 -1.95369627e-02 -6.74682796e-01 -3.23858559e-01
6.68879032e-01 -1.78845860e-02 1.95154324e-01 -2.72149205e-01
-6.85682416e-01 1.60675317e-01 6.51206017e-01 -1.08862147e-01
6.41455710e-01 -8.80838633e-01 -4.89119738e-01 -1.97039261e-01
-2.51451712e-02 4.42248762e-01 -2.27148026e-01 3.56291622e-01
-6.57535017e-01 3.33473563e-01 1.31967381e-01 -8.17481458e-01
-8.63159895e-01 3.23415667e-01 1.74364090e-01 -2.01456577e-01
-1.97816297e-01 8.19003403e-01 -1.05493836e-01 -5.42467415e-01
1.40516505e-01 -5.64647734e-01 1.34428903e-01 -2.45126128e-01
1.77654609e-01 4.02995199e-01 -2.85144359e-01 -5.20207882e-01
-2.43414924e-01 2.61716753e-01 1.21350154e-01 -9.34579253e-01
9.90778863e-01 -3.16094398e-01 1.28747895e-01 7.05179751e-01
5.30179739e-01 -1.40684918e-01 -1.74279869e+00 -1.09047271e-01
2.75488734e-01 -5.54833651e-01 8.34607556e-02 -7.74609029e-01
-2.22313821e-01 7.71056890e-01 5.82330108e-01 5.73539995e-02
4.74015743e-01 3.19817364e-01 4.28346157e-01 6.65695488e-01
9.41658318e-01 -1.11038530e+00 2.65991176e-03 3.84973675e-01
2.72581339e-01 -1.44701612e+00 2.56681383e-01 -4.77888077e-01
-4.64747697e-01 9.03763354e-01 2.91965991e-01 -1.59302831e-01
5.98682404e-01 2.42255762e-01 -1.28866211e-01 -1.91605766e-03
-3.78818929e-01 -5.18015027e-01 1.73255324e-01 6.04118049e-01
1.27128595e-02 4.79070693e-01 -1.61945865e-01 2.91651905e-01
-1.88043267e-01 -1.26227140e-01 5.77153683e-01 1.07611990e+00
-1.15571189e+00 -9.53711271e-01 -7.25390494e-01 3.21219116e-01
-4.75439906e-01 1.27134621e-01 4.84298654e-02 5.78003824e-01
2.00623214e-01 1.20560145e+00 6.73878640e-02 1.34335235e-01
5.94279803e-02 -1.78539485e-01 6.57418013e-01 -7.98092365e-01
2.40494952e-01 5.66157699e-01 4.09737438e-01 -2.35032097e-01
-4.67640072e-01 -1.11539042e+00 -1.53609705e+00 1.30651653e-01
-6.55794680e-01 3.31440955e-01 1.22155094e+00 1.11127031e+00
-1.21440269e-01 1.96472913e-01 1.82636499e-01 -1.13471174e+00
-6.37696147e-01 -9.80978191e-01 -5.10030508e-01 -1.01574048e-01
-3.85463685e-02 -8.03093731e-01 -1.39341012e-01 -8.32521245e-02] | [7.230166912078857, -1.078290581703186] |
8136ee75-a5f3-4f34-b05e-679f0d0a3493 | unsupervised-few-shot-learning-via | 2004.05805 | null | https://arxiv.org/abs/2004.05805v2 | https://arxiv.org/pdf/2004.05805v2.pdf | Diversity Helps: Unsupervised Few-shot Learning via Distribution Shift-based Data Augmentation | Few-shot learning aims to learn a new concept when only a few training examples are available, which has been extensively explored in recent years. However, most of the current works heavily rely on a large-scale labeled auxiliary set to train their models in an episodic-training paradigm. Such a kind of supervised setting basically limits the widespread use of few-shot learning algorithms. Instead, in this paper, we develop a novel framework called Unsupervised Few-shot Learning via Distribution Shift-based Data Augmentation (ULDA), which pays attention to the distribution diversity inside each constructed pretext few-shot task when using data augmentation. Importantly, we highlight the value and importance of the distribution diversity in the augmentation-based pretext few-shot tasks, which can effectively alleviate the overfitting problem and make the few-shot model learn more robust feature representations. In ULDA, we systemically investigate the effects of different augmentation techniques and propose to strengthen the distribution diversity (or difference) between the query set and support set in each few-shot task, by augmenting these two sets diversely (i.e., distribution shifting). In this way, even incorporated with simple augmentation techniques (e.g., random crop, color jittering, or rotation), our ULDA can produce a significant improvement. In the experiments, few-shot models learned by ULDA can achieve superior generalization performance and obtain state-of-the-art results in a variety of established few-shot learning tasks on Omniglot and miniImageNet. The source code is available in https://github.com/WonderSeven/ULDA. | ['Tiexin Qin', 'Yang Gao', 'Wenbin Li', 'Yinghuan Shi'] | 2020-04-13 | null | null | null | null | ['unsupervised-few-shot-learning', 'unsupervised-few-shot-image-classification'] | ['computer-vision', 'computer-vision'] | [ 1.35963336e-01 -1.99861363e-01 -4.39331114e-01 -2.28469506e-01
-4.77432847e-01 -4.68945056e-02 7.87416875e-01 7.13577569e-02
-5.11031985e-01 5.06674230e-01 1.38642967e-01 1.32457048e-01
-1.41975302e-02 -9.90482569e-01 -5.69562614e-01 -8.98873031e-01
2.50729471e-01 1.73519969e-01 4.82549310e-01 -4.60011870e-01
2.04192381e-02 1.44326106e-01 -1.77118468e+00 -5.22180572e-02
8.65103364e-01 7.97458768e-01 3.47867846e-01 1.62801549e-01
-3.19706410e-01 6.39110446e-01 -5.13713539e-01 -1.71389624e-01
8.81250426e-02 -6.02753818e-01 -4.77822304e-01 2.12924823e-01
2.32809875e-02 -3.38675320e-01 -5.04274011e-01 9.89344835e-01
6.57726765e-01 5.97677112e-01 5.56549370e-01 -1.29733658e+00
-7.07314730e-01 3.86566997e-01 -7.17971742e-01 4.51246113e-01
-1.88131537e-02 2.57382363e-01 8.35746884e-01 -1.18808365e+00
6.88822806e-01 8.59259069e-01 4.96462435e-01 6.31302714e-01
-1.09111643e+00 -7.19560623e-01 1.82966352e-01 5.10130107e-01
-1.43524134e+00 -4.23696041e-01 9.62671101e-01 -2.48346671e-01
6.25919819e-01 -2.93134525e-02 6.70129418e-01 1.18808198e+00
-1.90449432e-01 9.55947518e-01 7.99900651e-01 -5.92000782e-01
4.94758755e-01 7.20444769e-02 2.31851205e-01 5.59484661e-01
1.05374545e-01 2.33529266e-02 -5.93885839e-01 -5.70387840e-02
6.60347104e-01 7.35696375e-01 -2.38179862e-01 -7.43710339e-01
-9.74931121e-01 1.03320038e+00 5.08607805e-01 5.83733678e-01
-3.20925057e-01 -1.03366315e-01 5.08456528e-01 2.37592861e-01
5.77731669e-01 3.68785232e-01 -1.90409020e-01 -1.45241255e-02
-7.42570698e-01 2.87247058e-02 4.48639005e-01 1.00957847e+00
1.07355404e+00 2.64084071e-01 -4.33498889e-01 1.11093378e+00
-2.97404304e-02 2.32199892e-01 9.87598121e-01 -3.87235671e-01
1.46383926e-01 5.34924269e-01 -1.87381864e-01 -6.75102532e-01
-2.96862274e-01 -4.87199157e-01 -8.75232399e-01 -1.39152363e-01
5.16012833e-02 -2.74042487e-01 -1.21782398e+00 1.82686353e+00
4.00039226e-01 6.77007198e-01 -2.35748850e-02 8.51742864e-01
9.73734677e-01 5.12074113e-01 1.73564062e-01 -4.00100708e-01
1.22546828e+00 -1.14249098e+00 -8.55113208e-01 -2.34747574e-01
8.13335538e-01 -5.24895012e-01 1.54825377e+00 3.35319117e-02
-5.18675268e-01 -6.31975591e-01 -1.21599722e+00 2.29305297e-01
-5.51260650e-01 -2.84877598e-01 6.70682788e-01 5.07528901e-01
-5.08570254e-01 5.43713212e-01 -7.44878173e-01 -5.35177469e-01
6.44125998e-01 -5.63279502e-02 -2.16880575e-01 -5.84896386e-01
-1.42292833e+00 6.53367579e-01 4.01596487e-01 -5.11973083e-01
-8.23951066e-01 -8.36330831e-01 -1.06966341e+00 2.03090712e-01
8.31831515e-01 -4.99924809e-01 1.17486775e+00 -4.80322331e-01
-1.34037626e+00 6.06343269e-01 -9.34416614e-03 -3.95988703e-01
1.38054267e-01 -2.00924948e-01 -3.50333005e-01 9.32821110e-02
2.14806348e-02 6.29518092e-01 1.05461717e+00 -1.00178683e+00
-4.06863600e-01 -3.17113340e-01 -1.08719386e-01 2.41625547e-01
-9.54422712e-01 -2.36663312e-01 -6.08427942e-01 -9.28432167e-01
-8.74379873e-02 -7.69272327e-01 -3.07377309e-01 -3.91368009e-02
-1.72781218e-02 -2.78021604e-01 8.99498641e-01 -2.82296985e-02
1.30468702e+00 -2.43625569e+00 -8.44687410e-03 -1.35413900e-01
2.50506788e-01 6.98485196e-01 -3.63657713e-01 4.82524872e-01
-3.46488766e-02 -7.10484013e-02 -3.69562000e-01 -3.18824589e-01
-7.57037103e-02 3.35861146e-01 -3.16148639e-01 3.53924721e-01
1.32523417e-01 1.05311608e+00 -1.09108961e+00 -3.47211599e-01
4.80229527e-01 5.08683920e-01 -4.26601261e-01 1.29475355e-01
-2.03023240e-01 2.12948561e-01 -4.85993594e-01 6.26323402e-01
5.45671642e-01 -3.06559861e-01 -1.52647823e-01 9.02397111e-02
4.46832106e-02 -2.61284322e-01 -1.06610513e+00 2.08358908e+00
-4.50503051e-01 3.94028157e-01 -6.22175753e-01 -1.06540954e+00
9.84382510e-01 1.16240941e-01 3.94054741e-01 -9.16666627e-01
2.61528134e-01 -5.19484542e-02 2.49288045e-02 -4.29420143e-01
3.08209389e-01 -4.70248014e-01 2.12024413e-02 5.18913269e-01
4.97980326e-01 9.78316180e-04 2.43762746e-01 2.38939807e-01
1.01423275e+00 -3.06975562e-02 6.06420815e-01 1.13955691e-01
1.27344102e-01 -1.17301159e-01 6.44984722e-01 7.52427876e-01
-3.58857840e-01 7.59365201e-01 1.05299182e-01 -2.14144334e-01
-9.92481470e-01 -7.81365097e-01 -1.99109986e-01 1.56551754e+00
2.73106188e-01 -5.36337852e-01 -4.71693277e-01 -5.89178741e-01
-2.47699469e-02 8.64443421e-01 -8.01106632e-01 -7.48091161e-01
-2.23591030e-01 -1.00964916e+00 3.31473112e-01 6.56798959e-01
6.39043987e-01 -1.04635823e+00 -6.28390729e-01 1.73991889e-01
4.10990864e-02 -8.63565564e-01 -3.88168007e-01 3.73858362e-01
-8.07999551e-01 -8.88033509e-01 -1.14123309e+00 -6.48680866e-01
4.83269095e-01 8.15813661e-01 6.82925224e-01 3.89607400e-02
-4.25418437e-01 3.36180478e-01 -9.13936377e-01 -5.32820880e-01
2.60962665e-01 1.05755545e-01 1.00509919e-01 8.37933496e-02
6.98496699e-01 -8.29860270e-01 -5.24594545e-01 3.78617615e-01
-1.12790751e+00 -7.02617839e-02 6.76670671e-01 1.13229966e+00
7.00895309e-01 -8.46256241e-02 8.09628606e-01 -1.02008271e+00
6.35216475e-01 -7.54438400e-01 -8.15737396e-02 1.56957805e-01
-6.16937280e-01 1.01705687e-02 6.21069849e-01 -8.32692921e-01
-1.01549196e+00 -2.75941193e-01 -1.82314813e-02 -8.78402710e-01
-1.94511414e-01 6.68043315e-01 -1.40471280e-01 -6.48309290e-02
8.74095321e-01 4.87780243e-01 2.01799907e-02 -5.40060401e-01
6.15787804e-01 5.59292316e-01 2.29421243e-01 -2.73772448e-01
8.64726961e-01 5.00861824e-01 -2.77243435e-01 -9.87828612e-01
-1.03638196e+00 -7.83934355e-01 -6.47671282e-01 -1.49384756e-02
4.47509408e-01 -8.84279847e-01 2.71453056e-02 5.03851831e-01
-5.69759727e-01 -4.12456363e-01 -6.81180120e-01 5.16072690e-01
-4.68071163e-01 2.74254173e-01 -3.06168467e-01 -5.57341158e-01
-2.81996995e-01 -7.37713933e-01 6.82323933e-01 4.24368531e-01
-6.07038848e-02 -7.78911173e-01 2.54173636e-01 -7.56619722e-02
4.88446534e-01 1.04808278e-01 8.77024114e-01 -9.74006176e-01
-1.75019398e-01 -3.31350982e-01 -1.08887330e-01 1.19610749e-01
3.87951344e-01 -4.38172758e-01 -1.05611002e+00 -3.62979829e-01
1.02500454e-01 -6.15412831e-01 1.15039730e+00 2.24523872e-01
1.21388245e+00 1.49708195e-03 -2.33526453e-01 6.83272302e-01
1.34190989e+00 1.26774490e-01 6.57520533e-01 3.15363586e-01
6.52334154e-01 3.17843616e-01 9.63007748e-01 8.17026734e-01
1.39815152e-01 5.87096453e-01 2.39272565e-01 3.03480942e-02
-3.19410443e-01 -2.57457942e-01 -9.59721133e-02 6.97697103e-01
-7.74928927e-02 -4.26310040e-02 -8.65492225e-01 5.42591453e-01
-1.96880662e+00 -1.06953788e+00 4.81998801e-01 2.15615559e+00
9.26079452e-01 9.98051092e-02 7.06211999e-02 6.12056851e-02
7.51982093e-01 5.38070381e-01 -8.73904288e-01 2.13719800e-01
-9.60634723e-02 2.40104333e-01 7.11000860e-02 -4.54462282e-02
-1.13858342e+00 1.11384487e+00 5.26185465e+00 1.17099679e+00
-1.14969218e+00 3.30383986e-01 4.38220203e-01 -4.17834789e-01
-1.71589822e-01 -1.77129552e-01 -8.14839363e-01 4.91275430e-01
6.28213346e-01 -3.98300916e-01 2.90337622e-01 1.04455364e+00
-3.75141054e-02 -1.68426950e-02 -8.02905083e-01 1.10644722e+00
3.64535779e-01 -1.39151561e+00 4.77913357e-02 -6.84680045e-02
8.28131616e-01 1.29430085e-01 7.75240958e-02 8.81178916e-01
4.82622869e-02 -6.61762476e-01 2.46255115e-01 3.63479346e-01
8.75625491e-01 -8.25699210e-01 7.44123459e-01 5.42648613e-01
-1.12392008e+00 -2.23142192e-01 -5.92972755e-01 -5.40349558e-02
8.08718130e-02 5.34514666e-01 -4.84842867e-01 4.78245199e-01
5.40491164e-01 7.63020098e-01 -5.23564339e-01 1.32743943e+00
-3.70408148e-01 5.67725420e-01 -1.23455033e-01 -5.66602200e-02
3.03751707e-01 7.68185109e-02 4.58921909e-01 8.22177768e-01
1.99042425e-01 3.26177150e-01 2.40840405e-01 5.55823803e-01
-2.67186224e-01 2.55897939e-01 -7.23320842e-01 -9.47954133e-02
6.39193118e-01 1.25624526e+00 -6.68518126e-01 -5.32401741e-01
-5.99943340e-01 7.96370566e-01 4.13337469e-01 4.42013532e-01
-7.05526233e-01 -6.56581521e-01 4.48293805e-01 1.21675469e-01
5.56941271e-01 -1.73311513e-02 -6.78686723e-02 -1.34611106e+00
-1.12291589e-01 -6.56286418e-01 5.30192852e-01 -6.29020751e-01
-1.31840336e+00 3.97318453e-01 1.75024003e-01 -1.51407814e+00
-2.46422142e-01 -7.80127868e-02 -1.01943636e+00 4.95619804e-01
-1.59129715e+00 -1.03412044e+00 -5.59269071e-01 6.84209824e-01
8.28503251e-01 -4.18317586e-01 8.34802091e-01 3.45414251e-01
-6.80932224e-01 6.90221071e-01 2.17808828e-01 -2.27924995e-03
9.40367997e-01 -8.54356349e-01 1.93510443e-01 7.41498530e-01
3.62483650e-01 7.34614730e-01 5.67170799e-01 -5.48347652e-01
-1.16751862e+00 -1.17934191e+00 3.66762966e-01 7.64720291e-02
7.04540968e-01 -3.85416389e-01 -1.29524589e+00 5.72709441e-01
2.89398525e-02 4.77450818e-01 1.03074050e+00 2.70060718e-01
-4.24169689e-01 -1.54226124e-02 -9.90558684e-01 6.22502744e-01
1.09720862e+00 -3.93050820e-01 -7.31555820e-01 2.17994094e-01
9.59556460e-01 -4.73501720e-02 -5.88481605e-01 4.43117887e-01
3.48317474e-01 -6.89955533e-01 8.77116024e-01 -7.31297195e-01
3.22971910e-01 -1.70226187e-01 -2.39633083e-01 -1.51628315e+00
-3.82772475e-01 -4.34245020e-01 -5.84125102e-01 1.16916835e+00
-2.60436796e-02 -5.79314828e-01 7.33058333e-01 2.96933949e-01
-2.72332937e-01 -9.93245482e-01 -7.06375718e-01 -9.88243937e-01
-1.41766975e-02 -1.71635807e-01 6.39299273e-01 1.20214081e+00
7.88517222e-02 4.22135681e-01 -5.60603678e-01 -2.85124272e-01
4.93993372e-01 1.34802550e-01 8.47693563e-01 -1.24647844e+00
-1.77177504e-01 -2.01070756e-01 -4.38644975e-01 -7.57493377e-01
2.75029764e-02 -8.27093661e-01 4.57679518e-02 -1.37271011e+00
4.42909986e-01 -2.71224231e-01 -6.48345590e-01 7.56158173e-01
-4.53404427e-01 2.18655229e-01 3.33341420e-01 3.79797667e-01
-8.38115215e-01 1.23169672e+00 1.21980608e+00 -8.31513777e-02
-4.45781648e-01 -1.00773476e-01 -8.54800105e-01 7.61461079e-01
8.60136271e-01 -4.93795931e-01 -7.34978259e-01 -1.25858024e-01
-2.62272209e-01 -3.59844118e-01 2.37222850e-01 -1.07704139e+00
3.28680068e-01 -1.69018969e-01 4.17538881e-01 -3.75437021e-01
3.69271725e-01 -4.83752459e-01 -4.39515918e-01 3.94928247e-01
-1.57116354e-01 -4.98050690e-01 2.15896696e-01 8.26882958e-01
-2.99373776e-01 -3.68625075e-01 8.88700485e-01 -1.52868867e-01
-1.40325010e+00 6.78544939e-01 -5.03118597e-02 1.95565760e-01
1.27743065e+00 -1.82584733e-01 -3.65596861e-01 -2.31330365e-01
-7.42476702e-01 2.41558954e-01 4.44191635e-01 5.08581817e-01
8.19850862e-01 -1.54623711e+00 -3.73800427e-01 3.37263882e-01
7.19394445e-01 -1.11322910e-01 6.91269040e-01 9.02563989e-01
1.85600176e-01 1.16614431e-01 -3.97596061e-01 -3.47034723e-01
-1.02232456e+00 7.44540870e-01 -9.38490480e-02 -6.65341318e-02
-9.16038036e-01 7.89145589e-01 2.76748508e-01 -3.08580548e-01
2.83311993e-01 1.72375917e-01 -4.01034027e-01 3.35189402e-01
9.14150715e-01 2.15990096e-01 -9.32411775e-02 -2.51291215e-01
-1.55729756e-01 2.69018859e-01 -4.07580227e-01 1.10506587e-01
1.55670750e+00 -4.59401235e-02 4.58814830e-01 6.50461674e-01
1.05851078e+00 -4.38166559e-01 -1.32656705e+00 -7.44587362e-01
-1.91837743e-01 -6.01414680e-01 1.44648328e-01 -4.60918099e-01
-9.74335432e-01 1.05564129e+00 6.24687672e-01 -3.79380099e-02
1.03741026e+00 9.52356979e-02 9.12855566e-01 5.63189387e-01
4.24709827e-01 -1.14221430e+00 4.47396338e-01 4.90791589e-01
6.20053649e-01 -1.46877289e+00 4.24961001e-03 -1.61035746e-01
-8.43711019e-01 7.31041431e-01 7.47320712e-01 -9.99390036e-02
7.51162529e-01 -5.91658987e-02 -2.01506853e-01 -1.77450344e-01
-6.77132130e-01 -6.71691358e-01 1.72294408e-01 6.98977470e-01
1.88132763e-01 -1.36452883e-01 -2.22778693e-01 8.27840686e-01
1.83109567e-01 1.27262399e-01 3.93602580e-01 1.25761271e+00
-7.96269417e-01 -9.50238705e-01 -3.72628905e-02 6.43400729e-01
3.36128846e-02 -2.46200100e-01 -8.80608559e-02 8.79489601e-01
3.70753929e-02 7.23335743e-01 4.00069840e-02 -4.38493788e-01
3.06441277e-01 1.97762683e-01 3.13407540e-01 -1.09068072e+00
-3.43003795e-02 1.81490377e-01 -3.21427137e-01 -3.05759043e-01
-2.81899154e-01 -4.27258044e-01 -1.26233733e+00 -1.74169704e-01
-4.20708060e-01 1.94146577e-02 3.01728010e-01 1.11801004e+00
3.95729393e-01 6.39733911e-01 6.37363553e-01 -8.12083423e-01
-6.52738571e-01 -1.24608707e+00 -8.32525969e-01 5.16028583e-01
2.93575563e-02 -1.18281329e+00 -2.90987760e-01 -1.80094734e-01] | [9.99714183807373, 3.0037131309509277] |
1def2b99-4a6f-4d2c-99d2-79bec803ec73 | multi-scale-progressive-fusion-learning-for | 2011.11865 | null | https://arxiv.org/abs/2011.11865v1 | https://arxiv.org/pdf/2011.11865v1.pdf | Multi-Scale Progressive Fusion Learning for Depth Map Super-Resolution | Limited by the cost and technology, the resolution of depth map collected by depth camera is often lower than that of its associated RGB camera. Although there have been many researches on RGB image super-resolution (SR), a major problem with depth map super-resolution is that there will be obvious jagged edges and excessive loss of details. To tackle these difficulties, in this work, we propose a multi-scale progressive fusion network for depth map SR, which possess an asymptotic structure to integrate hierarchical features in different domains. Given a low-resolution (LR) depth map and its associated high-resolution (HR) color image, We utilize two different branches to achieve multi-scale feature learning. Next, we propose a step-wise fusion strategy to restore the HR depth map. Finally, a multi-dimensional loss is introduced to constrain clear boundaries and details. Extensive experiments show that our proposed method produces improved results against state-of-the-art methods both qualitatively and quantitatively. | ['Charlie C. L. Wang', 'Zitian Zhang', 'Kun Qian', 'Chuhua Xian'] | 2020-11-24 | null | null | null | null | ['depth-map-super-resolution', 'single-image-deraining'] | ['computer-vision', 'computer-vision'] | [ 5.10044336e-01 -1.41888455e-01 9.82078090e-02 -3.51685733e-01
-9.36948538e-01 1.08740823e-02 6.58523366e-02 -3.49016368e-01
-3.42593521e-01 9.21236694e-01 1.57001227e-01 2.72964954e-01
-2.47254387e-01 -9.92650032e-01 -4.06158537e-01 -7.70568609e-01
2.58451402e-01 -1.73828825e-01 6.78506076e-01 -2.52818227e-01
2.76928037e-01 5.86994350e-01 -1.78414214e+00 2.22086594e-01
1.08894467e+00 1.11213803e+00 5.59669733e-01 3.51040989e-01
-1.69521049e-01 6.87329054e-01 -2.07759529e-01 -4.11948375e-02
3.12731802e-01 -4.26674187e-01 -6.54171467e-01 1.91501081e-01
5.73480785e-01 -6.33374214e-01 -5.68113685e-01 1.41665685e+00
5.03275216e-01 1.75971836e-01 2.10298002e-01 -7.85163343e-01
-5.39827645e-01 1.61212921e-01 -1.15608156e+00 5.05609274e-01
3.31301808e-01 -2.57573396e-01 6.18194640e-01 -8.92331362e-01
6.00240529e-01 1.30035758e+00 4.24746841e-01 4.88120645e-01
-1.10729933e+00 -8.03043544e-01 1.85117930e-01 2.51968622e-01
-1.44072843e+00 -2.07536086e-01 1.07224894e+00 6.36792183e-02
3.32700551e-01 5.48563376e-02 4.47025120e-01 7.01111138e-01
1.23505235e-01 4.10131365e-01 1.55421567e+00 -2.09025264e-01
1.50652975e-01 4.05959412e-02 -2.33404696e-01 6.97762668e-01
2.06552446e-01 2.00393364e-01 -6.67533636e-01 2.35745206e-01
1.49019742e+00 4.02676985e-02 -6.07881665e-01 -3.54382217e-01
-7.70340204e-01 5.02145648e-01 6.42878473e-01 5.10640204e-01
-2.82758147e-01 -1.83636054e-01 -7.21133277e-02 -1.15884012e-02
6.08250260e-01 1.64495700e-03 -2.44266003e-01 1.93749979e-01
-9.09847438e-01 -1.34837568e-01 8.99971202e-02 7.14651346e-01
1.11469090e+00 4.43923008e-03 2.37668052e-01 9.41909730e-01
1.43059060e-01 1.89962909e-01 3.00676107e-01 -1.24479854e+00
5.04655838e-01 5.97473085e-01 1.92426309e-01 -1.12264454e+00
-4.22216475e-01 -3.69411588e-01 -1.33465266e+00 6.79523051e-01
1.95802957e-01 2.13829383e-01 -8.19841087e-01 1.40621150e+00
5.01389861e-01 2.18613639e-01 1.51154891e-01 1.23038423e+00
8.78782272e-01 6.91020787e-01 -3.43769550e-01 -4.81466144e-01
9.86916602e-01 -7.91542113e-01 -7.74449348e-01 -2.66196907e-01
-1.60739318e-01 -5.14641345e-01 1.00811124e+00 6.01014614e-01
-1.24477208e+00 -7.22214520e-01 -1.16048598e+00 -5.23573458e-01
-1.20050376e-02 -8.87216330e-02 5.25157750e-01 2.64799923e-01
-9.68331933e-01 7.24200308e-01 -6.90720975e-01 -1.49563670e-01
3.12118053e-01 9.84674394e-02 -5.58500290e-01 -5.61259091e-01
-1.22088695e+00 7.11320043e-01 4.31164354e-01 2.72675723e-01
-4.81130689e-01 -5.54449141e-01 -8.51634145e-01 -2.93697119e-01
2.54150391e-01 -5.50411761e-01 7.36695945e-01 -5.40743828e-01
-1.44494820e+00 7.14933634e-01 -1.88343659e-01 1.39595181e-01
5.83869219e-01 -1.03975199e-01 -4.23558086e-01 5.53794444e-01
7.38232732e-02 3.42752814e-01 8.69046628e-01 -1.57302213e+00
-1.18105817e+00 -7.86825955e-01 1.46631747e-01 6.66332364e-01
-3.56610239e-01 -5.45893013e-02 -5.04509032e-01 -2.29292855e-01
7.33517289e-01 -2.23184451e-01 -3.77822965e-01 1.21708743e-01
-1.30901337e-01 1.56441599e-01 7.65294254e-01 -7.71243989e-01
1.21389532e+00 -2.13361192e+00 4.63385761e-01 -1.99044839e-01
3.72136593e-01 -9.80927572e-02 1.55318469e-01 -1.73442915e-01
1.24848500e-01 -3.53101194e-02 -3.67597014e-01 -4.48787719e-01
-6.39486074e-01 1.82207763e-01 -1.20472707e-01 5.50411880e-01
-4.93417233e-02 4.43147421e-01 -8.46636534e-01 -7.21395493e-01
5.11651635e-01 1.01104164e+00 -4.97149415e-02 2.28197962e-01
3.53204578e-01 7.90226936e-01 -5.51285565e-01 7.31242120e-01
1.20961034e+00 -2.78118640e-01 -3.22538882e-01 -5.90988338e-01
-4.61353928e-01 -1.58951506e-01 -1.42443693e+00 2.02586699e+00
-5.65895259e-01 4.15845782e-01 3.83453369e-01 -4.78636295e-01
1.06708539e+00 -1.86125606e-01 4.36031640e-01 -9.68514502e-01
-3.57651114e-02 3.22781503e-01 -5.79075396e-01 -2.38409817e-01
7.30129957e-01 -3.80692929e-01 1.84869677e-01 -3.04549243e-02
-3.26078862e-01 -2.62153119e-01 -2.73798794e-01 -1.79955959e-01
8.60068977e-01 9.94469896e-02 1.84351668e-01 1.53605223e-01
9.24639404e-01 -2.96106368e-01 8.98759604e-01 2.82541245e-01
-1.54072493e-01 9.93979156e-01 1.21751994e-01 -3.85644078e-01
-1.17195141e+00 -1.03428125e+00 -4.07736331e-01 5.51447988e-01
7.64782250e-01 4.64131963e-03 -4.17446315e-01 -1.73826694e-01
-2.08015323e-01 1.40093639e-01 -5.50833702e-01 6.68406039e-02
-6.98224604e-01 -7.97756612e-01 7.48789236e-02 4.55277920e-01
1.11642635e+00 -8.12543929e-01 -7.14089274e-01 9.64340791e-02
-3.26074123e-01 -1.39110184e+00 -1.33236602e-01 -4.33514863e-02
-1.19025981e+00 -1.00562513e+00 -9.31313097e-01 -7.43241847e-01
6.24291360e-01 6.15967155e-01 8.84609699e-01 -2.45048717e-01
-3.22813839e-01 2.28704643e-02 -3.55395377e-01 2.95518160e-01
-6.79934546e-02 -1.64510638e-01 -6.94981739e-02 3.71946953e-02
-1.08911388e-01 -8.41691077e-01 -9.31783557e-01 2.81969070e-01
-1.13630533e+00 3.30468237e-01 9.38511252e-01 6.38178051e-01
9.91443694e-01 7.74897695e-01 3.41776639e-01 -4.36561048e-01
2.02621505e-01 -4.61318679e-02 -7.72843122e-01 1.04203083e-01
-6.90750897e-01 -1.23267859e-01 5.66577196e-01 -2.18163710e-02
-1.38410473e+00 7.42122456e-02 2.82875020e-02 -5.80453873e-01
-1.44734427e-01 1.10199206e-01 -3.90357494e-01 -4.98613149e-01
2.74462134e-01 5.96646428e-01 -1.50162801e-01 -6.41951680e-01
2.73503959e-01 7.08126366e-01 7.75108278e-01 -2.34026045e-01
9.91772115e-01 8.84760559e-01 3.22070718e-01 -9.34014499e-01
-1.19500506e+00 -3.76531512e-01 -7.05983400e-01 -1.40816376e-01
9.92890537e-01 -1.19985271e+00 -5.44078529e-01 6.45633578e-01
-8.59004021e-01 -5.69825992e-02 4.07891311e-02 2.91766614e-01
-4.53759253e-01 6.76444471e-01 -7.78705716e-01 -8.47181380e-01
-2.44010687e-01 -8.09243917e-01 1.10905468e+00 7.77048409e-01
7.46180236e-01 -6.71506584e-01 -1.39068663e-01 4.84701484e-01
4.44546700e-01 4.46495563e-01 5.19176126e-01 5.38216650e-01
-9.99795020e-01 2.49727339e-01 -8.05331171e-01 5.26727378e-01
3.96556437e-01 -2.91976899e-01 -8.65366280e-01 -3.35047334e-01
3.54971379e-01 -2.65654743e-01 8.64847660e-01 3.31890851e-01
1.24970877e+00 2.88125984e-02 -9.98096839e-02 1.17122281e+00
1.99474657e+00 1.15026329e-02 7.84907043e-01 6.27088964e-01
9.46024120e-01 6.97175622e-01 1.01828837e+00 4.49803799e-01
4.70525324e-01 6.82585120e-01 5.32701194e-01 -2.58357197e-01
-2.04912633e-01 -2.48295680e-01 1.07836038e-01 5.48974574e-01
-2.76833802e-01 2.37106711e-01 -4.39332873e-01 3.04095745e-01
-1.60320938e+00 -7.75336683e-01 -1.06870957e-01 2.26576376e+00
9.26044822e-01 1.21710062e-01 -1.06240861e-01 2.66097069e-01
8.12803626e-01 4.53038096e-01 -7.45373189e-01 1.97929516e-01
-5.98697484e-01 5.71766086e-02 5.88501334e-01 6.12591028e-01
-1.04250145e+00 7.66067445e-01 5.63371181e+00 9.75115359e-01
-1.09246647e+00 1.45529015e-02 7.07063317e-01 -4.45221066e-02
-4.07438487e-01 -2.38089010e-01 -9.86124218e-01 4.20013279e-01
4.02368397e-01 -2.59453386e-01 4.60141033e-01 6.09152615e-01
1.56285390e-01 -4.60623533e-01 -7.13713050e-01 1.41509819e+00
1.52664781e-01 -1.05028808e+00 -2.71348190e-02 -3.30406837e-02
8.32029104e-01 -1.96658149e-01 1.48795202e-01 -1.45429537e-01
-2.67002080e-02 -1.03874671e+00 3.25220823e-01 6.14730060e-01
1.12019384e+00 -1.13781011e+00 6.54814124e-01 2.50669539e-01
-1.51073444e+00 -3.32483470e-01 -7.66433895e-01 4.08057459e-02
2.76612669e-01 7.16779768e-01 2.10280985e-01 9.88958359e-01
1.03426266e+00 9.70643580e-01 -5.32018661e-01 9.04219031e-01
-2.58915305e-01 -2.62601703e-01 -1.63877860e-01 5.68718314e-01
-1.65962130e-02 -4.02856380e-01 4.03604984e-01 7.07586706e-01
4.39320892e-01 6.09181762e-01 -2.30501391e-04 8.52275789e-01
8.86855870e-02 -1.40629217e-01 -4.27861273e-01 5.37266612e-01
4.82665986e-01 1.49631643e+00 -6.07019842e-01 -9.91574228e-02
-5.69348276e-01 1.39775777e+00 2.63655156e-01 4.00971770e-01
-5.75812697e-01 -3.78230393e-01 4.83740985e-01 1.71559647e-01
2.20011204e-01 -1.83933958e-01 -4.20989841e-01 -1.24199212e+00
2.05763340e-01 -4.82356280e-01 2.54551023e-01 -1.03224099e+00
-1.21374202e+00 7.91651428e-01 -2.11400509e-01 -1.56352794e+00
1.69004038e-01 -3.36820364e-01 -1.48626134e-01 1.04108262e+00
-2.15860367e+00 -1.01325953e+00 -9.87734735e-01 7.51181841e-01
4.81709033e-01 3.04029614e-01 3.34193408e-01 5.11261046e-01
-4.79123980e-01 3.61363232e-01 2.08303630e-01 -5.49019463e-02
5.69559336e-01 -1.03660572e+00 -2.09951743e-01 9.47388649e-01
-4.16887879e-01 1.63074151e-01 6.95143163e-01 -5.15677571e-01
-1.16349196e+00 -1.01466322e+00 3.38826418e-01 -7.51584545e-02
3.56372952e-01 -8.23396221e-02 -1.11808777e+00 3.20527256e-01
-2.55353421e-01 3.68917435e-01 2.78101489e-02 -3.63314241e-01
-1.72525212e-01 -4.73003954e-01 -1.48850679e+00 2.28881001e-01
1.10048330e+00 -5.23032784e-01 -4.59692299e-01 -2.02810898e-01
9.06386316e-01 -6.98358536e-01 -1.08137286e+00 6.99078321e-01
5.04479825e-01 -1.58836174e+00 1.23909199e+00 1.75388530e-01
7.58084834e-01 -6.76680923e-01 -4.01052058e-01 -1.05172980e+00
-4.34614390e-01 -1.55504510e-01 -9.04414505e-02 1.19777584e+00
-1.36165813e-01 -4.69383329e-01 9.16720092e-01 5.18281162e-01
-9.57136899e-02 -7.06136286e-01 -1.14080560e+00 -7.10673928e-01
-1.16478994e-01 7.90329874e-02 3.46492112e-01 8.56505632e-01
-3.57907414e-01 2.98249722e-01 -4.60834205e-01 5.29881895e-01
1.32001686e+00 3.44463736e-01 3.73732299e-01 -1.17626965e+00
-7.86216632e-02 -3.13797891e-01 -3.91532332e-01 -1.02624166e+00
-2.43463397e-01 -3.07001173e-01 9.11446474e-03 -1.79316354e+00
3.89835209e-01 -4.79370117e-01 -4.85513091e-01 2.70008929e-02
-2.43698239e-01 4.52902198e-01 -1.19591624e-01 2.27096900e-01
-5.91344535e-01 8.05603862e-01 1.59698379e+00 2.55114436e-01
-3.31162453e-01 -2.42449969e-01 -6.12328351e-01 7.25107491e-01
5.54352164e-01 -1.49637297e-01 -3.48871261e-01 -4.00632769e-01
2.15056792e-01 4.94703203e-01 3.60235363e-01 -1.11236215e+00
2.71960139e-01 -1.88346460e-01 7.34906137e-01 -8.55959952e-01
6.32613182e-01 -7.81323075e-01 1.00636268e-02 6.24578074e-02
-8.17170367e-02 -2.47127771e-01 1.95789915e-02 7.40560055e-01
-5.20186245e-01 1.82900712e-01 1.26344454e+00 -3.24849069e-01
-9.08578575e-01 5.96995890e-01 2.16928720e-01 -1.50722042e-01
8.02182198e-01 -4.57373112e-01 -4.27807391e-01 -2.06839010e-01
-4.80434120e-01 1.20036721e-01 1.01607728e+00 3.09809089e-01
1.14962280e+00 -1.26520729e+00 -5.73162913e-01 1.70658261e-01
4.85663414e-02 6.24227345e-01 8.41410756e-01 6.91913426e-01
-4.65098768e-01 1.38834417e-01 -5.70711911e-01 -4.75116909e-01
-1.14760387e+00 4.84164834e-01 4.18155640e-01 -2.56013304e-01
-1.08177614e+00 9.34468925e-01 3.77863735e-01 -5.20449057e-02
1.53774187e-01 -1.72716096e-01 -4.15879905e-01 -1.47422180e-01
8.97825480e-01 5.70499122e-01 -2.52852321e-01 -7.22893059e-01
-2.33895198e-01 1.16386962e+00 -1.06801637e-01 -1.01094581e-01
1.46676099e+00 -8.16911459e-01 -3.58734369e-01 4.94995236e-01
1.07162237e+00 -1.13195188e-01 -1.65547192e+00 -3.76485378e-01
-3.99524271e-01 -1.05575299e+00 4.65095669e-01 -4.52795178e-01
-1.24189746e+00 8.32680285e-01 8.88196468e-01 3.36459198e-04
1.65826476e+00 -6.23509474e-02 9.62588549e-01 -1.78146482e-01
8.41457188e-01 -1.05954862e+00 1.09916300e-01 1.20788276e-01
7.50475049e-01 -1.55974078e+00 3.40141654e-01 -6.40736997e-01
-3.87662470e-01 1.10477221e+00 9.06845629e-01 -1.38050079e-01
3.94889146e-01 8.02230984e-02 -1.19138718e-01 -1.06986426e-02
-3.64478797e-01 -2.05992222e-01 9.68459621e-02 6.13797069e-01
6.78488687e-02 -3.30030501e-01 -2.55784899e-01 3.63512725e-01
3.85660715e-02 1.56444445e-01 8.15372109e-01 6.23337269e-01
-7.86743402e-01 -8.39297414e-01 -4.60242480e-01 1.60409793e-01
-4.06168818e-01 -7.98216648e-03 1.18062317e-01 6.76887155e-01
7.54562691e-02 9.80400622e-01 -7.50266248e-03 -5.22364557e-01
2.54818410e-01 -5.77445269e-01 7.06179678e-01 -4.49752033e-01
2.06980303e-01 1.19352601e-02 -2.88329601e-01 -9.25936162e-01
-6.42578781e-01 -5.29755354e-01 -1.42677724e+00 -1.77781418e-01
-7.63039812e-02 -1.15915395e-01 5.35469115e-01 5.47984123e-01
5.02548255e-02 5.05100608e-01 9.13416088e-01 -9.35721397e-01
-9.03823003e-02 -7.97534108e-01 -1.19789398e+00 1.07370615e-01
6.54332042e-01 -6.98214650e-01 -7.69138813e-01 -3.34310979e-01] | [9.861794471740723, -2.386183738708496] |
374c88ee-6141-4f7a-9575-f07497382425 | sinusoidal-wave-generating-network-based-on | 1901.02050 | null | http://arxiv.org/abs/1901.02050v1 | http://arxiv.org/pdf/1901.02050v1.pdf | Sinusoidal wave generating network based on adversarial learning and its application: synthesizing frog sounds for data augmentation | Simulators that generate observations based on theoretical models can be
important tools for development, prediction, and assessment of signal
processing algorithms. In order to design these simulators, painstaking effort
is required to construct mathematical models according to their application.
Complex models are sometimes necessary to represent a variety of real
phenomena. In contrast, obtaining synthetic observations from generative models
developed from real observations often require much less effort. This paper
proposes a generative model based on adversarial learning. Given that
observations are typically signals composed of a linear combination of
sinusoidal waves and random noises, sinusoidal wave generating networks are
first designed based on an adversarial network. Audio waveform generation can
then be performed using the proposed network. Several approaches to designing
the objective function of the proposed network using adversarial learning are
investigated experimentally. In addition, amphibian sound classification is
performed using a convolutional neural network trained with real and synthetic
sounds. Both qualitative and quantitative results show that the proposed
generative model makes realistic signals and is very helpful for data
augmentation and data analysis. | ['David K. Han', 'Sangwook Park', 'Hanseok Ko'] | 2019-01-07 | null | null | null | null | ['sound-classification'] | ['audio'] | [ 5.29694438e-01 -9.87201259e-02 6.35275781e-01 -2.34052926e-01
-4.83637452e-01 -4.33212548e-01 5.50091088e-01 -2.73511320e-01
-5.09006828e-02 6.66333735e-01 -9.46138501e-02 -1.17631769e-02
1.03403755e-01 -9.30270374e-01 -8.04358184e-01 -6.90514207e-01
-2.43427053e-01 3.00385207e-01 3.18289921e-02 -3.99398059e-01
2.37588827e-02 6.33755684e-01 -1.64762068e+00 1.90941229e-01
8.64690065e-01 9.00521219e-01 3.03282827e-01 9.68016267e-01
-1.12220589e-02 5.97694457e-01 -1.40627062e+00 -2.05237389e-01
2.75797457e-01 -8.84531021e-01 -1.45142138e-01 3.98106724e-02
2.68355981e-02 -7.87771791e-02 -1.56177461e-01 1.02321231e+00
8.22203696e-01 1.71736345e-01 8.34996462e-01 -1.45268464e+00
-5.44495642e-01 7.49360263e-01 6.85844943e-02 -1.40461594e-01
4.57518995e-01 4.67332512e-01 4.91371244e-01 -4.06876475e-01
2.68783957e-01 1.06735468e+00 7.75907695e-01 6.99364960e-01
-1.52297485e+00 -1.04246044e+00 -3.33307534e-01 1.08018391e-01
-1.12986553e+00 -2.56690204e-01 1.39393485e+00 -3.97439837e-01
2.26438791e-01 4.11911458e-01 9.05445755e-01 1.41428483e+00
3.54520082e-01 5.66116273e-01 1.05139530e+00 -3.65153968e-01
4.45909828e-01 2.10102171e-01 -4.56648499e-01 1.83179140e-01
-9.66754258e-02 4.59490567e-01 -3.85368854e-01 -1.20411739e-01
6.63796484e-01 -1.70523539e-01 -4.84004915e-01 9.04812757e-03
-7.49311209e-01 7.44302154e-01 5.23099065e-01 1.90712541e-01
-5.47440469e-01 2.68485725e-01 2.33945966e-01 5.04551232e-01
2.00406030e-01 9.86549795e-01 1.75581500e-01 2.63655670e-02
-1.04761195e+00 4.52592373e-01 8.30493510e-01 5.93291819e-01
4.07420814e-01 1.08780169e+00 3.55215549e-01 8.70926380e-01
1.94335610e-01 8.81530285e-01 8.34179580e-01 -6.41135275e-01
-5.64886257e-02 2.72242427e-01 3.73152345e-02 -1.14219916e+00
-3.90973330e-01 -8.51142824e-01 -1.02615941e+00 6.02530479e-01
3.53612065e-01 -4.49323595e-01 -8.40531051e-01 1.60744846e+00
1.50105998e-01 5.66537261e-01 3.35741252e-01 8.90875041e-01
8.87534380e-01 9.44727480e-01 -2.47089729e-01 5.47551699e-02
7.57118165e-01 -3.68472934e-01 -6.98377788e-01 -1.41195491e-01
-6.63922057e-02 -7.98883915e-01 1.07910717e+00 5.33865571e-01
-8.86782348e-01 -8.52655947e-01 -1.13162172e+00 6.56797886e-01
-2.86504179e-01 1.79665722e-02 3.21260095e-01 7.56020010e-01
-7.72666216e-01 6.19647205e-01 -6.26943111e-01 9.09667686e-02
1.44018009e-01 2.26936609e-01 -9.79973823e-02 3.32665503e-01
-1.26845658e+00 4.28016156e-01 4.64777723e-02 3.14003468e-01
-1.31118512e+00 -7.99302995e-01 -8.57527077e-01 2.95831338e-02
-2.51187801e-01 -3.92453253e-01 1.18526816e+00 -1.06827700e+00
-1.77273667e+00 2.06099927e-01 3.80993664e-01 -8.01980317e-01
4.53914136e-01 -6.58537075e-02 -6.74073398e-01 1.17379338e-01
-2.21775264e-01 4.08261657e-01 1.18818486e+00 -1.53036928e+00
-1.61656961e-01 3.18427175e-01 -8.58714953e-02 -1.85389459e-01
-1.83662042e-01 -2.40536585e-01 3.72992367e-01 -1.03791678e+00
-8.53683054e-02 -7.94532120e-01 -3.28535169e-01 -9.42275226e-02
-4.52777833e-01 4.46967095e-01 9.70148921e-01 -4.66901302e-01
8.36429715e-01 -1.97378504e+00 -1.34042829e-01 4.26016659e-01
-1.42446265e-01 5.04102588e-01 -3.76155198e-01 6.01279616e-01
-2.49224111e-01 -6.17457777e-02 -2.77365953e-01 -6.59228116e-02
-3.10484711e-02 -4.96652983e-02 -6.71546280e-01 2.34647483e-01
3.99859905e-01 5.89953542e-01 -8.81418765e-01 -5.12761027e-02
3.78855437e-01 7.02355683e-01 -6.70225084e-01 6.02159619e-01
-3.10909390e-01 8.42869043e-01 -3.01824093e-01 2.77289838e-01
4.28699791e-01 3.43825549e-01 -1.20819919e-01 -9.59247202e-02
1.48698494e-01 4.18265611e-02 -1.17185390e+00 1.16107416e+00
-8.95574450e-01 7.24548161e-01 1.01788528e-01 -1.38129199e+00
1.40514195e+00 4.07425374e-01 3.72772753e-01 -3.01866233e-01
3.92947614e-01 1.20906062e-01 4.15731490e-01 -5.00667036e-01
3.76978256e-02 -4.98574466e-01 -1.27592966e-01 5.09153843e-01
9.45611820e-02 -8.94370437e-01 -1.26136839e-01 -2.62166977e-01
1.00089383e+00 3.26451249e-02 -1.58273056e-02 1.45994812e-01
5.50204158e-01 -1.74350679e-01 4.91759181e-01 5.29554546e-01
2.41194382e-01 5.62683344e-01 2.43553936e-01 -4.05855477e-01
-9.29471552e-01 -1.03263342e+00 5.26940748e-02 4.68889296e-01
-6.59096017e-02 1.42838061e-01 -7.80973971e-01 7.04229157e-03
-2.98264503e-01 9.08323824e-01 -6.32965267e-01 -5.40987253e-01
-3.35561395e-01 -6.40358627e-01 9.76959229e-01 3.90733868e-01
4.32984084e-01 -1.53850269e+00 -7.57940888e-01 5.77159762e-01
3.41705084e-02 -7.66495585e-01 -2.30001226e-01 2.37515107e-01
-6.15768135e-01 -7.68950224e-01 -6.18250251e-01 -7.73988962e-01
5.27447701e-01 -4.44453880e-02 7.86936760e-01 5.16176261e-02
-3.09298992e-01 2.99405098e-01 -6.04268849e-01 -9.78034735e-01
-1.05964243e+00 -2.83575833e-01 2.53110081e-01 1.71027243e-01
-2.01383755e-01 -1.21321571e+00 -4.30044323e-01 2.58154392e-01
-1.28594506e+00 -1.80488020e-01 5.04395962e-01 1.18295860e+00
3.42243671e-01 3.50319296e-01 1.11407566e+00 -6.22365713e-01
9.45143878e-01 -3.47982734e-01 -6.05954409e-01 -8.53267219e-03
-6.54127449e-02 1.30294472e-01 1.37192678e+00 -9.19974685e-01
-1.06762242e+00 1.71142161e-01 -5.32123983e-01 -6.84054196e-01
-3.43136042e-01 4.51856822e-01 -3.45618188e-01 -1.12396456e-01
1.02399719e+00 5.45195401e-01 1.65300101e-01 -1.83100417e-01
2.50468612e-01 6.20869100e-01 7.37556696e-01 -3.95233065e-01
1.15670180e+00 2.20013484e-01 5.02976365e-02 -1.26179767e+00
-3.29922229e-01 2.96353757e-01 -2.25892618e-01 -5.51750958e-01
3.95103335e-01 -3.63447249e-01 -6.08440161e-01 7.05972314e-01
-1.01496685e+00 -3.45844597e-01 -4.40781087e-01 6.19230747e-01
-5.77122688e-01 3.84695716e-02 -4.05661672e-01 -1.03958702e+00
-3.78374785e-01 -1.14070952e+00 6.33566797e-01 3.04052591e-01
-3.79003882e-01 -1.02884543e+00 1.52894288e-01 2.41468884e-02
6.60652757e-01 5.94381034e-01 6.61455929e-01 -7.23297060e-01
-2.29783237e-01 -7.19981015e-01 5.33480406e-01 5.88834167e-01
3.28046590e-01 2.17515990e-01 -1.09768987e+00 -2.00249068e-02
3.07239830e-01 -3.97938430e-01 3.95366162e-01 3.58339041e-01
1.22859776e+00 -4.71872956e-01 9.37549546e-02 6.42535210e-01
1.13782299e+00 7.00327992e-01 6.75322354e-01 -3.88077684e-02
2.21270949e-01 5.60427904e-01 3.12550902e-01 5.07150650e-01
-3.85253370e-01 4.01010990e-01 6.09029233e-01 -7.30570331e-02
-1.94437087e-01 -5.39697528e-01 3.24800789e-01 7.35820532e-01
6.62989914e-02 -4.29930270e-01 -6.73502505e-01 3.57775420e-01
-1.03560340e+00 -1.29680276e+00 5.61744682e-02 2.11812043e+00
7.14553773e-01 2.20985159e-01 2.61368547e-02 7.54640639e-01
6.57492101e-01 -4.19827327e-02 -4.63181287e-01 -4.07906651e-01
-9.63528678e-02 6.64282858e-01 -3.92529145e-02 2.76968390e-01
-8.49191070e-01 6.44145608e-01 6.46921635e+00 6.73498511e-01
-1.83878279e+00 -4.22100395e-01 2.06705347e-01 1.40343606e-01
-3.80271822e-01 -4.49249953e-01 -1.96040809e-01 6.57782495e-01
9.12100315e-01 -3.41866553e-01 5.01766503e-01 7.66854227e-01
4.32201862e-01 2.36976460e-01 -9.04813170e-01 8.43031645e-01
-1.64314695e-02 -1.30655742e+00 2.41587922e-01 -4.02960479e-01
7.15528667e-01 -5.55594146e-01 3.54967117e-01 8.53795931e-02
4.11959946e-01 -1.25099969e+00 7.05113292e-01 6.90713167e-01
6.61358714e-01 -8.38820875e-01 8.75127971e-01 5.25993466e-01
-1.08885193e+00 8.18207785e-02 -2.91772783e-01 -1.52387053e-01
2.48703673e-01 3.45202595e-01 -1.10375631e+00 4.09950227e-01
2.66209453e-01 3.43259126e-01 -2.41554976e-01 1.24063563e+00
-3.59365880e-01 1.27478600e+00 -3.12790662e-01 -3.52980137e-01
-4.41707075e-02 -1.15877233e-01 7.58972943e-01 1.13362134e+00
4.86090451e-01 -1.93103492e-01 -9.72248092e-02 1.04451454e+00
-3.86386923e-02 4.11007293e-02 -9.10229743e-01 -2.80070722e-01
4.63766754e-01 1.06257117e+00 -5.49852014e-01 -1.32201100e-02
8.91927481e-02 5.76920331e-01 -2.73538351e-01 3.21767628e-01
-1.04538631e+00 -7.29569495e-01 5.67441940e-01 1.31138355e-01
-5.68805523e-02 5.45921996e-02 -2.27191791e-01 -7.39690900e-01
-3.14980775e-01 -1.04666293e+00 -3.31262946e-01 -8.69573414e-01
-1.33553469e+00 1.05889332e+00 -1.60532445e-01 -1.92324364e+00
-6.61383033e-01 -2.64580041e-01 -1.20202088e+00 8.21042955e-01
-9.28982854e-01 -8.71648431e-01 -4.83124942e-01 3.33731055e-01
5.13804793e-01 -5.05616665e-01 1.04027390e+00 1.46663412e-01
-1.76727623e-01 5.97004652e-01 5.45771383e-02 1.54568702e-01
3.35863650e-01 -1.03788304e+00 2.80815959e-01 9.10936475e-01
4.00728226e-01 3.46972466e-01 1.35744345e+00 -3.35364789e-01
-1.05045772e+00 -1.28666091e+00 1.43402159e-01 1.80729344e-01
6.76573634e-01 -2.83806294e-01 -1.03313458e+00 1.98805422e-01
1.84582263e-01 -9.50559899e-02 8.26749980e-01 -5.88409960e-01
-1.76955722e-02 -2.96770096e-01 -1.19051754e+00 5.29456615e-01
5.48693895e-01 -2.74793625e-01 -6.76662922e-01 3.07360813e-02
4.26109463e-01 -2.65094519e-01 -6.90668941e-01 2.35813290e-01
6.40025437e-01 -6.92265987e-01 8.86724055e-01 -6.45718277e-01
6.33009672e-01 -4.07415330e-01 8.74353498e-02 -2.10512495e+00
1.36032179e-01 -7.19469368e-01 3.15345168e-01 1.18606532e+00
3.47465485e-01 -6.24141037e-01 7.71976173e-01 4.93240356e-02
-1.66656539e-01 -3.91511440e-01 -5.83538890e-01 -1.01663911e+00
7.00887963e-02 -8.62897456e-01 7.99140573e-01 7.49181807e-01
-2.15179637e-01 2.72632360e-01 -6.16961002e-01 3.97883743e-01
6.48642302e-01 1.96277454e-01 1.15863657e+00 -1.16280663e+00
-4.29876685e-01 -2.73974657e-01 -7.84156859e-01 -8.31008554e-01
2.05535889e-01 -6.71947956e-01 4.51557368e-01 -1.28791380e+00
-6.93283617e-01 -5.12116373e-01 4.44053262e-02 1.21583201e-01
-5.44749238e-02 4.28849906e-01 6.93707168e-02 -2.67013013e-01
3.02888095e-01 8.97893906e-01 1.27773154e+00 -2.24863112e-01
-1.87951639e-01 5.08584499e-01 -2.93340117e-01 8.92674088e-01
1.15424061e+00 -5.02303779e-01 -6.61557615e-01 -1.05236880e-01
-4.62448634e-02 3.08199525e-01 5.50313413e-01 -1.46992683e+00
1.19948015e-01 -1.00051798e-01 3.48734260e-01 -3.38566720e-01
6.32944942e-01 -8.85578990e-01 6.66688383e-01 6.76146388e-01
-4.86656517e-01 -2.25698903e-01 3.22734177e-01 4.70634609e-01
-6.43278182e-01 -3.04423153e-01 9.31302428e-01 2.79214531e-02
-1.91768214e-01 -1.15295596e-01 -6.30173683e-01 -3.45863402e-02
8.60813737e-01 -1.09682284e-01 2.01303419e-02 -9.71878409e-01
-8.49128902e-01 -1.90454587e-01 1.00016356e-01 1.98748708e-01
9.68886971e-01 -1.26548779e+00 -6.82565272e-01 4.58453476e-01
-1.47163451e-01 -1.04251467e-01 2.83203661e-01 3.15482944e-01
-7.40797281e-01 1.45051837e-01 -6.04304731e-01 -3.27074736e-01
-1.14166558e+00 3.88221323e-01 5.42154551e-01 3.65031650e-03
-2.93470532e-01 8.15625548e-01 1.65046751e-01 -4.54171628e-01
1.33225098e-01 -3.31837893e-01 -3.28973353e-01 -1.65656835e-01
4.38031852e-01 3.43407393e-01 -1.32073581e-01 -4.74334270e-01
5.47600836e-02 2.51797169e-01 6.36759758e-01 -2.64320910e-01
1.58344913e+00 4.87205178e-01 3.38696837e-01 4.17027563e-01
8.68925810e-01 1.66681319e-01 -1.16871560e+00 6.10337704e-02
-5.27425885e-01 -3.38062346e-01 -1.61209479e-01 -5.66579163e-01
-1.31590843e+00 1.16468775e+00 5.96720695e-01 5.45957267e-01
1.40089738e+00 -4.81425047e-01 6.53387487e-01 3.01119328e-01
4.77956891e-01 -6.49125516e-01 3.28602642e-01 -7.99619034e-03
1.29617441e+00 -7.74743080e-01 -4.22436148e-01 -1.99938461e-01
-5.87058783e-01 1.14214480e+00 5.07082760e-01 -6.36760592e-01
6.79504573e-01 6.84949219e-01 2.60997266e-01 1.58720106e-01
-5.64042211e-01 8.88309777e-02 3.16190362e-01 9.95674849e-01
1.87237337e-01 -3.35663408e-02 -2.04484195e-01 8.50650311e-01
-8.27082217e-01 -8.43972340e-03 7.47119725e-01 6.26758337e-01
-2.90445805e-01 -1.19776869e+00 -6.97165430e-01 4.35602516e-01
-4.59115416e-01 4.02317755e-02 -2.45295510e-01 5.59861898e-01
1.35679498e-01 9.90518808e-01 -1.73316494e-01 -8.22324753e-01
3.43460381e-01 -8.15959927e-03 1.94423750e-01 -4.99579221e-01
-6.12570941e-01 -4.49538715e-02 -2.44507194e-01 4.21720557e-02
-3.72391045e-01 -4.23333198e-01 -1.16324806e+00 -1.17161907e-02
-2.97130644e-01 4.03667510e-01 7.77388394e-01 5.76827347e-01
-9.43720788e-02 9.50626493e-01 9.15000677e-01 -9.37589288e-01
-6.40113354e-01 -1.17407119e+00 -6.97580218e-01 4.02083099e-01
1.41162783e-01 -6.86169386e-01 -5.86660981e-01 4.58363444e-01] | [15.528802871704102, 5.962412357330322] |
3c6e0aaa-3a11-4762-b2e0-a02486f9aee6 | big-bird-transformers-for-longer-sequences | 2007.14062 | null | https://arxiv.org/abs/2007.14062v2 | https://arxiv.org/pdf/2007.14062v2.pdf | Big Bird: Transformers for Longer Sequences | Transformers-based models, such as BERT, have been one of the most successful deep learning models for NLP. Unfortunately, one of their core limitations is the quadratic dependency (mainly in terms of memory) on the sequence length due to their full attention mechanism. To remedy this, we propose, BigBird, a sparse attention mechanism that reduces this quadratic dependency to linear. We show that BigBird is a universal approximator of sequence functions and is Turing complete, thereby preserving these properties of the quadratic, full attention model. Along the way, our theoretical analysis reveals some of the benefits of having $O(1)$ global tokens (such as CLS), that attend to the entire sequence as part of the sparse attention mechanism. The proposed sparse attention can handle sequences of length up to 8x of what was previously possible using similar hardware. As a consequence of the capability to handle longer context, BigBird drastically improves performance on various NLP tasks such as question answering and summarization. We also propose novel applications to genomics data. | ['Anirudh Ravula', 'Santiago Ontanon', 'Manzil Zaheer', 'Chris Alberti', 'Avinava Dubey', 'Philip Pham', 'Joshua Ainslie', 'Amr Ahmed', 'Qifan Wang', 'Li Yang', 'Guru Guruganesh'] | 2020-07-28 | null | http://proceedings.neurips.cc/paper/2020/hash/c8512d142a2d849725f31a9a7a361ab9-Abstract.html | http://proceedings.neurips.cc/paper/2020/file/c8512d142a2d849725f31a9a7a361ab9-Paper.pdf | neurips-2020-12 | ['linguistic-acceptability'] | ['natural-language-processing'] | [ 2.55932398e-02 1.87826231e-01 -6.33845776e-02 -2.13328153e-01
-8.88005674e-01 -7.20582902e-01 2.47681081e-01 5.81823111e-01
-5.81404150e-01 6.70200706e-01 3.80781054e-01 -5.11165679e-01
1.19659156e-01 -7.20324218e-01 -1.21734738e+00 -5.72595179e-01
-2.55938563e-02 4.93399203e-01 1.34418368e-01 -3.63561869e-01
5.05664706e-01 3.96614760e-01 -1.37966967e+00 4.20641959e-01
8.50634575e-01 8.14736426e-01 4.86961007e-01 9.39555645e-01
-6.78906202e-01 8.86500001e-01 -8.18262875e-01 -4.15301383e-01
6.65232614e-02 -1.47442400e-01 -1.23215330e+00 -4.93788600e-01
6.60116494e-01 -5.69005549e-01 -5.86910903e-01 8.16001594e-01
5.13355136e-01 2.09428281e-01 2.80998498e-01 -9.92447913e-01
-7.25414038e-01 1.06231475e+00 -4.78938401e-01 4.82882202e-01
2.09577754e-01 9.83539820e-02 1.51475406e+00 -4.99056846e-01
5.27243555e-01 1.29562521e+00 6.32743835e-01 4.55023468e-01
-1.00406146e+00 -1.84368446e-01 6.71156570e-02 3.69112402e-01
-1.15707600e+00 -4.59760547e-01 2.83218384e-01 1.38764288e-02
1.72143507e+00 4.21342015e-01 3.89399767e-01 5.46534181e-01
2.68713415e-01 9.64944661e-01 3.61342520e-01 -6.13355935e-01
2.47237876e-01 -3.51742595e-01 5.35533309e-01 7.33282387e-01
2.52191961e-01 -5.65495014e-01 -4.61323500e-01 -2.90655196e-01
3.27883810e-01 -1.55688286e-01 -4.02495056e-01 1.47718713e-01
-9.66519177e-01 7.89518535e-01 5.64446270e-01 4.81651276e-01
-1.24857672e-01 7.12852478e-01 7.37415791e-01 3.00950885e-01
3.06291401e-01 7.25407720e-01 -7.26365745e-01 -3.40532333e-01
-6.97185159e-01 1.22805864e-01 8.99058461e-01 1.09635317e+00
8.13144922e-01 -5.39056119e-03 -1.14661530e-01 5.13512313e-01
-2.64122814e-01 2.83774346e-01 6.61590278e-01 -9.85813737e-01
5.50400496e-01 4.86175030e-01 -2.35564411e-02 -9.54609990e-01
-4.83147532e-01 -5.38975239e-01 -8.62592459e-01 -4.97691035e-01
4.36764061e-01 -5.47161326e-03 -7.22513378e-01 1.86433578e+00
1.45620346e-01 -2.98211575e-02 -9.00289789e-02 6.67459607e-01
5.87187052e-01 1.15038037e+00 -7.83725157e-02 -5.17184883e-02
1.47054315e+00 -9.65284586e-01 -5.30457973e-01 -2.99878895e-01
1.06879961e+00 -4.51337457e-01 1.40883410e+00 2.28136778e-01
-1.26825380e+00 -4.01813298e-01 -9.57398534e-01 -8.56099665e-01
-2.91793227e-01 -1.93294868e-01 9.02099311e-01 5.35548687e-01
-1.26197135e+00 7.29259610e-01 -7.52568126e-01 -3.23457360e-01
3.46661985e-01 4.36569184e-01 -2.03588679e-01 -2.78421700e-01
-1.07612658e+00 8.01284373e-01 4.42313939e-01 1.01024315e-01
-6.13739014e-01 -9.80420530e-01 -7.64509678e-01 9.12134111e-01
3.35162044e-01 -7.15905368e-01 1.34744525e+00 -6.42255843e-01
-1.23270345e+00 5.28714895e-01 -4.89652812e-01 -9.45827007e-01
-3.21676135e-02 -5.02976060e-01 3.02769721e-01 3.64328533e-01
-2.87515938e-01 6.35950267e-01 4.45491076e-01 -4.85486150e-01
-2.87281245e-01 -3.36359411e-01 3.38692516e-01 -5.13122929e-03
-4.32475299e-01 -7.11829960e-02 -4.38322723e-01 -3.91321898e-01
-3.99773210e-01 -7.88554907e-01 -2.80512154e-01 -2.94695079e-01
-2.63332009e-01 -5.39716840e-01 4.48901743e-01 -7.67329514e-01
1.17031205e+00 -2.02540731e+00 1.80616349e-01 -1.02132626e-01
3.07987899e-01 5.39117396e-01 -3.46046448e-01 6.65871084e-01
2.49207988e-01 4.29548770e-01 -3.02278697e-01 -1.31903380e-01
-4.21393216e-02 3.40903938e-01 -3.41247618e-01 3.17164361e-01
3.82063389e-01 1.28289855e+00 -8.14187586e-01 -2.79329687e-01
-1.22099847e-01 1.92840293e-01 -7.44081914e-01 2.37921719e-02
-6.94317997e-01 -1.10377133e-01 -2.50187129e-01 2.21702680e-01
5.97332835e-01 -4.53712106e-01 4.56307113e-01 -1.84584949e-02
-2.88074575e-02 7.28087902e-01 -5.29540777e-01 1.96445870e+00
-5.37973225e-01 7.79302657e-01 2.00163245e-01 -1.25927365e+00
6.21708572e-01 1.91010892e-01 1.78240955e-01 -5.21723032e-01
1.59897476e-01 2.71400928e-01 1.25115648e-01 -3.80373806e-01
6.54230058e-01 5.56026632e-03 -5.54311536e-02 6.13883674e-01
2.02216789e-01 -3.06879357e-03 3.45053107e-01 5.94599843e-01
1.37371910e+00 -2.18226597e-01 2.99544662e-01 -3.85191798e-01
3.70516807e-01 2.81409472e-02 5.24131179e-01 9.97972012e-01
6.00555949e-02 3.68786484e-01 9.31575060e-01 -5.36334515e-01
-1.40082800e+00 -4.21312541e-01 2.74694800e-01 1.34308600e+00
-2.54862994e-01 -6.69140935e-01 -9.40583289e-01 -3.89180750e-01
-1.39738442e-02 4.86119241e-01 -4.84938711e-01 -1.79233223e-01
-9.87441719e-01 -5.82618833e-01 8.38762939e-01 6.82172656e-01
2.08097279e-01 -1.06116068e+00 -7.82052338e-01 3.72416437e-01
-2.77878404e-01 -9.19424891e-01 -6.85727596e-01 3.59840900e-01
-8.80348623e-01 -7.54459858e-01 -6.41311049e-01 -7.83767343e-01
2.65325338e-01 1.93127543e-01 1.26864195e+00 4.01864886e-01
-4.24164563e-01 -9.08725522e-03 -3.16426694e-01 -3.38070571e-01
-3.70627701e-01 6.16941154e-01 -3.22953761e-01 -3.72774869e-01
2.31670275e-01 -6.82360947e-01 -3.96231323e-01 -3.97834569e-01
-1.16911674e+00 -1.68331981e-01 6.79950476e-01 9.89669144e-01
4.31125760e-01 -2.84931421e-01 7.71041989e-01 -1.03465199e+00
5.13821065e-01 -4.18529660e-01 -6.58032238e-01 2.97695458e-01
-2.21970364e-01 5.41732490e-01 9.76984143e-01 -2.33787015e-01
-7.03222156e-01 -1.11546300e-01 -5.04489958e-01 -6.84102699e-02
8.98259878e-02 8.16453636e-01 -1.65041655e-01 -7.17081269e-03
4.42850143e-01 3.53573352e-01 1.15039520e-01 -6.55013025e-01
5.13445497e-01 5.14830828e-01 4.27333266e-01 -4.69085038e-01
2.28563130e-01 2.42351681e-01 3.03749535e-02 -1.12828958e+00
-8.39254916e-01 -4.68356431e-01 -3.25038344e-01 4.30835605e-01
5.28281331e-01 -6.59058928e-01 -1.30470562e+00 3.40057671e-01
-1.38995826e+00 -3.62173319e-01 -3.78745228e-01 1.75247774e-01
-4.81691211e-01 8.46586823e-01 -1.18247437e+00 -6.77088201e-01
-8.84077966e-01 -7.97034562e-01 8.33468735e-01 3.97885777e-02
-1.73686877e-01 -6.50238335e-01 -1.59689024e-01 2.99916357e-01
5.64948559e-01 -1.10193908e-01 1.67699409e+00 -6.92416906e-01
-7.64651179e-01 -1.33830830e-02 -2.92566925e-01 4.10123289e-01
-3.57822716e-01 -4.13243771e-01 -7.04992592e-01 -2.88724542e-01
9.47693512e-02 -5.12457550e-01 1.19071710e+00 2.92831540e-01
1.36527896e+00 -5.81990123e-01 -4.06440869e-02 4.53782558e-01
1.55106699e+00 2.07683314e-02 6.79072261e-01 8.61157626e-02
6.30247533e-01 2.51372397e-01 1.07201502e-01 4.02336121e-01
2.20928103e-01 4.32915658e-01 4.52687979e-01 1.88341841e-01
-1.61984414e-01 -2.07007423e-01 2.68281192e-01 1.27163565e+00
3.51366937e-01 -4.51716810e-01 -9.23230886e-01 7.83264756e-01
-1.72399235e+00 -9.71974075e-01 -1.41052037e-01 2.05714369e+00
7.75331557e-01 1.22479247e-02 -6.44488400e-03 1.85635149e-01
4.75321651e-01 1.82013959e-01 -6.81415617e-01 -1.00329030e+00
-2.37846464e-01 6.01308227e-01 6.62308395e-01 5.79587638e-01
-7.07682073e-01 9.51692343e-01 6.73234177e+00 9.96454060e-01
-1.05688810e+00 1.04201868e-01 6.00857258e-01 -2.65772283e-01
-4.93236184e-01 -1.74201682e-01 -7.38635361e-01 4.00661767e-01
1.22209716e+00 -2.68644005e-01 6.07370675e-01 7.32771516e-01
-5.84170744e-02 -2.33749934e-02 -1.15050006e+00 6.66246176e-01
1.27561897e-01 -1.67276347e+00 3.23439658e-01 -1.95442401e-02
4.95197475e-01 2.22825363e-01 -2.37113163e-01 3.98011833e-01
2.79141515e-02 -1.20049572e+00 5.37159979e-01 -1.33408245e-03
6.60100639e-01 -1.01993883e+00 9.22244966e-01 6.89067364e-01
-7.83060968e-01 -1.98637381e-01 -8.62202048e-01 -3.56151074e-01
1.14531636e-01 7.03409433e-01 -1.00337470e+00 5.09264052e-01
4.91208553e-01 1.89417899e-01 -2.02383026e-01 8.79306197e-01
1.64022576e-02 8.10511827e-01 -6.94284260e-01 -4.25017029e-01
5.22175193e-01 1.62455037e-01 2.95783132e-01 1.36464489e+00
1.94835156e-01 1.23858057e-01 -9.15179104e-02 6.23128235e-01
-5.36424518e-01 1.82517394e-01 -3.58857602e-01 -4.25201058e-01
4.45343852e-01 8.74497056e-01 -4.44854021e-01 -5.28893054e-01
-2.99041927e-01 8.21088970e-01 9.39366043e-01 3.78266871e-02
-7.01392174e-01 -7.61974275e-01 5.43161333e-01 -6.25387877e-02
7.30291843e-01 -4.16553706e-01 -3.19562286e-01 -1.05685091e+00
2.01711282e-01 -8.67718995e-01 3.10626358e-01 -6.40816152e-01
-9.11840320e-01 4.04704839e-01 -4.49512839e-01 -3.34204614e-01
9.65117812e-02 -5.32543957e-01 -3.16248596e-01 8.60289276e-01
-1.48620069e+00 -7.99886048e-01 1.47453114e-01 2.29483753e-01
6.44060552e-01 3.72919053e-01 8.52325201e-01 3.13058615e-01
-3.70414793e-01 7.50177324e-01 3.72974217e-01 2.44269911e-02
2.35020757e-01 -1.21162736e+00 8.61932576e-01 6.60892129e-01
3.24284881e-01 8.96609604e-01 5.51676989e-01 -2.43844867e-01
-1.95779669e+00 -8.54759634e-01 1.32828462e+00 -1.71006307e-01
5.16360641e-01 -5.44704437e-01 -1.08284891e+00 7.80088007e-01
1.76870629e-01 -2.40504578e-01 5.01756072e-01 2.91975409e-01
-5.19120872e-01 -9.17201489e-02 -8.29452395e-01 4.30697232e-01
9.98201191e-01 -4.74622875e-01 -7.19857097e-01 4.15960699e-01
1.37778604e+00 -4.09052879e-01 -6.95922554e-01 3.96157987e-02
4.92494375e-01 -6.94201767e-01 8.23602557e-01 -8.62785697e-01
5.88794231e-01 7.12326840e-02 5.80558926e-03 -1.07656944e+00
-4.50805038e-01 -8.04327548e-01 -4.15482849e-01 1.08430648e+00
4.35140908e-01 -5.30501902e-01 1.03723085e+00 3.72456670e-01
-3.48026574e-01 -9.66474950e-01 -9.87797856e-01 -8.29378724e-01
3.40408117e-01 -2.41170958e-01 7.30861306e-01 6.84769392e-01
2.43416816e-01 7.36445606e-01 -4.61773634e-01 -8.37060809e-02
1.76742226e-01 3.26520950e-01 5.12157023e-01 -9.65748847e-01
-6.03703022e-01 -3.78552765e-01 -1.94713622e-01 -1.53551149e+00
1.31317198e-01 -1.06472528e+00 2.01813102e-01 -1.49844420e+00
2.77644694e-01 -2.59906143e-01 -2.41447344e-01 5.00381529e-01
-2.01230630e-01 -1.08518854e-01 4.00409907e-01 3.26242745e-02
-5.61492145e-01 4.25078630e-01 9.21461582e-01 -1.35101765e-01
1.07595650e-02 -4.34400588e-01 -8.88579726e-01 3.62886190e-01
1.01629639e+00 -4.99016672e-01 -1.64146960e-01 -1.05096769e+00
4.18274164e-01 1.18116088e-01 -6.45535737e-02 -7.37216830e-01
3.09831947e-01 1.22213922e-01 -8.53283480e-02 -6.66935205e-01
4.02431518e-01 -4.01082128e-01 -2.55975097e-01 6.66145027e-01
-7.65943587e-01 1.24149136e-01 4.14566666e-01 6.12150788e-01
-1.01474077e-01 -4.37852114e-01 5.74587941e-01 -4.21164900e-01
-3.47175956e-01 2.13187993e-01 -4.84570742e-01 4.07971859e-01
5.13095915e-01 2.05441937e-01 -6.00947499e-01 -3.15169573e-01
-2.53625929e-01 2.88888603e-01 2.12059304e-01 3.43402149e-03
3.50165337e-01 -7.37890303e-01 -6.96685672e-01 -1.24977291e-01
-3.52801204e-01 1.54353082e-01 5.17475247e-01 7.21169829e-01
-9.22507405e-01 1.10728407e+00 -1.10918619e-01 -2.20254138e-01
-1.21563172e+00 9.66318667e-01 -4.72309552e-02 -5.92272818e-01
-6.67200506e-01 1.09520948e+00 2.55009979e-01 -2.42790729e-01
3.01042080e-01 -6.94646299e-01 2.49131799e-01 -2.00906113e-01
7.42026210e-01 3.42595816e-01 3.71280819e-01 -7.33465701e-02
-4.75818992e-01 2.75968194e-01 -3.47918659e-01 1.42587975e-01
1.46198213e+00 4.10562903e-02 -4.52499330e-01 1.05542183e-01
1.37319243e+00 -1.00604072e-02 -7.67766416e-01 -1.56410396e-01
1.10460423e-01 -1.09667577e-01 -1.37968287e-01 -5.70901632e-01
-7.35969961e-01 1.20452511e+00 -1.39255226e-01 2.41399691e-01
9.52651680e-01 3.62471794e-03 1.45017326e+00 8.85352671e-01
2.38894254e-01 -8.04989636e-01 -1.57593817e-01 9.03697133e-01
6.28286123e-01 -6.64573789e-01 -2.14566112e-01 -3.34828109e-01
-1.86280593e-01 9.31632578e-01 2.60855228e-01 -2.66186506e-01
1.20460726e-01 3.58306915e-01 -4.98865455e-01 2.41914243e-02
-1.16212928e+00 -1.79414935e-02 -2.45881855e-01 4.29271907e-01
5.15629888e-01 -1.57217741e-01 -4.16815370e-01 5.04435778e-01
-2.36766726e-01 6.37006909e-02 6.05294883e-01 9.39353228e-01
-7.88967133e-01 -9.66284037e-01 -7.73338974e-02 5.58347106e-01
-9.24497664e-01 -5.75434208e-01 -2.08468422e-01 4.78129774e-01
-1.78320646e-01 8.34600866e-01 1.63837329e-01 1.09807318e-02
5.54882437e-02 2.58685142e-01 5.47011256e-01 -6.14162385e-01
-9.36581135e-01 -4.14864689e-01 9.70126465e-02 -4.72127110e-01
1.35025099e-01 -2.80573338e-01 -1.31542993e+00 -7.73492634e-01
-3.08921665e-01 3.66095215e-01 6.42970800e-01 8.24800372e-01
8.72204185e-01 3.75042528e-01 1.78644240e-01 -5.57984710e-01
-9.37103391e-01 -9.18047607e-01 -3.74586135e-01 2.66646713e-01
4.97908682e-01 -1.86070707e-02 -2.39778772e-01 -1.68130249e-01] | [10.838602066040039, 7.473899841308594] |
0e36f99c-f6b5-4bf6-8be6-175d5052534d | embedding-physics-domain-knowledge-into-a | null | null | https://www.nature.com/articles/s41524-020-0277-x | https://www.nature.com/articles/s41524-020-0277-x.pdf | Embedding physics domain knowledge into a Bayesian network enables layer-by-layer process innovation for photovoltaics | Process optimization of photovoltaic devices is a time-intensive, trial-and-error endeavor, which lacks full transparency of the underlying physics and relies on user-imposed constraints that may or may not lead to a global optimum. Herein, we demonstrate that embedding physics domain knowledge into a Bayesian network enables an optimization approach for gallium arsenide (GaAs) solar cells that identifies the root cause(s) of underperformance with layer-by-layer resolution and reveals alternative optimal process windows beyond traditional black-box optimization. Our Bayesian network approach links a key GaAs process variable (growth temperature) to material descriptors (bulk and interface properties, e.g., bulk lifetime, doping, and surface recombination) and device performance parameters (e.g., cell efficiency). For this purpose, we combine a Bayesian inference framework with a neural network surrogate device-physics model that is 100× faster than numerical solvers. With the trained surrogate model and only a small number of experimental samples, our approach reduces significantly the time-consuming intervention and characterization required by the experimentalist. As a demonstration of our method, in only five metal organic chemical vapor depositions, we identify a superior growth temperature profile for the window, bulk, and back surface field layer of a GaAs solar cell, without any secondary measurements, and demonstrate a 6.5% relative AM1.5G efficiency improvement above traditional grid search methods. | ['Ian Marius Peters & Tonio Buonassisi', 'Fen Lin', 'Shijing Sun', 'Qianxiao Li', 'Rolf Stangl', 'Christoph J. Brabec', 'Armin G. Aberle', 'Erik Birgersson', 'Thomas Heumueller', 'Mariya Layurova', 'Jose Dario Perea', 'Hansong Xue', 'Yue Wang', 'Siyu I. P. Tian', 'Maung Thway', 'Felipe Oviedo', 'Zekun Ren'] | 2020-01-31 | null | null | null | null | ['physics-informed-machine-learning', 'physical-simulations'] | ['graphs', 'miscellaneous'] | [ 5.66157758e-01 -1.48248255e-01 -1.47296980e-01 2.04482600e-02
-6.90419674e-01 -5.79164207e-01 3.54335994e-01 3.70537251e-01
-1.86513171e-01 1.23283839e+00 -3.46435279e-01 -3.94352019e-01
-2.12514594e-01 -7.32907176e-01 -4.26196575e-01 -1.35001504e+00
6.34135962e-01 6.04725778e-01 -7.23459795e-02 2.06734836e-01
5.35916209e-01 5.98006308e-01 -1.11448836e+00 -2.29000181e-01
1.22955191e+00 1.36771667e+00 1.83619663e-01 5.88691831e-01
6.84601739e-02 -1.18202344e-01 -6.83596790e-01 7.39495009e-02
-8.65109488e-02 -4.92572218e-01 1.73960272e-02 -1.99428439e-01
1.84827507e-01 -2.84125321e-02 5.63280331e-03 8.86959136e-01
4.86113966e-01 -3.25608224e-01 1.07085061e+00 -1.01450455e+00
-2.69587606e-01 1.97318226e-01 -3.88272494e-01 -2.22890899e-01
-1.31290266e-02 8.08537781e-01 5.81024945e-01 -5.16132057e-01
5.07712543e-01 5.60245931e-01 5.99649012e-01 4.26048130e-01
-1.72279024e+00 -9.07586515e-01 -9.08459201e-02 -1.39745995e-01
-1.39938128e+00 -5.98749638e-01 5.42422414e-01 -5.89042425e-01
1.42281365e+00 5.82760237e-02 9.95998621e-01 1.12816167e+00
6.99724495e-01 -1.19500287e-01 1.33238125e+00 -5.31995177e-01
6.50411129e-01 6.64500371e-02 1.36551797e-01 5.72135985e-01
8.65759969e-01 2.99007952e-01 -5.76945245e-01 -3.61257434e-01
6.65580213e-01 -2.93718159e-01 -2.50654578e-01 -5.33139706e-01
-5.89253247e-01 3.71653467e-01 2.62812436e-01 7.59650813e-03
-3.84942144e-01 6.38366103e-01 -1.04355060e-01 -2.25041568e-01
9.76512805e-02 7.78963923e-01 -3.97739410e-01 -2.26253688e-01
-1.14194632e+00 7.46677443e-02 1.01819241e+00 6.97606444e-01
7.60745406e-01 1.09330706e-01 -3.64628136e-01 3.24344486e-01
6.51480436e-01 1.13135278e+00 -3.31029058e-01 -1.02746224e+00
-2.86861993e-02 6.78668559e-01 5.63679338e-01 -1.24783710e-01
-2.53880709e-01 -3.95847142e-01 -8.09311211e-01 4.89663124e-01
4.17216510e-01 -2.99614310e-01 -1.36514819e+00 1.36507237e+00
2.04892933e-01 -3.31334680e-01 3.26896384e-02 2.66786665e-01
3.27608317e-01 8.60411227e-01 1.14666820e-01 -8.68408859e-01
1.19294178e+00 -5.92673182e-01 -7.63987720e-01 -3.01300406e-01
9.86113697e-02 -4.31851447e-01 6.86114013e-01 4.23902094e-01
-1.42081988e+00 1.59655288e-01 -1.42356002e+00 1.11762928e-02
-3.12855721e-01 2.62055695e-01 5.21281898e-01 1.17382646e+00
-8.08331370e-01 8.98275316e-01 -1.04965365e+00 -4.39450204e-01
5.96160114e-01 7.86510766e-01 4.19058800e-01 8.83414596e-02
-4.45318371e-01 7.18756795e-01 8.10961798e-02 1.87324777e-01
-9.34567928e-01 -1.17455208e+00 -1.71209052e-01 2.03688741e-01
7.16018975e-01 -1.07543242e+00 1.01298416e+00 -3.07315856e-01
-2.32616401e+00 3.52121919e-01 -7.05839157e-01 -2.70076901e-01
2.24434763e-01 -1.68030839e-02 -2.24784747e-01 -8.00087228e-02
-3.00524443e-01 2.41016209e-01 8.75145018e-01 -1.42285442e+00
-1.36604821e-02 -3.25501949e-01 -6.11418664e-01 -1.44330263e-01
-3.43526065e-01 -4.16170806e-02 -3.50029409e-01 1.47482499e-01
6.97842240e-02 -7.54634440e-01 -4.05309230e-01 4.79093567e-02
-5.55097878e-01 7.02193603e-02 5.19212246e-01 -5.32967031e-01
1.09721375e+00 -1.74550474e+00 2.11562201e-01 7.81732082e-01
2.13289052e-01 -6.45562559e-02 4.34265137e-01 6.31893277e-01
4.83038038e-01 3.75006765e-01 -5.93361557e-01 -7.87608624e-02
-7.27064684e-02 -8.79048631e-02 1.34543046e-01 4.56210941e-01
1.33482516e-01 9.31928635e-01 -5.56355536e-01 -2.95935170e-04
1.20263599e-01 4.65481341e-01 -3.84372503e-01 -3.39078940e-02
-5.73688209e-01 4.48349923e-01 -4.05800462e-01 9.79883611e-01
5.11879802e-01 -5.48677385e-01 5.57813883e-01 -6.04941249e-01
-6.04867995e-01 2.78372109e-01 -8.37561727e-01 1.45308483e+00
-3.79546046e-01 3.85805696e-01 3.13459754e-01 -4.02503788e-01
9.97140110e-01 -5.06580323e-02 4.48646873e-01 -7.01845884e-01
1.91583022e-01 4.89140213e-01 6.03818595e-02 -7.14795813e-02
2.47860432e-01 -2.65245855e-01 4.87569869e-02 2.08582476e-01
-6.34559691e-02 -8.13816547e-01 6.46680295e-02 -2.09994152e-01
1.31288934e+00 1.01931296e-01 3.06906968e-01 -6.53077006e-01
1.29713103e-01 -1.37916477e-02 7.81713307e-01 8.46637249e-01
1.48817435e-01 1.66926399e-01 6.89517319e-01 2.88190454e-01
-1.34735107e+00 -1.16662776e+00 -2.60938704e-01 2.13965684e-01
1.88308075e-01 -4.36618239e-01 -7.67364204e-01 6.08730800e-02
2.90830135e-02 8.44953477e-01 -2.04249159e-01 -5.27152009e-02
-2.87780046e-01 -1.33364797e+00 1.99192896e-01 5.55462837e-01
2.22170964e-01 -5.52034557e-01 -6.11415863e-01 2.66544312e-01
6.76821411e-01 -1.09976745e+00 1.06922328e-01 5.52839518e-01
-9.96389449e-01 -6.81502938e-01 -2.67600983e-01 -1.24797821e-01
7.61412919e-01 -5.42825520e-01 1.13044190e+00 -1.78624183e-01
-7.04058588e-01 1.28722280e-01 4.67430830e-01 -6.41296804e-01
-5.77231467e-01 5.87904975e-02 4.09477279e-02 -4.62155938e-01
-5.79816252e-02 -8.29056919e-01 -8.00303698e-01 1.93706155e-01
-4.21389937e-02 2.03574121e-01 6.92025304e-01 6.20997667e-01
7.58612335e-01 3.69767487e-01 2.25013241e-01 -5.99311769e-01
4.14638042e-01 -9.22219381e-02 -1.40595675e+00 4.72097933e-01
-1.21840560e+00 7.41032660e-02 2.62647003e-01 -1.35196894e-01
-1.24093568e+00 6.78700879e-02 2.63197124e-01 -7.69733191e-02
-6.30924338e-03 3.35306764e-01 -5.98231912e-01 -4.11262542e-01
6.26720607e-01 1.10729896e-02 -1.02795996e-01 -2.53306806e-01
7.44132474e-02 2.44901627e-01 2.18923181e-01 -7.19489574e-01
9.02966797e-01 4.27250803e-01 6.02480173e-01 -9.27405357e-01
-3.22132945e-01 -4.77873161e-02 -3.31069887e-01 -1.68122798e-01
8.96867275e-01 -8.20675969e-01 -1.20659387e+00 6.11237824e-01
-9.19660032e-01 -6.39663756e-01 -1.90209538e-01 3.77631374e-02
-1.73328385e-01 -2.37919405e-01 -3.45475256e-01 -1.33581078e+00
-5.90654850e-01 -1.10030234e+00 1.05034673e+00 7.17491150e-01
-2.14343324e-01 -9.46229339e-01 -2.94452220e-01 5.77182829e-01
8.48182678e-01 3.39902520e-01 1.52045906e+00 4.24309820e-02
-1.32233417e+00 -2.51899194e-02 -4.79344517e-01 1.62759468e-01
1.69111118e-01 3.89668941e-01 -1.06685853e+00 -3.95654321e-01
2.29035482e-01 1.06147930e-01 8.43852520e-01 9.34071004e-01
1.06359518e+00 1.11794613e-01 -8.33929896e-01 6.26058936e-01
1.78030920e+00 4.94927078e-01 6.20933533e-01 -2.91210592e-01
8.58368754e-01 5.12115844e-02 -1.25024421e-02 3.68505359e-01
-2.34909151e-02 5.47624528e-01 3.22649986e-01 5.03013991e-02
1.09400805e-02 -6.84549361e-02 3.91176581e-01 5.30570745e-01
-2.57189900e-01 -6.03641450e-01 -1.05372834e+00 2.47210115e-01
-1.33980954e+00 -3.94785881e-01 -1.94798425e-01 2.45257449e+00
7.72631288e-01 2.64316410e-01 -2.07976237e-01 -2.87001997e-01
5.09455621e-01 -1.32246420e-01 -1.21188581e+00 -1.31852388e-01
5.51140532e-02 5.89189708e-01 9.66456115e-01 4.40707803e-01
-3.87443990e-01 6.28092527e-01 6.76565456e+00 6.19425833e-01
-1.10403836e+00 -1.84401069e-02 6.35720968e-01 -4.24133837e-01
-5.41556835e-01 3.46031904e-01 -1.18079782e+00 3.10732722e-01
1.20097637e+00 -1.14710659e-01 6.96962297e-01 7.71549195e-02
1.84475318e-01 -6.18433177e-01 -1.39392328e+00 9.99303579e-01
-2.79291749e-01 -1.75869417e+00 -4.74297196e-01 5.57888210e-01
9.45023715e-01 -1.35078490e-01 -1.43095508e-01 -1.35062143e-01
1.50211647e-01 -1.28049338e+00 3.67140591e-01 8.75355244e-01
8.18897069e-01 -4.60359186e-01 2.56905705e-01 -8.74134675e-02
-6.01554394e-01 -3.46241176e-01 2.70918608e-02 1.56484708e-01
2.18622953e-01 1.11300755e+00 -7.13801563e-01 4.32755262e-01
5.84271133e-01 3.32185090e-01 -1.56902283e-01 1.03349745e+00
-4.90167029e-02 8.03527713e-01 -8.00151408e-01 -4.13943678e-01
-2.62016654e-01 -6.28741086e-01 5.87178171e-01 6.68650031e-01
6.79200172e-01 -8.18409622e-02 -1.63705885e-01 1.69059563e+00
-3.11531752e-01 -3.83255988e-01 -1.85142443e-01 -3.59899044e-01
8.14377129e-01 1.15872157e+00 -7.55081892e-01 7.61235356e-02
1.75836518e-01 5.71597219e-01 -2.99590141e-01 6.41439438e-01
-5.16172886e-01 -9.15107876e-03 6.98481500e-01 2.91212350e-01
4.98026490e-01 -3.09930086e-01 -8.91945839e-01 -5.87283969e-01
2.38999814e-01 -5.10349154e-01 -4.74241316e-01 -6.44457102e-01
-1.10658371e+00 -3.38627577e-01 6.87342808e-02 -1.27149224e-01
1.22791328e-01 -9.80549097e-01 -8.44558239e-01 1.30926323e+00
-1.49669075e+00 -6.85185373e-01 -4.90979910e-01 -2.93252319e-01
-8.95367861e-02 4.42035533e-02 8.64815414e-01 -1.72227379e-02
-9.13086057e-01 2.78250068e-01 7.53734589e-01 -6.17127299e-01
3.09830368e-01 -1.21947896e+00 3.17583412e-01 5.81772029e-01
-5.62423110e-01 4.46369022e-01 8.31203818e-01 -9.63386893e-01
-2.06385446e+00 -7.12080300e-01 1.97755143e-01 -5.22457361e-01
7.69887507e-01 -5.51826954e-01 -8.14801931e-01 -6.75659552e-02
3.62995595e-01 -3.53478044e-01 4.94771242e-01 1.96435541e-01
2.83239067e-01 -5.52533150e-01 -1.18517804e+00 6.27754509e-01
1.08174813e+00 -4.62790996e-01 3.35001409e-01 4.51872855e-01
4.09960687e-01 -4.03344899e-01 -1.04512501e+00 8.19717228e-01
6.95974886e-01 -6.47352993e-01 8.15910995e-01 4.02916037e-02
1.72751658e-02 -3.80964428e-01 -1.35937795e-01 -9.54922974e-01
8.06049854e-02 -1.03024828e+00 -3.83084983e-01 1.48273373e+00
7.48937011e-01 -6.89603150e-01 9.95104909e-01 8.80293012e-01
-9.35793892e-02 -9.50900793e-01 -9.01021481e-01 -9.29877698e-01
1.15271337e-01 -1.36796013e-01 3.70370775e-01 2.24649712e-01
-7.27795720e-01 3.75761658e-01 1.07523531e-01 4.19786066e-01
8.63692403e-01 4.67404798e-02 3.34955841e-01 -1.50356662e+00
-2.27548882e-01 -5.38060725e-01 1.00292310e-01 -6.67586923e-01
-5.56719936e-02 -4.08732802e-01 2.72722393e-01 -1.61061215e+00
1.80977643e-01 -7.14331448e-01 -3.09589863e-01 1.66967586e-01
-9.24143940e-02 -2.57574826e-01 -1.50877640e-01 -1.23868339e-01
-6.36612028e-02 6.15135074e-01 8.34998608e-01 -1.28454924e-01
-5.03013313e-01 -7.91711956e-02 -6.65523112e-01 4.14682448e-01
7.85926998e-01 -4.92905408e-01 -2.93369263e-01 -2.22029313e-01
5.21891117e-01 2.48336017e-01 4.98965651e-01 -1.39400244e+00
3.81899983e-01 -2.28482202e-01 5.06693065e-01 -3.25493425e-01
7.71814346e-01 -6.56580091e-01 4.57349092e-01 3.30312520e-01
2.77416587e-01 -4.97945189e-01 4.85645533e-01 7.31519282e-01
4.83475447e-01 -5.11347592e-01 6.50944769e-01 4.98205759e-02
-6.69110939e-02 9.27587077e-02 -4.65900689e-01 -2.22321525e-01
8.66736948e-01 -4.58383948e-01 -4.22770172e-01 1.37341529e-01
-2.75490493e-01 2.70203918e-01 9.55549479e-01 -4.35073644e-01
2.00917631e-01 -8.29464257e-01 -2.67922342e-01 1.51104957e-01
-1.19254567e-01 4.86419983e-02 2.74710327e-01 7.57207513e-01
-3.74948442e-01 1.74911007e-01 6.87259994e-03 -8.55075598e-01
-1.14871466e+00 -1.21108845e-01 6.80767775e-01 -4.31395739e-01
-2.11524427e-01 8.25237393e-01 -1.13809176e-01 2.17687994e-01
-2.16529816e-01 -6.24238789e-01 5.56301951e-01 -1.30225778e-01
-1.69533104e-01 3.63227755e-01 1.91350386e-01 2.12105915e-01
-3.20918620e-01 7.15880036e-01 9.14460048e-02 -1.24508053e-01
1.18318403e+00 1.18501402e-01 -3.26439083e-01 6.73599720e-01
6.15588129e-01 2.01901793e-03 -1.55617440e+00 1.98633984e-01
-7.51713663e-02 -3.15336660e-02 5.84186614e-01 -1.15982234e+00
-7.03585744e-01 6.50528610e-01 1.08468342e+00 6.89899176e-02
1.28431392e+00 -2.15112776e-01 4.68117058e-01 6.24953091e-01
4.71204631e-02 -1.40144980e+00 -2.34328955e-01 1.17665216e-01
4.74600196e-01 -1.03957951e+00 5.82364380e-01 -7.39562154e-01
9.30137746e-03 9.84433532e-01 6.10980630e-01 3.96910340e-01
6.12461507e-01 6.59451008e-01 -1.91865772e-01 -4.14912581e-01
-8.51081371e-01 3.10601890e-01 9.94317830e-02 3.86897773e-01
-8.17388017e-03 1.23523377e-01 1.17968082e-01 1.83633789e-01
1.69657007e-01 -3.11443340e-02 1.36690870e-01 9.49964643e-01
-3.56855720e-01 -1.26139665e+00 -2.55396962e-01 7.96688199e-01
5.91976643e-02 -3.69716793e-01 -2.92917699e-01 5.40177166e-01
7.63453990e-02 7.75166392e-01 6.01164736e-02 2.94727385e-01
2.32934371e-01 4.08012331e-01 8.25012743e-01 -4.87103581e-01
-4.81200099e-01 7.32873529e-02 1.38371363e-01 -4.25218880e-01
-1.91463575e-01 -7.02667117e-01 -1.12617218e+00 -1.73239522e-02
-6.54055774e-01 -2.35510886e-01 1.22509456e+00 9.86402988e-01
5.43987036e-01 9.05037463e-01 2.31707349e-01 -8.80120695e-01
-9.16102454e-02 -6.03266120e-01 -3.78329813e-01 -3.95782888e-01
2.39980444e-01 -6.50210738e-01 -2.09401667e-01 -4.14516479e-01] | [5.770608425140381, 4.464186191558838] |
fe625eba-1bf3-42b8-9253-4fd0b91abe5e | non-separable-multi-dimensional-network-flows | 2305.08628 | null | https://arxiv.org/abs/2305.08628v1 | https://arxiv.org/pdf/2305.08628v1.pdf | Non-Separable Multi-Dimensional Network Flows for Visual Computing | Flows in networks (or graphs) play a significant role in numerous computer vision tasks. The scalar-valued edges in these graphs often lead to a loss of information and thereby to limitations in terms of expressiveness. For example, oftentimes high-dimensional data (e.g. feature descriptors) are mapped to a single scalar value (e.g. the similarity between two feature descriptors). To overcome this limitation, we propose a novel formalism for non-separable multi-dimensional network flows. By doing so, we enable an automatic and adaptive feature selection strategy - since the flow is defined on a per-dimension basis, the maximizing flow automatically chooses the best matching feature dimensions. As a proof of concept, we apply our formalism to the multi-object tracking problem and demonstrate that our approach outperforms scalar formulations on the MOT16 benchmark in terms of robustness to noise. | ['Florian Bernard', 'Daniel Cremers', 'Viktoria Ehm'] | 2023-05-15 | null | null | null | null | ['multi-object-tracking'] | ['computer-vision'] | [ 1.60996929e-01 -1.12878524e-01 -2.59645909e-01 -2.53622234e-01
-1.60607949e-01 -7.27233410e-01 6.92097664e-01 3.25997740e-01
-5.03136873e-01 6.98907733e-01 1.57707203e-02 2.46818806e-03
-8.84957969e-01 -9.03773367e-01 -2.05890328e-01 -7.97668993e-01
-4.88603085e-01 4.43726540e-01 3.33991647e-01 -8.12717006e-02
1.22489616e-01 9.62822378e-01 -1.42658591e+00 -2.57332444e-01
7.75254667e-01 1.16005003e+00 -5.50416633e-02 6.29976630e-01
-1.26647115e-01 2.75380760e-01 -3.70012373e-01 -4.71457869e-01
6.54244959e-01 -3.91192406e-01 -8.13612580e-01 2.32569471e-01
7.18543947e-01 1.01455767e-02 -2.89909601e-01 1.32547569e+00
1.39174685e-01 2.67189682e-01 7.77290940e-01 -1.61210358e+00
-1.01384573e-01 1.69588923e-01 -3.95438105e-01 4.11843300e-01
-1.05597656e-02 6.70597851e-02 1.39572668e+00 -5.69092691e-01
8.15241337e-01 1.34894657e+00 6.23556077e-01 2.61875391e-01
-1.75324953e+00 -1.30355179e-01 9.75683033e-02 2.09406149e-02
-1.12461591e+00 -4.58006889e-01 8.94618094e-01 -6.49425030e-01
4.78993654e-01 2.84511268e-01 6.70407176e-01 6.50030851e-01
1.15488991e-01 3.64830732e-01 8.09752226e-01 -9.23915505e-02
2.97059417e-01 9.40755755e-02 2.23580599e-01 8.25057209e-01
6.16717458e-01 1.21406103e-02 -4.11736548e-01 -2.16575623e-01
6.96636200e-01 -3.96843292e-02 -2.56476909e-01 -9.13471520e-01
-1.35575461e+00 1.12480319e+00 5.51986635e-01 4.64501023e-01
-2.74314761e-01 1.55622378e-01 6.12161219e-01 6.23302042e-01
2.92548567e-01 6.10433102e-01 -1.23196803e-01 -2.02275738e-01
-7.15530276e-01 1.59969509e-01 8.95748913e-01 6.85240507e-01
9.58103359e-01 -2.11436749e-01 -7.21739233e-02 5.46103954e-01
2.18951464e-01 2.37517893e-01 1.48992315e-01 -9.23748791e-01
4.76806372e-01 6.83650255e-01 -1.31563777e-02 -1.48691952e+00
-6.71151936e-01 -3.93802404e-01 -1.12185121e+00 3.82299006e-01
7.69827008e-01 -7.47685283e-02 -4.47205603e-01 1.97054875e+00
5.56769311e-01 -1.18528858e-01 -1.09218225e-01 9.73684371e-01
3.32440287e-01 2.57039547e-01 5.02904393e-02 -3.04552436e-01
1.17794740e+00 -4.48809981e-01 -5.86374879e-01 8.19608271e-02
5.82698822e-01 -3.79652798e-01 8.10126185e-01 1.44506603e-01
-8.87447357e-01 -2.92629778e-01 -9.03099537e-01 1.58842355e-01
-4.95913357e-01 -2.23537624e-01 5.84937394e-01 6.45134628e-01
-1.00145555e+00 8.51219118e-01 -7.73820877e-01 -4.72402334e-01
4.32748854e-01 6.18188441e-01 -6.43147349e-01 2.78572768e-01
-9.89291728e-01 5.46033263e-01 4.27343935e-01 6.01560809e-02
-1.07585363e-01 -7.89297223e-01 -7.83170760e-01 1.86725885e-01
4.80749935e-01 -7.44305968e-01 5.93231499e-01 -8.71904731e-01
-1.30854678e+00 6.39068544e-01 1.88579008e-01 -5.45302749e-01
8.34885180e-01 2.10305303e-01 -2.68099576e-01 4.71968204e-01
5.83370961e-02 4.18719143e-01 9.30221498e-01 -8.25130761e-01
-4.84078020e-01 -4.27456260e-01 4.06950772e-01 1.30030727e-02
-7.00593591e-01 -9.84048024e-02 7.33263195e-02 -4.63482499e-01
-1.48273539e-03 -1.00116825e+00 -3.96808118e-01 5.97600818e-01
-4.97033775e-01 -7.36067966e-02 8.88581216e-01 -1.85780227e-02
1.11327028e+00 -2.11367559e+00 6.74776196e-01 6.17222607e-01
6.94786668e-01 2.61330873e-01 3.99477892e-02 7.33871162e-02
1.72750115e-01 1.39373690e-01 -3.86894971e-01 -1.91808730e-01
9.47651267e-02 2.32962042e-01 2.74082810e-01 8.52818251e-01
4.85528052e-01 7.04450071e-01 -1.01228821e+00 -7.38471746e-01
2.93738991e-01 5.47969639e-01 -5.37861586e-01 -1.73892871e-01
6.53806841e-03 3.80208045e-01 -9.00301754e-01 2.77886122e-01
5.46135724e-01 -4.64877069e-01 1.89589381e-01 -4.04806525e-01
-1.05461404e-01 -8.41393098e-02 -1.43967843e+00 1.67007720e+00
-4.16826218e-01 7.34269083e-01 6.62632287e-02 -1.16914999e+00
7.98092067e-01 1.96332894e-02 1.04507506e+00 -2.96978712e-01
1.20649174e-01 1.88653916e-01 3.45251381e-01 -2.41781667e-01
4.80289489e-01 -1.08977616e-01 -8.10101107e-02 2.91889548e-01
-5.18768802e-02 2.59570301e-01 7.03862011e-01 2.81037807e-01
1.25446022e+00 -3.67625177e-01 5.04083931e-01 -6.66292012e-01
9.23035562e-01 -1.06816992e-01 5.58618248e-01 5.58611631e-01
-2.39281163e-01 3.42335045e-01 1.05675781e+00 -3.68230939e-01
-9.22329783e-01 -9.45168436e-01 -2.73336023e-01 6.24928594e-01
2.70361267e-02 -4.79442239e-01 -4.24363345e-01 -9.95322585e-01
4.92069453e-01 3.68724838e-02 -6.98396623e-01 -1.17383599e-01
-5.06329775e-01 -5.80361843e-01 3.61448318e-01 2.10238785e-01
2.88866848e-01 -7.27563560e-01 -1.01264250e+00 2.45464951e-01
3.49973768e-01 -1.15544689e+00 -5.43799102e-01 1.62977278e-01
-9.37762976e-01 -1.11921394e+00 -5.19910932e-01 -2.47552663e-01
6.75881863e-01 1.37521341e-01 1.08682919e+00 -3.75568457e-02
-4.88652349e-01 5.64452708e-01 -9.63443890e-02 3.01340759e-01
-2.85848856e-01 2.74148345e-01 2.74119503e-03 3.44378024e-01
7.51752108e-02 -7.46862173e-01 -6.60744429e-01 2.21395224e-01
-1.02447295e+00 -3.70732039e-01 4.44939911e-01 9.05939698e-01
5.05625188e-01 -1.34375870e-01 7.77650476e-01 -8.35733831e-01
7.32922792e-01 -5.05974472e-01 -1.01452482e+00 2.38250881e-01
-6.23343289e-01 4.21757460e-01 8.59666586e-01 -5.15296400e-01
-5.90097964e-01 2.87273943e-01 3.97781640e-01 -4.00869489e-01
-1.27321631e-01 4.73981231e-01 -2.01096281e-01 -6.18639052e-01
4.57285583e-01 -2.64822513e-01 2.32800812e-01 -3.39815080e-01
5.36774218e-01 2.22416192e-01 1.15942895e-01 -5.15656114e-01
9.47895885e-01 5.63942850e-01 9.34343517e-01 -1.01263189e+00
-5.48148513e-01 -4.53316092e-01 -9.38062489e-01 -3.04216921e-01
7.83548594e-01 -2.67779440e-01 -9.81040061e-01 7.35947043e-02
-9.80756819e-01 1.23367682e-01 -6.01618588e-01 5.52710414e-01
-7.06193507e-01 3.08866888e-01 -3.27495754e-01 -7.23726392e-01
-4.12059911e-02 -1.13582325e+00 8.05184782e-01 2.43967414e-01
-1.14520386e-01 -1.41944480e+00 1.03153937e-01 -2.40172058e-01
4.58124042e-01 4.65056688e-01 1.06664479e+00 -7.35118210e-01
-5.16964376e-01 -9.12216753e-02 -4.72635686e-01 1.59064859e-01
2.10607469e-01 -8.91846698e-03 -6.14950240e-01 -2.87438840e-01
-1.67492613e-01 -1.36477306e-01 7.37362325e-01 2.81768590e-01
9.36253130e-01 -1.73171908e-01 -2.95824379e-01 6.18745148e-01
1.50569296e+00 -2.54811585e-01 2.18801033e-02 1.66579634e-01
7.46149361e-01 6.62640572e-01 3.98914665e-01 5.32549918e-01
-4.96641323e-02 9.36493099e-01 5.49141884e-01 1.71277270e-01
-2.16877326e-01 1.48619249e-01 -4.38179150e-02 4.44096506e-01
4.66625877e-02 -2.85285473e-01 -6.46583915e-01 4.93339330e-01
-1.84050035e+00 -9.67748523e-01 -2.02181116e-01 2.38913369e+00
4.45928872e-01 6.43291622e-02 5.26180744e-01 1.94696307e-01
7.13580966e-01 3.82347405e-01 -6.50271714e-01 -8.18114653e-02
-7.09079504e-02 -7.19506964e-02 6.57571614e-01 4.59048897e-01
-1.10610557e+00 4.67001855e-01 5.64907789e+00 4.72591609e-01
-1.19784594e+00 -1.10070311e-01 1.84493542e-01 -1.33324832e-01
-2.09209070e-01 -8.08508173e-02 -6.05282009e-01 3.96358162e-01
6.75074399e-01 -4.74307507e-01 3.41960877e-01 5.45671165e-01
-8.88227113e-03 7.66220540e-02 -1.23931086e+00 9.90005672e-01
-2.57522404e-01 -1.40255475e+00 -8.88493434e-02 5.55714667e-01
3.10948133e-01 -1.47177011e-01 9.81918573e-02 -2.87081748e-01
2.01099366e-02 -7.93253958e-01 3.08642149e-01 4.04691845e-01
6.82811081e-01 -8.73435318e-01 3.40619713e-01 1.70848100e-03
-1.40785837e+00 -5.45154139e-02 -2.95685470e-01 1.82775483e-01
3.35134745e-01 7.72506535e-01 -6.35139704e-01 6.99927807e-01
1.52627170e-01 8.95088255e-01 -4.31771338e-01 1.29020751e+00
3.35803866e-01 1.05501905e-01 -5.96003830e-01 -1.17473997e-01
3.98342997e-01 -4.99005824e-01 9.25460875e-01 1.25183916e+00
2.83264190e-01 -2.63803184e-01 3.45868886e-01 9.21005309e-01
-1.55140355e-01 2.80758649e-01 -7.16034234e-01 -1.55004099e-01
1.64059564e-01 1.55405414e+00 -1.12313092e+00 -2.77442224e-02
-5.65773487e-01 7.04721987e-01 3.50003481e-01 2.55082905e-01
-5.56377470e-01 -4.67114866e-01 1.24646556e+00 -2.95918137e-02
4.06806648e-01 -2.73351222e-01 -7.72899985e-02 -1.19191325e+00
9.31408778e-02 -4.91348177e-01 4.12253886e-01 1.67915434e-01
-1.44384885e+00 6.04964137e-01 7.22013265e-02 -1.31955326e+00
-3.67456853e-01 -6.87703073e-01 -3.11056882e-01 6.57220542e-01
-1.57448411e+00 -6.28323913e-01 -2.30490506e-01 7.49849081e-01
9.24614370e-02 -1.70998229e-03 5.93728900e-01 5.59470952e-01
-3.77953112e-01 4.11575764e-01 2.38802303e-02 1.00043096e-01
3.29618454e-01 -1.35831368e+00 2.66805857e-01 7.30509818e-01
3.64442825e-01 3.94930780e-01 7.42377043e-01 -4.98288572e-01
-1.58569121e+00 -8.89442742e-01 6.19136989e-01 -1.85824130e-02
1.03872275e+00 -3.00334632e-01 -7.70223200e-01 3.78532320e-01
-2.69839317e-01 6.63884163e-01 4.90397483e-01 -1.86119508e-02
-1.52972266e-01 -1.51594758e-01 -1.37616992e+00 3.72924626e-01
1.25411749e+00 -4.33483362e-01 -8.69641975e-02 3.10308278e-01
4.05055672e-01 1.00440890e-01 -1.28677154e+00 1.63408712e-01
6.02099597e-01 -1.05204237e+00 9.65091884e-01 -8.58486235e-01
3.99849117e-02 -3.20293754e-01 -1.45154685e-01 -1.37946939e+00
-3.27106237e-01 -7.76440680e-01 -1.72391459e-01 9.69433427e-01
1.76856160e-01 -7.67571747e-01 7.86550224e-01 4.96935606e-01
3.57872188e-01 -6.48587704e-01 -1.18299425e+00 -1.09759760e+00
-1.79501936e-01 -1.20963410e-01 5.39691985e-01 9.32304144e-01
-9.45699885e-02 2.73184568e-01 -3.09807390e-01 -4.03503291e-02
9.94680524e-01 1.51963696e-01 5.02332747e-01 -1.90027392e+00
-1.80712089e-01 -8.35208118e-01 -1.24695098e+00 -8.65112245e-01
4.11124140e-01 -9.91417050e-01 -2.94069856e-01 -1.09462035e+00
1.23680029e-02 -8.33499610e-01 -3.58900636e-01 1.12299345e-01
2.81300277e-01 1.64123982e-01 4.96249229e-01 -5.15673943e-02
-5.67363262e-01 4.58736926e-01 1.22952867e+00 1.86079275e-02
-2.94158995e-01 2.02047944e-01 -3.37160110e-01 4.71589297e-01
7.08349645e-01 -5.90571404e-01 -4.88760799e-01 -1.05038278e-01
2.67916292e-01 1.80764601e-01 5.02757549e-01 -8.76337707e-01
3.03527296e-01 -2.05808744e-01 5.79902111e-03 -5.48297015e-04
4.06651616e-01 -1.17016041e+00 1.17683060e-01 5.42063594e-01
-4.02076513e-01 8.95634443e-02 -2.15262234e-01 7.41507888e-01
-2.37008259e-01 -2.66691297e-01 7.43994594e-01 8.96404237e-02
-4.50436741e-01 5.54153442e-01 1.48039728e-01 1.72764614e-01
1.08226168e+00 -2.90874988e-01 -3.34515870e-01 -2.25864559e-01
-6.73149347e-01 1.30460724e-01 5.26395857e-01 4.18212026e-01
3.55408579e-01 -1.45501018e+00 -5.16874671e-01 2.60166317e-01
2.35447884e-02 -1.10623628e-01 -1.56970955e-02 1.17120636e+00
-1.63013682e-01 3.70450795e-01 -5.13003707e-01 -7.28568614e-01
-1.37442744e+00 4.16901886e-01 5.19785225e-01 -3.75895262e-01
-7.80758739e-01 3.71577442e-01 1.16610728e-01 9.91669446e-02
9.89620090e-02 -3.56823951e-01 -1.74230814e-01 4.76025075e-01
3.01780105e-01 6.61383688e-01 3.42774652e-02 -8.32280457e-01
-4.96589780e-01 7.40422606e-01 4.40878309e-02 -8.69484618e-02
1.33439803e+00 -1.31132305e-01 2.03725533e-03 4.35753047e-01
1.45223045e+00 1.90918576e-02 -1.49539232e+00 -3.57836902e-01
3.77457350e-01 -7.54846632e-01 7.35169575e-02 -1.32360116e-01
-1.55678785e+00 6.56626582e-01 4.76564556e-01 8.24990511e-01
9.23039377e-01 2.51072720e-02 4.01678324e-01 5.54017425e-01
1.50129780e-01 -7.96579242e-01 -3.57257575e-02 2.70215333e-01
5.23560524e-01 -1.27012837e+00 3.12308483e-02 -4.73090857e-01
-3.82900298e-01 1.51102090e+00 1.99062645e-01 -2.58539200e-01
7.82132387e-01 2.63739437e-01 -2.55965263e-01 -3.41514766e-01
-5.24405777e-01 -3.59362274e-01 4.48080808e-01 5.41471183e-01
2.75016904e-01 -2.59633875e-03 -4.01808679e-01 -2.49029830e-01
-7.23307133e-02 -2.43902475e-01 4.51096416e-01 7.15021074e-01
-2.76584178e-01 -1.13846529e+00 -1.17432795e-01 6.87459230e-01
-3.11084688e-01 2.97886848e-01 -2.48362616e-01 8.39695096e-01
-1.92364365e-01 7.30815470e-01 1.58303633e-01 -1.58873320e-01
3.06149334e-01 -8.80261660e-02 5.44063747e-01 -2.50851750e-01
-6.02139056e-01 -1.64747670e-01 -7.84557220e-03 -7.42841125e-01
-5.93904495e-01 -8.60660732e-01 -1.04020071e+00 -3.29326868e-01
-2.08880588e-01 7.08783269e-02 6.84606314e-01 8.55510056e-01
3.50398749e-01 5.08495688e-01 5.59424222e-01 -5.76631188e-01
-8.66622686e-01 -4.60968763e-01 -7.96265185e-01 6.49468064e-01
6.47760689e-01 -9.36438680e-01 -5.73953092e-01 -2.98059791e-01] | [7.5842604637146, 4.301497936248779] |
41de14d6-1f48-4a99-b30a-fe2fdce7db3a | bias-threat-and-aggression-identification | null | null | https://aclanthology.org/2022.trac-1.4 | https://aclanthology.org/2022.trac-1.4.pdf | Bias, Threat and Aggression Identification Using Machine Learning Techniques on Multilingual Comments | In this paper, we presented our team "IIITRanchi” for the Trolling, Aggression and Cyberbullying (TRAC-3) 2022 shared tasks. Aggression and its different forms on social media and other platforms had tremendous growth on the Internet. In this work we have tried upon different aspects of aggression, aggression intensity, bias of different forms and their usage online and its identification using different Machine Learning techniques. We have classified each sample at seven different tasks namely aggression level, aggression intensity, discursive role, gender bias, religious bias, caste/class bias and ethnicity/racial bias as specified in the shared tasks. Both of our teams tried machine learning classifiers and achieved the good results. Overall, our team "IIITRanchi” ranked first position in this shared tasks competition. | ['Rajiv Ranjan Suman', 'Shaury Srivastav', 'Kirti Kumari'] | null | null | null | null | trac-coling-2022-10 | ['aggression-identification'] | ['natural-language-processing'] | [-9.93076444e-01 -1.45456284e-01 -1.36686012e-01 -1.62668541e-01
-4.77732897e-01 -1.49209380e-01 5.73487103e-01 2.54381835e-01
-7.47507274e-01 8.70059609e-01 3.57938081e-01 3.03449072e-02
-3.91920954e-01 -3.54677975e-01 1.50578216e-01 -4.50576067e-01
-1.16899729e-01 7.56810665e-01 2.27034390e-01 -7.78669357e-01
1.02571130e+00 3.02814871e-01 -1.43390965e+00 3.85760009e-01
6.14182711e-01 4.04504001e-01 -7.41020441e-01 8.01847935e-01
-4.75041308e-02 1.19776487e+00 -8.85662854e-01 -1.03919137e+00
8.20627362e-02 -1.36417970e-01 -1.27582109e+00 -7.32040584e-01
8.77212882e-01 -3.65969926e-01 -2.29767174e-01 1.10171545e+00
6.60474122e-01 4.10757452e-01 8.45635474e-01 -1.52320671e+00
5.87474694e-03 7.44175851e-01 -1.01364172e+00 7.61213481e-01
6.96636915e-01 1.39815891e-02 7.27994025e-01 -6.09025478e-01
7.42587686e-01 1.61289227e+00 8.65153909e-01 9.86178815e-01
-9.05863702e-01 -1.27828979e+00 -3.18478465e-01 7.05532789e-01
-1.22098970e+00 -3.61396164e-01 6.11841917e-01 -9.11474884e-01
6.99171484e-01 6.74499691e-01 5.48040986e-01 1.79982972e+00
-3.02171241e-02 2.93014914e-01 1.43147457e+00 1.23238832e-01
-1.33031934e-01 2.45720670e-01 7.48890460e-01 2.75083482e-01
5.92740849e-02 -2.32635774e-02 -8.46366048e-01 -5.42601526e-01
1.35768607e-01 -6.96030796e-01 4.94354188e-01 6.06697142e-01
-1.48742795e-01 1.10856044e+00 1.64888501e-01 5.28877378e-01
8.31514746e-02 9.06097740e-02 1.23989582e+00 2.25555792e-01
6.71754658e-01 5.44601262e-01 1.59160178e-02 -9.10994530e-01
-4.77211416e-01 7.01396883e-01 5.37762463e-01 9.45229530e-02
8.25632140e-02 -2.10050225e-01 -2.28463158e-01 1.69180751e+00
7.15925321e-02 2.35127598e-01 6.91221237e-01 -9.23183084e-01
6.60849154e-01 4.42912906e-01 -3.25132668e-01 -1.33993423e+00
-1.01211882e+00 4.86973114e-02 -4.55904275e-01 5.51568031e-01
7.14516938e-01 -4.20285463e-01 -3.68768334e-01 1.61010289e+00
5.33168256e-01 -9.69811750e-04 -8.50382328e-01 9.82266665e-01
8.23945761e-01 1.45692095e-01 4.94882762e-01 2.50362724e-01
1.22516870e+00 -6.05933726e-01 -7.18637049e-01 2.58053184e-01
9.23445702e-01 -7.16001034e-01 7.61384487e-01 1.06042159e+00
-1.23110604e+00 -6.26485124e-02 -5.55103660e-01 -3.20061557e-02
-8.14530373e-01 -5.51557660e-01 5.54993033e-01 1.49402773e+00
-7.50772357e-01 5.87947488e-01 -3.01650912e-01 -5.26972950e-01
3.86321157e-01 4.40565228e-01 -7.06753969e-01 6.76580071e-01
-9.98042822e-01 1.20555532e+00 -9.47819352e-02 -6.05370343e-01
-6.00330174e-01 -6.95226729e-01 -3.57791036e-01 -5.85343182e-01
2.21267462e-01 -7.14251101e-02 9.24457610e-01 -6.10062599e-01
-1.61180544e+00 1.52046382e+00 5.59079587e-01 -2.09515423e-01
6.67830288e-01 -3.34006310e-01 -6.29438221e-01 -4.29593354e-01
2.28473783e-01 -4.56619896e-02 5.56386292e-01 -7.34306872e-01
-8.66275012e-01 -9.52973366e-01 8.09037685e-02 -9.07810703e-02
-4.11319464e-01 1.22963464e+00 5.07136881e-01 -4.31607604e-01
-3.23845804e-01 -7.97370255e-01 1.94416463e-01 -4.53017771e-01
-6.20958447e-01 -7.42404521e-01 7.15646505e-01 -6.39722347e-01
1.62183857e+00 -2.02944112e+00 4.86726165e-01 2.05555663e-01
5.85288465e-01 4.95532572e-01 5.29186785e-01 7.89162874e-01
-2.08004892e-01 6.36372685e-01 4.51170832e-01 -4.05910343e-01
1.38950154e-01 7.89091662e-02 7.21195117e-02 7.02349424e-01
-5.59659004e-01 1.50514126e-01 -7.14872539e-01 -6.15631223e-01
1.44041136e-01 -1.74298197e-01 -7.27587283e-01 9.23892111e-03
6.22184038e-01 5.54765761e-01 -2.66882330e-01 4.15351033e-01
7.88405120e-01 7.97317684e-01 -3.38008940e-01 4.15351033e-01
-5.13855338e-01 3.39157283e-01 -7.80099034e-01 9.18993533e-01
-4.27857816e-01 7.70801902e-01 5.84396720e-01 -5.61205268e-01
8.74390960e-01 -1.55644976e-02 4.60450530e-01 -7.75269687e-01
5.48675001e-01 6.73037946e-01 6.39424026e-01 -9.19574678e-01
5.48548639e-01 -1.49947211e-01 -2.47865424e-01 4.97601449e-01
1.92274094e-01 4.68443446e-02 3.69588971e-01 1.41256377e-01
1.25929570e+00 -4.40183133e-01 2.29907542e-01 -4.59622175e-01
7.38990664e-01 -1.77782103e-01 3.11937064e-01 6.77261233e-01
-7.52162158e-01 3.68373394e-01 9.46602404e-01 -5.82530439e-01
-8.87561858e-01 -6.45273864e-01 -4.51296866e-01 1.65074182e+00
-3.55113626e-01 -6.65929615e-01 -6.18332326e-01 -8.86526108e-01
8.34344234e-03 7.53790498e-01 -9.90017951e-01 -3.42572659e-01
-6.73788309e-01 -5.47928095e-01 1.35313749e+00 -1.01373449e-01
3.77809793e-01 -6.86549723e-01 -1.05159640e-01 -3.09212804e-01
1.91555142e-01 -7.45232999e-01 -1.52078882e-01 -3.92622173e-01
-3.08523215e-02 -1.42048287e+00 -4.56668139e-01 1.81695390e-02
-9.24770311e-02 -1.77880660e-01 6.53946400e-01 4.18233901e-01
-5.14728129e-01 -6.05506189e-02 -7.33521044e-01 -6.81305766e-01
-2.19402969e-01 2.22368285e-01 3.23894769e-01 1.11660957e-01
4.73301679e-01 -7.21397519e-01 -2.58482367e-01 3.98575485e-01
-1.71895787e-01 -2.62169182e-01 -3.82669836e-01 4.24943626e-01
-1.02449739e+00 -4.59843099e-01 2.35513762e-01 -1.03621101e+00
1.11986732e+00 -9.51198637e-01 -1.03535898e-01 -7.18281865e-01
-9.92240906e-02 -9.19726908e-01 2.06177279e-01 -1.19981386e-01
-7.70475149e-01 -1.01892519e+00 -7.38647938e-01 1.27519341e-02
-3.62137765e-01 1.12273708e-01 4.04576987e-01 -1.18980162e-01
1.01619852e+00 -5.57408810e-01 9.81833227e-03 -7.97217488e-01
-4.89988387e-01 1.02985299e+00 -1.01798803e-01 -6.69005334e-01
7.64181376e-01 1.85715556e-01 1.10435590e-01 -9.54714060e-01
-8.72170925e-01 -6.67556584e-01 -4.87898380e-01 -7.26861715e-01
1.22749555e+00 -4.33900654e-01 -1.34409535e+00 1.06296086e+00
-1.01606560e+00 -2.25519583e-01 3.00479084e-01 5.15905380e-01
-9.29207727e-02 1.29240185e-01 -6.86783493e-01 -1.36969936e+00
-2.35540703e-01 -7.97817767e-01 6.25245810e-01 7.53772026e-03
-7.42386341e-01 -8.57255936e-01 7.48519421e-01 8.57794285e-01
4.71880317e-01 2.62861431e-01 8.22869837e-01 -1.34440899e+00
8.87201965e-01 -8.22347030e-02 -1.93131402e-01 1.82795092e-01
-1.64651111e-01 4.65194046e-01 -6.63495123e-01 3.11590821e-01
-3.21611285e-01 -7.70726621e-01 5.31985879e-01 -1.98320463e-01
1.32269526e+00 -2.77017862e-01 3.52978855e-02 2.83627778e-01
7.33340800e-01 3.18769157e-01 5.81551254e-01 7.04986393e-01
7.30970740e-01 8.89475405e-01 4.54775244e-01 8.95194888e-01
1.59749895e-01 1.27868402e+00 6.57316864e-01 4.28038836e-01
-2.56755352e-01 -5.02836742e-02 1.81953847e-01 5.68232298e-01
-8.13035190e-01 1.26673788e-01 -1.38006473e+00 3.28885466e-01
-1.69407690e+00 -1.30758560e+00 -8.74339342e-01 1.99705923e+00
3.82891387e-01 -8.73265117e-02 1.34652734e+00 3.66780192e-01
7.90115416e-01 2.08915100e-01 8.36387947e-02 -1.65523672e+00
1.16736837e-01 3.75474215e-01 3.90502304e-01 6.18320942e-01
-8.81199598e-01 7.75758028e-01 6.49197340e+00 1.42791724e+00
-1.01081753e+00 4.85472560e-01 4.87692952e-01 -5.64524651e-01
4.46604043e-01 -4.17184114e-01 -8.42471540e-01 7.47197092e-01
9.50406671e-01 -2.51913428e-01 2.98287809e-01 6.44353986e-01
2.53227562e-01 -2.24769458e-01 -5.55004835e-01 1.13867736e+00
2.72320986e-01 -7.56372452e-01 -2.23458320e-01 4.94921237e-01
3.85595053e-01 3.95057127e-02 2.98285395e-01 7.79852390e-01
5.24241507e-01 -1.31884480e+00 5.37203312e-01 2.63744831e-01
2.19222426e-01 -7.38191187e-01 7.61399567e-01 2.80968994e-01
-1.59893677e-01 -6.22029305e-01 -7.93066174e-02 -6.37322009e-01
-1.20520838e-01 -2.75408514e-02 -3.61631781e-01 4.30689692e-01
1.02719736e+00 5.62052250e-01 -6.74166143e-01 1.16833746e+00
1.77489981e-01 8.08440208e-01 -4.12868828e-01 -4.30952698e-01
4.14656073e-01 -3.88282359e-01 1.05632389e+00 9.78907466e-01
-6.16367199e-02 1.13017410e-01 -1.43741846e-01 1.86720401e-01
4.20853138e-01 6.91070974e-01 -5.33076823e-01 2.81513840e-01
3.94825041e-02 1.30160546e+00 -1.37745604e-01 8.74918699e-02
-4.77238111e-02 2.05779463e-01 6.80696428e-01 -4.21278596e-01
-1.03514588e+00 -3.50645110e-02 1.09316587e+00 5.81812739e-01
-5.64557552e-01 -7.79564455e-02 -6.12394571e-01 -7.91713953e-01
-5.86464107e-01 -1.01858008e+00 8.35098803e-01 -3.53784740e-01
-1.72900999e+00 2.76100636e-01 3.78097445e-01 -9.88620520e-01
4.80592400e-02 -8.37487578e-01 -9.58530009e-01 4.75753307e-01
-3.10444951e-01 -9.39362347e-01 -7.41776093e-05 6.78811789e-01
6.36441946e-01 -7.95267582e-01 3.76597524e-01 7.90542960e-01
-1.33502102e+00 7.63406515e-01 -2.35520914e-01 2.65912384e-01
9.49827135e-01 -1.17260420e+00 -4.23387408e-01 -1.29279166e-01
-4.20597643e-01 3.64928693e-01 9.65601802e-01 -5.85348725e-01
-6.94213033e-01 -3.50307763e-01 4.79008555e-01 -5.58403492e-01
1.60320330e+00 -4.98744875e-01 -3.69257361e-01 3.84236366e-01
4.16206300e-01 -7.31549740e-01 1.26487529e+00 1.00395513e+00
-6.59488678e-01 1.18325762e-01 -1.31165457e+00 7.31024742e-01
1.39409137e+00 -1.50176004e-01 -4.71656621e-01 8.70137930e-01
-3.97742480e-01 -5.26857436e-01 -6.65136397e-01 -2.03115910e-01
5.98382175e-01 -1.62776148e+00 4.68526959e-01 -1.39137435e+00
1.08295596e+00 7.82443762e-01 5.08548021e-02 -1.23813212e+00
-2.88637847e-01 -4.48565781e-01 6.73059523e-01 1.28397453e+00
2.84294933e-01 -9.28109646e-01 6.96299493e-01 9.69157815e-01
-2.93024510e-01 -5.17359316e-01 -1.80269110e+00 -6.66739166e-01
8.90419304e-01 -6.13493562e-01 2.68171430e-01 1.19027734e+00
6.63762212e-01 3.06375235e-01 -7.40724623e-01 -5.15689671e-01
3.61639678e-01 -7.96974301e-01 1.45420098e+00 -1.33986545e+00
-4.97792624e-02 -1.20832610e+00 -1.01593292e+00 9.29056481e-02
1.92320064e-01 -7.59919167e-01 -1.01039290e+00 -8.41283739e-01
3.84746820e-01 -3.91935080e-01 1.12713590e-01 4.13725853e-01
2.03350917e-01 5.30499160e-01 3.51905733e-01 -8.56796578e-02
-5.92992663e-01 2.68248767e-01 8.55984807e-01 3.38245323e-03
7.94690698e-02 -4.11497727e-02 -4.76395607e-01 1.03259671e+00
9.24336493e-01 -6.31412983e-01 2.62103409e-01 -1.67921305e-01
4.67539459e-01 -7.87481740e-02 2.52217054e-01 -1.16638935e+00
5.20285629e-02 -6.15693450e-01 -2.21528396e-01 -1.64358273e-01
9.98355448e-01 -4.43621069e-01 -1.79115608e-01 5.04616797e-01
-2.17617035e-01 -6.96493685e-02 5.12396917e-02 -2.64076710e-01
-5.54569857e-03 -6.00358963e-01 7.68917859e-01 -7.39583522e-02
1.80680435e-02 1.61694899e-01 -7.80092418e-01 2.89391220e-01
1.33509111e+00 -1.89348236e-01 -7.62751102e-01 -4.68363732e-01
-1.02657390e+00 2.14744434e-01 9.02335942e-02 6.07851565e-01
1.13102451e-01 -9.64880645e-01 -1.28545392e+00 -2.46991903e-01
4.20482941e-02 -1.27777708e+00 4.90170807e-01 1.41038001e+00
-5.49605310e-01 1.28446907e-01 -9.18606341e-01 -8.43631402e-02
-1.93495047e+00 9.63434055e-02 5.09890556e-01 -2.93214113e-01
-3.05580981e-02 8.50601971e-01 -3.15865964e-01 -6.14527225e-01
-2.14204013e-01 9.92414474e-01 -8.30148220e-01 3.58095556e-01
5.65150678e-01 1.53574693e+00 -1.08993791e-01 -1.41771460e+00
-3.38369638e-01 5.77771008e-01 -3.89561892e-01 -3.39527518e-01
1.10649729e+00 2.62205839e-01 -5.59444368e-01 7.22362101e-01
1.23256052e+00 6.73280478e-01 2.34048083e-01 5.47160387e-01
1.62946016e-01 -1.38929486e+00 -1.52333364e-01 -7.51730442e-01
-1.00047398e+00 6.97482646e-01 3.54991287e-01 8.94602478e-01
2.89639026e-01 -1.69004217e-01 6.00803912e-01 -1.42019585e-01
5.57167351e-01 -1.48251057e+00 1.08145192e-01 8.61909747e-01
1.28754318e+00 -1.09763610e+00 9.76300240e-02 -4.90150928e-01
-6.01772666e-01 8.19073617e-01 1.30218446e+00 -5.22174299e-01
7.63514280e-01 -1.02605395e-01 2.05038041e-01 -6.47000611e-01
-6.08969331e-01 -6.39477596e-02 2.30592072e-01 8.46939623e-01
4.56998557e-01 1.11789159e-01 -1.51057410e+00 5.43920457e-01
-8.97192001e-01 -6.10024333e-01 6.35186255e-01 2.85010010e-01
-1.76853374e-01 -1.27617085e+00 -7.20564365e-01 6.89600348e-01
-9.75864768e-01 4.34958279e-01 -1.29129910e+00 1.16816521e+00
4.04081374e-01 1.20003200e+00 -6.89729080e-02 -8.84628654e-01
4.86691415e-01 -1.25156686e-01 2.72220105e-01 -6.74304247e-01
-1.65119731e+00 -4.87606138e-01 1.09254968e+00 -7.70594180e-01
2.43068829e-01 -9.74621236e-01 -6.59949899e-01 -9.13697064e-01
2.76833139e-02 2.16632962e-01 9.43153858e-01 7.75379777e-01
-7.62729347e-02 1.54688861e-02 5.79919577e-01 -8.58647287e-01
-4.07250762e-01 -1.36978662e+00 -9.17616010e-01 9.39839482e-01
-9.67069790e-02 -1.34091699e+00 -7.61173964e-01 -1.26914012e+00] | [8.789952278137207, 10.748431205749512] |
c2fe5dcd-eff2-4035-9386-81ef62fb7462 | mining-negative-temporal-contexts-for-false | 2305.18060 | null | https://arxiv.org/abs/2305.18060v1 | https://arxiv.org/pdf/2305.18060v1.pdf | Mining Negative Temporal Contexts For False Positive Suppression In Real-Time Ultrasound Lesion Detection | During ultrasonic scanning processes, real-time lesion detection can assist radiologists in accurate cancer diagnosis. However, this essential task remains challenging and underexplored. General-purpose real-time object detection models can mistakenly report obvious false positives (FPs) when applied to ultrasound videos, potentially misleading junior radiologists. One key issue is their failure to utilize negative symptoms in previous frames, denoted as negative temporal contexts (NTC). To address this issue, we propose to extract contexts from previous frames, including NTC, with the guidance of inverse optical flow. By aggregating extracted contexts, we endow the model with the ability to suppress FPs by leveraging NTC. We call the resulting model UltraDet. The proposed UltraDet demonstrates significant improvement over previous state-of-the-arts and achieves real-time inference speed. To facilitate future research, we will release the code, checkpoints, and high-quality labels of the CVA-BUS dataset used in our experiments. | ['LiWei Wang', 'Dong Wang', 'Dengbo Chen', 'Ziwei Zhao', 'Quanlin Wu', 'Youcheng Li', 'Haojun Yu'] | 2023-05-29 | null | null | null | null | ['real-time-object-detection'] | ['computer-vision'] | [ 5.43210983e-01 2.06847563e-01 -2.70087183e-01 -2.00466871e-01
-1.06344771e+00 -3.87159556e-01 3.38160455e-01 -1.57830492e-01
-3.54688674e-01 7.05357552e-01 1.26596196e-02 -6.41677380e-01
-7.18386769e-02 -2.95528233e-01 -5.71417451e-01 -6.73978925e-01
-2.23216996e-01 8.10119808e-02 5.47986805e-01 4.52309281e-01
2.39398643e-01 3.93922716e-01 -1.28367507e+00 4.58169103e-01
6.04182065e-01 8.10135841e-01 3.31409633e-01 1.07816303e+00
1.88657761e-01 1.05719435e+00 -4.38637078e-01 -2.73246288e-01
6.29760176e-02 -3.38647008e-01 -6.76179111e-01 2.90066719e-01
5.22438228e-01 -8.17845523e-01 -3.97094876e-01 7.80565679e-01
5.08319497e-01 -2.74405908e-02 3.61984611e-01 -1.11441839e+00
-2.08992809e-01 3.00265849e-01 -6.86394691e-01 7.84591436e-01
3.40119958e-01 5.53796172e-01 5.56844950e-01 -8.59134018e-01
7.42702901e-01 1.00813997e+00 5.67982912e-01 7.41333187e-01
-9.85127151e-01 -6.35424435e-01 1.84582487e-01 3.09949547e-01
-1.13854182e+00 -6.15144312e-01 3.63081694e-01 -2.41709024e-01
7.20307767e-01 5.60430944e-01 7.18202889e-01 1.21749675e+00
3.51500541e-01 9.72372413e-01 9.46880639e-01 -3.11336130e-01
1.72101468e-01 -2.46293738e-01 4.00069691e-02 1.11144316e+00
3.73406917e-01 2.91254818e-01 -5.58440089e-01 -2.58920908e-01
9.72095549e-01 3.57957110e-02 -5.08818030e-01 -8.07550326e-02
-1.43532956e+00 5.10757029e-01 1.23306431e-01 3.10935173e-02
-4.35568988e-01 4.08633649e-01 2.09421039e-01 -1.48781925e-01
1.10608377e-01 1.49699762e-01 4.06263098e-02 -5.93846917e-01
-8.06714475e-01 -5.88039346e-02 6.26198590e-01 7.38986492e-01
3.02548915e-01 -1.28562544e-02 -5.15987813e-01 3.25127721e-01
1.29672128e-03 4.59021419e-01 5.24414539e-01 -1.37528026e+00
-6.81899786e-02 1.97090894e-01 2.27328867e-01 -9.14057732e-01
-5.09400547e-01 -6.29962564e-01 -5.28282166e-01 -1.16269715e-01
4.79816318e-01 -1.17987446e-01 -1.22508037e+00 1.43032849e+00
5.69739938e-01 9.62844253e-01 -1.66210845e-01 1.06598175e+00
6.15475595e-01 2.18542412e-01 1.06373653e-01 -4.90779847e-01
1.63091004e+00 -8.19794774e-01 -8.50703061e-01 -2.24675953e-01
7.42286921e-01 -7.60874450e-01 9.84469950e-01 6.34775877e-01
-1.13929784e+00 -3.11712831e-01 -8.88848066e-01 1.64696053e-01
3.18344325e-01 2.42717415e-01 7.66241848e-01 7.11838067e-01
-8.78146648e-01 3.09639275e-01 -1.63291669e+00 -1.85213864e-01
5.55378020e-01 2.71858454e-01 -5.92388436e-02 -4.14949983e-01
-8.93035412e-01 7.32153714e-01 4.45745699e-03 2.38104597e-01
-1.07691956e+00 -9.00613010e-01 -7.85802662e-01 -2.64392167e-01
9.69100654e-01 -6.10380590e-01 1.70692146e+00 -4.30749834e-01
-1.09376633e+00 6.29919052e-01 -4.88970220e-01 -5.79274833e-01
8.31743360e-01 -3.54171097e-01 -4.34931129e-01 7.55734503e-01
1.07839309e-01 7.18439758e-01 9.22459126e-01 -1.04898548e+00
-9.40422893e-01 -3.07600554e-02 4.09847349e-02 -9.24554095e-02
-1.12975031e-01 -2.06031263e-01 -9.74973142e-01 -6.53112531e-01
2.65515596e-01 -1.21269679e+00 -3.74238163e-01 5.26012301e-01
-4.83194262e-01 2.11565420e-02 9.43784952e-01 -5.34040153e-01
1.38762963e+00 -2.26218629e+00 -5.21573246e-01 8.01693723e-02
4.30930436e-01 2.77150273e-01 2.04734251e-01 -4.56742018e-01
7.67562836e-02 -5.39056119e-03 -1.08551957e-01 -3.30811560e-01
-5.41171551e-01 5.85254967e-01 -4.34856527e-02 6.54888570e-01
3.92153502e-01 9.14645731e-01 -1.45073080e+00 -1.11065137e+00
5.11833131e-01 3.06785971e-01 -4.96727258e-01 -8.22993591e-02
2.41447184e-02 6.29017770e-01 -7.29460537e-01 7.98091233e-01
5.84941089e-01 -6.28398180e-01 1.65925115e-01 -4.12515730e-01
-2.88380329e-02 2.35088542e-01 -1.13551962e+00 1.53523934e+00
-2.90080279e-01 6.75041437e-01 1.60763022e-02 -7.78346181e-01
1.08909413e-01 4.30710018e-01 7.43833005e-01 -6.06279850e-01
4.36313450e-03 1.87140644e-01 2.28693932e-01 -8.56268466e-01
4.82256353e-01 7.61114955e-02 4.00769949e-01 4.46045220e-01
-3.74289066e-01 1.31316055e-02 3.93102288e-01 4.23293233e-01
1.44465709e+00 2.92989463e-02 2.52729505e-01 9.76716429e-02
3.73085111e-01 9.58163813e-02 6.86612189e-01 1.33320606e+00
-6.47855222e-01 7.91607380e-01 5.28558195e-01 -3.35914642e-01
-5.88658392e-01 -1.04307175e+00 -2.46658266e-01 5.88177383e-01
1.57973990e-01 -2.50094265e-01 -3.40204954e-01 -1.02919936e+00
-1.45189926e-01 6.26822829e-01 -6.56070828e-01 -6.82178661e-02
-9.97500479e-01 -5.86447060e-01 5.19632459e-01 5.78934491e-01
1.24575645e-01 -8.54169607e-01 -1.22647738e+00 4.92146343e-01
-5.68075240e-01 -1.48032773e+00 -4.72408950e-01 -6.82816654e-02
-9.24651325e-01 -1.22745228e+00 -8.08119178e-01 -3.68515670e-01
6.89975679e-01 6.69983923e-01 1.33093596e+00 4.39742088e-01
-8.25301349e-01 5.67590833e-01 -3.77724051e-01 -3.92581195e-01
-5.03506958e-01 -2.06597671e-01 -2.98387617e-01 -1.79318279e-01
9.72084254e-02 -2.74162799e-01 -1.01872313e+00 3.76682103e-01
-1.04145384e+00 4.10311460e-01 7.56056368e-01 9.27013516e-01
6.81931615e-01 -3.53960246e-01 3.73328209e-01 -1.12211049e+00
9.39427614e-02 -2.10037947e-01 -3.86202991e-01 2.92687297e-01
-2.92582572e-01 6.75004497e-02 2.72968382e-01 -5.17515004e-01
-1.03670001e+00 1.82884142e-01 -1.97551567e-02 -6.12202108e-01
-7.39416331e-02 2.74857610e-01 7.24004447e-01 -1.45566300e-01
3.85562330e-01 2.50417948e-01 -6.32124692e-02 7.70316944e-02
1.28086954e-01 3.99965644e-01 8.84056449e-01 -4.32723701e-01
3.02457660e-01 9.39594746e-01 1.29952118e-01 -7.16573834e-01
-1.15544617e+00 -6.70356274e-01 -1.65655196e-01 -5.97521603e-01
6.78782225e-01 -8.04126322e-01 -6.26653910e-01 -3.73610407e-02
-1.01989651e+00 -8.27931166e-02 -1.68969706e-01 7.28641212e-01
-4.71985400e-01 6.06934071e-01 -6.33299530e-01 -1.03084874e+00
-1.14671588e-01 -1.35323691e+00 1.19630742e+00 2.89681852e-01
-2.63147622e-01 -5.86901605e-01 -3.77400726e-01 2.59801298e-01
2.21906453e-01 2.19859764e-01 5.12274265e-01 -2.31014237e-01
-8.18951666e-01 -8.93125609e-02 -3.11682165e-01 -1.07163668e-01
2.16167331e-01 1.85883507e-01 -1.06131518e+00 -2.71580905e-01
6.78998306e-02 -1.13190904e-01 1.00518405e+00 6.86526954e-01
1.57647634e+00 4.61092852e-02 -6.55016840e-01 4.76993412e-01
9.13724959e-01 1.78514600e-01 4.61819470e-01 2.19897017e-01
4.69247997e-01 1.74828812e-01 9.14213598e-01 6.70478106e-01
2.08875924e-01 4.32943851e-01 5.97499192e-01 -1.73827007e-01
-3.21451098e-01 2.14039102e-01 -1.81774516e-02 5.79454005e-01
-8.71928260e-02 -1.96700975e-01 -9.03302252e-01 8.66085947e-01
-1.82163036e+00 -7.22251713e-01 -1.30097687e-01 1.94567358e+00
8.13001990e-01 4.42376494e-01 -3.16432863e-01 6.75972030e-02
5.64570904e-01 5.74945472e-02 -5.71378410e-01 1.13564067e-01
2.13971883e-01 4.94317412e-02 6.26687586e-01 3.47818345e-01
-1.10186279e+00 4.26504076e-01 6.07973480e+00 6.53558612e-01
-1.25790846e+00 2.84459978e-01 7.97907352e-01 -2.73492575e-01
-5.14556617e-02 -4.88007534e-03 -5.67657828e-01 4.65143144e-01
7.92861581e-01 1.87619641e-01 -3.98358181e-02 6.07172132e-01
5.49125671e-01 -6.47727728e-01 -1.13220787e+00 8.36431801e-01
9.54438597e-02 -1.25981677e+00 -2.50901431e-01 -7.41834715e-02
5.27883291e-01 -2.13465214e-01 1.47486493e-01 2.52068967e-01
2.61259586e-01 -6.48362756e-01 4.73780930e-01 5.52115023e-01
8.46392095e-01 -3.00657749e-01 6.18047297e-01 1.04590178e-01
-9.92805123e-01 2.17340980e-02 -8.91221166e-02 2.94349700e-01
4.52967256e-01 7.46397436e-01 -1.37803638e+00 6.40173078e-01
5.59736133e-01 7.22758412e-01 -5.88711619e-01 1.28276205e+00
-6.12594821e-02 9.85543609e-01 -5.08728862e-01 2.40010634e-01
3.84258389e-01 3.65703344e-01 6.46238089e-01 1.27676475e+00
4.89705324e-01 3.17050099e-01 2.50477076e-01 5.30045033e-01
3.72361153e-01 -5.30118763e-01 -1.40910894e-01 5.16221747e-02
5.37094057e-01 1.20615518e+00 -8.91559184e-01 -7.71740794e-01
-6.19804442e-01 7.80893743e-01 -1.32544279e-01 3.44287902e-01
-1.13172877e+00 2.45548889e-01 1.87945604e-01 3.02870590e-02
3.26749325e-01 -1.33280858e-01 -2.01780245e-01 -1.19717979e+00
4.11823532e-03 -6.69301510e-01 6.14863634e-01 -8.08995545e-01
-9.32527959e-01 2.71195620e-01 -8.72449800e-02 -1.44073558e+00
-3.84878188e-01 -5.58875322e-01 -3.31298679e-01 2.34035373e-01
-1.68048370e+00 -7.63384283e-01 -5.61555505e-01 4.50730592e-01
7.93433547e-01 5.21433771e-01 4.88855213e-01 3.66253167e-01
-5.42056859e-01 4.86208916e-01 -1.94428295e-01 8.33708271e-02
8.76481712e-01 -1.10931301e+00 1.42709017e-01 1.04824710e+00
-8.87955949e-02 4.26224917e-01 8.51616740e-01 -6.97642624e-01
-1.47209859e+00 -1.04757631e+00 4.08628047e-01 -4.25833881e-01
4.98769522e-01 2.56496549e-01 -8.48704338e-01 6.35953486e-01
-3.07933956e-01 6.79532468e-01 3.93424898e-01 -4.74959224e-01
2.57975549e-01 1.66461065e-01 -9.99743164e-01 6.86298847e-01
9.49541390e-01 -2.44023353e-01 -4.17368501e-01 3.78058553e-01
7.32362151e-01 -9.41139817e-01 -4.84755546e-01 4.11685675e-01
8.65172565e-01 -9.62621450e-01 1.06664240e+00 -3.89002025e-01
3.59805316e-01 -2.10650682e-01 1.76212326e-01 -6.85482681e-01
2.62596994e-03 -5.86540043e-01 -6.67409599e-01 5.21782398e-01
1.65417656e-01 -3.55552256e-01 1.10306883e+00 5.98866463e-01
-2.34071717e-01 -7.64758170e-01 -1.10045052e+00 -4.74482685e-01
-5.33514559e-01 -8.24483097e-01 -3.16069387e-02 7.47351587e-01
-4.02541429e-01 -3.95025223e-01 -3.62088919e-01 4.31063443e-01
7.20731735e-01 5.06401248e-03 5.41346908e-01 -7.61148453e-01
-4.60365593e-01 -1.48490027e-01 -1.61829367e-01 -1.10990202e+00
-3.73008549e-01 -2.14319617e-01 2.56433517e-01 -1.28941202e+00
3.12296540e-01 -3.96006972e-01 -4.08884555e-01 4.37142223e-01
-5.58101058e-01 5.54889321e-01 1.99615061e-01 1.44511998e-01
-8.86211336e-01 -1.16523849e-02 1.78852904e+00 1.92614794e-02
4.93849367e-02 1.23467751e-01 -4.26054746e-01 7.47457504e-01
4.76107121e-01 -7.33999550e-01 -2.17222497e-01 -3.33677799e-01
-1.31831199e-01 4.71101463e-01 6.12828314e-01 -1.01829493e+00
3.15127045e-01 -2.90540159e-01 3.34442317e-01 -9.40540969e-01
5.39987147e-01 -7.00366855e-01 -1.83821663e-01 1.01840401e+00
-2.05219179e-01 4.32161475e-03 1.01710938e-01 8.05005252e-01
-7.17034116e-02 -3.13901573e-01 5.93780756e-01 -3.32443744e-01
-7.81412542e-01 3.12174596e-02 -6.92973614e-01 2.17211731e-02
1.03395569e+00 -2.66729563e-01 -3.32150936e-01 -4.90655690e-01
-8.37186277e-01 4.09330815e-01 2.01146424e-01 1.32595643e-01
7.73900092e-01 -8.63547444e-01 -6.82053983e-01 1.46625400e-01
9.57862586e-02 2.32548267e-01 6.14252985e-01 1.53944361e+00
-6.15033627e-01 4.47936386e-01 3.65220457e-01 -1.07720327e+00
-1.23697019e+00 3.89749885e-01 2.93309689e-01 -3.60967785e-01
-8.68003249e-01 6.80621147e-01 2.91195661e-01 1.95173442e-01
2.48085409e-01 -6.86428785e-01 1.69981252e-02 -3.11767071e-01
7.28565395e-01 3.15193146e-01 2.93671072e-01 -2.51755178e-01
-4.72466558e-01 1.34423107e-01 -5.05801976e-01 -2.26506945e-02
8.77881110e-01 -2.75087327e-01 5.24905384e-01 4.92664017e-02
7.44873822e-01 -7.09338933e-02 -1.59124660e+00 -2.47942358e-01
-4.51557077e-02 -6.73329234e-01 2.59451479e-01 -7.66568899e-01
-1.11767066e+00 7.07460046e-01 8.39533865e-01 5.76929040e-02
1.07736695e+00 -7.40258023e-02 8.65657389e-01 2.67677158e-01
1.72276154e-01 -7.10402668e-01 2.72452414e-01 1.54644147e-01
3.53364438e-01 -1.34290779e+00 2.69125700e-01 -7.12769151e-01
-5.29492438e-01 1.16872525e+00 6.87562883e-01 9.70668048e-02
3.21787596e-01 5.49336076e-01 7.60697275e-02 -2.83734165e-02
-1.05324543e+00 -1.34912685e-01 1.93249986e-01 5.23963630e-01
3.55900496e-01 7.53260916e-03 -3.61544281e-01 2.45750502e-01
1.94826141e-01 3.08520317e-01 1.02385151e+00 1.49472201e+00
-3.35692465e-01 -7.07145393e-01 -6.60887718e-01 6.50473177e-01
-8.89803946e-01 7.21841259e-03 2.99091190e-01 8.84093940e-01
-6.55643549e-03 8.76889944e-01 1.79635271e-01 1.54664353e-01
1.60061479e-01 -3.54424059e-01 5.00999272e-01 -4.50685740e-01
-1.39965504e-01 3.91684473e-01 1.41551450e-01 -8.64899635e-01
-6.12610638e-01 -7.41423011e-01 -1.40312839e+00 -4.53564711e-02
-5.34729004e-01 -2.34422341e-01 3.76282364e-01 9.20555711e-01
3.50869179e-01 9.08012986e-01 2.63604492e-01 -8.16080570e-01
-5.70917666e-01 -6.77520573e-01 -1.49678230e-01 4.74279225e-01
8.08765888e-01 -8.33819509e-01 -4.02605623e-01 3.61372173e-01] | [9.084856033325195, -0.0822000503540039] |
bbe617a1-a611-427e-9149-cb5283eba252 | lm-vc-zero-shot-voice-conversion-via-speech | 2306.10521 | null | https://arxiv.org/abs/2306.10521v1 | https://arxiv.org/pdf/2306.10521v1.pdf | LM-VC: Zero-shot Voice Conversion via Speech Generation based on Language Models | Language model (LM) based audio generation frameworks, e.g., AudioLM, have recently achieved new state-of-the-art performance in zero-shot audio generation. In this paper, we explore the feasibility of LMs for zero-shot voice conversion. An intuitive approach is to follow AudioLM - Tokenizing speech into semantic and acoustic tokens respectively by HuBERT and SoundStream, and converting source semantic tokens to target acoustic tokens conditioned on acoustic tokens of the target speaker. However, such an approach encounters several issues: 1) the linguistic content contained in semantic tokens may get dispersed during multi-layer modeling while the lengthy speech input in the voice conversion task makes contextual learning even harder; 2) the semantic tokens still contain speaker-related information, which may be leaked to the target speech, lowering the target speaker similarity; 3) the generation diversity in the sampling of the LM can lead to unexpected outcomes during inference, leading to unnatural pronunciation and speech quality degradation. To mitigate these problems, we propose LM-VC, a two-stage language modeling approach that generates coarse acoustic tokens for recovering the source linguistic content and target speaker's timbre, and then reconstructs the fine for acoustic details as converted speech. Specifically, to enhance content preservation and facilitates better disentanglement, a masked prefix LM with a mask prediction strategy is used for coarse acoustic modeling. This model is encouraged to recover the masked content from the surrounding context and generate target speech based on the target speaker's utterance and corrupted semantic tokens. Besides, to further alleviate the sampling error in the generation, an external LM, which employs window attention to capture the local acoustic relations, is introduced to participate in the coarse acoustic modeling. | ['Yuping Wang', 'Qiao Tian', 'Lei Xie', 'Yuanzhe Chen', 'Zhichao Wang'] | 2023-06-18 | null | null | null | null | ['voice-conversion', 'audio-generation', 'disentanglement', 'voice-conversion'] | ['audio', 'audio', 'methodology', 'speech'] | [ 2.43394732e-01 2.27526203e-01 -1.43341199e-02 3.49575616e-02
-1.23044825e+00 -2.74731666e-01 2.98577726e-01 -7.89234266e-02
-1.13175847e-02 6.45597041e-01 6.57316506e-01 2.06609797e-02
2.58019298e-01 -6.29978478e-01 -5.50214112e-01 -7.72805572e-01
2.52041072e-01 6.03452474e-02 1.25206262e-01 -9.85653177e-02
-6.68605715e-02 5.95838986e-02 -1.84563005e+00 4.28165674e-01
9.55024779e-01 9.47873712e-01 7.69282222e-01 7.69864380e-01
-6.03313208e-01 7.08207965e-01 -1.09088135e+00 -1.50826007e-01
4.70802672e-02 -7.31520116e-01 -4.36571181e-01 -6.67252615e-02
-1.72752223e-03 -1.54012352e-01 -4.83953357e-02 1.12431574e+00
7.63045430e-01 4.36409622e-01 6.22495651e-01 -9.38542843e-01
-4.05113637e-01 1.06758904e+00 -1.93224013e-01 -1.73722103e-01
3.79367262e-01 1.92819372e-01 9.51995432e-01 -1.25042701e+00
2.11061016e-01 1.63848388e+00 4.64853376e-01 6.65928185e-01
-1.22116709e+00 -9.04435098e-01 1.88975468e-01 2.63898700e-01
-1.47965097e+00 -9.51274037e-01 1.11869657e+00 -2.56388545e-01
6.10649049e-01 3.99943084e-01 5.96020520e-01 1.32309783e+00
-2.03826860e-01 6.41893089e-01 7.47412026e-01 -6.66877389e-01
3.04860711e-01 3.55870992e-01 -2.89548725e-01 1.46644682e-01
-6.12446427e-01 4.75052983e-01 -7.43756294e-01 -8.71908851e-03
3.43147129e-01 -5.77745318e-01 -2.74743468e-01 3.97516280e-01
-9.01973665e-01 5.19725442e-01 6.28011227e-02 3.72776359e-01
-4.04110610e-01 -4.54998091e-02 3.93235832e-01 1.62325412e-01
4.47549909e-01 2.95507401e-01 -1.86183721e-01 -2.93772340e-01
-1.22730184e+00 1.39914975e-01 6.23174787e-01 9.61278617e-01
7.44842589e-01 7.42553890e-01 -3.78591180e-01 1.23309052e+00
5.76415002e-01 5.71373463e-01 7.34989285e-01 -6.84576929e-01
4.61855114e-01 -2.63436157e-02 -7.35868737e-02 -6.65997028e-01
1.39801487e-01 -5.25390744e-01 -8.29191625e-01 1.16400249e-01
7.48152882e-02 -1.59297585e-01 -8.67672801e-01 1.95197582e+00
3.63744825e-01 5.97581446e-01 3.33890855e-01 8.36728871e-01
8.28659296e-01 1.10894990e+00 1.03935145e-01 -4.75578517e-01
1.25364590e+00 -1.00996304e+00 -1.00800681e+00 -3.02256376e-01
2.12117329e-01 -1.12618446e+00 1.44095874e+00 2.08005890e-01
-1.23369873e+00 -1.13329506e+00 -1.10381556e+00 8.55075121e-02
-1.42100751e-01 7.66780674e-02 -1.33225054e-01 6.76372409e-01
-5.90645254e-01 4.51255471e-01 -5.46278715e-01 1.28881037e-01
-2.13223342e-02 -9.59652141e-02 3.15559179e-01 1.10060029e-01
-1.58749080e+00 5.41976333e-01 3.78228754e-01 9.37494114e-02
-1.17462468e+00 -9.39267755e-01 -8.61654639e-01 3.03509206e-01
3.37590754e-01 -3.30725074e-01 1.27245736e+00 -1.08852506e+00
-1.95832312e+00 9.00984854e-02 -4.00375575e-01 -2.22780377e-01
3.12964499e-01 -4.32646647e-02 -7.54575551e-01 6.16301782e-04
1.01536177e-01 6.17613435e-01 1.24781990e+00 -1.37185705e+00
-6.27438843e-01 3.99374925e-02 -4.65886891e-01 3.96255463e-01
-3.38851571e-01 -1.21967413e-01 -2.37273753e-01 -1.19409633e+00
-4.46016863e-02 -5.90524495e-01 -3.16556431e-02 -3.69708180e-01
-5.53335428e-01 -8.72082859e-02 7.80502737e-01 -8.37964058e-01
1.43468070e+00 -2.67716050e+00 4.09532562e-02 1.27886727e-01
-3.26282263e-01 3.85595679e-01 -3.96027237e-01 3.01264316e-01
-1.11197610e-03 3.05765197e-02 -1.08056515e-01 -8.25522184e-01
2.41664629e-02 3.27848978e-02 -8.17751288e-01 -2.15055749e-01
2.40852669e-01 4.60098267e-01 -8.93042862e-01 -5.65938354e-01
2.67604917e-01 6.59940600e-01 -5.94899237e-01 5.18740058e-01
-2.97904372e-01 5.87277710e-01 -1.31446451e-01 3.99650604e-01
5.11558056e-01 6.30887270e-01 -1.33294806e-01 -2.88266063e-01
-1.52474284e-01 7.52281368e-01 -1.36643291e+00 1.55127430e+00
-7.74022520e-01 2.63042122e-01 3.85300636e-01 -6.67040348e-01
1.05447173e+00 7.75423825e-01 1.94794685e-01 -4.33087379e-01
-3.59789841e-02 2.69710451e-01 2.51711924e-02 -4.48097765e-01
5.87530375e-01 -4.80642229e-01 1.22832105e-01 2.71963954e-01
4.03163619e-02 -5.14511526e-01 -1.94842666e-01 -1.74323648e-01
4.36756551e-01 1.38201430e-01 -1.49426493e-03 3.76321077e-02
5.87032735e-01 -4.44968075e-01 7.53424585e-01 5.18877029e-01
-2.26524435e-02 7.47920334e-01 1.11319860e-02 3.92501533e-01
-8.58658075e-01 -1.37806237e+00 1.39292568e-01 1.09781289e+00
-5.95500767e-02 -4.48303252e-01 -1.05903792e+00 -5.89733943e-03
-3.86411041e-01 1.21871519e+00 3.64534347e-03 -5.71655333e-01
-5.72279274e-01 -2.80812830e-01 8.28548372e-01 2.90340900e-01
1.36857703e-01 -1.37154448e+00 -6.51542936e-03 7.17611611e-01
-6.25291646e-01 -8.78391385e-01 -7.55900979e-01 -4.45518922e-03
-5.63175082e-01 -3.67035300e-01 -6.16177678e-01 -7.98110068e-01
3.39214087e-01 1.69029176e-01 6.34871185e-01 -2.96961546e-01
-7.06017911e-02 -1.95540309e-01 -4.51117218e-01 -4.97597992e-01
-1.07793248e+00 -4.04958278e-02 3.13179463e-01 2.80674517e-01
-5.05904816e-02 -7.32882500e-01 -1.79801106e-01 8.31941739e-02
-8.74600649e-01 1.99745193e-01 4.15478468e-01 9.28875685e-01
6.41032696e-01 4.53632534e-01 1.08625567e+00 -4.12701190e-01
7.14662671e-01 -3.99727464e-01 -2.75110006e-01 -5.29521443e-02
-3.16313565e-01 -1.90213054e-01 1.12003601e+00 -8.94705236e-01
-1.46802294e+00 -2.61743903e-01 -4.55019593e-01 -6.44614220e-01
-1.80227026e-01 5.35763681e-01 -7.27627099e-01 6.12893343e-01
5.57043016e-01 5.26221693e-01 -2.13923352e-03 -6.18346751e-01
5.24957597e-01 1.07821476e+00 6.91988468e-01 -5.86119294e-01
7.75772214e-01 -3.92488912e-02 -4.87239510e-01 -1.11717558e+00
-6.34875655e-01 -6.99893832e-02 -2.64188975e-01 -1.20335601e-01
7.47119963e-01 -9.97115791e-01 -1.78359881e-01 5.37216365e-01
-1.29245293e+00 -2.51792461e-01 -5.30064166e-01 7.24132717e-01
-6.11550868e-01 3.23002845e-01 -6.26030624e-01 -1.22580290e+00
-1.81789592e-01 -1.27531910e+00 8.53908241e-01 9.41180140e-02
-4.27103996e-01 -6.81204915e-01 -1.06766701e-01 3.12694579e-01
6.36864781e-01 -3.77428979e-01 1.12886512e+00 -4.17328060e-01
-5.19019723e-01 1.10904180e-01 2.18804002e-01 7.20481694e-01
5.33609211e-01 2.78574228e-02 -1.49906790e+00 -1.35068476e-01
2.62852192e-01 -1.13525674e-01 4.16489869e-01 3.15781415e-01
9.76079404e-01 -6.24604046e-01 -6.45561665e-02 3.91772002e-01
8.87565374e-01 3.86313975e-01 6.08963728e-01 -3.62559855e-01
6.77631199e-01 7.81085253e-01 8.26183438e-01 4.56510365e-01
1.69158518e-01 5.70314467e-01 4.21050899e-02 -1.74086973e-01
-7.30842292e-01 -6.87332690e-01 8.34716201e-01 1.49862647e+00
2.54461616e-01 -1.73380256e-01 -3.25446129e-01 5.47057807e-01
-1.38808250e+00 -9.76213694e-01 3.25170755e-01 2.34516931e+00
1.18291819e+00 2.60711849e-01 -3.63906175e-02 3.20754379e-01
9.61600065e-01 3.34621221e-01 -3.62135410e-01 -2.59392709e-01
2.57420149e-02 2.82791793e-01 -1.82104498e-01 7.53934920e-01
-4.77857798e-01 1.14363396e+00 5.47097969e+00 1.40739369e+00
-1.26795995e+00 2.43532851e-01 4.07776892e-01 -3.23844075e-01
-6.03504181e-01 1.82618257e-02 -8.13190877e-01 7.16558695e-01
1.12632775e+00 -3.44812959e-01 6.75802410e-01 6.02874041e-01
7.14282274e-01 2.32900798e-01 -1.06919622e+00 7.91073680e-01
-1.06249684e-02 -9.38455462e-01 2.72599310e-01 -1.65207878e-01
4.59481835e-01 -3.81180495e-01 1.58212900e-01 5.78897595e-01
-1.52879342e-01 -9.49995399e-01 1.30552077e+00 4.24394757e-01
1.07380605e+00 -8.25751126e-01 3.57324749e-01 5.43179274e-01
-1.55777562e+00 -5.84913418e-02 -3.17954302e-01 1.48454607e-01
3.81750762e-01 6.39965177e-01 -9.94849801e-01 4.95198935e-01
3.16207349e-01 1.66545570e-01 4.04211804e-02 7.66304791e-01
-2.67153025e-01 9.57585275e-01 -1.19053267e-01 1.37316868e-01
-8.59391093e-02 -4.77638468e-02 1.03632855e+00 1.21631837e+00
5.67018986e-01 -1.86034113e-01 2.90009558e-01 1.28284276e+00
1.20252036e-01 2.64678329e-01 -3.24923158e-01 -1.01812497e-01
1.07245755e+00 8.81260931e-01 -2.40229025e-01 -4.31884646e-01
-2.07392216e-01 8.71619642e-01 -9.60281193e-02 6.99501216e-01
-9.24376190e-01 -5.17155170e-01 7.60974944e-01 3.36558223e-02
1.56977117e-01 -1.10248765e-02 -8.89059305e-02 -9.24431860e-01
8.64280015e-03 -9.29379940e-01 -1.81561187e-01 -6.82741821e-01
-1.15035701e+00 6.80847883e-01 -1.80609763e-01 -1.48334980e+00
-5.93172193e-01 -2.02409178e-03 -9.06027913e-01 1.27259684e+00
-1.42395020e+00 -1.09891415e+00 1.04709737e-01 4.47347939e-01
1.16092432e+00 -2.32687920e-01 9.93636429e-01 4.42107290e-01
-4.76742029e-01 7.83527911e-01 -1.11486115e-01 -1.92526355e-01
8.01181734e-01 -9.13711846e-01 3.14776897e-01 7.71438301e-01
2.10215282e-02 4.26924914e-01 7.35562146e-01 -5.79116106e-01
-9.94428813e-01 -1.37612629e+00 9.99972045e-01 8.46982449e-02
5.29669821e-01 -5.37456632e-01 -1.15678668e+00 2.62167782e-01
-5.58801219e-02 -3.48493963e-01 7.61583209e-01 -2.30186790e-01
-2.13712633e-01 -3.71507585e-01 -9.41530764e-01 8.47668290e-01
7.09814131e-01 -9.32306349e-01 -7.85419643e-01 -1.68826029e-01
1.23516357e+00 -1.67361736e-01 -5.56894362e-01 8.29333216e-02
3.77843916e-01 -7.38119841e-01 9.07923400e-01 -1.50713623e-01
1.79316238e-01 -5.25222361e-01 -2.94836104e-01 -1.68217826e+00
-7.34761432e-02 -9.02925074e-01 -6.78653345e-02 2.04473996e+00
5.38045824e-01 -3.56206954e-01 3.06080908e-01 2.25916997e-01
-5.12637496e-01 -4.08859372e-01 -1.02876055e+00 -8.66248250e-01
3.02759781e-02 -7.53669441e-01 8.30335319e-01 6.69336140e-01
3.85417119e-02 3.25290769e-01 -4.28250194e-01 2.80125558e-01
4.90450710e-01 -1.34698883e-01 5.23748636e-01 -8.52995336e-01
-4.58566099e-01 -4.88822609e-01 2.81202555e-01 -1.01893282e+00
2.90914506e-01 -8.06313336e-01 5.97904384e-01 -1.00779974e+00
-4.28740799e-01 -6.00966811e-01 -3.59128624e-01 9.71048325e-02
-3.38483632e-01 -3.54954749e-02 4.52296317e-01 4.38760519e-02
6.91074952e-02 9.91045058e-01 1.30240417e+00 -1.89141691e-01
-5.04743576e-01 3.46652061e-01 -5.20495296e-01 6.41535878e-01
5.54838181e-01 -5.19838095e-01 -6.10677719e-01 -1.20904461e-01
-3.19194168e-01 4.89528179e-01 1.43108577e-01 -1.09975660e+00
6.37218729e-02 -1.53618902e-01 8.51904601e-03 -5.44054747e-01
7.62654841e-01 -7.03154624e-01 2.42573604e-01 3.01893204e-01
-4.66054738e-01 -6.07067406e-01 1.46066323e-01 3.80153716e-01
-5.76333404e-01 -4.04871732e-01 9.18427050e-01 -8.49237368e-02
-1.80472285e-01 1.06611259e-01 -7.55131841e-01 -1.84459984e-02
4.85902816e-01 -2.03492299e-01 4.01192904e-03 -4.52971369e-01
-7.61763275e-01 -1.38757050e-01 5.57784662e-02 5.94999731e-01
6.23700321e-01 -1.51919830e+00 -7.84426868e-01 5.33650756e-01
-9.05513763e-02 1.08545624e-01 5.21402121e-01 4.61706549e-01
3.08288038e-01 7.96189383e-02 2.89838105e-01 -2.83039331e-01
-1.04213703e+00 4.78048503e-01 2.74752229e-01 1.17445841e-01
-4.33765799e-01 8.03496003e-01 4.83184993e-01 -2.69397080e-01
4.56761658e-01 -3.68787706e-01 -8.66434053e-02 2.50275403e-01
6.16396427e-01 4.60653037e-01 -7.87204951e-02 -7.11556077e-01
-1.04694583e-01 3.59284639e-01 9.10336971e-02 -6.18042707e-01
8.66926372e-01 -4.95531052e-01 6.48745671e-02 8.88773382e-01
1.09265304e+00 5.63882649e-01 -1.26329339e+00 -2.77928859e-01
-2.87960976e-01 -2.88262486e-01 6.99895769e-02 -6.61706150e-01
-6.52263403e-01 1.10365343e+00 4.30075407e-01 2.74853259e-01
1.10478652e+00 -7.69331306e-02 1.15334964e+00 -1.84457138e-01
1.38973385e-01 -1.35023534e+00 2.44261652e-01 6.17278218e-01
9.01888371e-01 -7.26259470e-01 -6.27867937e-01 -5.15212953e-01
-5.85962534e-01 9.36808527e-01 6.33707404e-01 2.14461759e-01
5.09882867e-01 4.10183221e-01 2.80875802e-01 4.53432888e-01
-7.12454796e-01 -9.93349701e-02 2.22735405e-01 6.45493805e-01
3.03965002e-01 2.05210254e-01 2.17238769e-01 1.05631304e+00
-7.63304889e-01 -4.39106166e-01 4.00231421e-01 3.20811152e-01
-6.33877099e-01 -1.15818596e+00 -7.65854180e-01 -6.01862790e-04
-3.51044714e-01 -5.67168772e-01 -1.96263835e-01 1.00052133e-01
2.70465314e-01 1.28265262e+00 1.30041167e-01 -3.60917658e-01
3.51027638e-01 4.27403331e-01 9.42158177e-02 -9.04821515e-01
-4.09458190e-01 8.40638459e-01 9.11412984e-02 -1.28327921e-01
4.10549641e-02 -4.42763627e-01 -1.59882569e+00 9.46697667e-02
-5.60060978e-01 2.23066837e-01 6.60537779e-01 9.56346571e-01
1.54404223e-01 1.03484619e+00 1.02265525e+00 -1.03326082e+00
-6.88685834e-01 -1.26942110e+00 -6.88840628e-01 1.95899069e-01
6.44863069e-01 -3.43133301e-01 -6.67733788e-01 1.89526409e-01] | [15.094902992248535, 6.428300857543945] |
97556320-e27c-484d-b9fe-2dd91ee9cbe7 | mega-rst-discourse-treebanks-with-structure | 2011.03017 | null | https://arxiv.org/abs/2011.03017v1 | https://arxiv.org/pdf/2011.03017v1.pdf | MEGA RST Discourse Treebanks with Structure and Nuclearity from Scalable Distant Sentiment Supervision | The lack of large and diverse discourse treebanks hinders the application of data-driven approaches, such as deep-learning, to RST-style discourse parsing. In this work, we present a novel scalable methodology to automatically generate discourse treebanks using distant supervision from sentiment-annotated datasets, creating and publishing MEGA-DT, a new large-scale discourse-annotated corpus. Our approach generates discourse trees incorporating structure and nuclearity for documents of arbitrary length by relying on an efficient heuristic beam-search strategy, extended with a stochastic component. Experiments on multiple datasets indicate that a discourse parser trained on our MEGA-DT treebank delivers promising inter-domain performance gains when compared to parsers trained on human-annotated discourse corpora. | ['Giuseppe Carenini', 'Patrick Huber'] | 2020-11-05 | null | https://aclanthology.org/2020.emnlp-main.603 | https://aclanthology.org/2020.emnlp-main.603.pdf | emnlp-2020-11 | ['discourse-parsing'] | ['natural-language-processing'] | [ 2.48767093e-01 1.03914642e+00 -2.48251185e-01 -4.20820743e-01
-1.46218991e+00 -8.99723291e-01 1.01444256e+00 3.46233457e-01
-2.95745552e-01 1.31018627e+00 1.08536410e+00 -6.13046646e-01
3.65878046e-01 -7.25007057e-01 -5.21147490e-01 -4.10881341e-01
1.84861481e-01 9.11204219e-01 4.72553432e-01 -6.43881619e-01
4.68076289e-01 -2.51660794e-01 -1.11606312e+00 9.98961389e-01
8.84785891e-01 4.17888045e-01 2.57626057e-01 8.29253912e-01
-6.86727166e-01 1.45333600e+00 -1.20157814e+00 -6.80032909e-01
-2.14904130e-01 -8.69411051e-01 -1.46761799e+00 5.95396906e-02
3.91761549e-02 -2.31379509e-01 2.92759895e-01 5.30918717e-01
3.96529168e-01 -4.71229441e-02 6.44538224e-01 -3.44152510e-01
-4.61931199e-01 1.30958438e+00 -1.01377599e-01 2.31446862e-01
7.51416147e-01 -2.62361497e-01 1.39467931e+00 -4.44824487e-01
1.45719421e+00 1.44640791e+00 4.52693552e-01 7.62553394e-01
-1.22938001e+00 -2.52931677e-02 7.63558000e-02 9.15143713e-02
-3.66312891e-01 -4.22990590e-01 1.02270150e+00 -4.73835170e-01
1.29250979e+00 2.73870617e-01 2.76470691e-01 1.51058102e+00
-7.52635971e-02 9.46125388e-01 1.22485483e+00 -9.74152982e-01
3.53921741e-01 1.76554266e-02 2.76320875e-02 2.84237832e-01
-1.28521413e-01 -3.97576183e-01 -5.01853049e-01 -1.87654376e-01
2.16502890e-01 -1.21710396e+00 1.53473141e-02 1.12755753e-01
-9.30000067e-01 1.12702084e+00 -9.47188810e-02 7.32060850e-01
-2.14373633e-01 -2.65745819e-01 1.11620569e+00 4.49716985e-01
9.63855147e-01 6.72691822e-01 -2.35327944e-01 -5.29601872e-01
-6.46793067e-01 5.17285168e-01 1.29290116e+00 8.42053115e-01
1.93427458e-01 -1.09322488e-01 -2.71691918e-01 7.85134256e-01
2.52335042e-01 6.41281456e-02 5.30858278e-01 -1.28401029e+00
1.06395602e+00 7.10546494e-01 2.21530885e-01 -7.21790731e-01
-4.08546388e-01 3.14087182e-01 -5.13712823e-01 -1.41474321e-01
8.33183110e-01 -5.93060851e-01 -2.96243668e-01 1.46786666e+00
6.29216611e-01 -6.08121991e-01 6.22107029e-01 6.00552261e-01
1.06959105e+00 7.55041718e-01 1.77878276e-01 -4.90105569e-01
1.37695873e+00 -1.00332522e+00 -8.77189577e-01 -1.30829336e-02
1.20333862e+00 -7.47944474e-01 1.04688120e+00 3.40193182e-01
-1.42864752e+00 -2.08971232e-01 -1.02186739e+00 -3.61216486e-01
-5.81517108e-02 -1.49636522e-01 2.75821567e-01 7.55817711e-01
-8.79200399e-01 4.89602953e-01 -7.25873768e-01 -8.32903758e-02
5.14576793e-01 -2.56802231e-01 4.59013246e-02 2.48122111e-01
-1.13071907e+00 1.14914143e+00 8.30791950e-01 -2.31907725e-01
-5.98701179e-01 -3.48959923e-01 -9.92188334e-01 -2.96847790e-01
3.10637027e-01 -4.14920211e-01 1.81527936e+00 -1.09438968e+00
-2.04223990e+00 1.12533283e+00 -3.52926373e-01 -1.01916897e+00
6.09708905e-01 -3.05673093e-01 6.08624751e-03 2.39317343e-01
2.52317220e-01 2.48740271e-01 4.22432035e-01 -1.11166620e+00
-6.67378664e-01 -6.68011457e-02 3.14538419e-01 3.14433664e-01
-2.77303666e-01 5.65257251e-01 3.20831239e-01 -4.97767121e-01
-3.08580101e-01 -5.54231167e-01 -3.25244784e-01 -8.72786760e-01
-6.00749493e-01 -8.11886370e-01 5.37426353e-01 -9.21184838e-01
1.36796319e+00 -1.48104465e+00 4.74718690e-01 -2.27396831e-01
-1.07210897e-01 3.34499329e-01 -2.21836120e-02 7.39055276e-01
2.84820259e-01 1.73262402e-01 -2.95406789e-01 -4.96852636e-01
5.34545481e-02 4.89051878e-01 -4.63314563e-01 2.21382845e-02
5.38960993e-01 7.80868351e-01 -1.23315680e+00 -9.39391077e-01
7.42916539e-02 5.04038259e-02 -4.23605174e-01 4.03363377e-01
-1.12428844e+00 6.35715306e-01 -7.31962740e-01 3.27191263e-01
6.25409335e-02 -2.87887007e-01 1.05455172e+00 3.40364933e-01
-2.87957549e-01 1.19854510e+00 -5.70477724e-01 1.79091012e+00
-5.46223104e-01 8.64170074e-01 4.03344184e-02 -1.03523469e+00
1.05674493e+00 7.28441179e-01 -6.52453527e-02 -6.74217284e-01
4.03858185e-01 4.07213211e-01 -2.62046196e-02 -6.44373417e-01
8.12301457e-01 -1.86611429e-01 -5.93413591e-01 7.11989760e-01
1.32411763e-01 -6.56973496e-02 7.19633222e-01 2.00272858e-01
1.13683307e+00 3.87946665e-01 5.08216977e-01 -3.04049700e-01
7.99271584e-01 7.94585347e-01 3.89367014e-01 3.60238254e-01
1.68891639e-01 4.98195738e-01 8.86022270e-01 -3.47033679e-01
-1.38945115e+00 -4.75848436e-01 -1.78141311e-01 1.34199536e+00
-5.28126955e-01 -5.67977726e-01 -1.03881013e+00 -8.10078204e-01
-4.96062309e-01 9.97786939e-01 -4.61954862e-01 8.89394462e-01
-1.58120668e+00 -8.24574172e-01 6.31865025e-01 4.04276431e-01
2.87557006e-01 -1.31377518e+00 -6.86909676e-01 8.00694764e-01
-4.16723311e-01 -1.17110121e+00 2.57169276e-01 -4.69255112e-02
-6.66527152e-01 -1.21852684e+00 -5.52351594e-01 -9.23401058e-01
2.03159377e-01 -4.62738395e-01 1.47946918e+00 1.90027542e-02
3.17453653e-01 -2.28605419e-01 -7.68824100e-01 -4.53181267e-01
-1.30954254e+00 5.74964941e-01 -6.72312021e-01 -6.21417344e-01
2.87192672e-01 -2.32295081e-01 -1.01630285e-01 -2.96144038e-01
-4.74628061e-01 1.33836553e-01 -3.58121656e-02 1.24005556e+00
1.26566619e-01 -4.47514743e-01 9.82762337e-01 -1.50799286e+00
1.08195198e+00 -4.87789869e-01 -5.88862956e-01 2.69552559e-01
-1.64428368e-01 4.23756801e-02 5.60623348e-01 9.14088741e-04
-1.89918721e+00 -3.76603156e-01 -5.60343742e-01 4.89524692e-01
-1.42853167e-02 6.63714468e-01 -6.79342896e-02 7.79986858e-01
1.02134407e+00 -3.42291802e-01 -2.02276289e-01 -4.19358402e-01
7.98534393e-01 7.47174203e-01 4.97214645e-01 -8.85172963e-01
3.71039003e-01 2.31360763e-01 -3.35657299e-01 -8.69885802e-01
-1.34420562e+00 -2.82681406e-01 -7.81337321e-01 -4.74099582e-03
1.19433439e+00 -9.85642850e-01 -2.64635891e-01 1.59784436e-01
-1.78420043e+00 -5.54478586e-01 -4.54762995e-01 9.15246736e-03
-5.79479873e-01 2.24979013e-01 -1.01201069e+00 -9.56283748e-01
-5.99521041e-01 -6.18502557e-01 1.11929762e+00 2.08228022e-01
-9.77110267e-01 -1.43634284e+00 7.09404051e-01 8.04996610e-01
5.09858392e-02 8.13375175e-01 7.91705608e-01 -1.00118899e+00
-3.77711892e-01 3.58814001e-01 -3.08170263e-02 1.18828341e-01
-5.41034825e-02 1.21993143e-02 -9.63481665e-01 2.84962952e-02
-1.19806938e-01 -9.74239409e-01 4.91706997e-01 2.14390650e-01
4.81121153e-01 -5.61029851e-01 -9.12740380e-02 -9.14606228e-02
9.83265042e-01 4.30859402e-02 4.24972445e-01 1.07646751e+00
3.77675861e-01 8.99295807e-01 8.08587730e-01 4.49363500e-01
8.44550252e-01 2.49862462e-01 -9.87707525e-02 2.93299913e-01
-9.46729854e-02 -2.68450052e-01 4.92040843e-01 1.11977446e+00
-2.56639034e-01 -3.73818308e-01 -1.09381902e+00 1.04521108e+00
-1.89563406e+00 -1.02903450e+00 -4.73767042e-01 1.29379153e+00
1.57208085e+00 5.22086203e-01 2.62451380e-01 1.18735358e-01
5.25268376e-01 5.97430170e-01 -1.39168322e-01 -8.88214231e-01
-5.16760111e-01 5.03556967e-01 -4.55239862e-02 7.08302498e-01
-1.01594985e+00 1.08561540e+00 6.32285213e+00 5.92544973e-01
-5.77271879e-01 4.10069436e-01 6.60171092e-01 -2.92127207e-02
-5.85145414e-01 1.11426271e-01 -8.95843685e-01 2.47415811e-01
1.54421914e+00 -3.88517767e-01 -2.57262260e-01 1.02060938e+00
3.69125716e-02 -9.72787589e-02 -9.81175303e-01 -9.98426601e-03
1.01313321e-02 -2.13500094e+00 -2.27622330e-01 -2.39117220e-01
1.01502907e+00 9.66423824e-02 -6.37070417e-01 3.08450758e-01
1.06843865e+00 -8.60519946e-01 7.27560222e-01 -1.34109363e-01
5.09596825e-01 -5.04884899e-01 8.68771315e-01 5.44651091e-01
-7.43538141e-01 1.92745030e-01 -5.58868013e-02 -4.07538295e-01
6.38303161e-01 3.47958893e-01 -1.38210082e+00 7.45424628e-01
3.94074529e-01 5.35859525e-01 6.99833781e-03 8.51374418e-02
-7.16244757e-01 9.55661058e-01 -9.39233080e-02 -4.72321332e-01
4.37699080e-01 1.57826506e-02 5.80297887e-01 1.49240589e+00
-9.66495350e-02 2.53578424e-01 2.22200140e-01 5.88515639e-01
-1.68797523e-01 2.87512094e-01 -5.10110736e-01 -8.19865912e-02
6.85992241e-01 1.05679107e+00 -5.59972882e-01 -6.28490329e-01
-2.68670440e-01 3.82592708e-01 5.15769780e-01 -1.58178240e-01
-3.41062546e-01 -8.95907134e-02 2.83646397e-02 -1.16532281e-01
4.19965059e-01 -2.36132354e-01 -4.76259828e-01 -8.42877328e-01
1.19851559e-01 -1.21315300e+00 6.30752981e-01 -3.65356863e-01
-1.27531111e+00 1.00996399e+00 1.41453847e-01 -7.95366049e-01
-1.10713232e+00 -4.40980107e-01 -9.03877914e-01 7.38572180e-01
-1.67711651e+00 -1.19176626e+00 1.98674440e-01 1.99424084e-02
9.82911170e-01 -2.69451648e-01 1.04640365e+00 -1.66435242e-01
-4.89915282e-01 1.35259181e-01 1.29781276e-01 4.58865732e-01
5.31572223e-01 -1.60978127e+00 6.55779183e-01 4.91691083e-01
-8.77555981e-02 1.97456777e-01 1.12896979e+00 -6.46768212e-01
-7.87962317e-01 -7.36621916e-01 1.47858024e+00 -8.60974371e-01
1.12972260e+00 -4.12172914e-01 -1.18163621e+00 7.05190659e-01
1.10980260e+00 -8.34235966e-01 7.87962914e-01 4.40464735e-01
-6.89561516e-02 6.11248791e-01 -9.26051319e-01 3.43519866e-01
8.13735008e-01 -5.25368333e-01 -1.51053941e+00 6.11690164e-01
7.91125774e-01 -1.09926093e+00 -1.24308312e+00 1.02229096e-01
1.78956598e-01 -8.61599028e-01 5.58578610e-01 -7.77301192e-01
1.16417491e+00 2.05460951e-01 -1.09400146e-03 -1.12626767e+00
3.77787828e-01 -1.03687608e+00 -2.26213753e-01 1.57981515e+00
6.67288661e-01 -2.24290550e-01 9.04061258e-01 5.45978487e-01
-3.00813675e-01 -6.02790654e-01 -1.10546148e+00 -4.03413057e-01
7.88664043e-01 -6.22464456e-02 4.95949686e-01 9.49105620e-01
6.25644684e-01 8.95132661e-01 2.29583913e-03 -4.60817397e-01
3.34458321e-01 3.89478266e-01 8.92508388e-01 -1.13110912e+00
-2.63809472e-01 -3.66231441e-01 3.59811127e-01 -9.35517073e-01
6.50618553e-01 -6.73560917e-01 6.52716681e-02 -1.55619097e+00
-2.60077953e-01 -4.13533270e-01 5.44553399e-01 1.75242648e-01
-6.62454292e-02 -2.18620345e-01 4.59791869e-02 1.08518727e-01
-6.60189331e-01 6.99212432e-01 1.21770155e+00 -4.35085706e-02
-4.36104178e-01 -3.32664490e-01 -5.36900282e-01 1.05227542e+00
8.96710873e-01 -5.57642817e-01 -2.82355368e-01 -4.60605234e-01
2.39431635e-01 4.60494220e-01 -1.94745883e-01 -2.27048814e-01
9.85892788e-02 -1.75076693e-01 -1.72548499e-02 -8.67240310e-01
2.57308125e-01 1.53843135e-01 -3.77402365e-01 2.26964265e-01
-6.69295311e-01 6.22230209e-02 2.32034385e-01 2.33732164e-01
-5.22741199e-01 -6.78915501e-01 2.98806369e-01 -5.36863863e-01
-4.25341845e-01 -7.77656615e-01 -6.10291481e-01 7.15253770e-01
8.97988558e-01 -6.30693659e-02 -8.78164291e-01 -6.97001889e-02
-5.92401028e-01 2.65925646e-01 3.16249132e-01 2.50965923e-01
3.92250195e-02 -1.09427655e+00 -1.25788999e+00 -4.93211538e-01
-2.06315339e-01 5.87218285e-01 -8.40780511e-02 2.28701919e-01
-5.90131819e-01 7.07704306e-01 -1.50745928e-01 -4.92594033e-01
-1.39840019e+00 -9.09874216e-02 -1.27522796e-01 -1.11580276e+00
-7.42700517e-01 8.60344708e-01 -3.45316678e-01 -3.76564622e-01
-4.24115881e-02 -2.87461430e-01 -6.30868673e-01 4.76522118e-01
5.79139113e-01 6.73703477e-02 -3.24119441e-02 -7.15214849e-01
-8.20569769e-02 -8.95527154e-02 -1.38579771e-01 -4.11120653e-01
1.54090023e+00 -1.81497559e-01 -2.66438127e-01 5.81849635e-01
7.69888818e-01 6.14469051e-02 -1.13777375e+00 -2.38104895e-01
9.72368002e-01 -1.43630058e-01 -1.80647761e-01 -5.11718214e-01
-1.44060016e-01 5.18351674e-01 -3.11598331e-01 6.22935653e-01
6.82009399e-01 2.18636259e-01 8.74707282e-01 5.67804754e-01
2.16326520e-01 -1.44147909e+00 3.24521631e-01 9.78256106e-01
1.06231856e+00 -1.27796781e+00 -5.38303293e-02 -2.72797912e-01
-8.18600714e-01 1.11315882e+00 5.91415048e-01 -3.49567503e-01
1.63530260e-01 3.94799113e-01 3.07092667e-01 -2.42163464e-01
-1.23138845e+00 -3.36535238e-02 2.56760716e-02 6.28557563e-01
9.50236917e-01 -1.66872844e-01 -6.53923631e-01 5.38225651e-01
-6.21813357e-01 -1.44813865e-01 7.15169609e-01 1.10611475e+00
-5.53480804e-01 -1.80078006e+00 -3.80033255e-01 2.96024177e-02
-6.04264796e-01 -8.51608533e-03 -7.94614494e-01 9.37670469e-01
-2.50095755e-01 1.00895977e+00 1.98316291e-01 2.26443887e-01
1.64165452e-01 1.33084774e-01 5.49340487e-01 -1.00181639e+00
-9.62485254e-01 1.34330571e-01 1.56923819e+00 -1.08772628e-01
-1.17809629e+00 -1.16674578e+00 -1.44792557e+00 -2.21351027e-01
-1.86948121e-01 3.77586335e-01 4.53992456e-01 1.30209911e+00
3.73817161e-02 6.41196847e-01 2.97871441e-01 -8.02283883e-01
-6.36093676e-01 -1.25468993e+00 1.81825757e-01 4.02532369e-01
1.96369842e-01 -2.35725090e-01 9.15476829e-02 2.70293713e-01] | [10.847591400146484, 9.419922828674316] |
19d6083d-e5cf-4b77-a9c1-0f66bbefbd47 | event-based-visual-tracking-in-dynamic | 2212.07754 | null | https://arxiv.org/abs/2212.07754v1 | https://arxiv.org/pdf/2212.07754v1.pdf | Event-based Visual Tracking in Dynamic Environments | Visual object tracking under challenging conditions of motion and light can be hindered by the capabilities of conventional cameras, prone to producing images with motion blur. Event cameras are novel sensors suited to robustly perform vision tasks under these conditions. However, due to the nature of their output, applying them to object detection and tracking is non-trivial. In this work, we propose a framework to take advantage of both event cameras and off-the-shelf deep learning for object tracking. We show that reconstructing event data into intensity frames improves the tracking performance in conditions under which conventional cameras fail to provide acceptable results. | ['Carlos Sagues', 'Rodrigo Aldana-Lopez', 'Irene Perez-Salesa'] | 2022-12-15 | null | null | null | null | ['visual-tracking', 'visual-object-tracking'] | ['computer-vision', 'computer-vision'] | [ 2.27502152e-01 -7.35307693e-01 8.65737945e-02 -4.04049568e-02
-2.51355946e-01 -7.65551090e-01 6.45422637e-01 -1.22425914e-01
-5.96853077e-01 4.87446159e-01 -3.08231324e-01 -7.91354775e-02
2.53569514e-01 -3.75550002e-01 -8.75817895e-01 -6.08600855e-01
2.74475306e-01 -3.85195501e-02 8.06897402e-01 2.73834467e-01
3.36343907e-02 9.32839274e-01 -1.98138928e+00 1.94830313e-01
2.72957474e-01 1.06381094e+00 2.58136809e-01 1.03014600e+00
-6.53129369e-02 7.53682435e-01 -5.54135740e-01 -1.59321561e-01
3.71430069e-01 -2.40111545e-01 8.50392971e-03 3.98579001e-01
9.54409122e-01 -7.31672645e-01 -3.91902208e-01 1.06608367e+00
3.06133866e-01 4.61643748e-02 3.07758719e-01 -1.30746913e+00
-3.94409239e-01 -3.07469904e-01 -3.33085865e-01 4.85775083e-01
5.42622030e-01 6.16255224e-01 4.54558194e-01 -8.16625774e-01
4.97138441e-01 1.07932258e+00 7.84681857e-01 7.30674148e-01
-1.25738776e+00 -5.06441772e-01 9.93100256e-02 2.23727763e-01
-9.21872854e-01 -7.66624868e-01 5.55055499e-01 -7.06484377e-01
1.05420935e+00 -6.32353052e-02 6.78983092e-01 1.27384794e+00
3.27689797e-01 6.14936650e-01 9.13006127e-01 -1.90004781e-01
2.19373181e-01 8.62855930e-03 -2.89319772e-02 4.69383478e-01
5.89824080e-01 7.09356129e-01 -7.03615904e-01 1.00574397e-01
6.81746423e-01 2.43818402e-01 -4.64949489e-01 -6.14360094e-01
-1.18771720e+00 4.97069359e-01 3.66827220e-01 1.08653404e-01
-4.58228141e-01 4.31658447e-01 2.16698796e-01 1.50114968e-01
8.63474086e-02 2.28416443e-01 -2.03327522e-01 -1.50304943e-01
-1.06946337e+00 4.82580438e-02 6.69214249e-01 8.69257510e-01
3.78160596e-01 4.13296163e-01 -8.44504535e-02 6.67531136e-03
5.34069121e-01 6.66167855e-01 1.42020032e-01 -1.09998941e+00
3.24359052e-02 3.84708852e-01 4.96225268e-01 -6.80078149e-01
-3.63845885e-01 -1.25359967e-01 -4.28622514e-01 8.63661289e-01
9.14884984e-01 -1.05604656e-01 -9.15711343e-01 1.32498395e+00
3.03860158e-01 4.47756439e-01 7.72619098e-02 1.12377429e+00
6.72460496e-01 5.24720371e-01 1.09252065e-01 -4.07060027e-01
1.18932927e+00 -5.96802413e-01 -8.62401605e-01 -3.12091798e-01
-5.45648150e-02 -9.53058660e-01 7.89202511e-01 5.28180659e-01
-1.14775741e+00 -7.26674318e-01 -9.57938075e-01 8.00421983e-02
-3.16503286e-01 -7.44955614e-02 4.83498335e-01 7.97495604e-01
-9.60997283e-01 4.35579568e-01 -1.15584266e+00 -5.26375890e-01
4.59831804e-01 4.09826100e-01 -2.12100193e-01 1.56959668e-02
-6.23478651e-01 1.08160412e+00 4.55773354e-01 1.89095795e-01
-1.06972003e+00 -6.86739206e-01 -7.27409422e-01 -6.31983131e-02
3.96217406e-01 -6.48804784e-01 1.29815698e+00 -1.02998006e+00
-1.54802787e+00 7.18164086e-01 -1.32961750e-01 -6.17434025e-01
7.63514757e-01 -4.91048783e-01 -1.51497304e-01 3.73814672e-01
-2.79333204e-01 6.13465846e-01 1.26904273e+00 -9.89233911e-01
-6.42093897e-01 -2.72959113e-01 3.78745943e-02 -4.82980870e-02
-2.36110568e-01 2.66423821e-01 -5.00432253e-01 -2.23653838e-01
-4.28841412e-01 -8.87235880e-01 -1.31473094e-01 6.35574341e-01
1.33992121e-01 1.40716448e-01 1.39275920e+00 -2.75340885e-01
5.64227819e-01 -2.12568498e+00 -1.99721128e-01 -4.44478214e-01
1.11930687e-02 7.72837698e-01 1.37051463e-01 8.56489390e-02
2.08957449e-01 -5.24435043e-01 1.45593181e-01 -2.57182717e-01
-2.68296540e-01 2.00285971e-01 -3.64452511e-01 7.59320021e-01
2.79432744e-01 8.36979508e-01 -8.42688322e-01 -4.92644757e-01
9.81664598e-01 7.87092090e-01 -1.94053590e-01 4.15494561e-01
-4.33767706e-01 6.57373428e-01 -2.26926371e-01 7.08618641e-01
4.81999874e-01 -2.12397054e-01 -1.59348801e-01 -3.46474141e-01
-4.45893347e-01 -1.62990913e-01 -1.33275485e+00 1.29865563e+00
-1.33842245e-01 1.03590274e+00 2.71306872e-01 -4.94184524e-01
5.57843149e-01 3.46483082e-01 5.26183665e-01 -5.81053674e-01
3.17859352e-01 5.70995063e-02 -1.13693044e-01 -5.68733633e-01
4.33753401e-01 -2.31477126e-01 3.87500793e-01 1.86865568e-01
2.09490322e-02 4.24173055e-03 2.64961392e-01 -7.25279376e-02
1.07801938e+00 3.20914298e-01 2.26646930e-01 1.41621798e-01
3.53481501e-01 5.76336831e-02 3.17148060e-01 8.73756826e-01
-5.85508227e-01 7.11835861e-01 -1.63263425e-01 -5.34383476e-01
-1.09735823e+00 -1.27740622e+00 -1.87619984e-01 7.81275153e-01
2.74452597e-01 9.49272290e-02 -3.65563512e-01 -2.09505245e-01
8.57360661e-02 3.66425246e-01 -1.44756883e-01 -1.97305948e-01
-4.97311562e-01 -7.53867209e-01 4.07884926e-01 7.87037432e-01
3.12711000e-01 -9.39802587e-01 -1.46947837e+00 3.79660308e-01
1.82893261e-01 -1.53844857e+00 -3.73865992e-01 2.29173481e-01
-7.50665843e-01 -1.22185266e+00 -7.01087713e-01 -3.64408761e-01
3.57163787e-01 6.16603673e-01 8.96585107e-01 -7.52292722e-02
-6.00836277e-01 7.55242944e-01 -2.36341313e-01 -5.44351459e-01
-3.49177331e-01 -5.81427574e-01 1.18712604e-01 1.73148781e-01
4.26559687e-01 -1.31843686e-01 -7.24263847e-01 1.74798533e-01
-9.83708143e-01 -2.45848283e-01 2.93439388e-01 4.87808228e-01
2.82635927e-01 -1.27317086e-01 7.67884925e-02 -3.15867513e-01
4.80989441e-02 1.13530594e-04 -1.26148379e+00 1.48536684e-02
-3.08273405e-01 -3.03960919e-01 5.01144171e-01 -6.73664570e-01
-1.02813804e+00 7.12366164e-01 1.56595752e-01 -8.42089057e-01
-3.49694371e-01 -1.58902764e-01 1.32760406e-01 -4.71426159e-01
6.25647187e-01 -2.37576924e-02 4.62646894e-02 -2.75058210e-01
1.97331682e-01 4.68620241e-01 7.78246939e-01 -7.30990851e-03
7.80977964e-01 9.44037080e-01 1.55598968e-01 -8.27672243e-01
-9.36511576e-01 -7.08387613e-01 -6.11366928e-01 -6.50892913e-01
1.11587465e+00 -1.02998257e+00 -9.49478209e-01 6.76567376e-01
-1.29763615e+00 -1.43962160e-01 -2.78754592e-01 7.55971909e-01
-5.13177514e-01 4.28361446e-01 -6.54476345e-01 -1.06671953e+00
-8.59225318e-02 -1.05111456e+00 1.12092352e+00 5.48950434e-01
1.24616243e-01 -9.11423683e-01 -1.21178761e-01 1.88829467e-01
5.61737955e-01 3.48345727e-01 -4.36210819e-03 -1.21658891e-01
-1.05178750e+00 -3.59764278e-01 -3.62076789e-01 2.78837860e-01
8.52668881e-02 3.09529364e-01 -1.36518717e+00 -4.43297058e-01
8.02314505e-02 -1.94002584e-01 9.02451634e-01 7.23044336e-01
8.49302649e-01 2.11156890e-01 -3.30233544e-01 7.64068425e-01
1.58686411e+00 3.67887527e-01 4.78682637e-01 3.15325975e-01
7.23412752e-01 4.55670327e-01 4.09385324e-01 4.01983917e-01
3.01756281e-02 9.42498267e-01 5.35414457e-01 -1.76770110e-02
-4.07006294e-01 1.58181801e-01 6.61285102e-01 -7.06216171e-02
1.76527835e-02 -3.35597515e-01 -8.25676799e-01 5.62663317e-01
-1.71482253e+00 -1.19565690e+00 -4.64144826e-01 2.16366172e+00
5.33997893e-01 3.46492678e-01 2.35002711e-01 -4.56122644e-02
6.22530341e-01 -3.06227859e-02 -5.80566585e-01 1.43229151e-02
-1.75268024e-01 -1.20159253e-01 6.66824400e-01 2.35839322e-01
-1.13974404e+00 6.18725598e-01 7.08383226e+00 -3.47173326e-02
-1.24542260e+00 2.59508938e-02 -1.41573876e-01 -3.97496551e-01
3.34060729e-01 -8.92411843e-02 -1.04715383e+00 6.48796976e-01
9.90694404e-01 2.57211238e-01 2.54059613e-01 6.27717435e-01
2.75193870e-01 -2.39436105e-01 -1.29751992e+00 1.09915185e+00
2.08924741e-01 -1.05083072e+00 -2.33726978e-01 -1.99218273e-01
5.08716524e-01 7.62684494e-02 1.39155447e-01 -1.90516427e-01
2.19852090e-01 -8.81214261e-01 7.12772787e-01 5.45833886e-01
4.76018488e-01 -2.52431929e-01 5.90071976e-01 2.77727216e-01
-1.14472675e+00 -8.14283490e-02 -3.91114861e-01 -2.14594245e-01
4.85283613e-01 5.72038412e-01 -7.16509581e-01 7.65601248e-02
8.85548413e-01 8.04482937e-01 -5.48732638e-01 1.48428607e+00
9.61470008e-02 2.96490610e-01 -4.98779416e-01 1.81080937e-01
1.16888925e-01 1.14387609e-01 6.94566846e-01 1.24102533e+00
2.82732874e-01 -1.86474159e-01 3.08634430e-01 7.91127264e-01
2.24137798e-01 -7.22066104e-01 -9.34436619e-01 1.13577940e-01
2.44504735e-01 1.09917557e+00 -7.37920582e-01 -2.23249003e-01
-7.68097699e-01 1.03049624e+00 -3.44622470e-02 4.16758776e-01
-7.73890495e-01 -3.16019244e-02 6.08126581e-01 1.17525637e-01
7.49367595e-01 -5.58073282e-01 -6.48449138e-02 -1.22594917e+00
2.46019214e-02 -5.38892329e-01 3.59957367e-01 -9.78052497e-01
-1.15825200e+00 3.38153839e-01 4.32655960e-02 -1.37578416e+00
-1.63978830e-01 -1.09896398e+00 -5.06063282e-01 4.76034999e-01
-1.72689509e+00 -1.17800927e+00 -5.64112186e-01 7.48545229e-01
5.73840380e-01 -5.86650297e-02 4.59233046e-01 3.44107300e-01
-3.38651925e-01 7.29272068e-02 1.08195089e-01 -2.86021046e-02
6.03497446e-01 -1.20690191e+00 1.90167725e-01 1.24385309e+00
3.18415046e-01 3.06353986e-01 9.86379564e-01 -4.55404341e-01
-1.85191488e+00 -1.00801098e+00 4.33232069e-01 -6.93228602e-01
6.76410496e-01 -3.44909549e-01 -9.12123382e-01 6.36424422e-01
1.55260518e-01 4.49797273e-01 4.18898404e-01 -2.89250910e-01
-1.45055413e-01 -1.86167538e-01 -8.12904537e-01 4.41772014e-01
6.59492195e-01 -5.88021040e-01 -6.77250504e-01 1.91460654e-01
3.38770717e-01 -5.63217103e-01 -5.05139828e-01 3.16811353e-01
6.20725513e-01 -9.78618383e-01 1.19214499e+00 -3.62261117e-01
-2.49453709e-01 -7.34699786e-01 4.46611159e-02 -8.62947047e-01
-1.16922699e-01 -5.66253543e-01 -3.51687551e-01 1.09189117e+00
-8.48299861e-02 -3.74112338e-01 7.18927920e-01 4.84296560e-01
4.25093137e-02 1.42627269e-01 -1.00870240e+00 -9.68165457e-01
-4.03905749e-01 -4.93722677e-01 -5.35280779e-02 3.79237473e-01
-5.40007710e-01 1.80410355e-01 -5.09225011e-01 3.28651309e-01
9.87846494e-01 2.97541708e-01 7.05145836e-01 -1.32119298e+00
-2.70723581e-01 -2.45167524e-01 -6.13600791e-01 -8.32883537e-01
-4.23442526e-03 -2.48894081e-01 3.65521342e-01 -1.22317338e+00
1.38510177e-02 8.67233425e-02 -1.89053372e-01 -1.26227560e-02
-2.14817867e-01 6.19813502e-01 4.89210010e-01 2.93018132e-01
-9.36555386e-01 2.40942553e-01 8.71804595e-01 5.74670658e-02
1.13228530e-01 -2.36275103e-02 -2.99518295e-02 8.08221221e-01
5.62088609e-01 -4.00361389e-01 -1.01158530e-01 -5.19624770e-01
-1.39178736e-02 3.60033661e-02 9.98039544e-01 -1.39989579e+00
6.29220784e-01 -5.66208176e-02 8.39116871e-01 -7.53619134e-01
5.66424668e-01 -1.22020221e+00 4.49266583e-01 7.12455392e-01
-2.43471488e-02 1.73463076e-01 4.56893742e-01 8.44592154e-01
-1.52304128e-01 -2.91351020e-01 1.15876198e+00 -2.64592767e-01
-1.00727487e+00 7.89643452e-02 -6.27098322e-01 -1.31621197e-01
1.32366490e+00 -5.88477135e-01 -4.06431019e-01 -2.07779571e-01
-4.48388666e-01 -3.68782505e-03 7.70214975e-01 2.73674756e-01
6.74589574e-01 -1.10381126e+00 -3.13409388e-01 2.67698050e-01
-3.56483310e-02 -1.72830105e-01 7.01100528e-02 7.80689597e-01
-6.42122984e-01 5.18887401e-01 -4.77735430e-01 -9.87253249e-01
-1.37768209e+00 1.01442897e+00 4.39253509e-01 2.23034069e-01
-8.96399140e-01 6.23966873e-01 1.09887145e-01 3.38165790e-01
5.51842332e-01 -3.51188093e-01 1.47521989e-02 -1.04193330e-01
7.67623007e-01 2.69516140e-01 5.26068471e-02 -4.33429062e-01
-4.29158807e-01 7.25520492e-01 -1.37077679e-03 4.40450311e-02
9.60410833e-01 -2.28380501e-01 5.16548514e-01 5.49144149e-01
7.07323968e-01 -2.72817194e-01 -1.90038180e+00 -3.80856842e-02
1.78001858e-02 -6.59909666e-01 2.37708941e-01 -5.73428571e-01
-9.78418946e-01 1.02353394e+00 9.90972579e-01 1.54748023e-01
1.20455635e+00 -1.39230952e-01 7.51235604e-01 2.91602582e-01
2.70432442e-01 -8.58132005e-01 2.97454149e-01 4.09290731e-01
2.40525618e-01 -1.53510022e+00 -7.35994279e-02 -2.35098302e-01
-3.26718032e-01 1.25630689e+00 5.95423281e-01 -9.48944315e-02
4.00750250e-01 6.95515156e-01 1.95694536e-01 -1.44583359e-01
-8.16530824e-01 -7.40542233e-01 2.27627531e-01 9.61857557e-01
1.74630404e-01 -3.62785190e-01 1.82351917e-01 -3.02656114e-01
3.84009749e-01 2.91000843e-01 6.26196563e-01 1.02499413e+00
-5.40662885e-01 -6.26977861e-01 -7.50833452e-01 1.32089630e-01
-6.63570881e-01 2.18596503e-01 -2.28264123e-01 6.82374835e-01
1.22898415e-01 9.52737629e-01 3.20163816e-01 2.98318546e-02
5.07322848e-01 -6.06873110e-02 7.13283062e-01 -3.97570699e-01
-6.31553590e-01 9.94351506e-02 -2.07224324e-01 -5.87979555e-01
-8.91794562e-01 -9.18412626e-01 -9.76700604e-01 -1.31957754e-01
-3.42085212e-01 -5.62088788e-01 5.70933163e-01 8.41318905e-01
1.40523985e-01 5.28052092e-01 1.53243676e-01 -9.84964371e-01
-4.53251481e-01 -5.92687070e-01 -4.72078145e-01 5.46754301e-01
8.38466287e-01 -8.85712028e-01 -4.18758839e-01 5.20143032e-01] | [8.48865795135498, -1.206282615661621] |
dbe71648-245e-49ae-af52-6f007aceba31 | consensus-clustering-with-unsupervised-1 | 2010.01245 | null | https://arxiv.org/abs/2010.01245v2 | https://arxiv.org/pdf/2010.01245v2.pdf | Consensus Clustering With Unsupervised Representation Learning | Recent advances in deep clustering and unsupervised representation learning are based on the idea that different views of an input image (generated through data augmentation techniques) must either be closer in the representation space, or have a similar cluster assignment. Bootstrap Your Own Latent (BYOL) is one such representation learning algorithm that has achieved state-of-the-art results in self-supervised image classification on ImageNet under the linear evaluation protocol. However, the utility of the learnt features of BYOL to perform clustering is not explored. In this work, we study the clustering ability of BYOL and observe that features learnt using BYOL may not be optimal for clustering. We propose a novel consensus clustering based loss function, and train BYOL with the proposed loss in an end-to-end way that improves the clustering ability and outperforms similar clustering based methods on some popular computer vision datasets. | ['Urun Dogan', 'Eren Manavoglu', 'Aniket Anand Deshmukh', 'Jayanth Reddy Regatti'] | 2020-10-03 | consensus-clustering-with-unsupervised | https://openreview.net/forum?id=x9C7Nlwgydy | https://openreview.net/pdf?id=x9C7Nlwgydy | null | ['self-supervised-image-classification'] | ['computer-vision'] | [ 1.04205728e-01 -7.73870274e-02 -1.16985522e-01 -8.10821354e-01
-8.15732539e-01 -5.38199306e-01 6.29537702e-01 1.51585698e-01
-1.97316766e-01 2.33185381e-01 2.80840844e-01 2.41927221e-01
-3.81098181e-01 -3.21146220e-01 -7.19521821e-01 -1.09375477e+00
-1.88972384e-01 9.00545597e-01 -1.62081867e-01 4.35677528e-01
9.84253958e-02 5.05661331e-02 -1.66226840e+00 6.71356976e-01
9.52091038e-01 7.73959339e-01 2.37569630e-01 1.70976952e-01
1.80185974e-01 9.64759588e-01 -3.25130582e-01 1.00698039e-01
3.08131456e-01 -6.21457517e-01 -7.99283028e-01 4.93573517e-01
7.47906208e-01 1.78964883e-01 -5.77889793e-02 1.02411211e+00
3.41902554e-01 2.64808357e-01 1.04815102e+00 -1.50417209e+00
-6.79491758e-01 9.86526489e-01 -7.11533248e-01 -1.89634740e-01
-2.32396230e-01 -1.67769581e-01 1.04189062e+00 -7.68589675e-01
6.90595031e-01 1.27182972e+00 6.89002395e-01 5.65958738e-01
-1.62780344e+00 -9.24044788e-01 2.37834394e-01 3.42277944e-01
-1.54697657e+00 -2.41129726e-01 8.83528292e-01 -7.13582814e-01
4.60316896e-01 -2.31975153e-01 2.97789931e-01 1.04327059e+00
-2.53908575e-01 7.95158863e-01 1.44174314e+00 -3.62289846e-01
4.91804779e-01 5.69992214e-02 3.27888906e-01 6.88317835e-01
5.11754677e-02 -1.60713330e-01 -3.35380524e-01 1.29301533e-01
3.52365375e-01 3.48236054e-01 -1.44782186e-01 -1.22075307e+00
-1.01435935e+00 1.17164624e+00 9.55012679e-01 1.84031725e-01
-1.99680567e-01 4.62710232e-01 4.43914384e-01 2.60060966e-01
4.45999652e-01 4.75690126e-01 -1.57059833e-01 3.24252576e-01
-1.20236015e+00 -2.02484027e-01 5.31322718e-01 6.98297203e-01
9.19861555e-01 -2.26497114e-01 1.21629275e-01 8.69125009e-01
2.63988912e-01 9.31846052e-02 7.88113892e-01 -1.13937020e+00
1.72533244e-01 8.50088000e-01 -3.77502948e-01 -8.56594861e-01
-3.50764990e-01 -5.37708580e-01 -1.31969905e+00 5.73274374e-01
1.57303542e-01 4.73738648e-03 -1.14992559e+00 1.93484962e+00
-3.77920666e-03 3.02580804e-01 2.68535644e-01 7.85515249e-01
6.21724486e-01 4.65446264e-01 -3.78231257e-02 -1.08730368e-01
7.67041206e-01 -1.20985520e+00 -5.34305274e-01 -1.17930546e-02
5.80307543e-01 -5.00132442e-01 9.25989509e-01 5.50382733e-01
-6.54663086e-01 -7.85810888e-01 -1.08818054e+00 3.62002194e-01
-2.56266236e-01 3.53817970e-01 6.75592482e-01 5.08147538e-01
-1.10916460e+00 5.62114537e-01 -1.05759835e+00 -5.62125921e-01
7.36814499e-01 3.58038962e-01 -5.41012168e-01 -2.49983549e-01
-4.70374942e-01 5.07501245e-01 6.15304351e-01 -2.75176734e-01
-1.19391727e+00 -5.35857499e-01 -8.14003289e-01 6.21452853e-02
3.90382290e-01 -4.21132058e-01 5.74690878e-01 -1.39090514e+00
-1.17787611e+00 9.52602386e-01 4.31481488e-02 -8.01149428e-01
4.57207382e-01 -2.48089373e-01 1.08386777e-01 3.36565793e-01
3.04157466e-01 1.06812584e+00 1.03602922e+00 -1.79054463e+00
-3.14805895e-01 -4.73165572e-01 -4.62773114e-01 3.23437750e-01
-3.70322198e-01 -2.87371486e-01 -3.69851142e-01 -4.09094781e-01
3.38825405e-01 -1.20650053e+00 -5.21244824e-01 -9.46302265e-02
-5.75198710e-01 -4.12852585e-01 9.23887670e-01 -1.86507747e-01
7.40971625e-01 -2.24396467e+00 2.61426628e-01 4.59860295e-01
3.11486721e-01 1.54311538e-01 -6.00131527e-02 2.85792708e-01
-3.17005605e-01 1.48777843e-01 -4.39090371e-01 -7.02205300e-01
-7.33586326e-02 1.32368699e-01 -2.49535188e-01 6.38519824e-01
-6.19593114e-02 6.79457963e-01 -8.37450624e-01 -4.69603628e-01
4.99646753e-01 3.85334939e-01 -6.76058412e-01 2.53284276e-01
-4.69026342e-02 5.45121193e-01 8.12900364e-02 2.21745834e-01
6.15968287e-01 -5.08637428e-01 5.15128493e-01 -2.39772648e-01
3.32574785e-01 -1.23283148e-01 -1.10289168e+00 1.89741755e+00
-3.25299114e-01 6.94126010e-01 -2.11040899e-01 -1.37527454e+00
9.19953227e-01 9.74504054e-02 6.21247530e-01 -1.05496682e-01
-3.12432200e-02 -2.14841291e-01 1.47920907e-01 -1.63637817e-01
-1.11444928e-01 5.34654781e-02 -2.58701500e-02 6.50793135e-01
2.41316333e-01 2.15244114e-01 -2.11430285e-02 4.66069162e-01
8.95653307e-01 9.76848379e-02 2.20747013e-02 -3.99331689e-01
2.02567175e-01 -4.15089913e-02 5.37583709e-01 8.78737628e-01
6.49445876e-03 1.05168629e+00 3.13653499e-01 4.04030010e-02
-1.11498022e+00 -1.09433496e+00 -1.49014384e-01 1.02218282e+00
-2.30788183e-03 -3.58288467e-01 -8.91290367e-01 -8.91874433e-01
-6.97677508e-02 3.94278914e-01 -9.27360058e-01 -3.37252080e-01
-1.87520042e-01 -6.85993671e-01 4.76453453e-01 7.16651201e-01
6.25031531e-01 -9.86581683e-01 -4.60268617e-01 6.77398220e-02
-2.02669114e-01 -9.81715500e-01 -2.87270963e-01 6.04373336e-01
-8.36584926e-01 -1.19430196e+00 -4.15295929e-01 -1.05994844e+00
1.22025108e+00 3.84849340e-01 9.80053306e-01 9.00350045e-03
-3.01020771e-01 3.72600675e-01 -6.73020244e-01 5.52265048e-02
-3.38402212e-01 2.38959923e-01 1.85773686e-01 3.54111612e-01
3.59677792e-01 -5.68677664e-01 -6.45682752e-01 2.47647524e-01
-7.84222245e-01 9.20752436e-03 5.43720722e-01 9.53710020e-01
6.56705379e-01 4.32729959e-01 5.90022206e-01 -1.21235716e+00
2.97259808e-01 -6.52884126e-01 -2.19992325e-01 1.52259663e-01
-8.52034748e-01 3.01572889e-01 8.71424556e-01 -3.67256135e-01
-7.05360353e-01 4.91314858e-01 3.00103098e-01 -1.13018787e+00
-3.30395252e-01 5.71314931e-01 -9.68691260e-02 2.74610460e-01
7.10445881e-01 1.00706369e-01 4.28172022e-01 -5.40880382e-01
8.24904442e-01 4.65192646e-01 6.68246627e-01 -5.63707829e-01
8.17084849e-01 8.52094829e-01 -2.34052196e-01 -4.42207545e-01
-1.01654744e+00 -8.61409009e-01 -1.07041693e+00 -5.55013679e-02
1.05453038e+00 -1.09936512e+00 -4.44339901e-01 1.81405127e-01
-5.24475276e-01 -3.25456858e-01 -2.39751399e-01 4.78271753e-01
-6.89956903e-01 4.53442842e-01 -1.25072837e-01 -6.74871325e-01
-1.85055420e-01 -1.21611202e+00 7.37578630e-01 -1.80415455e-02
-1.59096897e-01 -9.00086045e-01 -5.56826591e-04 5.02865970e-01
1.36917219e-01 5.29572904e-01 7.63804555e-01 -7.75205791e-01
-4.24792051e-01 -1.15088649e-01 -6.82144165e-02 8.45709682e-01
1.77162021e-01 5.02344966e-02 -9.79242444e-01 -8.90669227e-01
-1.75802350e-01 -8.45384657e-01 1.60527873e+00 5.30769289e-01
1.47974181e+00 -3.02326858e-01 -4.87937540e-01 8.12902093e-01
1.73470914e+00 -9.45025086e-02 5.59422553e-01 6.66289553e-02
7.83921063e-01 7.48074472e-01 3.08008909e-01 1.57856539e-01
3.14140081e-01 4.91785914e-01 4.67140377e-01 -2.01317266e-01
-2.56375641e-01 -1.94177791e-01 3.68953735e-01 8.30865800e-01
1.16163887e-01 8.76210853e-02 -9.46890295e-01 6.87736094e-01
-2.32119584e+00 -1.05496728e+00 4.33134139e-02 2.14387298e+00
7.52115726e-01 2.09516566e-02 4.05580625e-02 6.06572516e-02
9.28978980e-01 -1.05550043e-01 -6.36550069e-01 7.36118555e-02
-1.04782559e-01 2.56711483e-01 3.01387995e-01 2.68338084e-01
-1.42577386e+00 1.07347143e+00 5.71682549e+00 8.08764935e-01
-1.02923357e+00 1.66191593e-01 9.00258064e-01 1.17245153e-01
1.90072328e-01 1.68760374e-01 -5.30280709e-01 4.46362197e-01
6.16979122e-01 1.33821175e-01 2.54887849e-01 1.12318325e+00
-1.51810292e-02 5.19562047e-03 -1.37783098e+00 9.46340263e-01
3.92966330e-01 -1.33867145e+00 1.77421361e-01 5.60896695e-02
1.20955753e+00 2.01689839e-01 4.24058944e-01 2.29868531e-01
8.42270136e-01 -1.33251143e+00 3.70510727e-01 4.32546407e-01
7.37916172e-01 -9.70237851e-01 7.64505148e-01 3.01568806e-01
-9.82045591e-01 -1.73872873e-01 -5.33049822e-01 2.23876551e-01
-4.35015321e-01 3.72018963e-01 -9.75785613e-01 3.66116196e-01
9.54424143e-01 1.02441728e+00 -1.12770164e+00 1.10150850e+00
-1.46908179e-01 1.04527175e+00 -2.69895494e-01 1.71337500e-01
5.68433881e-01 -3.15064520e-01 5.52729368e-02 1.10659611e+00
-1.60149693e-01 -4.55226481e-01 5.52446425e-01 9.53585029e-01
-2.45680884e-01 -1.50236160e-01 -7.02621520e-01 1.47137851e-01
5.45189142e-01 1.38295746e+00 -9.81978536e-01 -3.51297319e-01
1.78018641e-02 9.33626413e-01 6.85287535e-01 3.01257968e-01
-6.17254853e-01 -1.46523312e-01 4.78024811e-01 2.75574625e-03
4.82290655e-01 -5.54704554e-02 -2.21512794e-01 -1.02017486e+00
-2.78810441e-01 -7.36470819e-01 5.00458300e-01 -6.55574262e-01
-1.77635133e+00 4.63114262e-01 2.08857171e-02 -1.53387308e+00
-1.75118938e-01 -1.99421376e-01 -7.21099913e-01 1.27181813e-01
-1.07301271e+00 -1.26201510e+00 -4.45946127e-01 7.86990821e-01
5.78581214e-01 -5.93946397e-01 8.64395559e-01 1.36426678e-02
-4.42610085e-01 6.88946843e-01 8.47461224e-01 4.61531520e-01
8.73283684e-01 -1.64166224e+00 -8.17546993e-02 6.59649849e-01
4.35365230e-01 7.49236345e-01 3.67085427e-01 -3.24863315e-01
-7.91437864e-01 -1.58618093e+00 1.69098362e-01 -6.03920639e-01
2.57983953e-01 -5.88992596e-01 -8.32891345e-01 6.52568936e-01
5.34944415e-01 2.50733644e-02 1.01025188e+00 2.40394980e-01
-7.79098094e-01 -1.22477226e-01 -1.10183203e+00 4.38891530e-01
7.55408168e-01 -4.09320414e-01 -5.96485794e-01 3.63189012e-01
5.84924877e-01 2.85234690e-01 -7.61921525e-01 3.48209172e-01
2.59025216e-01 -9.60255623e-01 9.21692729e-01 -6.26139224e-01
5.73841572e-01 -6.34563565e-01 -1.80623412e-01 -1.51394188e+00
-5.58621287e-01 -2.92358190e-01 1.83959767e-01 1.31045854e+00
5.28725162e-02 -1.58591151e-01 1.02896631e+00 1.65993515e-02
-1.39292076e-01 -7.72469401e-01 -7.60492861e-01 -6.88408136e-01
4.03424025e-01 -6.10781647e-02 4.18517813e-02 1.41601264e+00
-9.27676260e-02 6.53341115e-01 -2.56576657e-01 4.83396612e-02
1.26091731e+00 3.17629248e-01 9.00724590e-01 -1.51970327e+00
3.80595475e-02 -3.51150244e-01 -6.73536718e-01 -6.60803437e-01
5.15392900e-01 -1.40471184e+00 8.21482688e-02 -1.64744020e+00
6.76896930e-01 -5.26088834e-01 -4.79706019e-01 5.33853412e-01
-8.45366046e-02 6.02175176e-01 3.55167150e-01 6.66196585e-01
-1.20061660e+00 5.98172009e-01 7.03531384e-01 -3.28878671e-01
3.46179344e-02 -3.30671400e-01 -7.02500403e-01 8.37960958e-01
8.07993770e-01 -5.86285889e-01 -5.88398635e-01 -1.17694810e-01
-2.24753901e-01 -4.68948871e-01 3.29367518e-01 -1.41067266e+00
4.11481053e-01 2.55791217e-01 7.17965603e-01 -7.82575190e-01
2.30674356e-01 -9.21136558e-01 -6.23929091e-02 4.43435758e-01
-8.40593159e-01 -1.86387703e-01 -4.49694097e-01 9.74411786e-01
-3.43978703e-01 -7.67113119e-02 1.22947717e+00 -2.98968285e-01
-5.93332231e-01 3.22437346e-01 -3.51531714e-01 1.43590316e-01
1.17039824e+00 -2.68417746e-01 -1.78664774e-01 -3.92367661e-01
-9.84046459e-01 4.97483760e-01 6.17755711e-01 3.89207691e-01
6.61187410e-01 -1.43206346e+00 -8.23284686e-01 5.54795340e-02
3.12734842e-01 1.05927162e-01 -1.01941396e-02 6.06728673e-01
-4.03770059e-01 8.27066973e-02 -2.35351756e-01 -1.09438109e+00
-1.23979580e+00 6.90224409e-01 2.74515271e-01 -2.95390040e-01
-7.51115024e-01 8.65496874e-01 2.86571354e-01 -7.89358854e-01
6.39091611e-01 -2.91195465e-03 -3.19311947e-01 2.21930444e-01
1.73052356e-01 1.17619984e-01 -7.88246319e-02 -5.78865528e-01
-2.53988475e-01 5.81876695e-01 -4.75164890e-01 2.59034559e-02
1.43314135e+00 -2.64263153e-02 -1.59484714e-01 5.58373809e-01
1.57164562e+00 -6.09235108e-01 -1.43248999e+00 -3.76652420e-01
1.12002358e-01 -1.96391329e-01 -1.21196382e-01 -7.28845894e-01
-1.24349868e+00 1.00460422e+00 1.07867789e+00 3.57763581e-02
8.14723134e-01 2.67669171e-01 4.45796072e-01 6.15548790e-01
2.24166915e-01 -1.31128347e+00 6.37925506e-01 2.51964927e-01
6.41730189e-01 -1.44937742e+00 3.66442204e-02 -1.26330420e-01
-1.00835204e+00 8.48476946e-01 8.71759534e-01 -6.72527790e-01
8.52073729e-01 -2.23530605e-02 3.29361618e-01 -4.20490474e-01
-8.50703716e-01 -4.51587081e-01 1.44938409e-01 8.45372260e-01
3.63878548e-01 1.05184704e-01 8.60518366e-02 2.25571543e-01
-2.36347675e-01 -4.45978284e-01 4.04564083e-01 5.61109900e-01
-3.55965912e-01 -7.79028773e-01 -2.85246521e-01 7.24974394e-01
-1.47556901e-01 2.97503714e-02 -6.28436446e-01 4.23093796e-01
2.14532644e-01 1.06630492e+00 2.95427293e-01 -4.93064672e-01
-3.08600008e-01 1.44895956e-01 3.22714478e-01 -9.26012814e-01
-4.67448205e-01 8.58225822e-02 -5.18440664e-01 -3.96231800e-01
-8.36716056e-01 -6.96415544e-01 -1.49474132e+00 1.45601794e-01
-5.06626189e-01 2.86971927e-01 4.96667713e-01 7.52778411e-01
3.15113962e-01 4.19130743e-01 1.04911172e+00 -9.22681630e-01
-4.48713839e-01 -1.03293705e+00 -4.88801688e-01 1.03914273e+00
2.01411605e-01 -7.30139375e-01 -5.26162088e-01 3.43451887e-01] | [9.319465637207031, 3.050753593444824] |
85117ebd-6ad2-4a97-bda4-297417c39093 | can-llms-express-their-uncertainty-an | 2306.13063 | null | https://arxiv.org/abs/2306.13063v1 | https://arxiv.org/pdf/2306.13063v1.pdf | Can LLMs Express Their Uncertainty? An Empirical Evaluation of Confidence Elicitation in LLMs | The task of empowering large language models (LLMs) to accurately express their confidence, referred to as confidence elicitation, is essential in ensuring reliable and trustworthy decision-making processes. Previous methods, which primarily rely on model logits, have become less suitable for LLMs and even infeasible with the rise of closed-source LLMs (e.g., commercialized LLM APIs). This leads to a growing need to explore the untapped area of \emph{non-logit-based} approaches to estimate the uncertainty of LLMs. Hence, in this study, we investigate approaches for confidence elicitation that do not require model fine-tuning or access to proprietary information. We introduce three categories of methods: verbalize-based, consistency-based, and their hybrid methods for benchmarking, and evaluate their performance across five types of datasets and four widely-used LLMs. Our analysis of these methods uncovers several key insights: 1) LLMs often exhibit a high degree of overconfidence when verbalizing their confidence; 2) Prompting strategies such as CoT, Top-K and Multi-step confidences improve calibration of verbalized confidence; 3) Consistency-based methods outperform the verbalized confidences in most cases, with particularly notable improvements on the arithmetic reasoning task; 4) Hybrid methods consistently deliver the best performance over their baselines, thereby emerging as a promising state-of-the-art approach; 5) Despite these advancements, all investigated methods continue to struggle with challenging tasks, such as those requiring professional knowledge, leaving significant scope for improvement of confidence elicitation. | ['Bryan Hooi', 'Junxian He', 'Jie Fu', 'Yifei Li', 'Xinyang Lu', 'Zhiyuan Hu', 'Miao Xiong'] | 2023-06-22 | null | null | null | null | ['benchmarking', 'arithmetic-reasoning', 'decision-making', 'benchmarking'] | ['miscellaneous', 'reasoning', 'reasoning', 'robots'] | [-2.09371254e-01 3.32095265e-01 -3.08357477e-01 -6.71441436e-01
-1.14436173e+00 -7.56190598e-01 9.64923799e-01 8.01102996e-01
-6.39463961e-01 8.57000649e-01 -2.40859184e-02 -7.46715009e-01
-3.60964745e-01 -6.79537356e-01 -6.36836469e-01 -1.25031561e-01
2.04252779e-01 5.56582332e-01 4.34636846e-02 1.14609562e-01
7.60679960e-01 1.83253020e-01 -1.23558283e+00 1.84403270e-01
1.43350291e+00 1.27365363e+00 -3.57125461e-01 3.64710152e-01
-4.90387470e-01 1.06491828e+00 -6.45544231e-01 -9.73443687e-01
-1.29740357e-01 8.25428814e-02 -8.12799931e-01 -6.28462493e-01
1.76658630e-01 -1.62743896e-01 7.03869641e-01 1.12442529e+00
1.17163293e-01 -1.31143048e-01 8.58167648e-01 -1.39215779e+00
-5.53932488e-01 9.68981147e-01 -4.46677983e-01 -1.92674473e-01
5.65008521e-01 1.95163816e-01 8.36926222e-01 -7.70010471e-01
1.60954729e-01 1.28803587e+00 9.79752541e-01 1.12277821e-01
-1.48614025e+00 -8.53275180e-01 2.37945423e-01 1.03305086e-01
-1.55563104e+00 -3.84478211e-01 4.04875576e-01 -7.13814974e-01
8.98787975e-01 3.61073792e-01 2.80321717e-01 1.48717606e+00
6.05969727e-02 3.90698671e-01 2.07880211e+00 -4.29779887e-01
6.35477066e-01 8.31004381e-01 4.64344114e-01 4.44500774e-01
6.44651115e-01 1.13710023e-01 -6.56144440e-01 -3.11410278e-01
4.86253947e-01 -4.96761382e-01 2.66412217e-02 -6.22745082e-02
-1.12350035e+00 1.06450498e+00 1.31353423e-01 2.27995783e-01
-3.74664456e-01 -1.28987087e-02 2.03266993e-01 2.79232949e-01
6.77465856e-01 6.96154475e-01 -6.08832300e-01 -6.03756487e-01
-1.32334566e+00 4.42578912e-01 1.01040721e+00 7.69071996e-01
4.81413424e-01 -1.94124028e-01 -4.46313351e-01 6.50448382e-01
5.86027026e-01 2.93520004e-01 2.60224521e-01 -8.31611156e-01
6.34839594e-01 6.36584044e-01 5.44760346e-01 -1.05734241e+00
-3.74087691e-01 -4.52817947e-01 -7.92648792e-01 5.19289255e-01
5.48746109e-01 -3.17179784e-02 -5.66991150e-01 1.79000521e+00
6.70118555e-02 1.19908109e-01 1.31114885e-01 5.08673489e-01
5.78220606e-01 3.10855120e-01 4.84332144e-01 -1.70520067e-01
1.20199347e+00 -6.52357101e-01 -5.80498397e-01 -1.92257375e-01
4.89817202e-01 -7.27416873e-01 1.36719441e+00 8.10322046e-01
-1.20852339e+00 -5.47928035e-01 -1.20010376e+00 3.53859454e-01
-6.26064420e-01 -3.25625807e-01 8.24555576e-01 8.74868155e-01
-8.56591046e-01 7.40791619e-01 -8.01906824e-01 2.41083540e-02
2.03567788e-01 4.89375629e-02 -2.42276132e-01 1.72009040e-02
-1.28659225e+00 1.37311721e+00 4.06103402e-01 1.49973452e-01
-5.31403124e-01 -9.75104272e-01 -8.60053062e-01 -1.59953032e-02
5.49607277e-01 -5.79577327e-01 1.41208637e+00 -7.40447462e-01
-1.51535487e+00 6.84103608e-01 1.91227034e-01 -5.47672212e-01
1.09318316e+00 -3.42405140e-01 -3.47889125e-01 -3.74597013e-01
-2.79592946e-02 5.89435458e-01 5.87547302e-01 -1.35194325e+00
-4.81213987e-01 7.07520768e-02 1.45851612e-01 -2.13679701e-01
-1.86129659e-01 2.60869324e-01 1.62133407e-02 -4.94758755e-01
-1.60587192e-01 -4.19914633e-01 -2.62274504e-01 -1.44422770e-01
-4.00956661e-01 -3.55556548e-01 8.10072049e-02 -4.21318054e-01
1.71031523e+00 -1.63854265e+00 -1.77425191e-01 4.52091396e-01
1.84393868e-01 4.02478576e-01 2.56881356e-01 4.84138489e-01
2.29517952e-01 5.88937879e-01 -2.46565402e-01 -5.94455302e-01
5.95100880e-01 1.11335084e-01 -2.57205933e-01 5.80012947e-02
4.71052110e-01 8.00822020e-01 -7.57628560e-01 -7.44358957e-01
4.10379857e-01 4.22363758e-01 -4.53451842e-01 4.19505835e-01
-3.49778444e-01 3.78604174e-01 -2.92515457e-01 5.18923938e-01
7.77852654e-01 -4.25895303e-01 -4.91656251e-02 1.00087542e-02
-4.41866100e-01 3.47446829e-01 -1.55574441e+00 1.27571404e+00
-5.96281350e-01 -6.00456595e-02 -7.17946962e-02 -6.60570920e-01
9.53595221e-01 2.40570724e-01 -1.54200807e-01 -3.59518677e-01
-3.28247845e-02 5.38307548e-01 -1.33078262e-01 -2.35145777e-01
3.07179332e-01 -9.17164516e-03 -3.69398177e-01 2.52074152e-01
-2.39027273e-02 -4.81731027e-01 4.55785245e-02 2.04312623e-01
7.04454660e-01 8.77176970e-02 5.82815349e-01 -4.99182999e-01
6.60567045e-01 -1.55270159e-01 3.13613415e-01 8.34864974e-01
-2.57359266e-01 2.82077312e-01 8.07690680e-01 -4.96643484e-02
-5.89345694e-01 -1.00418377e+00 -3.79625767e-01 8.26914728e-01
-3.28566700e-01 -5.13921797e-01 -9.12772596e-01 -5.17327845e-01
1.47692695e-01 1.15623569e+00 -5.56898236e-01 -6.10563494e-02
9.33666006e-02 -4.70702618e-01 6.95007861e-01 6.52491152e-01
6.31147504e-01 -8.00987363e-01 -6.73424244e-01 3.10129106e-01
-1.32630169e-01 -1.08762336e+00 4.35885116e-02 2.88519293e-01
-6.27573490e-01 -9.94320989e-01 -5.01746893e-01 3.79821435e-02
4.14723009e-01 -5.10029912e-01 1.48657835e+00 6.45386577e-02
5.08796036e-01 3.77577931e-01 -2.86404312e-01 -6.59732580e-01
-6.02933884e-01 1.95715169e-04 3.60592343e-02 -2.12948367e-01
4.58364218e-01 -4.71858174e-01 -5.14722168e-01 2.49746293e-01
-7.20951259e-01 1.84393212e-01 6.09072566e-01 4.87517595e-01
3.45558912e-01 -2.73594230e-01 8.34937572e-01 -1.08924389e+00
1.20682490e+00 -5.26463270e-01 -9.14825618e-01 3.57208490e-01
-1.45988488e+00 2.25379005e-01 5.07877231e-01 -4.43995804e-01
-1.06400967e+00 -6.93185806e-01 -2.30393499e-01 9.05789658e-02
-2.81628400e-01 1.02648962e+00 2.06181049e-01 -2.36833040e-02
6.68483555e-01 -3.09911668e-01 -3.00039411e-01 -2.97294915e-01
3.21137309e-01 7.62666345e-01 6.29161119e-01 -1.14863563e+00
6.04348719e-01 -1.54612735e-01 -3.28983128e-01 -6.42254427e-02
-9.51210678e-01 -6.49867579e-04 -4.11224127e-01 -7.49972165e-02
6.43470407e-01 -7.81012058e-01 -8.96386504e-01 3.06603104e-01
-9.98755932e-01 -4.56224054e-01 2.51412038e-02 4.89766538e-01
-3.93257201e-01 3.48818153e-01 -4.71722156e-01 -1.28232789e+00
-5.61250567e-01 -1.16607606e+00 7.52372086e-01 2.72930831e-01
-8.55484962e-01 -1.07114995e+00 -1.35992998e-02 3.92797530e-01
9.37670648e-01 4.49957281e-01 1.25349641e+00 -9.53681290e-01
-2.85622269e-01 -2.56825954e-01 -2.51542628e-01 5.21399200e-01
-3.62328887e-01 2.77359813e-01 -1.10112000e+00 -1.37933921e-02
-1.30964682e-01 -7.65901148e-01 3.08880031e-01 1.71567231e-01
1.17565227e+00 -3.37876737e-01 5.73024116e-02 2.95476049e-01
1.15667295e+00 -2.73734868e-01 3.54330689e-01 4.71859157e-01
-1.65701304e-02 6.79232121e-01 8.45720053e-01 5.61657488e-01
7.09263206e-01 4.38054413e-01 3.37883234e-01 1.16327263e-01
4.26075041e-01 -4.33918059e-01 2.45884508e-01 9.06622887e-01
-1.71787024e-01 1.04165822e-01 -1.17479908e+00 2.62094401e-02
-1.89696133e+00 -5.32731533e-01 -1.72703013e-01 2.54920816e+00
1.27854764e+00 7.18624055e-01 -2.00925946e-01 2.99389392e-01
1.89825729e-01 5.99251548e-03 -3.76382321e-01 -7.49602795e-01
1.27896935e-01 5.25688708e-01 4.24712524e-02 6.58799052e-01
-7.27698386e-01 8.34945500e-01 6.31297493e+00 8.23613644e-01
-8.30950320e-01 6.02889843e-02 1.01656318e+00 3.64933521e-01
-6.36023462e-01 1.05360657e-01 -7.24388540e-01 4.62714732e-01
1.28647637e+00 -1.40202582e-01 2.73808420e-01 8.44243467e-01
1.80065602e-01 -5.13758421e-01 -1.36251569e+00 8.04195285e-01
-3.91156137e-01 -1.11419201e+00 -3.57865900e-01 -3.27366590e-01
6.93257570e-01 -3.75784606e-01 1.00330539e-01 8.25453222e-01
6.26754224e-01 -1.48050153e+00 1.03709030e+00 8.42765927e-01
8.81121039e-01 -7.46820211e-01 1.04914522e+00 6.37321115e-01
-8.47966969e-01 1.69005349e-01 -1.75591156e-01 -2.32965633e-01
2.12908983e-02 7.63873219e-01 -4.41508681e-01 5.16622365e-01
6.22567177e-01 7.65017271e-02 -7.99920559e-01 6.74705267e-01
-4.01942134e-01 8.50160837e-01 -2.92010039e-01 -1.27254620e-01
3.39785397e-01 -1.37818754e-01 -1.97464958e-01 1.52073169e+00
1.57013908e-01 -1.07528307e-01 9.63834673e-03 1.47144759e+00
1.48453861e-01 7.21030980e-02 -1.73842311e-01 -1.09878175e-01
7.13632107e-01 1.24262941e+00 -4.18121845e-01 -4.13035929e-01
-4.30637509e-01 4.30651724e-01 6.11819685e-01 3.38604540e-01
-9.98521388e-01 7.71556869e-02 2.77583808e-01 -4.46207039e-02
-1.65383369e-01 -3.17813605e-01 -6.44690037e-01 -7.46530712e-01
3.30679081e-02 -1.24682462e+00 4.80846673e-01 -9.07061934e-01
-1.39561391e+00 6.14227772e-01 5.77057898e-01 -6.67147696e-01
-6.35023296e-01 -5.51438034e-01 -5.49658775e-01 1.18217528e+00
-1.67568481e+00 -1.07160974e+00 -5.60554147e-01 4.56741601e-01
2.18963757e-01 7.40742609e-02 1.16832447e+00 3.20867933e-02
-2.15587199e-01 7.63417542e-01 -3.69801551e-01 -3.21589887e-01
1.03012908e+00 -1.34919000e+00 2.23649949e-01 5.33200383e-01
6.37081079e-03 1.00539958e+00 7.21845567e-01 -6.25120163e-01
-9.68624651e-01 -8.04958403e-01 1.01847100e+00 -8.42457712e-01
8.22940946e-01 -5.81857324e-01 -1.11538506e+00 5.06639540e-01
9.87331793e-02 -4.25133184e-02 7.00751722e-01 3.45503688e-01
-6.39182806e-01 3.66663449e-02 -1.38247967e+00 4.79220390e-01
5.57154596e-01 -5.33954740e-01 -7.19581306e-01 6.14620149e-02
6.39328778e-01 -3.81322235e-01 -1.16761887e+00 5.39060116e-01
5.86175263e-01 -1.32933879e+00 7.67857730e-01 -4.37932789e-01
4.79907602e-01 -1.61526144e-01 -1.69607982e-01 -1.23026097e+00
-1.29317477e-01 -4.93126303e-01 -3.29766840e-01 1.66168547e+00
6.77646041e-01 -9.60507512e-01 3.22617769e-01 1.24853408e+00
1.04190931e-01 -8.76070559e-01 -8.55474293e-01 -4.79912847e-01
4.19787705e-01 -9.38979328e-01 7.45977104e-01 1.03923869e+00
4.49345671e-02 -4.03597914e-02 -7.55072087e-02 1.61201507e-01
4.73386586e-01 -1.81338117e-01 6.88375056e-01 -1.32656896e+00
-2.60134578e-01 -7.02767730e-01 -8.86600018e-02 -5.81317544e-01
8.81825164e-02 -5.64045131e-01 -1.34870246e-01 -1.33205163e+00
2.84664124e-01 -6.80495381e-01 -3.66644859e-01 3.64820987e-01
-3.95918339e-01 -1.45200416e-01 -4.64129038e-02 -1.27811179e-01
-6.88859165e-01 2.12591380e-01 6.51813805e-01 7.70491809e-02
1.63865507e-01 1.59808010e-01 -1.03530788e+00 8.89682353e-01
6.78155363e-01 -4.28270131e-01 -4.07726228e-01 -1.48165286e-01
6.21148407e-01 1.63248762e-01 5.23463190e-01 -1.12106526e+00
4.28463399e-01 -4.68383223e-01 2.42456883e-01 -4.74889129e-01
8.37348998e-02 -8.57133627e-01 2.88998455e-01 2.44418681e-01
-5.37180305e-01 2.94426143e-01 2.87973672e-01 3.57670158e-01
-2.75527894e-01 -4.74874437e-01 6.67272389e-01 -4.68684584e-02
-4.90293115e-01 -1.32806435e-01 -1.80419713e-01 1.74889982e-01
7.34308481e-01 -8.63074586e-02 -3.25269520e-01 -3.83341521e-01
-2.68122673e-01 2.86560893e-01 1.93663374e-01 3.58454883e-01
2.21128345e-01 -1.14201450e+00 -8.05586755e-01 -1.84188690e-02
3.01918060e-01 1.64615825e-01 -8.20228159e-02 1.01680565e+00
-1.97889283e-01 6.92149282e-01 1.22373790e-01 -6.74588621e-01
-7.91764975e-01 3.81006330e-01 1.72205567e-01 -6.83728099e-01
-1.32844120e-01 9.70689178e-01 -2.38858685e-01 -5.86326420e-01
5.48436046e-01 -9.68711495e-01 -1.22059502e-01 -6.00496568e-02
5.27138710e-01 4.04726475e-01 2.52745539e-01 -1.14009388e-01
-3.83786619e-01 4.13209319e-01 1.27324447e-01 -2.97272950e-01
9.39284444e-01 1.41512051e-01 6.57719299e-02 6.96355581e-01
5.56096792e-01 4.73681204e-02 -1.27111614e+00 3.82817886e-03
5.92764497e-01 -4.34161425e-01 -7.10378885e-02 -1.39187503e+00
-4.30479228e-01 8.40605021e-01 3.02377760e-01 1.41512558e-01
7.16720760e-01 -1.68135256e-01 -3.46120186e-02 2.20757946e-01
7.23205030e-01 -1.19265425e+00 -3.05240657e-02 3.95000577e-01
9.03168082e-01 -1.67797327e+00 -1.59774832e-02 -1.21030584e-01
-9.00061429e-01 9.17937458e-01 6.45908713e-01 4.01866764e-01
5.29459894e-01 4.25971597e-01 2.11107418e-01 4.70040329e-02
-9.04464900e-01 1.66581631e-01 4.55344588e-01 5.46404123e-01
8.39359641e-01 2.71947145e-01 -4.56533968e-01 1.45647740e+00
-3.63193750e-01 2.03180358e-01 7.14821815e-02 7.07518518e-01
-1.60269767e-01 -9.65974569e-01 -5.20759463e-01 3.86226475e-01
-4.77690667e-01 -2.23421961e-01 -2.08208933e-01 1.00820494e+00
-3.89333419e-03 1.39052272e+00 -4.14841652e-01 -2.11940899e-01
5.11954248e-01 2.97159493e-01 2.50101030e-01 -4.64789301e-01
-8.65758002e-01 -4.41103756e-01 3.49062324e-01 -6.31556928e-01
-1.90688699e-01 -4.49213326e-01 -1.02008057e+00 -6.36715293e-01
-4.43896532e-01 4.15520191e-01 8.34562957e-01 1.01206255e+00
5.92539385e-02 3.18372756e-01 3.48869562e-02 -5.11800945e-01
-1.39193594e+00 -1.26998699e+00 -2.02257127e-01 2.69984305e-01
-3.44248563e-02 -8.19267809e-01 -3.55464518e-01 -4.04076844e-01] | [9.903563499450684, 7.3771257400512695] |
bb19e14c-d522-4ae3-959e-85fc667878c3 | one-class-classification-for-wafer-map-using | 2107.08823 | null | https://arxiv.org/abs/2107.08823v1 | https://arxiv.org/pdf/2107.08823v1.pdf | One-Class Classification for Wafer Map using Adversarial Autoencoder with DSVDD Prior | Recently, semiconductors' demand has exploded in virtual reality, smartphones, wearable devices, the internet of things, robotics, and automobiles. Semiconductor manufacturers want to make semiconductors with high yields. To do this, manufacturers conduct many quality assurance activities. Wafer map pattern classification is a typical way of quality assurance. The defect pattern on the wafer map can tell us which process has a problem. Most of the existing wafer map classification methods are based on supervised methods. The supervised methods tend to have high performance, but they require extensive labor and expert knowledge to produce labeled datasets with a balanced distribution in mind. In the semiconductor manufacturing process, it is challenging to get defect data with balanced distribution. In this paper, we propose a one-class classification method using an Adversarial Autoencoder (AAE) with Deep Support Vector Data Description (DSVDD) prior, which generates random vectors within the hypersphere of DSVDD. We use the WM-811k dataset, which consists of a real-world wafer map. We compare the F1 score performance of our model with DSVDD and AAE. | ['Seong-Whan Lee', 'Ha Young Jo'] | 2021-07-15 | null | null | null | null | ['one-class-classification'] | ['miscellaneous'] | [-1.74385116e-01 -1.53786644e-01 -3.68040078e-03 -6.55383646e-01
-2.43513301e-01 -4.36731964e-01 2.18181804e-01 -1.92484692e-01
4.63759363e-01 5.50127804e-01 -2.65939087e-01 -3.44440848e-01
-2.15420574e-02 -1.24391699e+00 -6.54193759e-01 -7.05423474e-01
6.13078535e-01 5.29365242e-01 2.08714250e-02 -1.85137495e-01
4.72443163e-01 4.88364786e-01 -1.26823187e+00 4.19770211e-01
8.90855134e-01 1.42969775e+00 1.76198393e-01 2.46089235e-01
-4.02742028e-01 5.69371343e-01 -9.57833290e-01 -5.78884602e-01
4.31287587e-01 -1.45479009e-01 -4.43387628e-01 2.16612861e-01
-1.52282149e-01 -7.04347640e-02 -3.72592121e-01 1.43487525e+00
2.17853308e-01 -3.67849261e-01 1.03452373e+00 -1.60962021e+00
-9.45777655e-01 3.94922078e-01 -4.83824342e-01 -2.17885792e-01
9.65676382e-02 9.20264646e-02 5.66493809e-01 -7.43868530e-01
6.82016611e-01 9.82066453e-01 5.76724291e-01 3.11526656e-01
-7.07054138e-01 -8.68876457e-01 -1.94123492e-01 -3.18163410e-02
-1.35356534e+00 -1.11866944e-01 1.15498424e+00 -8.50751042e-01
5.02235770e-01 -4.48830798e-02 4.76627320e-01 1.28923702e+00
8.29600811e-01 6.61092639e-01 1.08677006e+00 -9.33054909e-02
5.56216598e-01 5.43776512e-01 1.01632826e-01 6.62439048e-01
2.37878338e-01 1.57533884e-01 -2.14047492e-01 2.37680987e-01
9.11169767e-01 4.79429901e-01 -1.10147253e-01 -4.64949399e-01
-9.60377634e-01 8.84210944e-01 3.84640396e-01 3.88124675e-01
-2.86614448e-01 -2.98748106e-01 3.32508564e-01 5.53518593e-01
5.36066771e-01 6.05176568e-01 -4.04468179e-01 -4.08297777e-03
-4.57637042e-01 2.59997956e-02 7.88552821e-01 1.24824333e+00
5.89607298e-01 2.02118114e-01 1.10095277e-01 6.84790969e-01
2.98664600e-01 5.90230286e-01 6.59656584e-01 -6.96801186e-01
4.17267472e-01 7.96576679e-01 2.61561990e-01 -1.27658498e+00
-1.41790330e-01 -3.23945701e-01 -1.03291297e+00 3.39685082e-01
-8.61344859e-02 -4.20542667e-03 -1.05550611e+00 9.96056497e-01
2.04379782e-01 -3.42510864e-02 2.82649189e-01 8.55180860e-01
7.91329384e-01 1.00855303e+00 -5.85217416e-01 1.43255606e-01
8.83976579e-01 -8.56912196e-01 -1.02551663e+00 7.49548748e-02
4.20720220e-01 -5.23936689e-01 9.74744916e-01 8.54406536e-01
-4.90739256e-01 -7.56991982e-01 -1.63542485e+00 2.68293679e-01
-7.15714157e-01 4.60344225e-01 8.64510775e-01 1.05707085e+00
-4.89913970e-01 1.00553107e+00 -8.35239947e-01 -2.33011190e-02
5.72685719e-01 3.32379192e-01 -5.62770784e-01 -3.37425172e-01
-8.76091540e-01 6.05465472e-01 2.63098657e-01 -2.21127477e-02
-8.20327759e-01 -5.31153321e-01 -1.05475485e+00 -3.80289257e-02
6.95315599e-02 -2.34682143e-01 9.44079220e-01 -7.19026923e-01
-1.74441421e+00 6.34513736e-01 2.24251062e-01 -6.80182362e-03
2.33121336e-01 9.04243365e-02 -8.89485121e-01 -2.09968403e-01
8.72512311e-02 3.86868834e-01 1.16848171e+00 -1.23276365e+00
-3.36357027e-01 -4.10347730e-01 -2.45696619e-01 -3.89261276e-01
-2.62204409e-01 -2.01118171e-01 -7.00331628e-02 -4.01612341e-01
4.82575268e-01 -6.63158059e-01 -5.15780188e-02 -2.58406520e-01
-7.33481050e-01 -1.33738875e-01 9.59558189e-01 -7.79803038e-01
7.60988951e-01 -2.36111665e+00 -5.50673604e-02 3.10153157e-01
9.79489982e-02 -4.61960174e-02 2.24294782e-01 2.55478472e-01
-1.49349585e-01 3.61142009e-02 -2.04669580e-01 -1.07968695e-01
1.43462405e-01 9.59707871e-02 -3.22352022e-01 4.52722371e-01
5.07004499e-01 6.95444584e-01 -7.21833289e-01 1.73423141e-02
3.91417682e-01 1.55501468e-02 -2.81789005e-01 1.45351842e-01
-1.82069261e-02 3.22459906e-01 -5.15664458e-01 1.28174412e+00
8.91440570e-01 -3.40375125e-01 -2.03235090e-01 -3.13908547e-01
2.04760537e-01 -1.87058166e-01 -1.11162961e+00 1.62582302e+00
-3.63071442e-01 4.86742198e-01 -4.83044423e-02 -1.04876971e+00
1.48687625e+00 5.02170138e-02 3.74374330e-01 -5.72047293e-01
4.71195549e-01 3.36920947e-01 -7.40008801e-02 -4.80668455e-01
4.77150291e-01 -3.63717258e-01 -3.19608748e-01 7.87400827e-03
-3.20576727e-02 -6.58568144e-01 -2.81072378e-01 -9.40808430e-02
1.32828736e+00 -2.41917729e-01 -2.67606694e-02 -3.22209328e-01
1.93173558e-01 3.12021691e-02 8.15011680e-01 2.09136590e-01
-1.91873968e-01 7.36568689e-01 5.54222465e-01 -5.61938167e-01
-1.13543940e+00 -1.25563383e+00 -2.14454547e-01 -2.56039321e-01
4.90039080e-01 3.29629332e-02 -6.62199616e-01 -7.81027615e-01
2.47103289e-01 6.68253064e-01 -4.34624910e-01 -4.52613354e-01
-6.50796294e-02 -5.01157403e-01 -3.17018591e-02 7.59096622e-01
5.05895674e-01 -1.01514590e+00 1.72606885e-01 4.66468297e-02
4.79483277e-01 -1.13113475e+00 -7.97500834e-02 7.65831098e-02
-5.71584880e-01 -9.36110258e-01 -5.00183821e-01 -8.77579749e-01
1.00263500e+00 -1.06005624e-01 9.98823822e-01 -5.40291965e-01
-4.89447445e-01 -2.31822968e-01 -4.26601201e-01 -9.01607990e-01
-6.09540462e-01 -9.37194377e-02 5.12599587e-01 1.37726039e-01
5.39450169e-01 -2.58717060e-01 -2.57727325e-01 6.96430564e-01
-5.83023310e-01 -5.95218986e-02 5.68672180e-01 9.45921361e-01
7.00384974e-01 8.74496102e-01 9.13145781e-01 -9.47276294e-01
5.28493762e-01 -7.00798810e-01 -7.48845816e-01 1.49400411e-02
-4.51734096e-01 -3.07637006e-01 6.94074273e-01 -9.71064717e-02
-7.97275066e-01 1.42722517e-01 -1.74095526e-01 -9.46817875e-01
-4.17026021e-02 3.88250113e-01 -7.74370670e-01 -8.57484490e-02
5.05492628e-01 9.06051882e-03 1.32566601e-01 -2.56053448e-01
-1.50009051e-01 1.17801654e+00 4.03040826e-01 -1.60179645e-01
8.46721709e-01 3.54106814e-01 -1.71695709e-01 -6.17776811e-01
-5.40912390e-01 -3.03215355e-01 -3.22560430e-01 -2.87389994e-01
8.73662829e-01 -7.31065333e-01 -7.86882818e-01 6.52428150e-01
-1.01635969e+00 -1.42202213e-01 -1.67582601e-01 3.96897376e-01
-4.53977495e-01 1.93617474e-02 -6.69148624e-01 -6.66868925e-01
9.51402006e-04 -1.41747737e+00 1.26352382e+00 2.49625787e-01
2.81630899e-03 -7.56253958e-01 -3.71121734e-01 3.29922289e-01
3.07098508e-01 4.89390999e-01 9.48111832e-01 -5.77572048e-01
-7.65305758e-01 -8.47011387e-01 -1.17293172e-01 8.30282807e-01
6.39855623e-01 -1.57524765e-01 -8.87996852e-01 -1.94626957e-01
4.86086249e-01 -1.12448134e-01 5.29378653e-01 3.90554488e-01
1.75704181e+00 1.70729905e-01 -5.20707071e-01 4.93855119e-01
1.30173886e+00 8.39612782e-01 7.51825631e-01 -1.21547811e-01
8.13044190e-01 6.64163947e-01 1.00093842e+00 1.67352691e-01
-1.31103963e-01 3.19399536e-01 5.79801738e-01 1.13745511e-01
3.77921462e-01 -2.70283133e-01 2.48359680e-01 1.25880885e+00
4.55455184e-01 -4.45103168e-01 -9.07176077e-01 4.27153885e-01
-1.58077598e+00 -4.45420474e-01 5.55357151e-02 1.89763427e+00
2.38867745e-01 4.39508975e-01 -4.45553154e-01 5.86419821e-01
7.72387207e-01 -8.94306898e-02 -6.92340910e-01 -4.17062879e-01
9.98819396e-02 2.89413184e-02 5.02443075e-01 1.15336746e-01
-1.18412662e+00 6.89155400e-01 5.81594992e+00 9.17503178e-01
-1.13622963e+00 -1.82561018e-02 7.62248516e-01 3.26577783e-01
-3.89340729e-01 -4.80247378e-01 -7.30646610e-01 8.88158381e-01
7.48804212e-01 1.49135485e-01 2.56601423e-01 1.17607212e+00
-3.98731351e-01 1.63126633e-01 -1.38184226e+00 1.37260139e+00
8.80406350e-02 -1.48358393e+00 -4.02747840e-01 1.22160412e-01
8.96621168e-01 -3.41293663e-01 2.16707662e-01 4.19366330e-01
2.86004394e-01 -1.37145412e+00 3.67401928e-01 5.26315153e-01
7.15374172e-01 -9.08769846e-01 1.23525059e+00 2.77490318e-01
-7.92600334e-01 -2.12732777e-01 -6.50050044e-01 4.94004227e-02
6.16495237e-02 1.21161020e+00 -9.16553974e-01 7.35986888e-01
6.58854723e-01 9.36596513e-01 3.19230407e-02 5.19099295e-01
-1.74984708e-02 3.09896141e-01 -9.15729776e-02 -3.46230030e-01
-1.19931005e-01 -6.40450776e-01 2.77230024e-01 5.24135470e-01
7.56923497e-01 -2.24387795e-01 1.96291894e-01 1.14180005e+00
-2.56174356e-01 -1.34176686e-01 -1.12261641e+00 -4.84402090e-01
4.87660080e-01 9.89613712e-01 -8.17312181e-01 4.78765666e-02
-3.35444570e-01 1.13197327e+00 -1.89136744e-01 -6.98292209e-03
-7.00629175e-01 -7.98174560e-01 8.44421148e-01 7.07985908e-02
2.03631669e-01 -1.05621055e-01 -5.58187485e-01 -9.52046514e-01
3.77859354e-01 -7.50966132e-01 -2.44842559e-01 -9.11541045e-01
-1.75983679e+00 4.24610376e-01 -4.94508445e-01 -1.58358920e+00
9.71667618e-02 -1.29834187e+00 -6.20937943e-01 8.06269586e-01
-1.16362190e+00 -7.99893320e-01 -4.58391547e-01 1.32113501e-01
6.50160432e-01 -7.76889026e-01 7.94643819e-01 3.68638188e-01
-6.70431852e-01 4.49384600e-01 3.14735532e-01 2.53412277e-01
4.61310983e-01 -1.19952130e+00 7.24366188e-01 4.49005127e-01
-1.82917073e-01 2.69203961e-01 5.41346729e-01 -9.76972640e-01
-1.56785774e+00 -1.34343600e+00 2.79437095e-01 -6.88094437e-01
5.37997663e-01 -6.29109442e-01 -6.93037629e-01 6.50085986e-01
-8.67460296e-02 2.22657084e-01 4.95925725e-01 -9.62685719e-02
1.03472307e-01 -3.69260430e-01 -1.68393207e+00 2.20328853e-01
8.73506725e-01 -5.17545879e-01 -1.91751599e-01 7.05880940e-01
9.76626992e-01 -5.44812560e-01 -1.14557099e+00 6.04221523e-01
1.00362413e-01 -8.33872736e-01 7.02115119e-01 -3.65168810e-01
9.54201221e-01 -3.92855495e-01 -3.69502097e-01 -1.68698955e+00
-1.34629726e-01 -1.80708155e-01 2.76507474e-02 1.27361333e+00
3.79225135e-01 -8.14632475e-01 1.31610906e+00 3.74036342e-01
-5.16958594e-01 -1.00968468e+00 -6.67712688e-01 -1.11673498e+00
-1.14755832e-01 -3.67661566e-01 1.18731844e+00 1.13780010e+00
-1.41884670e-01 -2.29497440e-03 -2.42529988e-01 4.55297798e-01
3.16888064e-01 9.10804123e-02 7.03967452e-01 -1.46430218e+00
-3.29459816e-01 -8.41484368e-02 -1.00154245e+00 -5.82611144e-01
2.62729168e-01 -8.36127520e-01 7.92297721e-02 -1.45266175e+00
-2.11562842e-01 -6.90864325e-01 -3.29217702e-01 8.09495375e-02
2.26987943e-01 -9.38717052e-02 -2.89369822e-01 -5.11314496e-02
-1.13171920e-01 6.83400989e-01 1.29243064e+00 -6.93711638e-01
-2.15770677e-02 2.03568619e-02 -5.58249712e-01 7.03794241e-01
9.39244449e-01 -3.28081489e-01 -5.89712441e-01 -2.13742003e-01
2.30286986e-01 1.17665432e-01 -6.19168542e-02 -1.11790478e+00
2.06014916e-01 -4.73163873e-02 6.83417201e-01 -6.94304287e-01
4.39227849e-01 -9.45687294e-01 2.98286438e-01 3.80743325e-01
3.04038137e-01 -1.47805631e-01 -9.41585004e-02 6.92868292e-01
-4.44918126e-01 -3.52570891e-01 4.96751815e-01 -9.76040959e-02
-5.50546885e-01 5.19627392e-01 -1.73590943e-01 -6.03461146e-01
1.26529133e+00 -3.77391458e-01 -3.74565721e-01 -1.57873198e-01
-4.34087783e-01 2.71964282e-01 6.20877206e-01 6.80123448e-01
8.25063050e-01 -1.65497196e+00 -3.24642688e-01 6.43491387e-01
4.09269065e-01 2.89759189e-01 3.91638845e-01 2.09051237e-01
-7.00230598e-01 2.44162172e-01 -4.02760684e-01 -5.25184572e-01
-6.03354216e-01 7.33094573e-01 1.57112896e-01 -8.82039145e-02
-5.51528275e-01 9.53615010e-01 7.93296099e-02 -8.49627435e-01
-1.89434756e-02 -5.20106256e-01 -9.08989981e-02 -2.36961365e-01
4.07643974e-01 2.27608502e-01 5.22513568e-01 -4.67786863e-02
-1.76889196e-01 4.11371440e-01 -1.75552055e-01 3.01653415e-01
1.21767974e+00 4.61285621e-01 7.33279288e-02 5.51522017e-01
1.36583316e+00 -4.19783108e-02 -9.46711898e-01 3.08723807e-01
-2.04482734e-01 -6.24285758e-01 1.08226709e-01 -6.47018909e-01
-1.43810582e+00 8.08721304e-01 6.72456980e-01 4.55533743e-01
8.06216478e-01 -1.81469932e-01 9.24825013e-01 4.45489377e-01
9.56841171e-01 -1.20006323e+00 1.83438316e-01 2.05923647e-01
9.55134094e-01 -1.51446939e+00 -2.75813073e-01 -9.54031825e-01
-7.40810335e-01 9.43351984e-01 6.99006319e-01 -3.00746351e-01
9.89870965e-01 6.03916883e-01 1.34491220e-01 -3.98343682e-01
-2.11462051e-01 5.24071932e-01 -1.35917857e-01 7.84531891e-01
-1.79424122e-01 3.57187837e-01 3.29059511e-01 1.20108807e+00
-5.49564004e-01 -1.35678083e-01 4.81592447e-01 1.01728129e+00
-2.73018122e-01 -1.15909874e+00 -2.73832411e-01 8.27700913e-01
-2.23331749e-01 3.92354488e-01 -3.39036494e-01 5.08474469e-01
8.30274224e-02 9.26377892e-01 4.21709955e-01 -9.55651522e-01
2.96917588e-01 1.23322019e-02 2.96317190e-01 -6.56501412e-01
-2.99669914e-02 -7.51285911e-01 1.95957590e-02 -5.76815248e-01
1.96211189e-01 -5.01184225e-01 -1.07980371e+00 -2.36568332e-01
-3.32426578e-01 -5.62107414e-02 1.25086391e+00 7.74142504e-01
3.99456531e-01 9.15438592e-01 1.10181892e+00 -6.59166694e-01
-5.46210170e-01 -8.86119664e-01 -1.23549557e+00 2.90205598e-01
-8.37352350e-02 -9.17394161e-01 -1.69746906e-01 -3.06596488e-01] | [7.3827924728393555, 2.015512228012085] |
bcd7d73a-f689-446b-822a-0b56c368c2cf | black-box-variational-inference-converges | 2305.15349 | null | https://arxiv.org/abs/2305.15349v1 | https://arxiv.org/pdf/2305.15349v1.pdf | Black-Box Variational Inference Converges | We provide the first convergence guarantee for full black-box variational inference (BBVI), also known as Monte Carlo variational inference. While preliminary investigations worked on simplified versions of BBVI (e.g., bounded domain, bounded support, only optimizing for the scale, and such), our setup does not need any such algorithmic modifications. Our results hold for log-smooth posterior densities with and without strong log-concavity and the location-scale variational family. Also, our analysis reveals that certain algorithm design choices commonly employed in practice, particularly, nonlinear parameterizations of the scale of the variational approximation, can result in suboptimal convergence rates. Fortunately, running BBVI with proximal stochastic gradient descent fixes these limitations, and thus achieves the strongest known convergence rate guarantees. We evaluate this theoretical insight by comparing proximal SGD against other standard implementations of BBVI on large-scale Bayesian inference problems. | ['Jacob R. Gardner', 'Yian Ma', 'Jisu Oh', 'Kaiwen Wu', 'Kyurae Kim'] | 2023-05-24 | null | null | null | null | ['bayesian-inference'] | ['methodology'] | [-1.47051916e-01 8.41824338e-02 -1.63427263e-01 -3.74642104e-01
-1.11163962e+00 -6.69628501e-01 6.04249537e-01 -3.32899809e-01
-3.56306374e-01 1.14982033e+00 3.82712670e-02 -5.83152890e-01
-3.39016229e-01 -7.25730062e-01 -9.14678216e-01 -8.75815153e-01
-7.58691356e-02 8.05103362e-01 1.19042076e-01 2.88592458e-01
2.23306596e-01 4.07776952e-01 -1.13524950e+00 -5.62550187e-01
8.76976192e-01 6.09330177e-01 -5.26953377e-02 8.45251977e-01
6.64231852e-02 2.81217098e-01 -3.38576972e-01 -5.68128645e-01
1.36808932e-01 -4.21730459e-01 -3.10407132e-01 -1.49382278e-01
6.10639513e-01 -6.36448681e-01 1.07185412e-02 1.18438876e+00
4.92992818e-01 3.20397139e-01 9.14305806e-01 -1.14151371e+00
-5.43145478e-01 7.46661901e-01 -8.28023612e-01 2.05286160e-01
4.86903861e-02 -1.29911406e-02 1.16979110e+00 -8.46203804e-01
5.01511216e-01 1.37370026e+00 9.90182817e-01 4.64778394e-01
-1.72284377e+00 -3.17959428e-01 1.04579277e-01 -4.02676821e-01
-1.61286974e+00 -6.25117600e-01 4.83783692e-01 -4.19196188e-01
5.99253714e-01 2.72824913e-01 3.29399705e-01 1.22819138e+00
-1.24905817e-01 9.08535957e-01 1.05582190e+00 -2.76542246e-01
6.46902084e-01 2.04120532e-01 3.12500626e-01 6.54227972e-01
6.13898695e-01 3.70097645e-02 -3.15031826e-01 -8.52325618e-01
1.02509665e+00 -2.46159285e-01 -1.48761302e-01 -4.16099221e-01
-9.34410691e-01 1.13653576e+00 -1.24363139e-01 -5.36783226e-02
-1.41589120e-01 7.32884407e-01 2.78070748e-01 -2.52858978e-02
6.66800976e-01 -1.13187745e-01 -4.85441118e-01 -3.71050090e-01
-1.35147846e+00 7.71723270e-01 7.91999400e-01 1.08319008e+00
7.62419999e-01 2.41534129e-01 -3.63856077e-01 7.38381267e-01
7.11079895e-01 1.03247595e+00 -3.64077300e-01 -1.47120094e+00
3.34133774e-01 -5.23374438e-01 6.35184288e-01 -5.71588993e-01
-2.41789714e-01 -3.20410043e-01 -6.29367650e-01 1.59968987e-01
1.02281582e+00 -5.86803317e-01 -7.66875386e-01 2.17708063e+00
6.08088791e-01 3.53541106e-01 -3.52509946e-01 8.17126632e-01
1.42135859e-01 6.96255267e-01 -1.13328427e-01 -5.45579016e-01
1.10437369e+00 -3.82089943e-01 -6.83884859e-01 2.75460146e-02
3.93559009e-01 -5.08388937e-01 1.28660858e+00 3.92915010e-01
-1.52829349e+00 -1.31596461e-01 -8.10405672e-01 -8.87670293e-02
1.29929855e-01 -8.74891356e-02 8.24434519e-01 1.10720038e+00
-1.06662667e+00 6.71387136e-01 -1.30012202e+00 -1.25236958e-01
3.10958385e-01 -1.84057392e-02 3.68637592e-01 9.24294367e-02
-7.68740594e-01 4.85809058e-01 4.37476858e-02 1.80870667e-01
-1.01777816e+00 -1.09547067e+00 -5.62366784e-01 1.61026865e-01
3.91193271e-01 -8.20216358e-01 1.30305147e+00 -4.59005237e-01
-1.54716086e+00 5.53487420e-01 -6.24748766e-01 -3.71574938e-01
1.05289268e+00 -1.82029933e-01 2.14913636e-01 -1.54663146e-01
1.02655858e-01 1.44189730e-01 1.01559889e+00 -1.14135826e+00
-2.84258664e-01 -5.18935323e-01 -2.09445860e-02 -1.09967634e-01
-1.04907922e-01 6.28294004e-03 -4.88301307e-01 -3.69587928e-01
-1.55210227e-03 -6.78883255e-01 -2.93547124e-01 1.42202497e-01
-3.63383532e-01 -3.63581777e-01 3.52948129e-01 -6.14451289e-01
1.07391715e+00 -2.05543041e+00 3.22763883e-02 4.64667439e-01
9.80623588e-02 -2.03701004e-01 2.98214853e-01 3.01742643e-01
5.94205320e-01 1.36540055e-01 -5.42222738e-01 -7.42713392e-01
6.46477997e-01 3.46613973e-01 -4.73637462e-01 8.84417892e-01
-3.22802693e-01 8.93046498e-01 -9.21269894e-01 -5.83210230e-01
-1.84784047e-02 3.64718765e-01 -1.06105566e+00 -4.47207652e-02
-4.91686940e-01 3.89055580e-01 -5.69797933e-01 4.46479619e-01
1.01195920e+00 -5.14027417e-01 2.67830551e-01 1.97995350e-01
-6.65260702e-02 9.61984247e-02 -1.39673090e+00 1.57119882e+00
-3.72605741e-01 4.69749779e-01 6.78506196e-01 -9.89787281e-01
3.21647942e-01 2.04897135e-01 2.50581294e-01 4.60289940e-02
-5.01513481e-02 3.28609437e-01 -3.83865774e-01 -1.51951626e-01
3.20813298e-01 -4.44653153e-01 2.06012309e-01 4.17316645e-01
-9.90955383e-02 -4.21987921e-02 1.42009974e-01 2.76958257e-01
5.51078260e-01 4.47360009e-01 6.34329244e-02 -5.66799164e-01
1.93933677e-02 -2.28150353e-01 6.14732921e-01 1.36929810e+00
-2.41072968e-01 7.72701502e-01 6.72039866e-01 1.49513230e-01
-1.27821624e+00 -1.61118555e+00 -7.18463242e-01 1.29789948e+00
-1.25948831e-01 -2.32171714e-01 -8.87851894e-01 -2.92926818e-01
3.77365023e-01 8.94570231e-01 -5.73550284e-01 4.49340284e-01
-4.84011978e-01 -1.17087662e+00 6.81807816e-01 5.13381124e-01
1.59828931e-01 -4.96648759e-01 -2.22276539e-01 2.10067704e-01
1.80438533e-01 -8.43810856e-01 -4.75676715e-01 -1.10095695e-01
-1.11422217e+00 -6.12614214e-01 -9.23307300e-01 -1.34907469e-01
4.26146954e-01 -1.19929366e-01 1.10246670e+00 -2.72513837e-01
1.81680135e-02 4.02888745e-01 1.82928026e-01 -7.27777407e-02
-2.66915023e-01 -2.09572256e-01 2.43903734e-02 -2.19683260e-01
1.35697678e-01 -8.89926672e-01 -5.70384324e-01 3.60277034e-02
-5.90591013e-01 -2.02278480e-01 1.06351025e-01 9.31899190e-01
5.44038117e-01 -2.11240813e-01 4.54051673e-01 -9.89615381e-01
6.37784481e-01 -6.55033290e-01 -1.17918289e+00 2.75524378e-01
-5.99231780e-01 1.43855736e-01 4.79706526e-01 -2.92808056e-01
-1.40337181e+00 -4.94057477e-01 -3.83272588e-01 -4.78037775e-01
9.92372707e-02 6.49227381e-01 8.19856897e-02 1.86531007e-01
7.13351130e-01 9.82850417e-02 -8.98823068e-02 -5.38184583e-01
6.02953732e-01 4.84833091e-01 4.62690681e-01 -1.14160001e+00
6.86867058e-01 8.82984757e-01 9.55524966e-02 -8.98001611e-01
-9.86582637e-01 -1.16935305e-01 -1.77064136e-01 1.65183470e-01
6.87901735e-01 -7.56556451e-01 -9.70633626e-01 2.42593721e-01
-9.41272378e-01 -6.11916065e-01 -2.57413149e-01 4.43699688e-01
-8.19315851e-01 5.61069727e-01 -7.13300765e-01 -1.43711817e+00
-1.47920385e-01 -8.76844943e-01 1.18741846e+00 -9.53435339e-03
-1.80773348e-01 -1.43983352e+00 2.83011049e-01 2.44199201e-01
4.47488755e-01 1.37452215e-01 6.82621241e-01 -2.79849231e-01
-5.63792825e-01 2.06497610e-01 -2.13794753e-01 2.14609444e-01
-2.80947059e-01 3.24730903e-01 -9.82695699e-01 -4.23112363e-01
2.31788158e-02 -3.18810493e-01 9.29909408e-01 1.19869316e+00
1.10881615e+00 -2.59850532e-01 -3.28449160e-01 9.64588165e-01
1.39763439e+00 -3.96531969e-01 3.47252309e-01 -2.05802321e-01
6.42928243e-01 1.37869969e-01 3.21166396e-01 9.45745826e-01
1.68433964e-01 5.21018326e-01 1.04081342e-02 2.12773010e-01
1.98274314e-01 -4.25026923e-01 3.98296446e-01 5.00579655e-01
-1.72280625e-01 -7.05448017e-02 -6.75161839e-01 5.59981108e-01
-1.99080050e+00 -1.00820959e+00 -1.88649327e-01 2.33326173e+00
1.03957450e+00 1.00177407e-01 3.08827132e-01 -3.18719208e-01
5.51917851e-01 1.70773149e-01 -7.95725524e-01 -2.89300293e-01
-1.27954632e-01 3.64058733e-01 8.66875231e-01 1.01419973e+00
-8.44813049e-01 8.02158952e-01 7.92206812e+00 1.37913561e+00
-6.11450672e-01 6.14793539e-01 4.72309202e-01 -4.13499147e-01
-8.02188337e-01 1.89738274e-01 -1.05454826e+00 5.63466251e-01
9.63182926e-01 -6.42213272e-03 8.19936812e-01 9.21257257e-01
5.26351690e-01 -5.20274378e-02 -9.87028241e-01 8.53623807e-01
-4.69058663e-01 -1.39932239e+00 -4.53494251e-01 2.69384772e-01
9.97585833e-01 2.20630988e-01 4.27641161e-02 1.44120052e-01
9.00298893e-01 -9.77729797e-01 7.85209835e-01 3.72281641e-01
6.30309224e-01 -8.02039921e-01 3.99797559e-01 3.55901748e-01
-6.93107069e-01 3.18130702e-01 -7.92273283e-01 -3.14690918e-02
5.17999887e-01 1.05946732e+00 -3.34276229e-01 1.69634670e-01
4.87536281e-01 4.86962140e-01 1.84359744e-01 7.86635160e-01
-2.89605349e-01 1.24167621e+00 -1.02547503e+00 -1.43118739e-01
2.13323936e-01 -7.82971859e-01 7.63627291e-01 1.38892531e+00
3.64512384e-01 -1.83065966e-01 -3.20082866e-02 1.54876840e+00
7.40166232e-02 -2.85553783e-01 -3.79678041e-01 -2.32495759e-02
6.97739303e-01 7.93139577e-01 -5.47116518e-01 -4.43681598e-01
-3.71443599e-01 6.44465685e-01 3.62936407e-01 9.04609442e-01
-1.07248735e+00 -4.51686345e-02 1.03663206e+00 -1.46541903e-02
6.86014473e-01 -4.86479223e-01 -7.40497768e-01 -1.35813832e+00
2.43793912e-02 -4.16118324e-01 4.48702931e-01 -4.32008207e-01
-1.51873183e+00 -3.39325637e-01 3.76327634e-01 -2.13257805e-01
-5.20807803e-01 -5.79658151e-01 -6.17662072e-01 9.10548210e-01
-1.06172049e+00 -1.05203569e+00 4.18192238e-01 6.02302015e-01
1.33952126e-01 3.04156333e-01 4.17292178e-01 1.29176185e-01
-5.10202706e-01 7.67285466e-01 8.43660474e-01 -1.94489211e-01
2.67787814e-01 -1.47904778e+00 3.69693220e-01 8.40290070e-01
2.05194831e-01 1.06063569e+00 1.24266124e+00 -5.95172286e-01
-1.70098913e+00 -5.79565942e-01 4.53849584e-01 -4.77726161e-01
9.79334414e-01 -6.74280405e-01 -7.84526765e-01 9.83490169e-01
-1.23994544e-01 -1.25166640e-01 4.12754536e-01 6.77354693e-01
-2.58812159e-01 1.02723695e-01 -1.26741791e+00 7.85532534e-01
1.21466506e+00 -4.68370557e-01 -3.44878733e-01 5.56155622e-01
5.19266307e-01 -3.85707438e-01 -9.81605709e-01 1.06426187e-01
7.27977812e-01 -9.00595069e-01 1.17615604e+00 -6.86238527e-01
1.43563196e-01 -2.76659936e-01 -3.55781138e-01 -9.17266786e-01
-1.31125644e-01 -1.03272617e+00 -4.49396849e-01 1.25391579e+00
3.15074801e-01 -7.72033095e-01 9.09009993e-01 7.21994042e-01
8.15272890e-03 -4.78609383e-01 -1.19314790e+00 -9.06925738e-01
6.11013114e-01 -8.84256482e-01 3.17613572e-01 6.32691026e-01
-2.02141970e-01 -8.96841362e-02 -6.16084397e-01 2.98483819e-01
1.34040236e+00 2.30853930e-01 8.69283736e-01 -9.48359191e-01
-7.91758478e-01 -5.69438279e-01 7.99440071e-02 -1.55673873e+00
2.54549235e-01 -7.24264860e-01 1.93344969e-02 -1.24821198e+00
4.39850926e-01 -5.88023782e-01 -7.23390132e-02 9.94368941e-02
-3.41248244e-01 1.08094424e-01 -2.23426104e-01 4.21034880e-02
-3.24247539e-01 5.94768465e-01 1.05521667e+00 1.42137095e-01
-6.53920919e-02 8.96519572e-02 -6.05444014e-01 8.23127091e-01
4.85044181e-01 -3.68972093e-01 -5.10457337e-01 -3.62570941e-01
5.07879376e-01 1.39829800e-01 3.56137127e-01 -4.11277443e-01
5.99189177e-02 -5.32178879e-01 1.55931011e-01 -5.74944615e-01
3.01187843e-01 -2.38976374e-01 2.09680751e-01 6.48471415e-02
-2.81456500e-01 -5.86532831e-01 2.48841289e-02 6.62984729e-01
3.20772409e-01 -5.32403767e-01 8.77453625e-01 1.57820918e-02
-1.06397077e-01 2.95391053e-01 -3.34047437e-01 6.07315004e-01
3.42057496e-01 -6.12653829e-02 -1.01178639e-01 -5.12141943e-01
-8.17424834e-01 2.00650021e-01 5.22234321e-01 -4.94964302e-01
3.58133018e-01 -1.17436433e+00 -9.55844104e-01 -1.30921498e-01
-3.62816870e-01 -1.79341324e-02 1.18456185e-01 1.11282802e+00
-3.47769856e-01 1.77976817e-01 4.94054586e-01 -6.82788253e-01
-8.18966150e-01 3.51222068e-01 2.21259683e-01 -1.10975623e-01
-6.78882360e-01 9.43033874e-01 2.60110170e-01 -4.12995458e-01
2.11751178e-01 -2.16030627e-01 5.80408216e-01 -1.67377561e-01
4.73109663e-01 6.11483514e-01 -3.44593883e-01 -3.15660350e-02
-2.82441527e-01 4.07851577e-01 2.09615260e-01 -7.69146383e-01
1.15148437e+00 -3.97400916e-01 -8.10275376e-02 6.86840594e-01
1.12183869e+00 3.45812440e-01 -1.74054086e+00 -1.45342857e-01
-1.97653875e-01 -5.13979435e-01 1.36387587e-01 -3.85451764e-01
-8.21373522e-01 8.67194772e-01 1.99116945e-01 1.40144125e-01
4.45459872e-01 6.56952336e-02 5.40071249e-01 4.89536971e-01
5.20804584e-01 -1.02255940e+00 -5.84750593e-01 4.44989085e-01
4.76493537e-01 -1.10540056e+00 2.32864574e-01 -6.19634949e-02
-2.36909702e-01 6.82847142e-01 -9.19107199e-02 -3.20573658e-01
9.73364234e-01 3.09770882e-01 -3.22170824e-01 -1.57818928e-01
-5.83963454e-01 -1.69544704e-02 9.56406593e-02 4.15977478e-01
3.93696427e-01 2.00745568e-01 -4.36016768e-01 2.28908405e-01
-3.26266170e-01 -5.88970147e-02 2.59324372e-01 7.30621576e-01
-5.14418006e-01 -6.80371523e-01 -5.12858391e-01 4.78693485e-01
-7.08158076e-01 -2.78810740e-01 1.74599886e-01 7.82654643e-01
-4.49990004e-01 8.98339510e-01 1.78268969e-01 4.16599780e-01
-2.77768135e-01 1.03332475e-01 8.77176821e-01 -1.36305064e-01
-1.71885099e-02 4.99913275e-01 2.38816477e-02 -4.28367019e-01
-2.80014813e-01 -1.00850272e+00 -1.07557738e+00 -8.75831783e-01
-3.93736362e-01 2.26968110e-01 6.17987573e-01 1.10370636e+00
6.27605990e-02 1.89671181e-02 1.85004085e-01 -7.22615242e-01
-1.25949848e+00 -1.05501521e+00 -9.46119010e-01 2.09121153e-01
4.56967086e-01 -6.94620311e-01 -6.91425443e-01 -8.21036547e-02] | [6.942061424255371, 3.9629313945770264] |
e0ef9276-a8b0-43cd-b21a-c587688281ef | masked-face-recognition-with-latent-part | null | null | https://dl.acm.org/doi/10.1145/3394171.3413731 | https://dl.acm.org/doi/pdf/10.1145/3394171.3413731 | Masked Face Recognition with Latent Part Detection | This paper focuses on a novel task named masked faces recognition (MFR), which aims to match masked faces with common faces and is important especially during the global outbreak of COVID-19. It is challenging to identify masked faces for two main reasons. Firstly, there is no large-scale training data and test data with ground truth for MFR. Collecting and annotating millions of masked faces is labor-consuming. Secondly, since most facial cues are occluded by mask, it is necessary to learn representations which are both discriminative and robust to mask wearing. To handle the first challenge, this paper collects two datasets designed for MFR: MFV with 400 pairs of 200 identities for verification, and MFI which contains 4,916 images of 669 identities for identification. As is known, a robust face recognition model needs images of millions of identities to train, and hundreds of identities is far from enough. Hence, MFV and MFI are only considered as test datasets to evaluate algorithms. Besides, a data augmentation method for training data is introduced to automatically generate synthetic masked face images from existing common face datasets. In addition, a novel latent part detection (LPD) model is proposed to locate the latent facial part which is robust to mask wearing, and the latent part is further used to extract discriminative features. The proposed LPD model is trained in an end-to-end manner and only utilizes the original and synthetic training data. Experimental results on MFV, MFI and synthetic masked LFW demonstrate that LPD model generalizes well on both realistic and synthetic masked data and outperforms other methods by a large margin. | ['Yonghong Tian', 'Mengyue Geng', 'Yangru Huang', 'Peixi Peng', 'Feifei Ding'] | 2020-10-01 | null | null | null | proceedings-of-the-28th-acm-international | ['robust-face-recognition'] | ['computer-vision'] | [ 3.84415507e-01 -2.34811723e-01 3.54776904e-03 -6.36733294e-01
-5.46198964e-01 -5.29654801e-01 5.05119085e-01 -7.59670794e-01
-4.25261185e-02 6.46250308e-01 5.11265211e-02 2.80317087e-02
1.71615407e-01 -5.26478171e-01 -7.02944696e-01 -7.91766405e-01
-1.52312443e-01 3.14142346e-01 -2.65043348e-01 -1.00782223e-01
-1.26318902e-01 7.92162955e-01 -1.96806610e+00 5.56085527e-01
6.82213366e-01 1.11359453e+00 1.93484843e-01 1.28309369e-01
2.00293168e-01 3.32196206e-01 -5.79532146e-01 -7.68935204e-01
7.23122299e-01 -3.29186082e-01 -3.23240370e-01 5.10749042e-01
6.94594204e-01 -6.18871093e-01 -1.68312073e-01 1.04250228e+00
6.87531650e-01 -3.05658191e-01 5.79109132e-01 -1.51691139e+00
-7.27802992e-01 2.70632267e-01 -8.28846574e-01 -1.03761442e-01
2.63788313e-01 4.03807342e-01 2.31626734e-01 -1.40962315e+00
4.98023629e-01 1.60637546e+00 7.23335505e-01 9.69539285e-01
-1.03581905e+00 -1.25270832e+00 2.16615628e-02 -1.42845184e-01
-1.85971653e+00 -1.03356051e+00 5.59967816e-01 -5.86537242e-01
4.17882830e-01 2.49559730e-01 1.32877648e-01 1.17559218e+00
-3.17553699e-01 2.99614668e-01 1.41044331e+00 -2.10356176e-01
-6.47584647e-02 4.46314156e-01 -3.55992496e-01 7.05845952e-01
3.35127562e-01 2.96503067e-01 -5.74614465e-01 -3.29572141e-01
7.62727439e-01 1.67829365e-01 -4.31569964e-01 1.55825555e-01
-8.03960145e-01 8.02214265e-01 1.44215167e-01 -2.78544873e-02
-3.16794693e-01 -4.19997841e-01 7.63829648e-02 2.51824170e-01
5.69374740e-01 -8.48183557e-02 -2.46981904e-01 3.99233997e-01
-9.22960818e-01 8.86323527e-02 4.03485775e-01 7.46042728e-01
1.07057786e+00 1.53766334e-01 -2.61046708e-01 1.08570516e+00
3.60910028e-01 7.85011768e-01 3.46067131e-01 -5.82343638e-01
4.16067690e-01 6.61045849e-01 3.45095694e-01 -1.26689661e+00
1.09985135e-02 -1.37992784e-01 -1.10315537e+00 1.64454818e-01
1.65427804e-01 -3.01008597e-02 -1.12761521e+00 1.87618148e+00
4.40225303e-01 4.71267700e-01 1.03805903e-02 9.60674345e-01
1.02947891e+00 4.86807078e-01 -1.95819989e-01 -6.09444439e-01
1.49827611e+00 -5.06386817e-01 -7.38412440e-01 -3.73912811e-01
3.46460193e-02 -1.04128516e+00 8.77949357e-01 -4.87627275e-02
-8.07974041e-01 -8.24741721e-01 -8.31311047e-01 4.06530738e-01
-2.51444370e-01 4.89345670e-01 3.84624869e-01 1.03884399e+00
-1.11465764e+00 8.61454755e-02 -1.82958215e-01 -1.88827395e-01
8.40866804e-01 6.42621756e-01 -8.32837760e-01 -4.89232749e-01
-1.07613826e+00 5.70524037e-01 2.77716294e-02 5.42328835e-01
-1.22525299e+00 -5.44706225e-01 -1.00701201e+00 -2.15735659e-01
2.29712442e-01 -4.07475904e-02 7.34580398e-01 -1.10231566e+00
-1.08334935e+00 1.32457817e+00 -6.17861032e-01 -3.60748754e-03
5.53526461e-01 1.04506314e-01 -7.94420242e-01 5.51482439e-02
3.03896099e-01 6.28833950e-01 1.41672742e+00 -1.62360203e+00
-1.30329758e-01 -6.40724301e-01 -3.85994136e-01 -2.82858253e-01
-4.12346601e-01 6.46838546e-01 -4.28024799e-01 -7.81363606e-01
-1.08080417e-01 -7.93468595e-01 1.18818521e-01 -1.08513556e-01
-3.72546971e-01 3.31547596e-02 9.99846756e-01 -9.77786005e-01
8.06897283e-01 -2.26969552e+00 -4.06951070e-01 2.25681126e-01
2.72495896e-01 5.75228333e-01 -5.16234338e-01 1.03010662e-01
-4.16179597e-01 8.46207365e-02 -4.14560854e-01 -5.50059676e-01
-1.65240392e-01 7.59402290e-02 -5.33227026e-01 6.14246368e-01
7.02355444e-01 7.57328808e-01 -3.47968966e-01 -4.14454192e-01
-1.44212872e-01 6.60220623e-01 -3.90537381e-01 5.91824472e-01
5.99206751e-03 7.74031341e-01 -1.33208662e-01 1.06799495e+00
1.46063662e+00 7.42937103e-02 4.75038774e-02 -1.94133565e-01
4.62822057e-02 -3.96063477e-01 -1.30074024e+00 1.02473056e+00
-1.73212111e-01 6.62368461e-02 3.76015037e-01 -6.67430997e-01
1.16319752e+00 5.77853322e-01 2.36391500e-01 -4.24167097e-01
2.31374219e-01 2.72979885e-01 -2.00050727e-01 -5.61567008e-01
-3.77635732e-02 -7.20644370e-02 1.98327512e-01 4.80813771e-01
-1.56602859e-01 4.06354845e-01 2.17054561e-02 -1.39099866e-01
7.21190453e-01 -1.21775523e-01 -1.18166558e-01 -7.65389726e-02
8.41279268e-01 -3.33911151e-01 1.02363575e+00 3.05887848e-01
-1.40542433e-01 7.20495403e-01 1.77942395e-01 -5.56547344e-01
-6.45842910e-01 -9.79285538e-01 -2.52239347e-01 7.50535965e-01
5.72064612e-03 1.51322335e-02 -8.66629243e-01 -6.90151095e-01
1.64428800e-01 7.07194507e-02 -8.00543845e-01 -1.14230320e-01
-5.07676899e-01 -9.74324167e-01 5.33717215e-01 2.92679518e-01
7.22385705e-01 -1.17269230e+00 5.37940450e-02 -1.83726519e-01
-2.53371954e-01 -1.32481098e+00 -8.15004468e-01 -6.13934457e-01
-2.74370402e-01 -1.24407303e+00 -6.22155428e-01 -1.00180089e+00
1.13531208e+00 5.21047771e-01 8.86582971e-01 2.96533525e-01
-6.73432350e-01 -1.73148528e-01 -1.99714258e-01 -3.54046732e-01
-5.14475226e-01 -7.34986961e-01 6.33623958e-01 9.81876314e-01
4.94399607e-01 -3.71289223e-01 -5.06809115e-01 8.87867391e-01
-7.79540122e-01 -1.97711140e-01 8.42187524e-01 7.87856340e-01
4.73867327e-01 5.36565185e-02 9.18394864e-01 -7.68251956e-01
2.50337541e-01 -4.35957402e-01 -7.48701990e-01 4.59515691e-01
-2.59298176e-01 -5.88214934e-01 4.72820073e-01 -6.33686304e-01
-1.19711292e+00 2.76174724e-01 5.25456890e-02 -7.85067797e-01
-1.82264924e-01 8.89396947e-03 -7.88023770e-01 -2.81313479e-01
5.35154939e-01 3.16845387e-01 3.80268246e-01 -5.94166994e-01
1.60641864e-01 1.10077274e+00 6.04686499e-01 -2.80868143e-01
1.24543524e+00 6.18056655e-01 -2.23097071e-01 -8.46554816e-01
-5.45012951e-01 -2.83119351e-01 -5.96590936e-01 -1.77978218e-01
7.12833643e-01 -1.17418885e+00 -8.61721396e-01 8.65551293e-01
-1.15736496e+00 1.34477273e-01 1.27132043e-01 3.66474628e-01
-2.41329670e-01 1.88075542e-01 -3.57615352e-01 -1.10668099e+00
-3.18983108e-01 -1.26202226e+00 1.27418387e+00 1.34033069e-01
1.93098277e-01 -2.33684197e-01 -1.86357155e-01 5.99344313e-01
3.91300946e-01 3.35010529e-01 4.78065014e-01 -3.54744315e-01
-6.22515023e-01 -4.26859081e-01 -4.81490761e-01 5.60001850e-01
7.60732830e-01 -4.31128070e-02 -1.43736005e+00 -6.21112585e-01
2.50672370e-01 -6.27441704e-01 8.18249464e-01 2.89549120e-02
1.23837328e+00 -5.95965445e-01 -3.73963952e-01 6.96260214e-01
1.24137020e+00 1.48659110e-01 6.91082716e-01 -4.64735389e-01
6.93633258e-01 8.66262078e-01 5.48925459e-01 3.50335389e-01
1.25841305e-01 6.74820781e-01 2.93864399e-01 -2.03810990e-01
-9.94312614e-02 -1.46285564e-01 4.73799348e-01 4.01961058e-01
-1.16854668e-01 5.81270270e-02 -6.65209353e-01 4.21150565e-01
-1.33447087e+00 -1.07457602e+00 2.44165823e-01 2.43917537e+00
8.63198459e-01 -4.53379124e-01 9.17872712e-02 1.44256532e-01
1.19625878e+00 -3.08817364e-02 -5.46629548e-01 7.31434226e-02
-2.89411992e-01 3.58279347e-01 -2.44302861e-02 2.72732466e-01
-1.15369260e+00 8.33206773e-01 5.47656918e+00 7.91053116e-01
-1.20411336e+00 3.10261101e-01 9.46268439e-01 1.44882292e-01
1.45074189e-01 -3.36336762e-01 -1.09359348e+00 6.53391361e-01
5.48924983e-01 2.79273152e-01 5.87567270e-01 6.11334026e-01
3.06515843e-01 3.67588282e-01 -8.74656379e-01 1.28606558e+00
3.53884727e-01 -1.06429482e+00 -1.38017731e-02 1.13912322e-01
9.10298407e-01 -4.20455515e-01 3.35058719e-01 5.20775467e-02
9.90894064e-02 -1.57626355e+00 3.87251794e-01 2.84940481e-01
1.37598407e+00 -6.75782263e-01 7.10923612e-01 2.28154972e-01
-1.39375901e+00 -1.33950710e-01 -6.37478173e-01 1.89992875e-01
-1.50881410e-01 5.12501240e-01 -8.63621533e-01 5.40558398e-01
7.52590656e-01 3.91071945e-01 -7.68313527e-01 6.21189475e-01
-4.35293093e-02 3.21033478e-01 -5.54471537e-02 6.14515841e-01
-3.70397180e-01 -1.31552622e-01 1.67631567e-01 7.68982768e-01
3.15367967e-01 -9.43811983e-03 3.03436518e-01 9.04789984e-01
-4.17333812e-01 -9.24844481e-03 -6.90303504e-01 -8.04504231e-02
7.13244140e-01 1.50258362e+00 -4.09453779e-01 -2.93366518e-02
-4.86360937e-01 9.61746156e-01 1.89262837e-01 3.19114417e-01
-8.74509692e-01 9.44925323e-02 9.89822149e-01 2.39624724e-01
2.39790142e-01 1.77780613e-01 1.60635248e-01 -1.12263811e+00
3.65108758e-01 -1.25783646e+00 1.97254479e-01 -4.19314325e-01
-1.74046671e+00 1.07999730e+00 -1.53740242e-01 -1.24984503e+00
-1.25113606e-01 -5.63002050e-01 -6.23672009e-01 1.33404016e+00
-1.57555795e+00 -1.47641432e+00 -6.78850651e-01 7.48168826e-01
2.35756800e-01 -5.76713979e-01 8.88323128e-01 4.81556565e-01
-8.83554459e-01 1.02091813e+00 -1.23116225e-01 4.23263401e-01
7.17232525e-01 -3.84184837e-01 4.87100393e-01 9.55660462e-01
-1.37138963e-01 7.86000669e-01 2.24762008e-01 -8.35925937e-01
-1.42031062e+00 -1.66948414e+00 7.34759569e-01 -4.59032565e-01
7.77305439e-02 -7.20292687e-01 -9.72731650e-01 5.92894316e-01
-1.13295399e-01 4.26363140e-01 6.96698785e-01 -5.04416943e-01
-6.15290940e-01 -3.56056690e-01 -1.59225273e+00 3.32489431e-01
1.19796968e+00 -5.81199527e-01 -1.78313002e-01 5.50355971e-01
5.43040395e-01 3.34639065e-02 -6.05409563e-01 8.30540061e-01
4.15878445e-01 -9.11305845e-01 9.91062820e-01 -3.36176127e-01
3.07137929e-02 -5.02288342e-01 -3.48242521e-01 -8.29447985e-01
-1.15677500e-02 -7.84617782e-01 1.54787049e-01 1.74780166e+00
1.98677793e-01 -9.64463174e-01 8.76587689e-01 5.00196099e-01
3.32262307e-01 -6.27940953e-01 -8.48330081e-01 -1.05061269e+00
-4.47801769e-01 5.73269688e-02 1.06019962e+00 1.11706865e+00
-9.30369735e-01 -1.24026179e-01 -6.87976360e-01 5.91894865e-01
9.23940301e-01 2.81242609e-01 8.96590114e-01 -1.14597309e+00
3.05607473e-03 -5.62564097e-02 -3.96236926e-01 -4.71998960e-01
4.14221257e-01 -8.56690228e-01 3.41711007e-02 -9.98712838e-01
3.22478205e-01 -5.70851505e-01 5.42703494e-02 6.67254686e-01
-4.08623546e-01 8.88971567e-01 -8.98233652e-02 3.73636514e-01
2.12447047e-01 5.94606102e-01 1.08216810e+00 -1.73521429e-01
8.17577019e-02 9.64548066e-02 -6.68198109e-01 5.46120763e-01
6.97003603e-01 -3.71314019e-01 -3.70833397e-01 -2.66773641e-01
-4.68770534e-01 -1.16417885e-01 5.33080280e-01 -7.13840425e-01
-1.33573309e-01 -2.08800972e-01 7.73334265e-01 -4.53243941e-01
6.23450398e-01 -6.13406122e-01 4.13155943e-01 3.57330173e-01
2.59583890e-01 1.20404422e-01 2.32447311e-01 3.00067991e-01
-2.28642553e-01 4.24626358e-02 1.10812843e+00 -7.83150345e-02
-4.95635986e-01 8.05803478e-01 3.11744064e-02 -2.39499323e-02
1.19695449e+00 -4.00883704e-01 -3.67997825e-01 -1.41410515e-01
-4.35213983e-01 1.49872527e-01 5.62228680e-01 6.71888590e-01
8.97127271e-01 -1.48873842e+00 -1.25405014e+00 8.86664629e-01
2.10576326e-01 -1.45074055e-01 4.96467471e-01 4.51349705e-01
-3.22502077e-01 9.39351395e-02 -3.09935719e-01 -5.13357580e-01
-1.58332717e+00 9.23418820e-01 1.74843669e-01 2.44314626e-01
-9.35543776e-02 1.18291605e+00 6.17072225e-01 -6.01699293e-01
1.96786404e-01 4.54788357e-01 -1.19540587e-01 6.51492327e-02
1.03377891e+00 -8.36438835e-02 9.63260531e-02 -1.31550002e+00
-6.58342540e-01 7.55151331e-01 -1.15894318e-01 2.32658416e-01
1.14192069e+00 5.68044819e-02 -5.97837687e-01 -3.38093191e-01
1.21369445e+00 2.44915783e-01 -1.16880679e+00 -2.83895671e-01
-2.96728551e-01 -1.01554728e+00 -5.62006652e-01 -5.62127650e-01
-1.33688283e+00 8.18414927e-01 9.43629444e-01 -2.73786455e-01
1.37346625e+00 -8.36758167e-02 5.34995556e-01 -8.67176354e-02
4.44146961e-01 -4.71152782e-01 1.87838018e-01 7.84815550e-02
1.25099373e+00 -1.40172398e+00 -2.25942820e-01 -8.51643860e-01
-5.04638553e-01 5.89404345e-01 8.71317148e-01 3.75945985e-01
6.98184550e-01 1.92719594e-01 3.59549910e-01 -1.37162611e-01
-4.30017322e-01 -1.40960947e-01 1.97157994e-01 7.66330242e-01
1.75142344e-02 -2.64609363e-02 1.67649522e-01 6.80851996e-01
-2.02045962e-01 -8.72395262e-02 6.72902390e-02 7.08485305e-01
-1.95376694e-01 -1.26373482e+00 -7.48159230e-01 4.23703074e-01
-5.16848207e-01 1.10575452e-01 -4.38383281e-01 4.79425073e-01
5.97524822e-01 1.20226729e+00 -3.18340920e-02 -5.95278144e-01
4.03900333e-02 1.16205268e-01 4.02516544e-01 -7.15932310e-01
-3.84935796e-01 -2.50068069e-01 -2.32302561e-01 -4.12072986e-01
-4.07037944e-01 -3.71293396e-01 -6.84853852e-01 -4.03939724e-01
-3.57042998e-01 3.17911431e-02 4.95147228e-01 7.60527313e-01
4.46899414e-01 -1.27789140e-01 1.34699738e+00 -8.91985476e-01
-5.17104447e-01 -1.04660916e+00 -5.60837805e-01 5.93692064e-01
2.77226806e-01 -8.55105698e-01 -3.03276867e-01 1.78909689e-01] | [13.108011245727539, 0.5103166699409485] |
8d96aa38-c24a-41de-9daf-550758cfc2d2 | supervised-speech-separation-based-on-deep | 1708.07524 | null | http://arxiv.org/abs/1708.07524v2 | http://arxiv.org/pdf/1708.07524v2.pdf | Supervised Speech Separation Based on Deep Learning: An Overview | Speech separation is the task of separating target speech from background
interference. Traditionally, speech separation is studied as a signal
processing problem. A more recent approach formulates speech separation as a
supervised learning problem, where the discriminative patterns of speech,
speakers, and background noise are learned from training data. Over the past
decade, many supervised separation algorithms have been put forward. In
particular, the recent introduction of deep learning to supervised speech
separation has dramatically accelerated progress and boosted separation
performance. This article provides a comprehensive overview of the research on
deep learning based supervised speech separation in the last several years. We
first introduce the background of speech separation and the formulation of
supervised separation. Then we discuss three main components of supervised
separation: learning machines, training targets, and acoustic features. Much of
the overview is on separation algorithms where we review monaural methods,
including speech enhancement (speech-nonspeech separation), speaker separation
(multi-talker separation), and speech dereverberation, as well as
multi-microphone techniques. The important issue of generalization, unique to
supervised learning, is discussed. This overview provides a historical
perspective on how advances are made. In addition, we discuss a number of
conceptual issues, including what constitutes the target source. | ['Jitong Chen', 'DeLiang Wang'] | 2017-08-24 | null | null | null | null | ['speaker-separation', 'speech-dereverberation'] | ['speech', 'speech'] | [ 3.01669300e-01 -2.84343064e-01 -3.53138172e-03 -5.23686290e-01
-1.32205570e+00 -4.18803662e-01 3.54448497e-01 -1.57992974e-01
-8.18571821e-02 5.39748549e-01 5.29915273e-01 -1.42236039e-01
-1.37007236e-01 5.37046641e-02 -3.47175777e-01 -1.30238581e+00
1.26221171e-02 6.09566793e-02 -3.76572877e-01 -1.48281544e-01
-8.58380720e-02 6.32584035e-01 -1.76292157e+00 3.32917303e-01
8.79630625e-01 7.65770555e-01 2.86550224e-01 1.10451615e+00
-2.35090807e-01 5.89045763e-01 -1.21825576e+00 -1.89682897e-02
-9.41977054e-02 -9.08228874e-01 -5.40495336e-01 3.21885675e-01
6.20158613e-01 2.01498885e-02 -4.94849056e-01 1.06893265e+00
1.17469144e+00 5.43123364e-01 6.02994144e-01 -1.03541660e+00
-4.52910423e-01 7.37399280e-01 -3.55536371e-01 7.92907953e-01
-3.21376435e-02 -3.41180593e-01 8.12907577e-01 -1.16526675e+00
-4.81859803e-01 1.08094466e+00 6.16378665e-01 7.78088152e-01
-1.11232400e+00 -7.36702442e-01 1.55054897e-01 2.05152959e-01
-1.06138504e+00 -1.26306558e+00 9.16659236e-01 -3.20066392e-01
1.21393251e+00 6.77678823e-01 3.93518746e-01 1.07632184e+00
-3.35262150e-01 1.33950925e+00 7.33240485e-01 -8.32952559e-01
2.11789787e-01 1.05785690e-01 5.01509845e-01 2.12487057e-02
-1.07219517e-01 3.39232624e-01 -7.87305892e-01 4.42457944e-02
2.55331188e-01 -3.95885348e-01 -5.05345523e-01 5.08239605e-02
-6.86897814e-01 5.11413693e-01 -2.36189023e-01 8.19403470e-01
-7.12400675e-02 -3.15599710e-01 4.43401664e-01 3.17624629e-01
6.18206680e-01 2.55695611e-01 -2.79128224e-01 -1.19790025e-01
-1.41215324e+00 2.54949797e-02 8.45052481e-01 7.15931654e-01
1.08450301e-01 1.11251152e+00 2.14355811e-02 1.69172263e+00
1.66065976e-01 9.25735295e-01 7.75986791e-01 -6.37953997e-01
2.77814537e-01 -7.41768539e-01 -2.55900204e-01 -4.60258186e-01
-2.26869538e-01 -7.48406589e-01 -7.18162000e-01 2.68257886e-01
2.86430437e-02 -5.28908193e-01 -1.04130077e+00 1.45483208e+00
-6.53547645e-02 3.81909341e-01 4.20454353e-01 8.99982035e-01
1.31090164e+00 9.65617478e-01 -2.72803962e-01 -7.20957220e-01
9.06983495e-01 -1.48795462e+00 -1.37392259e+00 -6.11292720e-01
-1.16533265e-01 -1.16494262e+00 2.72786111e-01 5.88008404e-01
-1.29994404e+00 -8.26739013e-01 -1.04618108e+00 1.68562829e-01
-3.75561893e-01 1.02920502e-01 2.57386833e-01 1.14396226e+00
-1.06543016e+00 2.70232707e-01 -8.15262914e-01 -2.51579940e-01
1.41752928e-01 4.63537097e-01 -1.57870784e-01 2.84010321e-01
-1.08575070e+00 8.06426585e-01 -1.89571291e-01 1.94878876e-01
-1.10338700e+00 -4.64441419e-01 -9.81355608e-01 2.25352779e-01
1.81697626e-02 -2.92376995e-01 1.66383994e+00 -1.25560951e+00
-1.74480271e+00 8.82396221e-01 -8.22919548e-01 -4.47342396e-01
-4.12982106e-01 -4.03959900e-01 -1.26013768e+00 3.65716182e-02
-1.50711223e-01 -3.39841768e-02 1.38940799e+00 -1.43099439e+00
-8.34401846e-01 -2.50732750e-01 -5.30118644e-01 4.55589354e-01
-9.09351334e-02 7.00424969e-01 4.81558219e-02 -9.99857128e-01
1.42501652e-01 -4.29358393e-01 1.35216847e-01 -8.53619277e-01
-3.40387791e-01 9.02802777e-03 7.47609735e-01 -9.82303500e-01
1.20076358e+00 -2.85959530e+00 2.15783820e-01 -2.37987220e-01
2.06433848e-01 6.99669957e-01 -1.40488610e-01 4.61478889e-01
-8.07040870e-01 -3.88912290e-01 -4.27663118e-01 -9.90227818e-01
-7.78159276e-02 7.18978874e-04 -6.46511257e-01 5.65832734e-01
-2.06578195e-01 5.21787584e-01 -7.68552363e-01 3.58818397e-02
3.70940596e-01 6.79635108e-01 -5.68893962e-02 4.88206059e-01
6.54109895e-01 4.27835554e-01 4.76662219e-01 7.44478405e-01
9.40796494e-01 4.69884157e-01 -1.13879517e-01 5.20830825e-02
-3.83138418e-01 8.36024284e-01 -1.04917431e+00 1.29346752e+00
-4.55550969e-01 1.33908141e+00 1.06821990e+00 -1.51237357e+00
8.40319812e-01 8.86456549e-01 3.79792869e-01 -2.68984675e-01
6.17050158e-04 5.54295063e-01 1.41560361e-01 -4.83043522e-01
1.90174833e-01 -7.13213027e-01 4.05721188e-01 1.12024620e-01
4.89692092e-01 -2.18138516e-01 -9.11439508e-02 -1.21361263e-01
6.03699088e-01 -8.00354362e-01 2.47896045e-01 -7.63548464e-02
6.94528759e-01 -4.59808111e-01 3.66011262e-01 7.38030314e-01
-5.49906552e-01 7.29105294e-01 -3.07594419e-01 1.95260480e-01
-2.75264531e-01 -1.41910458e+00 -1.78349137e-01 1.35276461e+00
-8.08812901e-02 -1.96154729e-01 -7.51200914e-01 -2.40637019e-01
-1.81125119e-01 7.12610960e-01 -1.32184625e-01 -1.59410372e-01
-7.26567864e-01 -8.66223752e-01 7.79081523e-01 7.16156304e-01
1.88023765e-02 -1.01059222e+00 2.12343439e-01 -2.21367069e-02
-4.12821531e-01 -9.32562590e-01 -5.57084620e-01 9.21930075e-01
-7.37537920e-01 -5.17006099e-01 -1.00733840e+00 -1.56266129e+00
3.32355857e-01 8.79213870e-01 8.62298071e-01 -4.73894596e-01
-7.73720220e-02 7.14090824e-01 -1.86022162e-01 -6.95791602e-01
-5.32454491e-01 -4.50760424e-01 6.91627145e-01 1.24089770e-01
7.24681139e-01 -7.75774300e-01 -8.08642805e-02 -9.09323096e-02
-5.59710324e-01 -6.36500120e-01 4.38612640e-01 8.66403937e-01
3.11067551e-01 5.44979215e-01 7.61681557e-01 -1.48283139e-01
8.30090344e-01 -2.28733853e-01 -1.10475890e-01 -3.67346480e-02
-9.91335660e-02 -4.94054556e-01 5.62416494e-01 -4.44929689e-01
-1.34049654e+00 -3.63781273e-01 -5.11044085e-01 -5.11573136e-01
-7.45485723e-01 2.48347446e-01 -5.82429349e-01 4.92852628e-02
5.90112448e-01 5.90567887e-01 -3.68540511e-02 -9.33873534e-01
2.01353341e-01 1.24090767e+00 5.60676098e-01 -1.99978426e-01
5.63805521e-01 2.47935787e-01 -6.56751335e-01 -1.69154096e+00
-7.61736453e-01 -1.03914046e+00 -5.67753673e-01 -9.68668684e-02
7.51297593e-01 -9.03592825e-01 -1.14742823e-01 8.20571780e-01
-1.11088824e+00 -3.31892103e-01 -5.31441212e-01 9.49655294e-01
-4.02078569e-01 3.35395873e-01 -7.01290309e-01 -1.29427910e+00
-3.15433443e-01 -1.20171118e+00 9.71447885e-01 3.53891522e-01
-9.13759917e-02 -1.20664895e+00 4.19420540e-01 5.92706501e-01
4.66533780e-01 -4.87885326e-01 4.38970089e-01 -1.06490088e+00
1.81056365e-01 3.07032354e-02 3.85625958e-01 1.14700592e+00
7.90005863e-01 -2.71405607e-01 -1.70167208e+00 -3.54918271e-01
8.74833107e-01 2.35437546e-02 9.47658062e-01 1.10296309e+00
8.20721209e-01 -4.80113663e-02 -3.36040854e-01 7.32644260e-01
6.83950365e-01 8.32797766e-01 1.43719450e-01 -5.62895536e-02
5.47209620e-01 7.11837649e-01 1.93005860e-01 -1.27900198e-01
-3.13196778e-01 4.60251808e-01 -9.44749266e-02 -5.06867945e-01
-6.84656501e-01 3.50667179e-01 7.51025140e-01 1.66154826e+00
1.89098626e-01 -2.06193268e-01 -7.24924207e-01 7.45249152e-01
-1.21586812e+00 -1.23914504e+00 2.62679793e-02 2.22074819e+00
6.09992743e-01 -1.33819982e-01 2.97188848e-01 7.21808076e-01
1.12446475e+00 4.94195729e-01 -4.02443528e-01 -5.41282773e-01
-5.49163580e-01 5.86192846e-01 9.44866166e-02 1.11056912e+00
-1.38170135e+00 6.87954307e-01 7.87349319e+00 9.50789273e-01
-1.31740332e+00 1.74597815e-01 2.35933900e-01 -5.00660777e-01
1.86267972e-01 -6.45710349e-01 -7.77832091e-01 3.20803404e-01
1.18058002e+00 -8.45412537e-02 5.50697267e-01 5.70186436e-01
3.21824163e-01 1.58723444e-01 -1.07075524e+00 1.55304086e+00
6.65777147e-01 -8.06807518e-01 -2.94137180e-01 -2.33826488e-01
4.76842016e-01 2.15107784e-01 3.44284922e-01 4.58641142e-01
-1.92636952e-01 -9.59210217e-01 5.65654337e-01 8.66242684e-03
5.00399232e-01 -8.11272979e-01 3.87809873e-01 3.27857763e-01
-1.09699166e+00 -1.78701907e-01 -1.38109982e-01 9.37369186e-03
3.85743529e-01 9.21172976e-01 -5.51348150e-01 5.37394643e-01
7.46561408e-01 8.22392821e-01 1.05648287e-01 1.19666553e+00
-2.22632289e-01 1.05079210e+00 -4.76558134e-02 4.04788762e-01
-5.23844063e-02 -1.72950745e-01 1.22787583e+00 1.91371894e+00
1.19507061e-02 1.61644742e-01 -1.44676849e-01 3.48254204e-01
1.55712619e-01 -6.08700141e-02 -3.76630217e-01 -1.96224064e-01
3.88364613e-01 9.86138701e-01 -3.83089930e-01 -3.84384990e-01
-6.72598362e-01 8.90948117e-01 -1.36151984e-01 7.90589094e-01
-7.44622171e-01 -8.68901134e-01 1.15888155e+00 -1.86748788e-01
2.93236673e-01 -2.53191769e-01 -2.87301451e-01 -1.01541495e+00
-2.20052689e-01 -1.30047524e+00 1.63149118e-01 -6.41331851e-01
-1.26930058e+00 7.49296904e-01 -4.68390323e-02 -1.10561955e+00
3.17469798e-02 -6.63268268e-01 -7.76371598e-01 1.17062795e+00
-1.36918414e+00 -3.80665511e-01 1.34484068e-01 3.76846671e-01
1.22333372e+00 -6.51884913e-01 8.71270418e-01 6.65775955e-01
-7.87784636e-01 5.79085231e-01 6.91026986e-01 2.29865447e-01
9.83617783e-01 -1.33773386e+00 3.71731997e-01 9.95408297e-01
4.41498786e-01 6.45718038e-01 8.07425022e-01 -1.53044254e-01
-8.42833638e-01 -6.03908956e-01 9.58702862e-01 -2.02705413e-01
5.30880511e-01 -5.27056694e-01 -9.41580296e-01 6.35804176e-01
6.09387696e-01 -3.65590423e-01 1.48740315e+00 1.22287229e-01
-1.92842484e-02 -4.99053836e-01 -7.44736135e-01 3.30077678e-01
8.40858757e-01 -8.11366200e-01 -1.15775704e+00 2.40082070e-01
6.37456656e-01 -4.03071463e-01 -2.12639406e-01 1.32096350e-01
3.95333678e-01 -9.51400161e-01 1.12094748e+00 -6.40496612e-01
-1.90205768e-01 -1.10204481e-01 -2.78660566e-01 -1.73082519e+00
-5.42849660e-01 -9.19774830e-01 -4.49440330e-01 1.33717525e+00
3.14584285e-01 -6.40681863e-01 3.50433290e-01 2.45774567e-01
-6.88630342e-01 -2.98112482e-01 -8.84324968e-01 -1.10795498e+00
5.29109016e-02 -4.16194677e-01 2.08505064e-01 1.03839839e+00
3.26287061e-01 5.41442811e-01 -3.77120554e-01 4.15286660e-01
6.05422556e-01 1.49819583e-01 4.16274279e-01 -8.41526806e-01
-5.74785352e-01 -8.23847294e-01 -1.91534758e-01 -1.55983186e+00
4.44132507e-01 -7.88619041e-01 3.95273477e-01 -1.57869899e+00
-4.33193028e-01 2.57053763e-01 -5.05137563e-01 6.80496171e-02
-2.73983479e-01 -2.60851115e-01 -2.93005984e-02 9.62556601e-02
-8.71351436e-02 5.37267983e-01 7.97125876e-01 -4.61216748e-01
-4.64694619e-01 5.50774395e-01 -8.80089343e-01 7.67768443e-01
1.02457321e+00 -3.51732105e-01 -4.39927965e-01 -7.26555586e-01
-6.90870821e-01 -3.91432643e-02 -1.47346362e-01 -1.08294439e+00
3.49597782e-01 1.58907861e-01 1.11926839e-01 -6.07573748e-01
8.70159388e-01 -7.54222929e-01 -3.17181289e-01 6.48880261e-04
-4.04986113e-01 -5.73995531e-01 5.66910148e-01 4.06840444e-01
-9.37512457e-01 -3.42682242e-01 1.03132474e+00 1.75939500e-01
-5.25776327e-01 -2.36878172e-01 -1.04917169e+00 -6.01653531e-02
6.40390992e-01 -2.02742979e-01 6.83237389e-02 -5.58272779e-01
-1.33916044e+00 -1.81570455e-01 -5.74686170e-01 4.23988461e-01
5.70167184e-01 -1.12097597e+00 -7.35110521e-01 4.68960553e-01
-4.34461236e-01 -2.19295338e-01 3.34263891e-01 1.08811641e+00
8.01548287e-02 6.64386988e-01 3.33988756e-01 -5.75916111e-01
-1.79809260e+00 5.83586931e-01 8.38366568e-01 4.87496346e-01
-3.64891499e-01 1.46304560e+00 5.59562743e-01 -2.68998832e-01
7.93797970e-01 -2.04440892e-01 -3.81497204e-01 -3.59313600e-02
1.00893247e+00 7.98247933e-01 4.02434558e-01 -8.13649058e-01
-5.33897102e-01 3.38160217e-01 1.74192078e-02 -3.56872410e-01
1.14746559e+00 -3.26604605e-01 -3.05624396e-01 8.07286978e-01
1.22760475e+00 7.27626085e-01 -7.36098409e-01 -1.66846126e-01
-3.17491829e-01 -3.19832712e-01 2.83036262e-01 -8.23843896e-01
-1.02144086e+00 1.24339890e+00 5.84024668e-01 5.81018209e-01
1.47978783e+00 1.74784586e-02 7.20299363e-01 1.30930334e-01
-2.15412900e-01 -1.04552841e+00 -4.93141375e-02 7.61787236e-01
9.18506205e-01 -1.18557179e+00 -5.50453484e-01 -4.73994076e-01
-3.50518286e-01 1.01709700e+00 4.35331047e-01 1.61593869e-01
1.11097896e+00 7.50589728e-01 5.15915990e-01 7.53523484e-02
-1.57223240e-01 -4.12046343e-01 4.51047182e-01 1.08893836e+00
8.21821094e-01 5.74411601e-02 3.57660919e-01 9.50055540e-01
-4.30210888e-01 -7.05107808e-01 1.94301922e-02 7.38650501e-01
-8.89611185e-01 -1.02834618e+00 -1.08407199e+00 1.80847347e-01
-5.00943184e-01 -3.82871687e-01 -5.78114510e-01 6.16199262e-02
2.26827562e-02 1.57037818e+00 1.31063178e-01 -1.92381769e-01
4.86601263e-01 4.04004633e-01 3.83447796e-01 -9.43753779e-01
-4.36501235e-01 6.99686229e-01 1.33276120e-01 1.03924975e-01
-7.29579329e-01 -6.56613171e-01 -1.05364215e+00 -4.26720008e-02
-5.51350236e-01 7.21065402e-01 6.78210318e-01 1.05634594e+00
3.55580263e-02 9.90699947e-01 5.24229348e-01 -1.08311844e+00
-2.90700436e-01 -1.03284752e+00 -1.12302077e+00 -1.78781867e-01
1.30509686e+00 -4.39005613e-01 -1.06645119e+00 2.60951549e-01] | [14.822742462158203, 5.84196662902832] |
bca8cb5a-c44c-4d29-af84-3262c82024bc | a-tube-and-droplet-based-approach-for | 1609.03058 | null | http://arxiv.org/abs/1609.03058v2 | http://arxiv.org/pdf/1609.03058v2.pdf | A Tube-and-Droplet-based Approach for Representing and Analyzing Motion Trajectories | Trajectory analysis is essential in many applications. In this paper, we
address the problem of representing motion trajectories in a highly informative
way, and consequently utilize it for analyzing trajectories. Our approach first
leverages the complete information from given trajectories to construct a
thermal transfer field which provides a context-rich way to describe the global
motion pattern in a scene. Then, a 3D tube is derived which depicts an input
trajectory by integrating its surrounding motion patterns contained in the
thermal transfer field. The 3D tube effectively: 1) maintains the movement
information of a trajectory, 2) embeds the complete contextual motion pattern
around a trajectory, 3) visualizes information about a trajectory in a clear
and unified way. We further introduce a droplet-based process. It derives a
droplet vector from a 3D tube, so as to characterize the high-dimensional 3D
tube information in a simple but effective way. Finally, we apply our
tube-and-droplet representation to trajectory analysis applications including
trajectory clustering, trajectory classification & abnormality detection, and
3D action recognition. Experimental comparisons with state-of-the-art
algorithms demonstrate the effectiveness of our approach. | ['Mingliang Xu', 'Yang Zhou', 'Zicheng Liu', 'Junchi Yan', 'Weiyao Lin', 'Jianxin Wu', 'Hongteng Xu'] | 2016-09-10 | null | null | null | null | ['3d-human-action-recognition'] | ['computer-vision'] | [-3.91469970e-02 -7.86657810e-01 -2.53034145e-01 6.81603793e-03
-5.10898113e-01 -8.88514042e-01 5.99922836e-01 2.90552586e-01
8.65716636e-02 1.14823073e-01 5.01739323e-01 -2.81784534e-01
-7.93754160e-02 -7.71810830e-01 -4.88953769e-01 -1.01336849e+00
1.99063960e-02 -2.56663840e-02 5.15794158e-01 1.58008859e-01
3.58524978e-01 7.04924226e-01 -1.39814866e+00 2.51260281e-01
9.43543017e-01 8.67407680e-01 1.16721474e-01 6.88331068e-01
-4.44224060e-01 6.98852897e-01 -3.30135018e-01 2.83306509e-01
-2.19419301e-02 -6.50800765e-01 -7.16073036e-01 4.28838164e-01
-6.01879247e-02 -3.32072288e-01 -4.74336803e-01 7.70798385e-01
-3.94813977e-02 5.04382670e-01 9.34756219e-01 -1.25519598e+00
-2.92791456e-01 -9.58334580e-02 -5.06797671e-01 2.16283396e-01
5.81755936e-01 2.33156636e-01 6.23803794e-01 -7.47882485e-01
8.67294252e-01 1.25541234e+00 3.66066426e-01 2.15814859e-01
-9.38461781e-01 -3.52437377e-01 2.39877596e-01 1.12791523e-01
-1.16868806e+00 -1.05922647e-01 9.46622729e-01 -7.66280591e-01
6.72153115e-01 5.01376688e-01 1.08631694e+00 8.41588557e-01
1.44382447e-01 1.18058467e+00 4.81377572e-01 -7.44308680e-02
2.34373927e-01 -4.01296794e-01 3.01357478e-01 7.46598899e-01
-1.61866874e-01 -3.61474603e-01 -2.50748992e-01 -1.47563830e-01
7.23423600e-01 5.63405871e-01 -3.67920369e-01 -5.25080740e-01
-1.52609384e+00 2.96512574e-01 2.64394462e-01 3.69959444e-01
-2.59176761e-01 1.08742788e-01 3.98375005e-01 -6.70356676e-02
5.79431295e-01 6.37540892e-02 1.98744088e-01 -6.43361330e-01
-8.21459591e-01 5.12364984e-01 4.39219624e-01 1.03873396e+00
9.40447867e-01 -2.65536666e-01 -4.78505731e-01 3.18708509e-01
2.10058108e-01 6.17461443e-01 1.40224352e-01 -1.04750645e+00
5.49581110e-01 1.24927402e+00 2.03945264e-01 -1.16135871e+00
-2.69235343e-01 1.03077322e-01 -9.20589387e-01 -1.68322369e-01
2.96995431e-01 1.50828108e-01 -7.19272017e-01 1.36385441e+00
8.33839774e-01 5.77320814e-01 -1.80115893e-01 8.51064444e-01
5.36469579e-01 1.05309725e+00 -2.84730405e-01 -1.15053952e-01
1.06560266e+00 -7.18038857e-01 -7.77856708e-01 4.37527806e-01
9.10083234e-01 -2.86613464e-01 9.75165784e-01 1.04176648e-01
-8.89840066e-01 -4.91306782e-01 -8.34197104e-01 -5.75475916e-02
-3.85069519e-01 2.37417161e-01 1.10696517e-01 3.14671606e-01
-7.03217983e-01 6.64001584e-01 -1.36331499e+00 -5.36651790e-01
5.61718166e-01 -3.43292207e-02 -3.50802511e-01 -8.91100094e-02
-6.13982141e-01 3.87330711e-01 2.97423214e-01 4.96957414e-02
-6.11420095e-01 -7.06429183e-01 -1.11347389e+00 -2.56374300e-01
4.47937578e-01 -6.32119536e-01 8.95020187e-01 -9.41484049e-03
-1.25885737e+00 3.40160459e-01 -8.57270896e-01 1.03507765e-01
3.47839385e-01 -1.35180980e-01 -3.21351498e-01 5.10440290e-01
1.63260043e-01 7.16997460e-02 8.37316036e-01 -1.19191134e+00
-7.26677060e-01 -4.10939932e-01 -1.33583277e-01 1.85397848e-01
-2.61529446e-01 -9.71302465e-02 -7.52842903e-01 -6.43330872e-01
4.04354990e-01 -8.77788901e-01 -1.95164621e-01 1.35240242e-01
-5.35319746e-01 -2.48270363e-01 1.34449434e+00 -3.30623299e-01
1.98745537e+00 -2.31778407e+00 3.87227386e-01 4.07445401e-01
5.51218092e-01 2.95151889e-01 2.67318010e-01 8.48168910e-01
2.37080798e-01 2.24922404e-01 -7.87013292e-01 -4.27530855e-01
-3.77133153e-02 1.79319978e-01 -5.22163570e-01 6.76475644e-01
3.74118060e-01 1.08917272e+00 -1.25584435e+00 -5.44076681e-01
6.52153552e-01 3.29120427e-01 -3.63598078e-01 2.10560992e-01
-1.59172028e-01 8.40850651e-01 -1.11623144e+00 7.22167313e-01
6.10835314e-01 -1.33709297e-01 -2.43495032e-01 -2.08180442e-01
-4.48879063e-01 -2.16537844e-02 -1.10944891e+00 1.85137737e+00
6.75903857e-02 6.13376200e-01 -2.58264482e-01 -8.25585067e-01
1.01337481e+00 2.45122910e-02 9.67216611e-01 -2.98344523e-01
-1.95233617e-02 -3.73644531e-02 -6.28450096e-01 -8.83539259e-01
6.37672186e-01 3.39191616e-01 -1.76459864e-01 7.19380915e-01
-4.83009398e-01 1.27717033e-01 1.27969310e-01 2.67895043e-01
1.19145608e+00 5.26249111e-01 2.15217769e-02 -1.48795098e-01
6.98059440e-01 1.53991073e-01 4.73509163e-01 2.51833260e-01
-4.12345946e-01 6.40498996e-01 6.21678889e-01 -5.17455518e-01
-9.55049872e-01 -1.10876691e+00 1.31448150e-01 3.48612696e-01
5.31793416e-01 -8.80396366e-01 -9.20149863e-01 -6.19154751e-01
5.48267737e-02 3.68641108e-01 -7.76335955e-01 -1.12614967e-01
-8.34938824e-01 -3.39350045e-01 3.06616932e-01 7.07863927e-01
5.18800676e-01 -7.81278670e-01 -8.66424441e-01 1.65993154e-01
-3.62694561e-01 -9.98897374e-01 -8.56394231e-01 -5.27443409e-01
-9.31326330e-01 -1.08100510e+00 -5.52531421e-01 -4.01189834e-01
7.78220594e-01 8.46534312e-01 5.48892140e-01 7.93996304e-02
-3.65359783e-01 6.38128579e-01 -6.54273868e-01 1.68987647e-01
-1.84086204e-01 -3.53793710e-01 6.74732476e-02 5.29603899e-01
4.68388498e-01 -6.00499928e-01 -8.12965870e-01 3.77081215e-01
-9.55330968e-01 -1.04096703e-01 1.19983189e-01 2.84855098e-01
8.73281658e-01 2.65181631e-01 1.59218237e-01 -4.99114543e-01
4.66902137e-01 -6.10668361e-01 -2.04244018e-01 2.32085481e-01
1.04091413e-01 -1.95506290e-02 5.62512159e-01 -3.65194529e-01
-8.25856268e-01 2.82735050e-01 2.22825631e-01 -1.04537284e+00
-4.06789899e-01 2.96185017e-01 -3.72662336e-01 3.08864355e-01
2.48008639e-01 5.82511723e-01 1.43380906e-03 -5.07677555e-01
7.85815656e-01 6.06031179e-01 5.19395471e-01 -6.97139204e-01
9.50501978e-01 9.55809355e-01 2.34551802e-01 -1.06860816e+00
-3.20637822e-01 -9.96832252e-01 -1.31292617e+00 -6.07863843e-01
1.29145145e+00 -3.83102149e-01 -9.43447769e-01 6.01241410e-01
-1.07342279e+00 -2.71112055e-01 -1.58506006e-01 4.38197374e-01
-6.51840866e-01 7.13446379e-01 -3.67700219e-01 -9.70937133e-01
2.50096656e-02 -1.15833962e+00 1.43516731e+00 5.54061309e-02
-2.93074638e-01 -1.16517770e+00 3.71643364e-01 -1.26070067e-01
-1.55456141e-01 1.04437280e+00 9.47115719e-01 -1.12194628e-01
-8.79193425e-01 -1.71370357e-01 -6.64169192e-02 -1.37352362e-01
4.72636551e-01 3.17724138e-01 -7.15105176e-01 -5.81422970e-02
-1.55299157e-01 1.95829630e-01 8.96003962e-01 3.41334194e-01
1.32027316e+00 -1.88029379e-01 -8.09368074e-01 7.71926880e-01
1.07981181e+00 2.91624010e-01 4.60132003e-01 2.41498463e-03
1.14809191e+00 8.44946861e-01 7.77739227e-01 4.06936526e-01
4.28099573e-01 4.67651844e-01 2.41210654e-01 9.18490291e-02
1.74166933e-01 -6.50051236e-01 4.00209993e-01 8.79147768e-01
-6.98802546e-02 -1.88498005e-01 -8.30450892e-01 5.26232898e-01
-2.28038859e+00 -1.18206871e+00 -4.70565915e-01 2.05781484e+00
1.47832274e-01 -2.56183654e-01 5.93408227e-01 2.77460605e-01
7.31426239e-01 5.28813899e-01 -7.94773281e-01 4.11881730e-02
2.84378380e-01 -3.54843646e-01 3.66011411e-02 2.63083488e-01
-1.22939634e+00 8.55390370e-01 5.81181240e+00 9.14722383e-01
-8.24801981e-01 -3.87237012e-01 2.13945821e-01 4.01329249e-02
-4.17493790e-01 -8.09748247e-02 -4.73547697e-01 5.19562721e-01
6.11576736e-01 -3.14239264e-01 6.26241341e-02 5.64196825e-01
5.54735422e-01 -5.38350046e-02 -1.19603741e+00 1.15291989e+00
-2.31119111e-01 -1.54101634e+00 3.73277038e-01 2.59452999e-01
3.96048129e-01 -4.90339369e-01 6.72527775e-02 -7.77123943e-02
-1.45855278e-01 -8.27406228e-01 7.51328409e-01 7.28827953e-01
8.48124206e-01 -9.00557876e-01 -8.21599737e-02 6.64630413e-01
-2.05835891e+00 -2.02690507e-03 -2.77136445e-01 1.26448780e-01
2.95549929e-01 3.76465976e-01 -5.26560426e-01 9.69370008e-01
5.85045099e-01 1.66652691e+00 -3.72444957e-01 1.06184697e+00
1.56915352e-01 5.91778338e-01 -1.33482441e-01 -1.24599740e-01
5.74648619e-01 -8.52222860e-01 8.78804386e-01 1.52120602e+00
4.92013931e-01 5.67787707e-01 3.18836749e-01 1.16892254e+00
2.89031714e-01 2.89290454e-02 -1.09083116e+00 -1.28038853e-01
5.64766884e-01 1.16741574e+00 -9.90236402e-01 -2.71657020e-01
-2.08723158e-01 1.04248941e+00 1.76124573e-01 5.44607460e-01
-6.51261330e-01 -4.07957792e-01 8.70854557e-01 4.53466885e-02
3.03422689e-01 -6.06580555e-01 1.18401140e-01 -1.21215892e+00
4.08484280e-01 -2.39400059e-01 1.24869674e-01 -6.20983064e-01
-1.01145685e+00 4.09765810e-01 1.98251858e-01 -2.04450917e+00
-1.68507397e-01 -4.50208038e-01 -1.08292413e+00 7.01505542e-01
-1.11968660e+00 -9.79672492e-01 -3.90356511e-01 8.22417676e-01
4.02754664e-01 2.91098207e-01 5.81404030e-01 5.25123253e-02
-9.44250941e-01 4.64410186e-02 3.60920191e-01 2.61763453e-01
2.60854751e-01 -1.27468193e+00 6.86602354e-01 9.77466762e-01
3.69993597e-02 7.90478945e-01 1.36671796e-01 -8.08088183e-01
-1.83023000e+00 -1.51593518e+00 3.28626364e-01 -9.82959688e-01
7.60279059e-01 -4.47580636e-01 -1.12465811e+00 5.26510298e-01
-3.33910197e-01 -5.55150248e-02 7.09852695e-01 -4.57069933e-01
-2.54161954e-01 3.63637619e-02 -8.49835455e-01 7.16369867e-01
1.37902153e+00 -7.03776002e-01 -6.31874621e-01 5.76928630e-02
8.70474696e-01 -3.64986748e-01 -8.71932924e-01 1.08459570e-01
5.06124020e-01 -9.33589339e-01 1.02664912e+00 -7.17062294e-01
3.56692731e-01 -7.22458601e-01 2.69809254e-02 -1.18632424e+00
-2.75825799e-01 -1.10688806e+00 -4.45492089e-01 1.12274146e+00
-1.25051225e-02 -3.36737156e-01 7.42169321e-01 4.60595459e-01
-4.07470077e-01 -1.08319080e+00 -7.88419247e-01 -7.23652959e-01
-2.48041824e-02 -7.80223668e-01 8.99438083e-01 7.56863832e-01
3.01007599e-01 -2.47435585e-01 -7.80625343e-02 1.04460448e-01
6.34418309e-01 3.73814970e-01 9.33658063e-01 -8.85980010e-01
3.79104525e-01 -5.40447176e-01 -5.03880084e-01 -1.56716657e+00
1.23132020e-01 -9.56356227e-01 1.65801365e-02 -1.67375529e+00
1.68773428e-01 -2.92774379e-01 2.56760623e-02 -6.44623935e-02
-3.13497990e-01 -2.99866647e-01 2.32603788e-01 5.97647488e-01
-7.52262831e-01 8.60555947e-01 1.48876393e+00 3.99730168e-02
-5.27195036e-01 7.06400424e-02 -2.21276253e-01 7.14897811e-01
5.38359344e-01 -8.38386938e-02 -5.95929325e-01 -1.82038635e-01
-3.62949938e-01 1.45067945e-01 3.81748527e-01 -8.60987544e-01
3.57768416e-01 -2.62486696e-01 2.75507420e-01 -1.29243016e+00
3.18646997e-01 -8.68912697e-01 -3.49167213e-02 2.49271199e-01
-7.64297098e-02 8.34890157e-02 4.26491536e-02 9.51149106e-01
-1.41619459e-01 2.02996537e-01 2.88194209e-01 2.99760252e-02
-6.56215847e-01 7.92526782e-01 -3.97823930e-01 -8.22415575e-02
1.28814733e+00 -6.24027133e-01 -1.77681237e-01 -2.33630359e-01
-5.93141317e-01 5.46757102e-01 7.38561392e-01 3.89442831e-01
1.02381325e+00 -1.71804333e+00 -3.56418818e-01 3.21617007e-01
2.97050595e-01 3.46898854e-01 3.35414946e-01 7.92337537e-01
-5.57437003e-01 3.37483048e-01 3.72148827e-02 -1.19525766e+00
-1.03798246e+00 6.74895644e-01 1.55608937e-01 6.68305755e-02
-1.26751518e+00 3.28162134e-01 3.69922012e-01 -1.61670789e-01
8.68934169e-02 -9.85708117e-01 -3.11952353e-01 1.18530661e-01
8.75781417e-01 8.39538157e-01 -4.15235698e-01 -9.39209223e-01
-4.47019488e-01 1.39191186e+00 4.21171099e-01 -2.03732714e-01
1.04428995e+00 -2.96506256e-01 -1.67741075e-01 8.32719922e-01
1.51020408e+00 -1.30586728e-01 -1.56121635e+00 -2.12238412e-02
4.45945486e-02 -7.38772452e-01 -3.57680082e-01 3.99779603e-02
-7.71540463e-01 1.12585104e+00 -9.10596400e-02 2.83493340e-01
1.17828584e+00 8.32285546e-03 1.19308031e+00 3.54990035e-01
2.16521487e-01 -6.28272235e-01 1.12092003e-01 2.46210337e-01
5.93728125e-01 -7.88355529e-01 -2.04393417e-01 -5.01568913e-01
-7.21956491e-01 1.28968465e+00 1.83091104e-01 -8.16068649e-02
6.99623227e-01 2.99599748e-02 -2.84601927e-01 -3.70804042e-01
-4.07794446e-01 -2.79463857e-01 5.21016121e-01 6.66070044e-01
-1.11365907e-01 5.17223999e-02 3.03317130e-01 3.60031784e-01
8.54326189e-02 -2.04294696e-01 3.60771447e-01 1.05636370e+00
-5.89618206e-01 -8.90253484e-01 -3.53121966e-01 1.34799600e-01
1.37065113e-01 4.52469915e-01 -5.06965876e-01 6.50769830e-01
5.28557673e-02 8.80655944e-01 2.14636892e-01 -7.97276437e-01
4.44446653e-01 8.89064148e-02 2.30892479e-01 -4.77606267e-01
-2.76178241e-01 2.83577293e-01 -3.07851255e-01 -1.02204466e+00
-5.35420716e-01 -9.82037187e-01 -1.62175035e+00 -2.28150055e-01
2.18549907e-01 4.10004072e-02 4.25217748e-01 9.50541735e-01
4.90650803e-01 3.64691556e-01 8.64556134e-01 -8.98446739e-01
5.88485077e-02 -5.23289442e-01 -5.85603595e-01 4.92756665e-01
8.67231071e-01 -7.15014875e-01 -3.24475527e-01 2.45140880e-01] | [8.107708930969238, 0.2822382152080536] |
0163cde5-7e9a-4aa8-a702-15ab50a8974f | identifying-adversarial-attacks-on-text | 2201.08555 | null | https://arxiv.org/abs/2201.08555v1 | https://arxiv.org/pdf/2201.08555v1.pdf | Identifying Adversarial Attacks on Text Classifiers | The landscape of adversarial attacks against text classifiers continues to grow, with new attacks developed every year and many of them available in standard toolkits, such as TextAttack and OpenAttack. In response, there is a growing body of work on robust learning, which reduces vulnerability to these attacks, though sometimes at a high cost in compute time or accuracy. In this paper, we take an alternate approach -- we attempt to understand the attacker by analyzing adversarial text to determine which methods were used to create it. Our first contribution is an extensive dataset for attack detection and labeling: 1.5~million attack instances, generated by twelve adversarial attacks targeting three classifiers trained on six source datasets for sentiment analysis and abuse detection in English. As our second contribution, we use this dataset to develop and benchmark a number of classifiers for attack identification -- determining if a given text has been adversarially manipulated and by which attack. As a third contribution, we demonstrate the effectiveness of three classes of features for these tasks: text properties, capturing content and presentation of text; language model properties, determining which tokens are more or less probable throughout the input; and target model properties, representing how the text classifier is influenced by the attack, including internal node activations. Overall, this represents a first step towards forensics for adversarial attacks against text classifiers. | ['Daniel Lowd', 'Sameer Singh', 'Sabrina Reis', 'Carter Perkins', 'Kalyani Asthana', 'Wencong You', 'Adam Noack', 'Jonathan Brophy', 'Zhouhang Xie'] | 2022-01-21 | null | null | null | null | ['adversarial-text', 'abuse-detection'] | ['adversarial', 'natural-language-processing'] | [ 4.26382959e-01 -2.15218097e-01 -1.75813392e-01 -2.84168720e-01
-8.69916499e-01 -1.48597991e+00 9.57600057e-01 5.67500472e-01
-2.22805932e-01 3.89994770e-01 2.34114587e-01 -8.02598894e-01
1.64183140e-01 -8.98576081e-01 -5.34783840e-01 -5.04463017e-01
-4.86937240e-02 3.51708084e-01 1.05997689e-01 -3.69237959e-01
4.23928261e-01 7.69331813e-01 -9.64692295e-01 8.82055163e-01
1.93074510e-01 7.12872028e-01 -9.16064262e-01 1.12889552e+00
-1.67199284e-01 1.21700656e+00 -1.16672325e+00 -1.13408267e+00
1.44738749e-01 -1.75618306e-01 -1.08014238e+00 -4.69153970e-01
5.05949140e-01 -3.63904804e-01 -6.51178777e-01 1.13370609e+00
6.20155156e-01 -2.96727031e-01 6.85709953e-01 -1.46273887e+00
-4.05096143e-01 1.17267990e+00 -3.19076091e-01 7.23694265e-01
6.63396180e-01 4.18236643e-01 1.00035346e+00 -4.67142344e-01
4.66488779e-01 1.49686623e+00 7.24416137e-01 7.70544410e-01
-1.02625215e+00 -9.90972519e-01 -5.37049957e-02 3.31951194e-02
-8.41417968e-01 -6.30698562e-01 7.81397879e-01 -4.30309951e-01
1.04845321e+00 5.62864602e-01 1.42690703e-01 1.95529974e+00
4.12650257e-01 9.01510894e-01 1.07724988e+00 -4.77776945e-01
1.87992394e-01 2.64642984e-01 3.89114261e-01 5.96067965e-01
2.07361147e-01 1.23752676e-01 -2.92493105e-01 -8.81328821e-01
-2.27469429e-01 -8.25368911e-02 5.41632511e-02 2.43178979e-01
-4.54261750e-01 1.32092392e+00 -2.14868486e-02 3.10869217e-01
7.52286911e-02 2.36469954e-02 1.10609126e+00 5.49779117e-01
4.89943117e-01 5.47536850e-01 -5.70430517e-01 -3.22946429e-01
-7.39665151e-01 4.88926947e-01 1.25669086e+00 4.87411678e-01
1.55858412e-01 2.92465478e-01 -4.92931381e-02 5.45073330e-01
2.55820543e-01 6.78061068e-01 5.07990777e-01 -1.26540005e-01
8.07224929e-01 4.53382373e-01 -3.83960456e-01 -1.14288032e+00
-2.73594707e-01 1.81073286e-02 -3.48954082e-01 2.18852699e-01
5.10665476e-01 -3.50841552e-01 -8.19924533e-01 1.41778672e+00
2.41012856e-01 -1.07153773e-01 1.07420355e-01 1.38823539e-01
5.55828691e-01 5.78078985e-01 4.41852808e-01 2.40841523e-01
1.34017372e+00 -4.97254759e-01 -4.93341476e-01 -2.86546409e-01
1.10030317e+00 -9.54363942e-01 9.17941630e-01 5.98908484e-01
-9.15955424e-01 1.37340367e-01 -1.11102390e+00 3.60794574e-01
-1.17297685e+00 -6.45827174e-01 4.57425535e-01 1.55636060e+00
-3.81641835e-01 6.44019008e-01 -5.58461785e-01 -1.24274559e-01
7.38659561e-01 1.79569870e-01 -4.01487350e-01 1.91166382e-02
-1.64717126e+00 1.07526386e+00 2.81640589e-01 -4.91102338e-01
-1.09956694e+00 -7.40731239e-01 -8.95484626e-01 -3.63545716e-02
1.38023704e-01 2.35644784e-02 1.31916773e+00 -9.55926299e-01
-1.06876421e+00 9.83501613e-01 2.60834306e-01 -6.04563415e-01
5.12242258e-01 -1.75512895e-01 -8.85527790e-01 1.51561245e-01
-1.25430211e-01 -2.42309913e-01 1.27914548e+00 -9.57058966e-01
-3.11554968e-01 -6.06188118e-01 3.29117715e-01 -1.83529750e-01
-9.17789698e-01 8.64903986e-01 1.21656299e-01 -1.09587920e+00
-6.21732354e-01 -7.70506918e-01 -2.36442313e-02 -2.77374536e-01
-8.40152502e-01 -9.82198715e-02 1.40334821e+00 -7.01217711e-01
1.51203322e+00 -2.04679394e+00 -5.30449338e-02 4.52418149e-01
2.75959224e-01 7.57223427e-01 8.00776035e-02 8.18755805e-01
-5.32279491e-01 7.66190350e-01 -3.37485373e-01 -3.19001019e-01
2.11242318e-01 9.69418809e-02 -1.14567828e+00 6.48292482e-01
1.91822261e-01 7.62892067e-01 -6.29973471e-01 -3.77478778e-01
1.88139364e-01 2.69606650e-01 -3.77013117e-01 9.98614132e-02
-1.14473484e-01 -3.52056026e-02 -5.69319606e-01 7.83560216e-01
4.59841847e-01 3.78525734e-01 -7.57218078e-02 1.35224313e-01
4.61035162e-01 4.44761992e-01 -1.03552985e+00 8.66078615e-01
-3.75111610e-01 7.50957549e-01 2.22023174e-01 -8.33638906e-01
6.53361559e-01 3.24605405e-01 2.49289200e-01 -2.23903164e-01
5.86256564e-01 -7.97818452e-02 -1.09627306e-01 -4.52014685e-01
2.54673690e-01 8.87299851e-02 -6.54401958e-01 9.70398426e-01
7.58117586e-02 -1.65929541e-01 1.73765793e-01 6.53961718e-01
1.66974866e+00 -4.84649777e-01 2.37465858e-01 1.83492899e-01
7.64123023e-01 -2.84284595e-02 -6.89134970e-02 1.03222024e+00
-3.34393859e-01 7.28856921e-02 7.71672428e-01 -7.31451035e-01
-9.95336533e-01 -8.86285186e-01 -1.73208743e-01 1.37399805e+00
-5.02581775e-01 -8.01954865e-01 -1.07427049e+00 -1.39357853e+00
1.01840086e-01 8.47388566e-01 -6.34847701e-01 -5.69456100e-01
-6.53106928e-01 -8.65370810e-01 1.71635854e+00 6.10044420e-01
1.74167767e-01 -1.29163408e+00 -1.96803644e-01 -1.83259952e-03
8.28112736e-02 -1.05165279e+00 -3.29813123e-01 5.10078549e-01
-4.90773439e-01 -1.26793671e+00 9.03931782e-02 -3.60825181e-01
4.64459985e-01 -3.07268083e-01 1.09643245e+00 3.34206492e-01
-6.59373343e-01 4.50494856e-01 -6.45235479e-01 -8.27846646e-01
-1.06602383e+00 2.70839959e-01 8.08696374e-02 -9.86874774e-02
5.60461342e-01 -2.84144342e-01 5.62713891e-02 9.96500850e-02
-1.33249879e+00 -8.34258378e-01 3.33408862e-01 5.49498022e-01
-2.56953329e-01 3.42508227e-01 2.41933167e-01 -1.50361574e+00
9.88573134e-01 -9.24030662e-01 -1.18565097e-01 5.12546115e-02
-3.58599007e-01 -1.52921706e-01 1.21920824e+00 -7.59435773e-01
-6.83989108e-01 -1.08710259e-01 -6.69796526e-01 -3.30772400e-01
-3.65637451e-01 2.84426540e-01 -4.13547128e-01 -1.62313476e-01
1.22063005e+00 1.05000108e-01 -2.95351386e-01 -1.51981622e-01
4.87614602e-01 8.41560960e-01 2.29864269e-01 -7.02299654e-01
1.42240453e+00 2.63469011e-01 -3.04262251e-01 -7.71100461e-01
-8.70455861e-01 -2.91706443e-01 -5.20772636e-01 -2.75310203e-02
5.75350702e-01 -2.69242018e-01 -7.40264118e-01 8.30548942e-01
-9.68916237e-01 -2.99188673e-01 1.04756303e-01 -2.14278460e-01
-4.93379831e-02 6.31213665e-01 -9.62006152e-01 -8.08706462e-01
-7.02168345e-01 -1.14837718e+00 6.45994067e-01 -2.78570801e-01
-5.32457352e-01 -1.37162530e+00 -8.56335834e-03 3.62592220e-01
3.81191850e-01 5.03422916e-01 1.12059367e+00 -1.59220767e+00
4.09152746e-01 -9.55491543e-01 1.42710060e-01 3.75129044e-01
-1.44653574e-01 3.02709639e-01 -1.35572314e+00 -5.19059777e-01
1.51728302e-01 -7.30558753e-01 5.93966782e-01 -1.84872612e-01
1.29427183e+00 -6.54535711e-01 -4.00005519e-01 5.40487051e-01
1.04483855e+00 1.31995648e-01 5.50152183e-01 3.63412678e-01
8.32820117e-01 5.60998321e-01 1.60361603e-01 4.08053577e-01
-2.15946317e-01 4.30841923e-01 6.04039431e-01 6.02682419e-02
3.14327240e-01 -4.38086003e-01 7.56097555e-01 1.02664374e-01
6.61632776e-01 -6.24679029e-01 -1.11056054e+00 1.26387656e-01
-1.25362539e+00 -1.26470363e+00 -1.41981706e-01 1.77107573e+00
8.65062237e-01 7.84119666e-01 2.24694982e-01 7.60676026e-01
5.52917719e-01 4.37162071e-01 -4.43767339e-01 -9.77813363e-01
-6.97082058e-02 6.50099874e-01 6.23178840e-01 5.06756783e-01
-1.54340243e+00 1.10149658e+00 6.68837929e+00 9.83517706e-01
-1.06250048e+00 -8.07101578e-02 5.54251671e-01 -1.17893666e-01
-7.20623210e-02 -1.86820537e-01 -9.98538971e-01 5.71375132e-01
1.52913821e+00 -4.34650555e-02 4.01462913e-01 9.99744594e-01
-2.67604858e-01 3.49479139e-01 -1.01938355e+00 5.68570137e-01
4.49254453e-01 -1.24930489e+00 3.40577126e-01 3.73856388e-02
2.62312323e-01 -1.52886540e-01 3.17435235e-01 4.48995262e-01
9.16982710e-01 -1.21336317e+00 8.25862348e-01 -1.11401498e-01
5.50619006e-01 -9.87934649e-01 6.21967733e-01 3.69452566e-01
-9.93154407e-01 -4.47297782e-01 7.20374361e-02 9.21622217e-02
-2.05423787e-01 2.09266230e-01 -7.59954929e-01 1.80019259e-01
7.08088160e-01 4.93893206e-01 -9.86987114e-01 2.50588804e-01
-3.73044372e-01 1.18517578e+00 -1.55949041e-01 -1.64282262e-01
3.54717076e-01 5.85890293e-01 7.00686097e-01 1.61565828e+00
-5.25379956e-01 -2.53649279e-02 2.63081759e-01 4.34893966e-01
-2.31178597e-01 6.73471466e-02 -1.04501736e+00 -4.45121169e-01
6.77324474e-01 1.38331926e+00 -6.52716696e-01 -1.83157802e-01
-1.81331202e-01 8.62657309e-01 1.83844119e-01 -3.43118459e-02
-9.51481998e-01 -7.98824131e-01 6.38105094e-01 -4.16394882e-02
-6.33360222e-02 8.09362978e-02 -2.89017528e-01 -1.08921850e+00
-7.88279101e-02 -1.58867657e+00 9.45581853e-01 -3.36365312e-01
-1.61468506e+00 7.26108372e-01 -8.69263709e-03 -7.23395407e-01
-3.78948092e-01 -9.24618602e-01 -9.14104760e-01 9.00494754e-01
-9.36574876e-01 -1.37802112e+00 2.69978046e-01 9.29404020e-01
5.49874604e-01 -5.67646801e-01 1.17978442e+00 3.17607373e-01
-7.38684952e-01 1.09964585e+00 -8.77476186e-02 9.19252455e-01
6.86941028e-01 -1.30688834e+00 1.00529182e+00 9.37732160e-01
3.10747951e-01 6.41888380e-01 6.99527621e-01 -7.63394296e-01
-1.42949069e+00 -1.01956546e+00 4.74834532e-01 -1.25206864e+00
1.33227217e+00 -8.38572502e-01 -9.59945738e-01 8.32418382e-01
-7.65120238e-02 6.24029152e-02 9.45638418e-01 -6.10214472e-02
-9.51381922e-01 3.18304092e-01 -1.46830142e+00 6.35417998e-01
5.86812675e-01 -9.26800013e-01 -5.66849411e-01 4.87166971e-01
5.47304749e-01 -3.55824739e-01 -7.04306066e-01 -4.81462926e-02
3.32747132e-01 -6.95400655e-01 9.75609243e-01 -1.24993491e+00
4.91539598e-01 2.28275999e-01 -2.12667674e-01 -1.13182151e+00
-3.67486812e-02 -9.74119723e-01 -3.84483606e-01 1.58776879e+00
4.99676973e-01 -7.33054638e-01 8.02038193e-01 4.67606366e-01
5.80895916e-02 -6.20382369e-01 -7.45177448e-01 -3.52587909e-01
6.65009141e-01 -8.72766018e-01 4.35464442e-01 1.31646907e+00
9.13586318e-02 2.53228337e-01 -1.26202062e-01 2.07430378e-01
4.88511205e-01 -7.05580831e-01 8.06942880e-01 -1.02370870e+00
-1.34773239e-01 -6.29165471e-01 -4.60810453e-01 -3.14844072e-01
5.48471510e-01 -1.05676579e+00 -4.93415415e-01 -6.73790574e-01
-9.56690162e-02 -2.95671254e-01 2.22523473e-02 8.01671743e-01
-2.15705782e-01 3.51207405e-01 1.58418804e-01 1.81282833e-01
-1.29431531e-01 -1.16596192e-01 2.88735718e-01 -4.17496741e-01
2.00280517e-01 1.82888940e-01 -7.89598644e-01 9.15996730e-01
8.50776494e-01 -8.87417614e-01 -1.60485044e-01 -2.03348204e-01
3.02281499e-01 -3.22111487e-01 2.74566680e-01 -8.02917898e-01
1.57160312e-02 8.66155699e-02 6.30203843e-01 -3.40345204e-01
1.37958393e-01 -6.98473573e-01 -5.30563653e-01 5.50851941e-01
-6.17426932e-01 4.32175338e-01 5.26459932e-01 6.01337016e-01
1.04401708e-01 -6.07394278e-01 1.04982042e+00 -9.03859735e-02
-3.59452128e-01 3.39603901e-01 -7.18139410e-01 4.82101172e-01
1.19827855e+00 8.08270574e-02 -6.37629747e-01 -3.20144325e-01
-7.01375425e-01 -1.48426101e-01 2.82840192e-01 6.51293755e-01
5.07598996e-01 -8.47149134e-01 -7.76914835e-01 3.34845513e-01
1.00103974e-01 -8.43098998e-01 -1.55382767e-01 1.22355781e-01
-4.71833646e-01 3.90883982e-02 -7.89178684e-02 -8.17878470e-02
-1.70289016e+00 8.16159546e-01 4.26382750e-01 -6.12127364e-01
-4.66610163e-01 7.59758532e-01 -4.29610640e-01 -5.34296930e-01
3.45191658e-01 3.68810117e-01 -2.35978022e-01 9.36995521e-02
8.99024665e-01 2.94086576e-01 2.72977591e-01 -7.41178930e-01
-3.08400273e-01 4.13122289e-02 -5.85039198e-01 -1.47899106e-01
9.90552187e-01 5.79994023e-01 -1.39403477e-01 2.67420918e-01
1.22916114e+00 4.04576838e-01 -5.15711904e-01 -1.06984481e-01
1.55853972e-01 -4.22810078e-01 -1.44385308e-01 -9.96225059e-01
-8.41118634e-01 9.45558429e-01 3.89453709e-01 9.13863122e-01
7.49961853e-01 -1.35072306e-01 8.13006997e-01 3.89038563e-01
-5.92604056e-02 -7.29308546e-01 2.62185633e-01 7.42019951e-01
5.81544101e-01 -1.00281882e+00 1.15473196e-01 -2.43570819e-01
-7.07212329e-01 1.26026201e+00 4.92740631e-01 -8.70268047e-02
7.53553629e-01 7.31151938e-01 6.65186644e-02 -3.01907092e-01
-5.65661311e-01 4.39341426e-01 7.69623742e-02 6.29287541e-01
3.18461508e-01 -1.01139553e-01 3.24051797e-01 4.52897668e-01
-6.00186944e-01 -1.01786077e+00 4.68512595e-01 1.27669442e+00
-1.27657995e-01 -1.32330525e+00 -6.49685383e-01 6.76841080e-01
-1.24894261e+00 -2.78785616e-01 -1.18516648e+00 6.85222447e-01
-1.23097435e-01 1.28156257e+00 -3.30665946e-01 -6.41977072e-01
4.74296302e-01 2.57226080e-01 1.41163886e-01 -8.06659698e-01
-1.46557570e+00 -5.98396003e-01 3.44795972e-01 -3.64701658e-01
2.94379413e-01 -6.74367726e-01 -1.03334975e+00 -8.40800166e-01
-4.20209557e-01 7.87698850e-02 7.61096358e-01 9.83398676e-01
-4.08385955e-02 2.43288979e-01 9.37995255e-01 -5.51894546e-01
-1.05300581e+00 -1.00391328e+00 -2.95298517e-01 8.10506225e-01
1.60852775e-01 -2.73155034e-01 -8.18344653e-01 1.30866215e-01] | [5.988375663757324, 8.011249542236328] |
4da3a484-4c71-4fb6-ba9f-693a87ab4450 | nasnet-a-neuron-attention-stage-by-stage-net | 1912.03151 | null | https://arxiv.org/abs/1912.03151v2 | https://arxiv.org/pdf/1912.03151v2.pdf | NASNet: A Neuron Attention Stage-by-Stage Net for Single Image Deraining | Images captured under complicated rain conditions often suffer from noticeable degradation of visibility. The rain models generally introduce diversity visibility degradation, which includes rain streak, rain drop as well as rain mist. Numerous existing single image deraining methods focus on the only one type rain model, which does not have strong generalization ability. In this paper, we propose a novel end-to-end Neuron Attention Stage-by-Stage Net (NASNet), which can solve all types of rain model tasks efficiently. For one thing, we pay more attention on the Neuron relationship and propose a lightweight Neuron Attention (NA) architectural mechanism. It can adaptively recalibrate neuron-wise feature responses by modelling interdependencies and mutual influence between neurons. Our NA architecture consists of Depthwise Conv and Pointwise Conv, which has slight computation cost and higher performance than SE block by our contrasted experiments. For another, we propose a stage-by-stage unified pattern network architecture, the stage-by-stage strategy guides the later stage by incorporating the useful information in previous stage. We concatenate and fuse stage-level information dynamically by NA module. Extensive experiments demonstrate that our proposed NASNet significantly outperforms the state-of-theart methods by a large margin in terms of both quantitative and qualitative measures on all six public large-scale datasets for three rain model tasks. | ['Zhilin Wang', 'Xu Qin'] | 2019-12-06 | null | null | null | null | ['single-image-deraining'] | ['computer-vision'] | [-2.09621191e-01 -5.75506508e-01 5.31032681e-01 -7.24587142e-01
-1.25109777e-01 -1.64236054e-01 1.55005381e-01 -4.44542110e-01
-5.20445883e-01 7.51330853e-01 2.41964743e-01 -9.27492231e-02
5.71884438e-02 -7.38382995e-01 -7.13775396e-01 -1.13153934e+00
2.41066958e-03 3.00182477e-02 3.13948005e-01 -5.10433614e-01
-2.28944095e-03 6.11278415e-01 -1.56615257e+00 2.54869938e-01
1.18209767e+00 6.99752867e-01 6.93619370e-01 5.88022292e-01
-1.63862407e-01 9.68473852e-01 -5.73081434e-01 4.63923141e-02
1.65402308e-01 -4.36566621e-01 -1.76937610e-01 -2.27331128e-02
5.94700992e-01 -3.77809018e-01 -3.84737194e-01 1.09809864e+00
8.28146338e-01 6.72154352e-02 3.21436286e-01 -9.12958384e-01
-9.14385855e-01 5.19933283e-01 -7.73600280e-01 8.76046777e-01
-3.70802075e-01 5.08580506e-01 8.24746907e-01 -1.27555490e+00
1.36371195e-01 1.33442473e+00 6.31892920e-01 4.86652523e-01
-8.52831006e-01 -8.73124957e-01 7.87910759e-01 2.45950550e-01
-1.21821284e+00 -1.51163623e-01 4.48610008e-01 -5.00606652e-03
8.59681308e-01 2.99596161e-01 8.74786973e-01 7.64326870e-01
3.63645643e-01 8.40713918e-01 1.28132915e+00 2.28534207e-01
1.53520722e-02 -2.81854123e-01 4.78436381e-01 6.10667229e-01
6.00039899e-01 -7.88066015e-02 1.57801975e-02 4.07540649e-01
8.16790283e-01 5.12247264e-01 -7.77003169e-01 2.30673492e-01
-9.41833436e-01 6.50820673e-01 1.14094865e+00 2.53474385e-01
-5.20872533e-01 1.23964593e-01 -6.50324523e-02 5.40420890e-01
5.39300859e-01 3.14033210e-01 -4.96801317e-01 5.67630053e-01
-9.19393718e-01 4.42215532e-01 5.23560584e-01 5.00417173e-01
9.98233140e-01 4.47022587e-01 -3.72437000e-01 6.90711498e-01
3.54169399e-01 9.40980434e-01 2.32571021e-01 -5.68038523e-01
3.26287866e-01 4.47291732e-01 7.30902031e-02 -7.99339771e-01
-7.24690616e-01 -8.39364827e-01 -1.26211798e+00 4.78890866e-01
-7.59369507e-02 -2.13595733e-01 -1.54468632e+00 1.64578569e+00
8.71905163e-02 4.92447138e-01 2.36911461e-01 1.36862421e+00
1.42898107e+00 9.08012033e-01 2.92337593e-02 -4.01885062e-01
1.24932981e+00 -1.25662005e+00 -7.61433721e-01 -4.05669659e-01
-1.93360336e-02 -4.56535280e-01 1.09753919e+00 1.89802259e-01
-9.82825637e-01 -6.54101074e-01 -1.01693058e+00 6.07136339e-02
-2.91594684e-01 -6.45315647e-03 4.40530449e-01 1.45285919e-01
-1.10639966e+00 3.34640920e-01 -5.96009314e-01 -3.85918915e-01
6.43095672e-01 2.61838287e-01 -2.13992208e-01 -4.33419347e-01
-1.14629173e+00 8.44538927e-01 2.32251622e-02 9.43195283e-01
-1.18310285e+00 -6.91846967e-01 -6.28924131e-01 2.09262341e-01
-1.38973713e-01 -7.23453999e-01 9.06282544e-01 -1.03674924e+00
-1.16559911e+00 4.66872633e-01 -3.24198365e-01 -3.89145344e-01
1.41900316e-01 -5.65572262e-01 -4.94718760e-01 -2.15006806e-02
-2.75722414e-01 5.72717249e-01 7.89094925e-01 -1.52338910e+00
-8.24312687e-01 -2.15859368e-01 2.08434105e-01 5.56419551e-01
-6.50421456e-02 -2.75278557e-02 -4.42709029e-01 -6.20691061e-01
9.70863476e-02 -5.59857428e-01 -4.74951863e-01 1.23752296e-01
7.30163511e-03 1.93252265e-02 9.82304096e-01 -5.08756876e-01
1.06822693e+00 -2.02995253e+00 1.60072207e-01 -1.20428555e-01
5.92109084e-01 7.16284156e-01 -3.16832691e-01 -4.13314402e-02
3.61129157e-02 -2.06914637e-02 -5.20534456e-01 -3.88905048e-01
-4.81582969e-01 5.40251315e-01 -4.30348843e-01 4.91389364e-01
2.80230105e-01 9.83185947e-01 -7.83012986e-01 -3.13480109e-01
6.30527511e-02 5.24361193e-01 -1.99789718e-01 5.46423852e-01
-7.85630792e-02 2.61416078e-01 -2.14004055e-01 9.88621891e-01
1.11961019e+00 -8.59357417e-03 -4.88689691e-01 -3.64569157e-01
-2.55633086e-01 -2.18807250e-01 -8.51918101e-01 1.30784917e+00
-4.63476270e-01 5.72008491e-01 2.17110664e-01 -6.27737522e-01
1.06538939e+00 1.20631650e-01 -2.46697098e-01 -6.24119937e-01
2.91119516e-01 1.94875121e-01 3.30733657e-01 -7.76360095e-01
1.38633296e-01 -1.70335189e-01 3.32429737e-01 1.49554074e-01
-9.41508114e-02 2.51498222e-01 -1.41264960e-01 -6.28722012e-02
9.21562612e-01 -1.09830096e-01 1.46020786e-03 -3.77489626e-01
2.63537198e-01 -2.09731147e-01 7.61286855e-01 8.50001693e-01
-3.37126732e-01 1.05722106e+00 8.54357705e-02 -6.11058891e-01
-7.16086924e-01 -8.33935738e-01 -9.25450996e-02 9.45360601e-01
4.39864725e-01 3.23887914e-01 -5.99709928e-01 -5.83253026e-01
-5.29154204e-02 4.04164612e-01 -1.06960905e+00 3.43069434e-02
-6.96840823e-01 -1.42312038e+00 5.07356524e-01 6.65943861e-01
1.08736682e+00 -1.69422245e+00 -5.44676006e-01 1.45643666e-01
5.11285178e-02 -8.68488967e-01 -3.64969641e-01 4.12608236e-01
-8.22871923e-01 -7.61519909e-01 -9.03691173e-01 -9.91648793e-01
6.96144640e-01 8.19393158e-01 1.13081348e+00 3.42006266e-01
-2.05552921e-01 -3.86101097e-01 -4.78330135e-01 -4.72646296e-01
5.00295341e-01 -3.64346541e-02 -1.10247239e-01 1.58866435e-01
3.23727667e-01 -9.20167029e-01 -9.80901182e-01 1.86648309e-01
-9.53819573e-01 -2.06543542e-02 1.02157593e+00 8.18196476e-01
5.67640662e-01 -2.48707607e-01 4.94566917e-01 -8.79543722e-01
6.39268875e-01 -7.82427967e-01 -3.21070820e-01 1.21473216e-01
-4.64119107e-01 -1.34389192e-01 5.90061367e-01 -1.29062757e-01
-1.13462269e+00 -1.25512660e-01 -1.05684757e-01 -7.91651428e-01
-2.35449359e-01 5.80715716e-01 -4.52160358e-01 -1.96156114e-01
4.63416010e-01 5.17963171e-01 -2.82882780e-01 -5.91614962e-01
1.92948893e-01 5.14129579e-01 7.23778784e-01 3.10007595e-02
1.05907881e+00 5.58606088e-01 -3.80271673e-01 -5.69507301e-01
-1.01153660e+00 -3.96218807e-01 -3.64371449e-01 -1.76265061e-01
8.33788753e-01 -1.36835194e+00 -5.57638228e-01 1.02161252e+00
-1.09854031e+00 -5.63877106e-01 -2.92918868e-02 2.74874777e-01
2.39983678e-01 6.01530783e-02 -7.26366937e-01 -7.17482507e-01
-9.03720140e-01 -8.68777573e-01 6.43991768e-01 9.08573806e-01
5.66236258e-01 -6.42200828e-01 3.46688293e-02 -1.24128796e-01
8.99938166e-01 7.55142421e-02 4.31014597e-01 -1.86582357e-01
-6.56077206e-01 1.59155682e-01 -7.02216685e-01 4.38930273e-01
1.80115670e-01 -1.40863303e-02 -1.06336427e+00 -4.16238666e-01
1.04523711e-01 -2.39415631e-01 1.65472865e+00 4.79204983e-01
1.08590567e+00 -2.51485616e-01 -9.88898352e-02 1.16275859e+00
1.67686844e+00 2.93414630e-02 1.01939464e+00 3.63313943e-01
1.11086857e+00 3.86972427e-01 4.97654319e-01 1.35239676e-01
5.28242648e-01 1.09642714e-01 9.70914245e-01 -6.58395112e-01
-2.92606711e-01 2.51295030e-01 4.14584398e-01 7.03442335e-01
-3.55516583e-01 -5.86259604e-01 -5.02830207e-01 8.43917310e-01
-1.91710210e+00 -9.81742799e-01 -3.26042235e-01 1.72174227e+00
5.89932382e-01 6.03069589e-02 -2.52523720e-01 -3.19096327e-01
4.82304484e-01 4.06737059e-01 -6.03662610e-01 -1.43691644e-01
-5.98862231e-01 2.98199803e-01 5.90486646e-01 5.41828752e-01
-1.01950729e+00 1.07359779e+00 5.77263832e+00 5.81754863e-01
-1.27814555e+00 8.59948322e-02 5.14474511e-01 -2.57529408e-01
-3.00399572e-01 -2.77622193e-01 -1.00940359e+00 5.87822795e-01
4.60758299e-01 3.65173757e-01 2.71995753e-01 5.83638489e-01
4.78350937e-01 5.33819534e-02 -3.71913135e-01 8.79801512e-01
2.20029101e-01 -1.14878273e+00 2.09220782e-01 -2.72706479e-01
7.30003655e-01 6.49423897e-01 -5.59084155e-02 4.92071658e-01
4.13517803e-01 -1.29659927e+00 4.38306361e-01 7.66718745e-01
4.94141877e-01 -7.29389191e-01 1.24439859e+00 1.23288557e-01
-1.34386218e+00 -3.90576646e-02 -7.24929035e-01 -1.40602678e-01
-9.22484417e-03 7.81144738e-01 -8.75380337e-02 4.43328053e-01
1.22855759e+00 8.92120361e-01 -5.38885891e-01 1.38026369e+00
-4.61011261e-01 8.43954921e-01 -4.38719004e-01 8.80264565e-02
4.82174695e-01 -1.87934846e-01 5.22843540e-01 1.55223584e+00
3.11224073e-01 5.90211332e-01 3.70226242e-02 6.73109949e-01
-1.14098206e-01 -3.10091466e-01 -3.91662300e-01 5.91596246e-01
3.54842901e-01 1.52153862e+00 -6.72531545e-01 -2.82106698e-01
-3.17143768e-01 1.11708009e+00 3.17900151e-01 6.21291578e-01
-9.45004165e-01 -7.22396970e-01 9.48278129e-01 -1.65993229e-01
8.02291870e-01 -4.01972458e-02 -3.18517506e-01 -1.19744778e+00
1.14893816e-01 -8.86256933e-01 6.63451105e-02 -1.06212974e+00
-1.35843551e+00 1.16008174e+00 -2.78622568e-01 -1.07171810e+00
6.10770345e-01 -4.13090557e-01 -1.25622499e+00 1.08388698e+00
-2.13669515e+00 -1.20487010e+00 -1.22221696e+00 7.64315605e-01
7.64233172e-01 -3.70117947e-02 5.10587692e-01 4.01074886e-01
-8.95385146e-01 6.05365753e-01 1.09294794e-01 1.13817155e-01
6.54218256e-01 -1.16032374e+00 4.79247600e-01 1.17173731e+00
-6.77282438e-02 3.44664931e-01 8.26968670e-01 -4.62349832e-01
-1.01293802e+00 -1.53771162e+00 7.59089708e-01 -3.51683907e-02
2.60388434e-01 -2.48410881e-01 -1.44110966e+00 5.56025088e-01
2.64623463e-01 5.55131793e-01 6.32247776e-02 1.85829382e-02
-4.35230166e-01 -5.61011493e-01 -1.00402236e+00 5.40974736e-01
1.07423866e+00 1.00935833e-03 -6.43585026e-01 2.51670897e-01
9.16416883e-01 -4.41169560e-01 -2.69763649e-01 5.66736042e-01
3.43257040e-01 -1.07834840e+00 7.14042842e-01 -4.28628325e-01
5.00244856e-01 -7.93199539e-01 -3.35334361e-01 -1.50292206e+00
-8.21461439e-01 -2.60693610e-01 -1.06729180e-01 1.21052337e+00
3.53587419e-01 -6.72118127e-01 5.01994610e-01 -9.96352509e-02
-5.24137974e-01 -1.05426061e+00 -5.96275747e-01 -5.36658525e-01
-5.67981862e-02 -2.95035215e-03 6.28479660e-01 8.58793676e-01
-7.28887439e-01 3.72783124e-01 -6.56624079e-01 6.66614890e-01
5.43339908e-01 5.11602223e-01 5.20140350e-01 -1.03491843e+00
-2.46619582e-01 -3.69430482e-01 -1.78735614e-01 -1.23031831e+00
-2.66682357e-01 -5.08147061e-01 5.81665397e-01 -1.90985668e+00
4.51068759e-01 -3.67219865e-01 -7.97533154e-01 7.95394599e-01
-6.51750326e-01 5.40362775e-01 2.25836262e-01 5.76428175e-01
-6.53890193e-01 8.17214012e-01 1.33876956e+00 -7.80845731e-02
-2.55338937e-01 -1.08593650e-01 -8.53110790e-01 8.56210113e-01
9.09872890e-01 -5.77238500e-01 -2.13708535e-01 -1.01563239e+00
-6.41249493e-02 -3.77168655e-01 3.89960408e-01 -1.29333746e+00
7.78564066e-02 -2.46282056e-01 4.56648260e-01 -5.16322017e-01
2.27610052e-01 -5.97985625e-01 -1.20162368e-01 4.21402156e-01
4.61129509e-02 1.18439183e-01 2.94136047e-01 7.09670246e-01
-3.78565997e-01 1.52391300e-01 9.98077929e-01 -2.33444408e-01
-9.02036190e-01 7.04221010e-01 -4.09137934e-01 -1.59022480e-01
6.66650295e-01 -1.35971546e-01 -7.61246204e-01 -2.15560675e-01
-6.36935353e-01 7.62760341e-01 9.01909247e-02 3.83870304e-01
8.16220760e-01 -9.26615536e-01 -1.29596293e+00 1.97410047e-01
-3.06392945e-02 2.86887079e-01 5.41195869e-01 9.88776982e-01
-7.12196887e-01 -4.06021737e-02 -5.16061068e-01 -4.02040660e-01
-1.30579364e+00 1.29789501e-01 7.38470912e-01 -2.73506433e-01
-8.36846530e-01 1.33715594e+00 8.63351524e-01 -2.01605409e-01
1.68891177e-01 -5.07467210e-01 -6.43486202e-01 3.57172661e-03
7.83641815e-01 -1.10980943e-01 -2.17127595e-02 -3.83320838e-01
-3.73743236e-01 5.29468000e-01 -4.14811224e-01 3.62138957e-01
1.55677557e+00 -2.26550668e-01 -1.29154295e-01 4.66634125e-01
6.88098729e-01 -3.16023380e-01 -1.53291082e+00 -3.64942193e-01
-9.25186694e-01 -3.53074282e-01 2.06074700e-01 -9.78672028e-01
-1.89050221e+00 1.07658887e+00 9.61545825e-01 -1.86784968e-01
1.50463533e+00 -2.91471690e-01 9.90392864e-01 5.72794139e-01
-1.68236736e-02 -3.45284641e-01 -9.79178473e-02 8.67904603e-01
9.72792327e-01 -1.22999132e+00 -3.61147299e-02 -8.28629434e-02
-7.29717910e-01 7.78711379e-01 1.08832645e+00 -6.58630669e-01
8.62981975e-01 3.31923842e-01 4.69365567e-01 -3.86421233e-01
-7.97849953e-01 -6.73664451e-01 -1.09410331e-01 5.75477779e-01
1.42581537e-01 -3.58060822e-02 -2.15857983e-01 8.49394381e-01
2.20170572e-01 -6.40112609e-02 4.40495521e-01 8.40199947e-01
-1.08673227e+00 -4.82844025e-01 -3.45262706e-01 5.57603538e-01
-3.53275448e-01 -7.02579200e-01 -1.52046289e-02 6.91613674e-01
5.00912368e-01 8.07786405e-01 2.76505828e-01 -5.13447762e-01
5.03053248e-01 -5.34228921e-01 3.39976549e-01 -3.62835735e-01
-8.89028311e-01 -5.81698045e-02 -2.81614304e-01 -3.06618482e-01
-5.42822719e-01 -4.97866243e-01 -1.18175638e+00 -2.96607316e-01
-2.35370412e-01 4.12357040e-02 1.39574021e-01 1.01761520e+00
4.13007051e-01 7.98883140e-01 6.25512004e-01 -1.13611591e+00
-7.48458728e-02 -1.25842416e+00 -7.48039782e-01 6.09770156e-02
6.65297985e-01 -4.22305137e-01 -4.83896434e-01 -1.25340477e-01] | [10.92706298828125, -3.2225069999694824] |
2a5ca32b-4965-4cf0-a8d2-395e1ae0cce3 | ruccmu-system-report-for-dense-captioning | 1806.08854 | null | http://arxiv.org/abs/1806.08854v1 | http://arxiv.org/pdf/1806.08854v1.pdf | RUC+CMU: System Report for Dense Captioning Events in Videos | This notebook paper presents our system in the ActivityNet Dense Captioning
in Video task (task 3). Temporal proposal generation and caption generation are
both important to the dense captioning task. Therefore, we propose a proposal
ranking model to employ a set of effective feature representations for proposal
generation, and ensemble a series of caption models enhanced with context
information to generate captions robustly on predicted proposals. Our approach
achieves the state-of-the-art performance on the dense video captioning task
with 8.529 METEOR score on the challenge testing set. | ['Alexander Hauptmann', 'Shizhe Chen', 'Yuqing Song', 'Yida Zhao', 'Qin Jin', 'Jiarong Qiu'] | 2018-06-22 | null | null | null | null | ['dense-captioning', 'dense-video-captioning'] | ['computer-vision', 'computer-vision'] | [ 2.36311764e-01 2.55495548e-01 -2.71042079e-01 -3.62507343e-01
-1.26421320e+00 -2.96765029e-01 9.65753734e-01 -5.49113929e-01
-1.70082510e-01 1.05779183e+00 1.05993640e+00 2.29833573e-01
5.97651601e-01 -2.70924360e-01 -1.02592516e+00 -3.97068769e-01
-2.01992974e-01 6.45603895e-01 5.20123005e-01 -1.00283302e-01
1.06053762e-01 -1.86081842e-01 -1.38406968e+00 9.88669693e-01
3.71640891e-01 1.07340932e+00 3.47442269e-01 8.81915331e-01
1.22233808e-01 1.05701387e+00 -6.39145672e-01 -4.22387660e-01
1.65958002e-01 -6.01299167e-01 -6.92066908e-01 1.01904497e-01
9.34233069e-01 -6.87325180e-01 -7.55697072e-01 6.14661515e-01
6.21537566e-01 2.60911644e-01 7.92725384e-01 -1.34531105e+00
-7.93512404e-01 7.19664931e-01 -5.03869474e-01 8.04740250e-01
8.06319296e-01 3.97978067e-01 1.13820660e+00 -1.53289998e+00
8.83375764e-01 1.16098261e+00 2.91385412e-01 1.06284082e+00
-6.35976076e-01 -7.11644948e-01 3.82638633e-01 6.18112206e-01
-1.37426639e+00 -7.06965148e-01 4.68266189e-01 -5.05399585e-01
1.05194449e+00 3.26830335e-02 7.93772042e-01 1.88594055e+00
-8.55795816e-02 1.20050561e+00 1.55171469e-01 1.59447491e-01
1.81393519e-01 -2.76098907e-01 -5.16022965e-02 5.32686889e-01
2.98106849e-01 5.84881343e-02 -7.88197100e-01 -1.58180565e-01
8.92891347e-01 -2.60245621e-01 -4.18661922e-01 -3.82316634e-02
-1.83143914e+00 9.66130912e-01 3.47470552e-01 -1.64231267e-02
-8.53422761e-01 6.94875419e-01 4.87226248e-01 -3.01327139e-01
4.02857453e-01 5.32930434e-01 -2.15061486e-01 -2.60578632e-01
-1.20668852e+00 6.38014495e-01 4.00405288e-01 1.18669367e+00
2.07074001e-01 1.62748396e-01 -1.47325802e+00 6.40532196e-01
6.79440558e-01 5.58901012e-01 5.01775622e-01 -8.57474208e-01
9.98550177e-01 -6.04830123e-02 3.88777703e-01 -6.40976369e-01
1.46628439e-01 -3.48056495e-01 -5.23263335e-01 -6.99207962e-01
-1.04412556e-01 -2.66035944e-01 -1.22063208e+00 1.78745174e+00
-9.64943692e-02 1.08162093e+00 2.08259448e-01 1.27132213e+00
1.47242033e+00 1.46862757e+00 6.42799616e-01 -2.46390998e-01
1.24075377e+00 -1.67824066e+00 -8.16635668e-01 -2.98256785e-01
1.86002314e-01 -7.05985129e-01 4.92853284e-01 -2.83095352e-02
-1.32857406e+00 -7.47913539e-01 -7.99843848e-01 4.66931462e-02
2.21508741e-01 1.84402436e-01 5.45949101e-01 -6.76360130e-02
-1.41756618e+00 7.04255626e-02 -4.19534236e-01 -4.30034310e-01
6.19682431e-01 1.54479846e-01 -3.59511673e-01 -3.41765493e-01
-1.37709951e+00 8.19810688e-01 7.66425192e-01 1.71464935e-01
-1.55652523e+00 -5.81159890e-01 -9.94415939e-01 1.85548544e-01
6.71824813e-02 -1.15172517e+00 1.61966467e+00 -9.89373684e-01
-1.21643817e+00 5.70283473e-01 -1.73896983e-01 -8.40915918e-01
4.54955995e-01 -2.17174888e-01 -4.01080459e-01 5.90298116e-01
3.40328008e-01 1.90534925e+00 9.75403428e-01 -9.76411164e-01
-6.77872598e-01 5.27978241e-01 -1.39555812e-01 4.77716744e-01
1.57241318e-02 2.09190264e-01 -8.11690688e-01 -7.64331639e-01
-3.18176538e-01 -8.13494623e-01 -2.33978480e-01 -1.66094139e-01
-1.08997568e-01 -5.99600434e-01 8.74541223e-01 -8.52330744e-01
1.05937791e+00 -1.72888839e+00 1.63150981e-01 -2.80762017e-01
1.03149869e-01 4.12888288e-01 -7.08939791e-01 7.56459534e-02
2.09103730e-02 1.13219827e-01 1.97868589e-02 -3.39243203e-01
-1.14298433e-01 -8.07992667e-02 -5.84304094e-01 1.13919325e-01
6.91999555e-01 1.35217333e+00 -1.28350425e+00 -8.78793061e-01
1.55821279e-01 4.58692640e-01 -6.00269914e-01 8.58897567e-01
-5.45271635e-01 2.12859198e-01 -6.22702003e-01 6.33211434e-01
3.04061681e-01 -6.38986766e-01 -5.24747968e-02 -3.67169231e-01
2.96154886e-01 9.31789353e-02 -5.19828200e-01 2.06116366e+00
1.35626227e-01 8.29054236e-01 -5.49302161e-01 -5.37905931e-01
7.72390783e-01 9.54577625e-01 5.42045474e-01 -2.99162954e-01
-1.30390927e-01 -2.49537509e-02 -2.25929841e-01 -5.69234371e-01
7.57615805e-01 2.77135879e-01 -2.68388212e-01 2.57593781e-01
5.24591565e-01 2.02956855e-01 4.02501255e-01 6.43265486e-01
1.30884337e+00 5.97303629e-01 3.07705998e-01 5.67709543e-02
4.00799721e-01 -9.22551826e-02 4.46931601e-01 8.24791551e-01
-5.15754402e-01 1.40918493e+00 2.92603225e-01 -7.30303228e-01
-1.28395700e+00 -1.20371854e+00 4.44032639e-01 1.00719702e+00
-3.77079443e-04 -5.27072370e-01 -4.95192111e-01 -9.66120839e-01
-5.46447515e-01 6.07133627e-01 -7.82436132e-01 -8.69156122e-02
-7.31224179e-01 -5.41264415e-01 3.89915496e-01 7.31704533e-01
5.36135197e-01 -1.55329275e+00 -1.71069458e-01 4.30349559e-01
-9.09883916e-01 -1.38587987e+00 -9.00224209e-01 -6.28849447e-01
-7.01206386e-01 -8.39399159e-01 -1.35574663e+00 -9.44250584e-01
5.75159609e-01 2.73599714e-01 1.49709904e+00 -5.85845038e-02
1.61652938e-01 5.29593050e-01 -5.46890199e-01 -6.43290132e-02
-2.57282555e-01 8.00293833e-02 -3.63655202e-03 -2.92423721e-02
2.14239374e-01 -9.00071114e-02 -6.71320379e-01 3.18224847e-01
-7.05800056e-01 6.03796721e-01 7.59745061e-01 7.37851202e-01
4.26777333e-01 -1.30403435e+00 9.47278798e-01 -1.16902022e-02
4.79822874e-01 -9.66516614e-01 -3.21990609e-01 3.25821042e-01
-5.74715156e-03 5.38085289e-02 6.05193563e-02 -4.29865927e-01
-1.11388075e+00 3.80363077e-01 -2.20182478e-01 -8.19734633e-01
1.90041307e-02 9.02682990e-02 3.17440964e-02 2.84346640e-01
5.41844904e-01 4.53949660e-01 -3.11581641e-01 -9.04703960e-02
5.56939125e-01 3.09450567e-01 7.19424486e-01 -5.30013144e-01
6.91867709e-01 1.78860798e-02 -4.27065969e-01 -4.20321167e-01
-1.16865599e+00 -5.66064358e-01 -2.56075293e-01 -6.39887929e-01
1.16294241e+00 -1.60008824e+00 -1.17281124e-01 -1.67076305e-01
-1.96117747e+00 1.20259458e-02 -1.64538309e-01 5.22190869e-01
-8.28438580e-01 2.51088083e-01 -4.61398154e-01 -3.06341320e-01
-8.65609586e-01 -1.10438073e+00 1.35248852e+00 3.42356533e-01
-1.80902869e-01 -4.66137171e-01 4.51713622e-01 4.40752387e-01
3.99490058e-01 1.61301091e-01 -8.93784240e-02 -7.31931627e-01
-1.05402744e+00 -2.27984730e-02 -5.22246599e-01 7.91938752e-02
-4.48666543e-01 -1.38679385e-01 -9.76403117e-01 -8.36057365e-02
-6.73389494e-01 -5.43311417e-01 1.11435962e+00 5.84586561e-01
1.16822231e+00 -5.30028582e-01 -3.51447791e-01 3.44415307e-01
9.16125834e-01 -1.62627876e-01 1.10968506e+00 4.24500741e-02
6.50832593e-01 2.47293547e-01 8.81181061e-01 5.69980860e-01
3.58008504e-01 9.12020802e-01 5.38618147e-01 1.52855337e-01
-4.13489610e-01 -5.42000890e-01 6.23189807e-01 5.23003161e-01
-1.71657354e-01 -7.44176567e-01 -6.03387356e-01 8.18121433e-01
-2.32501030e+00 -1.49303138e+00 -4.03034315e-02 1.52951229e+00
5.37933826e-01 -2.01606192e-02 1.58714309e-01 -7.41242766e-01
1.28405094e+00 5.19659638e-01 -1.66565046e-01 -2.27801520e-02
7.00245723e-02 -1.70460775e-01 1.10499784e-01 7.11260587e-02
-1.22489321e+00 1.18378568e+00 6.98871994e+00 6.11768305e-01
-6.89033389e-01 3.42980713e-01 7.07902014e-01 -1.03446424e-01
-2.10263282e-01 -3.18045765e-01 -1.01051247e+00 6.94777787e-01
1.40248656e+00 -3.50399077e-01 -2.47596636e-01 9.73797441e-01
4.52039450e-01 2.35183820e-01 -1.37609649e+00 1.38989472e+00
6.92364573e-01 -1.91169131e+00 7.51970351e-01 -1.35054618e-01
1.08175063e+00 3.10830355e-01 1.40626505e-02 8.34158778e-01
-2.54873652e-02 -9.04604673e-01 8.20482194e-01 5.92156231e-01
5.97771168e-01 -1.48358002e-01 9.96253371e-01 -2.86170512e-01
-1.24700248e+00 1.21178947e-01 -3.70403796e-01 1.01930566e-01
9.88291502e-01 2.92786092e-01 -1.47395873e+00 1.98151380e-01
3.51514369e-01 9.65210855e-01 -5.52672029e-01 1.85016978e+00
-4.57163215e-01 8.18433464e-01 2.22948510e-02 -5.47474734e-02
6.01663888e-01 5.24018109e-01 5.59269726e-01 1.35865450e+00
6.57512724e-01 2.23883748e-01 2.02940509e-01 7.04798698e-01
-6.86719835e-01 -6.87225536e-02 -5.47629893e-01 -1.17814377e-01
3.01852554e-01 1.28394198e+00 -4.28558230e-01 -8.64205599e-01
-2.72374123e-01 9.20513988e-01 9.85172540e-02 3.77366483e-01
-1.47634423e+00 2.81694502e-01 5.55534601e-01 5.94394580e-02
5.27402163e-01 4.19137860e-03 5.36427140e-01 -1.41093862e+00
-1.60640419e-01 -5.64523578e-01 4.25064802e-01 -1.33779573e+00
-1.34676814e+00 7.97599494e-01 3.87089550e-01 -1.47555363e+00
-6.40675068e-01 -1.93860427e-01 -9.56543982e-01 3.68804336e-01
-1.39320123e+00 -1.22158062e+00 -5.35638928e-01 4.44894105e-01
1.26809645e+00 -2.07346320e-01 5.01172602e-01 3.84452581e-01
-3.81901562e-01 3.57840568e-01 -5.96122622e-01 7.21993148e-02
8.31828535e-01 -6.72693670e-01 8.95587504e-01 9.99009371e-01
3.03519398e-01 3.00852321e-02 7.56611645e-01 -7.87871897e-01
-8.32335591e-01 -1.67458546e+00 1.03340614e+00 -6.12762690e-01
3.75346690e-01 -2.66737491e-01 -6.28311396e-01 7.73772717e-01
5.51782370e-01 4.41541433e-01 6.13805652e-01 -4.37335581e-01
-2.62504727e-01 2.28711441e-01 -9.25866067e-01 5.15575469e-01
1.24371254e+00 -3.60939391e-02 -1.17290604e+00 7.07898438e-01
1.20945334e+00 -3.19594502e-01 -5.32295287e-01 2.95022815e-01
5.32929480e-01 -3.91712606e-01 1.03109324e+00 -7.86044002e-01
7.95052469e-01 -4.56360906e-01 -6.83944821e-02 -7.99160719e-01
-6.32505357e-01 -7.43643403e-01 -7.34971642e-01 1.02408302e+00
7.02407300e-01 3.12302411e-01 9.09455121e-01 2.81754375e-01
-5.10608256e-01 -3.93229991e-01 -9.57805812e-01 -5.00662982e-01
-5.80548167e-01 -2.64679223e-01 4.18849975e-01 4.74145532e-01
-2.09042922e-01 6.49329901e-01 -9.53176856e-01 -4.56934869e-02
5.33324122e-01 -3.20188820e-01 5.60561121e-01 -8.76866579e-01
-1.08527683e-01 -1.55255068e-02 -5.32858372e-01 -1.36409855e+00
4.21472460e-01 -6.45770967e-01 5.34726799e-01 -2.00319624e+00
5.45720160e-01 2.01371312e-01 -4.45263803e-01 4.51336384e-01
-2.33136490e-01 7.04786181e-01 3.56459588e-01 3.61764550e-01
-1.76370597e+00 1.04601777e+00 1.18365002e+00 -3.45627755e-01
9.32493582e-02 -1.37357414e-01 -5.59500217e-01 2.78526425e-01
4.48860526e-01 -5.68055391e-01 -4.68964130e-01 -5.87238848e-01
-8.65715817e-02 2.19774470e-01 2.42910430e-01 -1.30564857e+00
-8.23447630e-02 -2.07438737e-01 5.42260528e-01 -9.13529754e-01
7.29579747e-01 -2.73266226e-01 9.38276649e-02 2.83297002e-01
-4.88338321e-01 2.13970914e-01 1.12532243e-01 8.65031123e-01
-2.40291670e-01 -3.40926647e-02 3.62971395e-01 -3.02125633e-01
-1.23893738e+00 8.11745346e-01 -4.42151725e-01 1.23843543e-01
1.28772283e+00 -4.52373624e-02 -5.65798461e-01 -6.47436380e-01
-6.27363563e-01 4.23729002e-01 6.68263510e-02 1.06441724e+00
1.12702525e+00 -1.82902396e+00 -1.34294331e+00 -3.50215495e-01
4.52697277e-01 -1.85038328e-01 2.06713244e-01 4.89142388e-01
-4.18920368e-01 7.93710411e-01 -3.99050653e-01 -7.34137714e-01
-9.32222009e-01 3.77548873e-01 -3.62756215e-02 -3.36347669e-01
-4.34462786e-01 1.06159329e+00 3.21803689e-01 4.73025948e-01
1.90247566e-01 -2.08439633e-01 -5.64463317e-01 -8.01996589e-02
1.04287696e+00 -4.60838974e-02 -4.72930074e-01 -8.42068374e-01
-3.96283478e-01 2.33396232e-01 -4.26958710e-01 -2.04954281e-01
1.13215613e+00 9.37277824e-02 3.75435203e-01 -3.37197393e-01
1.14210463e+00 -7.86722660e-01 -1.49402499e+00 2.12281331e-01
-2.06574097e-01 -3.91912818e-01 -1.17735103e-01 -7.20640481e-01
-8.09880793e-01 5.59830189e-01 5.17770886e-01 -4.56623286e-01
6.97715342e-01 3.46022189e-01 8.90424550e-01 4.86359298e-01
1.93987265e-01 -7.80962050e-01 8.09822083e-01 6.33409023e-01
1.32596684e+00 -1.43785393e+00 -1.28828093e-01 -1.35997936e-01
-9.76607382e-01 9.36426282e-01 1.24365127e+00 -2.16482580e-01
2.00421363e-02 -4.57313061e-01 -3.88625592e-01 -5.71381003e-02
-1.36280274e+00 -8.46728310e-02 5.28591216e-01 6.84704900e-01
4.60481405e-01 -2.74352819e-01 -3.99179578e-01 5.18570960e-01
4.00785923e-01 2.55246133e-01 8.07091177e-01 5.71555197e-01
-7.87530243e-01 -6.16158545e-01 -3.79968137e-01 4.49508041e-01
-3.35059911e-01 -4.15226310e-01 -2.14217246e-01 2.25131989e-01
-7.00482279e-02 7.67654836e-01 2.16263890e-01 -3.76174212e-01
-3.41106579e-02 9.43540186e-02 3.61477673e-01 -9.44976091e-01
-3.35819662e-01 -1.08565293e-01 3.56501430e-01 -7.11362123e-01
-7.15251744e-01 -5.70449173e-01 -9.33847070e-01 2.54028171e-01
-9.52636823e-02 5.40401936e-01 5.22546291e-01 8.96866679e-01
5.34119487e-01 4.88822818e-01 2.73581624e-01 -1.37876999e+00
-2.72866070e-01 -1.37310493e+00 1.28691062e-01 3.07015747e-01
9.05974954e-02 -6.81384444e-01 -1.98216587e-01 3.72631103e-01] | [10.458306312561035, 0.6539652943611145] |
5b5ad782-45c7-4a91-b1f7-1cb4e9d15c4c | wav2code-restore-clean-speech-representations | 2304.04974 | null | https://arxiv.org/abs/2304.04974v2 | https://arxiv.org/pdf/2304.04974v2.pdf | Wav2code: Restore Clean Speech Representations via Codebook Lookup for Noise-Robust ASR | Automatic speech recognition (ASR) has gained a remarkable success thanks to recent advances of deep learning, but it usually degrades significantly under real-world noisy conditions. Recent works introduce speech enhancement (SE) as front-end to improve speech quality, which is proved effective but may not be optimal for downstream ASR due to speech distortion problem. Based on that, latest works combine SE and currently popular self-supervised learning (SSL) to alleviate distortion and improve noise robustness. Despite the effectiveness, the speech distortion caused by conventional SE still cannot be completely eliminated. In this paper, we propose a self-supervised framework named Wav2code to implement a generalized SE without distortions for noise-robust ASR. First, in pre-training stage the clean speech representations from SSL model are sent to lookup a discrete codebook via nearest-neighbor feature matching, the resulted code sequence are then exploited to reconstruct the original clean representations, in order to store them in codebook as prior. Second, during finetuning we propose a Transformer-based code predictor to accurately predict clean codes by modeling the global dependency of input noisy representations, which enables discovery and restoration of high-quality clean representations without distortions. Furthermore, we propose an interactive feature fusion network to combine original noisy and the restored clean representations to consider both fidelity and quality, resulting in even more informative features for downstream ASR. Finally, experiments on both synthetic and real noisy datasets demonstrate that Wav2code can solve the speech distortion and improve ASR performance under various noisy conditions, resulting in stronger robustness. | ['Eng Siong Chng', 'Qiushi Zhu', 'Chen Chen', 'Yuchen Hu'] | 2023-04-11 | null | null | null | null | ['speech-enhancement'] | ['speech'] | [ 1.89045608e-01 -5.48853338e-01 1.30118176e-01 -2.89385885e-01
-9.28975821e-01 -2.26182014e-01 2.94846773e-01 -7.52119496e-02
-3.97640131e-02 3.71998906e-01 6.62348807e-01 -2.23328426e-01
-1.45624459e-01 -5.66156507e-01 -4.52543199e-01 -8.95067871e-01
1.95606664e-01 -3.58783424e-01 4.01086174e-02 -5.12310147e-01
-1.68972518e-02 3.34454536e-01 -1.72259617e+00 3.19495946e-01
1.15127051e+00 9.81017649e-01 6.90966785e-01 5.63705027e-01
-1.77897528e-01 6.80087090e-01 -7.65168309e-01 -2.74123368e-03
1.27080157e-01 -5.17283201e-01 -1.58578753e-01 1.60739377e-01
-2.24696204e-01 -1.96635634e-01 -8.67725253e-01 1.36114454e+00
8.80213797e-01 3.41161102e-01 3.60781521e-01 -6.75157487e-01
-7.56697237e-01 7.62803614e-01 -2.77084291e-01 2.77126551e-01
3.87209624e-01 2.07169577e-01 6.93038285e-01 -1.10479331e+00
8.02509487e-02 1.38522851e+00 6.03659928e-01 5.89372158e-01
-9.95765865e-01 -8.66294742e-01 5.01818880e-02 5.33876061e-01
-1.52401698e+00 -1.15789497e+00 9.60335970e-01 1.63768619e-01
9.99422312e-01 5.02079129e-01 4.70169127e-01 1.09762621e+00
-1.78741619e-01 7.24244118e-01 1.08708775e+00 -3.46720368e-01
1.65950745e-01 -1.12716429e-01 -4.99258339e-02 3.83545160e-01
-1.87060326e-01 4.77709651e-01 -4.93147224e-01 2.42267102e-01
4.64588702e-01 4.73155677e-02 -7.72171140e-01 2.37884685e-01
-1.06873727e+00 4.61501390e-01 6.34584546e-01 6.49743378e-01
-2.84246683e-01 -2.66127110e-01 4.01957512e-01 5.19309938e-01
2.73044735e-01 4.01533693e-02 -2.76999354e-01 -1.88524812e-01
-1.05444741e+00 -1.45974264e-01 3.60626131e-01 7.53633380e-01
6.63548827e-01 6.50193751e-01 -2.67012805e-01 1.43361473e+00
4.97318476e-01 7.45801985e-01 1.04474139e+00 -5.07966459e-01
6.83847308e-01 1.40462905e-01 -2.09227681e-01 -1.11862493e+00
-1.14116721e-01 -9.21898305e-01 -1.52652347e+00 1.56396478e-01
-3.35065097e-01 1.22250289e-01 -1.12356699e+00 1.49330056e+00
1.20121089e-03 4.34793025e-01 4.70979005e-01 1.05217946e+00
8.92960548e-01 1.19419336e+00 -2.81628251e-01 -5.78293741e-01
8.91435683e-01 -9.73308682e-01 -1.07427537e+00 -2.13645324e-01
3.59633237e-01 -1.01082945e+00 8.46690178e-01 5.93286753e-01
-9.32857931e-01 -1.01106715e+00 -1.22976410e+00 6.38327748e-02
-1.32767677e-01 2.33939782e-01 -6.35557771e-02 7.57394612e-01
-1.07563615e+00 6.42061532e-01 -6.41613126e-01 1.51542230e-02
3.27854306e-01 2.33945012e-01 -3.47903103e-01 -2.85981387e-01
-1.36038017e+00 7.42804408e-01 3.34170848e-01 5.01654208e-01
-1.02633727e+00 -4.74403858e-01 -9.36012268e-01 2.86716312e-01
2.59589106e-01 -1.85546309e-01 1.03281379e+00 -8.32668602e-01
-1.83043146e+00 1.69169933e-01 -3.75951856e-01 -3.72096241e-01
2.50120372e-01 -8.54289010e-02 -1.12544560e+00 -1.68469742e-01
-2.33656734e-01 4.42819521e-02 1.09241080e+00 -1.30292976e+00
-4.31686461e-01 -2.11811364e-01 -3.89906168e-01 3.33809853e-01
-4.81553346e-01 -6.62714383e-03 -3.29005390e-01 -1.13942838e+00
5.97101152e-01 -4.04063880e-01 -1.81849837e-01 -4.91858065e-01
-2.33998135e-01 -7.59764910e-02 1.00985694e+00 -1.06851923e+00
1.57601619e+00 -2.51628208e+00 1.71616003e-01 2.56316811e-01
-1.08339161e-01 9.50928032e-01 -3.75829250e-01 3.22207242e-01
-3.92914295e-01 7.02203833e-05 -3.91209334e-01 -4.47078258e-01
-2.55553812e-01 2.37321198e-01 -5.25042534e-01 3.40898633e-01
2.77251303e-01 5.03818631e-01 -7.97742009e-01 -2.91802078e-01
4.87543195e-01 7.80874491e-01 -3.24174643e-01 4.39760447e-01
2.52120703e-01 4.64564115e-01 -2.75645614e-01 4.52192366e-01
9.32268023e-01 4.27340299e-01 -1.87450647e-01 -3.48011583e-01
-2.44712979e-01 5.10451972e-01 -1.50244057e+00 1.68891656e+00
-8.11710238e-01 3.40501875e-01 4.80443060e-01 -1.09561872e+00
1.40776157e+00 4.54219341e-01 -2.51091681e-02 -1.00125611e+00
3.86578850e-02 4.56224293e-01 -1.02100238e-01 -5.40622056e-01
1.45326540e-01 -4.70641889e-02 3.62986445e-01 -1.51534125e-01
3.67018953e-02 -2.19075039e-01 -3.41275245e-01 -2.55585425e-02
1.02289391e+00 -1.18369423e-01 7.91330934e-02 5.49752042e-02
1.13162875e+00 -6.16885304e-01 7.08098233e-01 4.92519855e-01
-5.28703108e-02 9.55949664e-01 -3.17670912e-01 5.09882905e-02
-9.78660703e-01 -9.64487135e-01 -1.31228834e-01 6.85066760e-01
2.13094473e-01 -3.76688421e-01 -6.89976692e-01 -3.10969263e-01
-3.76076043e-01 6.30301535e-01 8.56849626e-02 -5.46304166e-01
-7.71364510e-01 -5.00702858e-01 5.23009300e-01 2.70798802e-01
6.84190452e-01 -1.00004196e+00 3.72421920e-01 4.63855445e-01
-3.26375425e-01 -8.08288574e-01 -5.52574456e-01 2.23112077e-01
-5.80642521e-01 -6.20472610e-01 -8.28446627e-01 -9.18297589e-01
3.56611878e-01 7.23591149e-01 3.71703357e-01 3.68396699e-01
1.07702548e-02 -1.20732173e-01 -8.36685359e-01 9.28712264e-02
-9.23238516e-01 -1.47124335e-01 3.47816616e-01 3.66486728e-01
-5.27470596e-02 -8.69561791e-01 -4.10444856e-01 3.36837560e-01
-1.01700068e+00 -1.05070956e-01 5.79426646e-01 1.04577696e+00
5.17864883e-01 4.25567031e-01 8.72970760e-01 -1.54972792e-01
6.63131416e-01 -4.53307718e-01 -3.90453964e-01 1.90342590e-01
-5.09280503e-01 -8.66870862e-03 1.26180148e+00 -2.02639669e-01
-1.31734598e+00 -2.28427097e-01 -7.71084726e-01 -6.73195958e-01
-2.57460773e-01 5.20315289e-01 -6.21073484e-01 1.10647485e-01
5.31655371e-01 8.86201739e-01 -6.82040080e-02 -9.43603694e-01
2.06754223e-01 1.44958746e+00 6.15419447e-01 -2.13337094e-01
1.15976477e+00 -6.26751333e-02 -3.54557931e-01 -9.80126381e-01
-5.95782340e-01 -5.01239896e-01 -2.86786437e-01 7.72721097e-02
3.64966750e-01 -1.22012234e+00 -3.90128613e-01 7.54101336e-01
-1.10612619e+00 2.48825494e-02 -6.52225614e-02 6.14170671e-01
-3.43120277e-01 7.90695131e-01 -5.42348325e-01 -9.59679127e-01
-4.90252286e-01 -1.40066099e+00 7.84338236e-01 2.51533359e-01
4.32422251e-01 -4.74298030e-01 -1.68486819e-01 2.79402941e-01
7.53977418e-01 -4.88414764e-01 6.20061636e-01 -4.35879916e-01
-4.31798160e-01 1.11748450e-01 -1.03540227e-01 9.83789861e-01
5.08825064e-01 -2.98587233e-01 -1.13811588e+00 -5.04195392e-01
3.92058104e-01 -5.34430705e-02 1.01232636e+00 4.85144779e-02
1.27883554e+00 -3.79563302e-01 -1.55511051e-01 8.83497775e-01
1.23477471e+00 5.08278012e-01 9.10316050e-01 7.42846727e-02
6.20171666e-01 1.62607804e-01 5.45923591e-01 2.73106307e-01
-2.22602324e-03 7.81903982e-01 2.22832069e-01 2.79447045e-02
-8.11116755e-01 -4.29450363e-01 6.68805718e-01 1.75693214e+00
1.25057772e-01 -1.43817365e-01 -5.09338081e-01 4.99466538e-01
-1.55602574e+00 -1.03774941e+00 1.44405320e-01 2.18776512e+00
9.68824327e-01 3.77282314e-02 -3.30906808e-01 6.80087149e-01
9.98292804e-01 3.77131522e-01 -4.01453823e-01 -1.97836876e-01
-4.54902560e-01 4.61504042e-01 1.05780795e-01 4.94664907e-01
-9.37259972e-01 7.62357831e-01 5.15653181e+00 1.54513454e+00
-1.23492301e+00 1.85433701e-01 5.03797293e-01 1.79007545e-01
-3.37076396e-01 -1.04908824e-01 -5.76460004e-01 6.50783300e-01
9.89965022e-01 -1.56853259e-01 9.46297467e-01 7.17164874e-01
6.17073178e-01 4.77635801e-01 -5.61052740e-01 1.34141278e+00
9.08915550e-02 -1.11451268e+00 6.78320825e-02 -4.28401351e-01
6.52528822e-01 -2.68131196e-02 1.35070994e-01 3.72931093e-01
-1.66875087e-02 -1.01064515e+00 7.40138710e-01 4.78600293e-01
9.35766935e-01 -8.99111807e-01 7.36515820e-01 4.75689143e-01
-1.47148108e+00 -4.50645834e-01 -6.18932486e-01 1.44514203e-01
5.06401733e-02 7.71055877e-01 -5.77905595e-01 9.37419116e-01
6.75164461e-01 9.13600624e-01 -3.50022048e-01 1.19341755e+00
-3.43171328e-01 8.90451789e-01 -2.39683539e-01 2.12279081e-01
5.88327125e-02 -8.01107734e-02 8.05359900e-01 1.32534766e+00
6.49846137e-01 2.44376987e-01 2.71297060e-02 6.56144083e-01
-4.25669812e-02 1.37946516e-01 -2.70885557e-01 1.97149634e-01
7.01524615e-01 9.57603455e-01 -1.62812725e-01 -1.56345487e-01
-2.10556492e-01 1.04178333e+00 7.15434998e-02 5.29749393e-01
-6.43090189e-01 -7.54205048e-01 8.59038651e-01 -1.43446848e-01
4.60019916e-01 -2.31835440e-01 2.05280874e-02 -1.43909156e+00
1.81190208e-01 -1.27245307e+00 -1.28440395e-01 -8.40746582e-01
-1.22536564e+00 8.16488683e-01 -6.18045032e-01 -1.46040320e+00
8.54580253e-02 -2.46190727e-01 -5.82499206e-01 1.08301735e+00
-1.92891610e+00 -9.41536367e-01 -2.28886232e-01 7.69476831e-01
9.82432544e-01 -4.17774975e-01 8.03954899e-01 7.02314317e-01
-7.45295405e-01 8.14974070e-01 6.04207039e-01 7.55674765e-02
5.26627779e-01 -7.82250583e-01 4.92920816e-01 1.11237824e+00
2.45412454e-01 5.01542628e-01 4.91958231e-01 -5.83779633e-01
-1.44597995e+00 -1.33113873e+00 5.43481410e-01 3.77238512e-01
3.67060453e-01 -2.39355922e-01 -1.21222889e+00 1.41985655e-01
1.29285390e-02 1.50262326e-01 1.40264407e-01 -3.81654620e-01
-5.52040696e-01 -6.59552634e-01 -1.01570165e+00 3.59936059e-01
1.22456205e+00 -7.50198364e-01 -8.24371636e-01 6.64468035e-02
1.17349803e+00 -3.54213089e-01 -6.02982104e-01 3.76074851e-01
2.51633488e-02 -1.03154182e+00 9.53858793e-01 -1.14336342e-01
2.04988614e-01 -7.33144283e-01 -4.41189975e-01 -1.73602176e+00
-4.44009036e-01 -8.34226966e-01 -5.18644266e-02 1.70020974e+00
1.68014839e-01 -4.78058785e-01 2.55012453e-01 -2.86964595e-01
-7.28883564e-01 -4.28713173e-01 -1.16912103e+00 -8.69713306e-01
-2.18849838e-01 -6.33327901e-01 9.28769052e-01 8.16949844e-01
-5.97894192e-02 7.57554872e-03 -5.48510313e-01 4.72432554e-01
4.93248314e-01 -1.60196915e-01 4.25263226e-01 -7.28335619e-01
-3.06673378e-01 -2.45847210e-01 -4.86980051e-01 -1.32944810e+00
1.33424371e-01 -1.00638247e+00 3.63952845e-01 -1.48734200e+00
-2.32306451e-01 -6.46120608e-01 -5.45878708e-01 3.06128561e-01
-3.01276714e-01 1.87814084e-03 2.14661896e-01 2.18434602e-01
-2.85804987e-01 1.05057001e+00 1.12114573e+00 -3.24283749e-01
-2.69895583e-01 1.28185987e-01 -5.88234901e-01 4.23218548e-01
6.27513468e-01 -3.52012247e-01 -2.99698204e-01 -5.61112642e-01
-3.30035836e-01 3.05120468e-01 1.70145750e-01 -1.45437098e+00
4.09033835e-01 2.44085379e-02 3.21741700e-01 -4.38197643e-01
3.35968792e-01 -9.09605145e-01 1.03565216e-01 2.61194557e-01
-3.02046955e-01 -5.18184125e-01 1.14028074e-01 6.96404457e-01
-6.95340991e-01 -2.30237320e-01 9.79351521e-01 6.48427382e-02
-5.32564104e-01 1.47752926e-01 -2.94073761e-01 -2.59063721e-01
4.79550004e-01 -1.53378293e-01 -4.61535715e-02 -4.89988387e-01
-7.68888712e-01 -3.29151303e-02 6.62859306e-02 6.20836377e-01
1.00533259e+00 -1.41190314e+00 -8.49427342e-01 6.48345828e-01
-9.44500640e-02 -1.49424300e-01 7.06966758e-01 2.83596665e-01
-1.79375678e-01 1.58773988e-01 1.77899867e-01 -4.20410931e-01
-1.12322640e+00 6.61388099e-01 4.29858565e-01 9.04759616e-02
-6.54735863e-01 7.89941907e-01 -1.64424390e-01 -3.68737042e-01
3.88873160e-01 -1.94767579e-01 -3.41466278e-01 -4.02596921e-01
9.33965147e-01 3.75500441e-01 4.61755216e-01 -9.75433946e-01
-1.52383193e-01 6.13806725e-01 7.01257437e-02 1.23161241e-01
1.43448055e+00 -5.79943359e-01 -3.54285985e-02 -8.42500944e-04
1.50879097e+00 1.66404426e-01 -9.56846297e-01 -4.15059298e-01
-3.97248089e-01 -6.19212568e-01 3.84524345e-01 -7.20250070e-01
-1.30209756e+00 1.10461938e+00 9.62726712e-01 1.50072485e-01
1.50165534e+00 -3.77412230e-01 1.21140110e+00 3.29386324e-01
3.22111964e-01 -8.44895184e-01 1.87577531e-02 4.90470290e-01
1.02495325e+00 -1.03631628e+00 -3.86731178e-01 -2.98186570e-01
-3.61920029e-01 1.29004645e+00 3.42948258e-01 5.24511039e-02
7.11789787e-01 3.18938226e-01 1.73274621e-01 4.67398286e-01
-3.04822654e-01 -2.63420850e-01 3.24417092e-02 8.02485645e-01
3.71989459e-02 -1.01952739e-01 -1.07912011e-01 9.03096914e-01
-3.17536831e-01 -1.55490592e-01 3.54214042e-01 5.46648741e-01
-7.24555969e-01 -1.25823617e+00 -7.34375954e-01 2.60345638e-01
-2.04589203e-01 -2.87392288e-01 2.77402550e-01 -1.08664110e-01
2.92565674e-01 1.53539002e+00 -1.94819406e-01 -8.84994328e-01
6.20402038e-01 -1.46466032e-01 3.92952375e-02 -5.11471748e-01
-4.84854639e-01 3.39422464e-01 -2.35821947e-01 -3.64165962e-01
-2.35405758e-01 -3.49877834e-01 -1.24632633e+00 -1.68347321e-02
-6.50080264e-01 2.90723860e-01 6.90817714e-01 9.46042776e-01
4.45874780e-01 7.39289820e-01 1.34305501e+00 -7.22361028e-01
-8.96349370e-01 -1.03274930e+00 -5.98555267e-01 4.45170075e-01
7.90409565e-01 -4.45616364e-01 -7.33490586e-01 3.08848750e-02] | [14.905423164367676, 5.985017776489258] |
dd5e2e37-0600-411d-945f-9f745afdd789 | cyber-attack-detection-in-discrete-nonlinear | 2102.01166 | null | https://arxiv.org/abs/2102.01166v1 | https://arxiv.org/pdf/2102.01166v1.pdf | Cyber-Attack Detection in Discrete Nonlinear Multi-Agent Systems Using Neural Networks | This paper proposes a distributed cyber-attack detection method in communication channels for a class of discrete, nonlinear, heterogeneous, multi-agent systems that are controlled by our proposed formation-based controller. A residual-based detection system, exploiting a neural network (NN)-based observer, is developed to detect false data injection attacks on agents communication channels. A Lyapunov function is used to derive the NN weights tuning law and the attack detectability threshold. The uniform ultimate boundedness (UUB) of the detector residual and formation error is proven based on the Lyapunov stability theory. The proposed methods attack detectability properties are analyzed, and simulation results demonstrate the proposed detection methodologys performance. | ['Rastko R. Selmic', 'Kiarash Aryankia', 'Amirreza Mousavi'] | 2021-02-01 | null | null | null | null | ['cyber-attack-detection'] | ['miscellaneous'] | [-1.66942149e-01 2.40482464e-01 -2.06186138e-02 5.54855347e-01
-2.07993761e-01 -6.89166903e-01 5.49345791e-01 3.93902719e-01
-1.37877658e-01 7.97480524e-01 -4.87221986e-01 -3.02692860e-01
-3.64650220e-01 -6.92210257e-01 -3.01506162e-01 -9.72418487e-01
-6.02975368e-01 -1.98866889e-01 3.32014084e-01 -2.93483466e-01
3.23675014e-02 3.83080125e-01 -9.13208842e-01 -5.71435571e-01
5.87506533e-01 1.04549611e+00 -3.90169889e-01 8.41905653e-01
1.03594232e+00 6.95212603e-01 -1.08729577e+00 2.34379649e-01
5.22319019e-01 -3.65659714e-01 1.47355869e-01 2.25819230e-01
-4.90181297e-01 -6.58113003e-01 -5.06197751e-01 1.76567507e+00
3.23401779e-01 5.09130657e-02 6.85426235e-01 -2.01906943e+00
-2.87748158e-01 6.49435103e-01 -4.01090354e-01 1.75054863e-01
3.68588835e-01 1.69722721e-01 2.35338718e-01 -4.99535173e-01
5.10906637e-01 1.29442203e+00 4.47150469e-01 6.01803958e-01
-9.22748804e-01 -1.09596503e+00 7.63239712e-02 -1.83947816e-01
-1.55120564e+00 1.27561986e-01 4.40683931e-01 -3.33801240e-01
7.00835705e-01 3.00915927e-01 6.02368653e-01 7.75847912e-01
1.08505583e+00 1.39120176e-01 1.06346560e+00 -2.63263136e-01
5.55250049e-01 3.14093262e-01 1.06782213e-01 9.06283498e-01
1.09419405e+00 8.80372107e-01 2.19338402e-01 -6.00948453e-01
1.07031214e+00 -3.14542837e-03 -3.11334103e-01 9.93358530e-03
-7.57316589e-01 1.14224148e+00 1.16302766e-01 2.33421206e-01
-5.63839316e-01 -1.84197202e-02 6.49062932e-01 8.57728720e-01
1.56157628e-01 1.22470200e-01 -9.93864462e-02 2.89049715e-01
3.79409224e-01 1.32733986e-01 1.39291000e+00 1.21018851e+00
1.20295204e-01 8.08684051e-01 2.63409764e-01 -1.96675658e-01
7.79912412e-01 9.92853940e-01 2.55950302e-01 -5.26312351e-01
1.13683322e-03 5.64243078e-01 3.27434957e-01 -1.31438684e+00
-4.65450168e-01 -3.11383754e-01 -1.04822767e+00 6.97253704e-01
-1.68402363e-02 -1.01272357e+00 -3.81924868e-01 1.51937687e+00
6.75270319e-01 3.23262185e-01 6.87368393e-01 9.18761849e-01
-7.26743415e-02 7.10955858e-01 -1.64840162e-01 -8.58865857e-01
8.70162189e-01 -2.78056264e-01 -1.04490542e+00 4.24977541e-01
1.90516114e-01 -4.83921111e-01 -8.09088126e-02 5.02026737e-01
-7.28817523e-01 -5.00589097e-03 -1.59169698e+00 1.30896664e+00
-3.47015381e-01 -1.72731966e-01 1.60139978e-01 9.60689783e-01
-6.14026010e-01 -1.25612319e-01 -6.39262080e-01 -2.07173958e-01
-4.92779464e-01 5.57194054e-01 -8.25005546e-02 7.75033295e-01
-1.63765204e+00 9.94340539e-01 6.02436483e-01 4.57644224e-01
-1.41747439e+00 -1.83951631e-01 -7.49307275e-01 -3.78088206e-01
3.81984442e-01 -2.40196794e-01 9.34743822e-01 -8.58900309e-01
-1.88071156e+00 -1.54034272e-01 9.86403525e-01 -8.29499304e-01
4.89918977e-01 2.67093480e-01 -7.68532097e-01 4.00338173e-01
-3.59212041e-01 -6.31769657e-01 1.13441265e+00 -1.29174256e+00
-9.23549771e-01 -2.35142745e-02 1.72025368e-01 3.65646817e-02
-4.15300816e-01 2.35124677e-02 8.20338070e-01 -3.86517584e-01
-1.89236522e-01 -8.12910020e-01 -5.93195856e-01 -2.79755801e-01
-5.20042300e-01 1.73229128e-01 1.38421679e+00 -2.08294526e-01
1.13895774e+00 -2.07383275e+00 -4.37396429e-02 9.97187436e-01
2.22945079e-01 4.29642767e-01 3.91324656e-03 9.19825971e-01
3.87091964e-01 -2.91611284e-01 2.66740203e-01 3.85416120e-01
2.43510023e-01 -2.74011999e-01 -4.83254701e-01 1.18314540e+00
-3.91870318e-03 -2.13252604e-01 -6.76811576e-01 1.18131809e-01
3.24960142e-01 4.48609084e-01 -9.44347158e-02 2.32724339e-01
1.66563615e-01 1.23926386e-01 -8.26250553e-01 5.16371787e-01
6.87036157e-01 3.73503059e-01 3.29097122e-01 6.69552013e-03
-4.27022040e-01 -6.32200718e-01 -1.79596961e+00 3.82700384e-01
-3.30918372e-01 3.08510274e-01 1.14814174e+00 -8.09310257e-01
1.15750122e+00 7.60220289e-01 3.26325029e-01 -2.30833590e-01
1.03194451e+00 3.71220887e-01 3.51071596e-01 -1.73094302e-01
-1.14212833e-01 -9.61268172e-02 -3.21635008e-01 3.79440755e-01
-2.47878611e-01 -8.97166580e-02 -7.66877010e-02 2.75731623e-01
1.24600959e+00 -9.41153646e-01 6.13799036e-01 -4.68502790e-01
1.16956782e+00 3.53200845e-02 6.03217602e-01 6.89718246e-01
-5.89613080e-01 -7.78202116e-01 2.88332820e-01 -1.07409677e-03
-7.80964017e-01 -8.28974009e-01 2.25431975e-02 3.06528986e-01
6.85301065e-01 8.41623023e-02 -7.29510784e-01 -3.91445667e-01
4.40913320e-01 2.36252904e-01 -5.77744901e-01 -9.14777756e-01
-2.44198129e-01 -4.84525353e-01 9.86712635e-01 1.00981042e-01
4.93209124e-01 -4.32536751e-01 -7.03498483e-01 5.91763794e-01
8.16794395e-01 -8.92240286e-01 -2.03700066e-01 9.07977596e-02
-1.23842269e-01 -1.62400830e+00 -2.75213160e-02 -1.00683284e+00
7.85610735e-01 9.06081647e-02 -9.42124724e-02 2.66217351e-01
-2.75658637e-01 1.07190073e+00 -2.12916121e-01 -9.47936535e-01
-1.02369750e+00 -6.88001275e-01 7.67787218e-01 2.03727186e-01
7.89283141e-02 -6.33603558e-02 -2.68841416e-01 3.87180567e-01
-8.16643298e-01 -8.46649647e-01 5.81579208e-01 9.18495834e-01
-3.75630297e-02 5.53456903e-01 1.19822657e+00 -3.63962680e-01
1.38923061e+00 -4.35850590e-01 -1.69052172e+00 -9.12937000e-02
-7.34647393e-01 -4.67605829e-01 1.11011100e+00 -8.16370487e-01
-7.78487742e-01 -1.55161813e-01 5.39142668e-01 -4.25428152e-01
-4.54699472e-02 5.77038467e-01 -6.43351376e-02 -6.78610504e-01
6.76007569e-01 1.75654218e-01 8.18150580e-01 4.09208357e-01
8.18792731e-02 7.89535046e-01 2.80289233e-01 -1.20686591e-01
1.28489745e+00 3.69573593e-01 3.75173181e-01 -9.91334379e-01
3.19166571e-01 -1.68501347e-01 1.19183369e-01 -5.04517674e-01
2.81489313e-01 -9.16274071e-01 -1.67475212e+00 6.60051167e-01
-1.10654628e+00 5.97683340e-02 2.39770055e-01 8.75852048e-01
-4.58007425e-01 1.69342667e-01 -1.21243358e+00 -1.57976711e+00
-5.71870208e-01 -1.01185989e+00 2.60126740e-01 3.05136234e-01
3.37263525e-01 -1.28558552e+00 3.03230792e-01 -5.64560950e-01
5.73580384e-01 8.80071163e-01 3.35649490e-01 -9.51238096e-01
-5.70260406e-01 -8.70966911e-01 2.25197613e-01 3.35962683e-01
-6.18046820e-02 1.83421806e-01 -2.05173522e-01 -1.06470120e+00
5.54234624e-01 -1.47735253e-01 -6.60937428e-02 2.95288354e-01
-2.52779514e-01 -7.65190661e-01 -4.61052924e-01 2.44386364e-02
1.85001707e+00 9.03393328e-01 1.16033472e-01 4.86352682e-01
2.22204372e-01 2.35454440e-01 7.58340776e-01 9.31169629e-01
-5.33369593e-02 1.14358075e-01 9.11841512e-01 1.51581407e-01
7.77470291e-01 2.48722970e-01 9.16228592e-01 8.29373062e-01
3.39738309e-01 -2.60922164e-01 -3.55720371e-01 8.07152987e-02
-1.84233296e+00 -7.16225743e-01 -2.59570360e-01 2.18368721e+00
4.38474119e-01 1.96250886e-01 2.72271693e-01 4.22991812e-01
1.27420866e+00 -5.21535873e-01 -4.26014632e-01 -5.72356820e-01
-1.72792345e-01 -4.46683109e-01 9.97927248e-01 8.92800391e-01
-1.12929773e+00 2.86660254e-01 5.96280289e+00 4.83567953e-01
-1.12227488e+00 8.14272985e-02 -1.68001145e-01 6.27442539e-01
5.53440988e-01 -1.04953095e-01 -7.00271249e-01 2.07894847e-01
1.02430809e+00 -8.57512414e-01 4.10804719e-01 7.87790775e-01
5.19673526e-01 3.64557058e-02 -5.09683251e-01 6.63159132e-01
7.90864900e-02 -8.68459940e-01 -3.09744120e-01 1.15158536e-01
6.55485392e-01 -5.67608118e-01 2.81632572e-01 1.65748909e-01
5.76583087e-01 -3.33281755e-01 5.01046121e-01 1.87533990e-01
8.09572339e-02 -1.07552385e+00 1.02959764e+00 3.23480368e-01
-1.40310228e+00 -6.95934057e-01 -2.73510844e-01 -4.21381801e-01
7.62508214e-02 1.57015651e-01 -8.42120647e-01 6.86517000e-01
-3.00651371e-01 3.84971738e-01 -7.81741664e-02 1.07392502e+00
-1.70709103e-01 2.75181532e-01 -4.57734555e-01 -7.16464400e-01
3.29715103e-01 -2.79109061e-01 1.19115686e+00 8.40498090e-01
-4.81318729e-03 3.54634792e-01 7.85274804e-01 5.96955061e-01
4.79719847e-01 -1.64701685e-01 -7.93221474e-01 1.10637255e-01
9.04383421e-01 1.12240744e+00 -4.43374336e-01 -3.10714006e-01
-1.83855504e-01 2.42214993e-01 -4.05931413e-01 3.47295761e-01
-1.02566886e+00 -1.03532970e+00 5.54357767e-01 -4.96966243e-01
-9.98389572e-02 -2.97344774e-01 2.21745253e-01 -7.25085855e-01
-3.01913708e-01 -9.48660016e-01 4.06449556e-01 1.34092450e-01
-1.33869553e+00 5.47421157e-01 -5.20000756e-02 -2.00833130e+00
-4.76238430e-01 -4.64674592e-01 -6.44647956e-01 3.90338898e-01
-8.30978930e-01 -7.32786715e-01 8.89266878e-02 9.96920764e-01
-5.69521859e-02 -7.25971162e-01 7.62660980e-01 1.98831171e-01
-9.85073328e-01 4.47336376e-01 3.91944319e-01 1.89730838e-01
2.05883607e-01 -8.70293319e-01 -8.01974595e-01 1.21694958e+00
-9.47049558e-01 6.97877645e-01 1.18424022e+00 -8.06224644e-01
-2.05446720e+00 -1.20435929e+00 -2.19207942e-01 5.06855071e-01
1.32110775e+00 -1.60842553e-01 -3.34979117e-01 6.88880920e-01
5.49429297e-01 -2.16317959e-02 1.34713039e-01 -9.01446939e-01
-7.51695409e-02 -1.54101595e-01 -1.54030943e+00 5.29674888e-01
4.96346317e-02 -2.34256238e-01 -2.53712177e-01 2.03883693e-01
6.44878030e-01 -4.74679619e-01 -1.17984927e+00 9.15946322e-04
3.57280493e-01 -2.04805523e-01 5.60646713e-01 -3.45561087e-01
-7.90911555e-01 -6.41287804e-01 -7.03981370e-02 -1.49699199e+00
-2.06258819e-01 -1.21531725e+00 -2.11333126e-01 8.85904312e-01
3.33861798e-01 -1.47912955e+00 2.61993408e-01 1.47313073e-01
1.30972415e-01 -9.65606570e-02 -1.20390332e+00 -1.40696514e+00
-2.69903570e-01 3.65631491e-01 3.72151472e-02 9.03052807e-01
9.21473563e-01 2.97083676e-01 -2.16900826e-01 1.15121734e+00
1.09262812e+00 -7.06816435e-01 7.11976111e-01 -1.01695251e+00
-4.26470339e-02 -2.46872589e-01 -8.70983839e-01 -6.17940426e-02
1.67736709e-01 -1.82077467e-01 1.41308710e-01 -8.48116577e-01
-6.68007493e-01 5.23840189e-02 -4.03860331e-01 1.59876794e-02
3.11571866e-01 -1.88531220e-01 1.69535086e-01 1.46707892e-02
-4.50235665e-01 4.66907889e-01 6.47987604e-01 -3.20860237e-01
-4.36615318e-01 4.93398460e-04 2.51458883e-01 7.02811956e-01
9.49793875e-01 -3.46928298e-01 -7.28353024e-01 3.20495099e-01
-3.70526344e-01 7.48649716e-01 4.32114840e-01 -1.18028927e+00
5.73594928e-01 -3.94891292e-01 -2.25058198e-01 -2.54804641e-01
2.15012375e-02 -1.37258291e+00 2.46647999e-01 1.73023891e+00
-2.31309161e-01 1.56486630e-01 -4.48660459e-03 1.14748812e+00
-2.01615453e-01 9.12716910e-02 9.59311783e-01 5.11810839e-01
-3.00993234e-01 5.94712161e-02 -1.13554871e+00 -5.08749366e-01
1.74838769e+00 1.29452795e-01 -6.75644040e-01 -4.49388355e-01
-6.42148852e-01 2.65858889e-01 -8.99004117e-02 1.50982425e-01
9.62129235e-01 -1.14484227e+00 -5.52753985e-01 3.82485092e-01
-2.43264034e-01 -7.31931508e-01 -1.15715869e-01 1.00505972e+00
-2.87590176e-01 3.69798303e-01 -2.76207119e-01 -1.76457688e-01
-1.13738728e+00 7.63960421e-01 6.71908855e-01 1.58679366e-01
-2.73955554e-01 2.94776320e-01 -3.49077046e-01 1.34887442e-01
4.16718870e-01 -2.48506203e-01 -6.49038628e-02 -3.35711271e-01
7.72850692e-01 7.06171155e-01 -3.84681940e-01 -3.89780879e-01
-3.83786291e-01 5.71230464e-02 6.72707930e-02 -3.55706990e-01
7.31320381e-01 -2.87126005e-01 -2.33025789e-01 9.64917056e-03
9.06432569e-01 -2.41509557e-01 -1.19427335e+00 3.37719694e-02
-3.37158144e-01 -2.73958653e-01 -9.80926156e-02 -5.61094761e-01
-9.15727437e-01 -1.18188970e-01 6.75786257e-01 1.12312293e+00
8.48661900e-01 -7.28232801e-01 4.11748618e-01 3.62051070e-01
7.82505512e-01 -1.11157203e+00 4.36843634e-02 5.51849782e-01
8.15035999e-01 -6.80539072e-01 -1.00931183e-01 -8.12688589e-01
-1.88194871e-01 1.28390598e+00 8.95984054e-01 -1.14159524e+00
1.02120101e+00 8.83956552e-01 -7.02488199e-02 -2.49861926e-02
-1.12867749e+00 5.14941216e-02 -5.85207880e-01 1.10592628e+00
-4.32598382e-01 1.94002330e-01 -8.56779218e-01 3.96536708e-01
5.55079162e-01 -4.13344830e-01 1.40286601e+00 1.03645527e+00
-8.07836890e-01 -4.59277183e-01 -9.19221640e-01 1.23310788e-02
-2.91764617e-01 5.35549700e-01 -3.24245214e-01 1.12894571e+00
-3.49384815e-01 1.44511914e+00 -1.34989828e-01 -6.44613087e-01
6.57050312e-01 -8.13107133e-01 9.99700353e-02 -2.01662213e-01
-6.73079073e-01 3.45384777e-01 1.26384929e-01 -1.30701527e-01
-3.15259010e-01 -1.41443729e-01 -1.48513436e+00 -2.34893411e-01
-9.57001984e-01 7.04625487e-01 4.83274788e-01 6.79638743e-01
1.78256541e-01 4.89033759e-01 1.25595725e+00 -3.84917140e-01
-1.26463675e+00 -9.11296964e-01 -9.76475000e-01 -1.98657155e-01
6.00928366e-01 -9.74107504e-01 -1.17800367e+00 -5.51188171e-01] | [5.284797668457031, 2.5780701637268066] |
eea05c1d-bf21-4722-a8cf-df5f964ba845 | transform-retrieve-generate-natural-language | null | null | http://openaccess.thecvf.com//content/CVPR2022/html/Gao_Transform-Retrieve-Generate_Natural_Language-Centric_Outside-Knowledge_Visual_Question_Answering_CVPR_2022_paper.html | http://openaccess.thecvf.com//content/CVPR2022/papers/Gao_Transform-Retrieve-Generate_Natural_Language-Centric_Outside-Knowledge_Visual_Question_Answering_CVPR_2022_paper.pdf | Transform-Retrieve-Generate: Natural Language-Centric Outside-Knowledge Visual Question Answering | Outside-knowledge visual question answering (OK-VQA) requires the agent to comprehend the image, make use of relevant knowledge from the entire web, and digest all the information to answer the question. Most previous works address the problem by first fusing the image and question in the multi-modal space, which is inflexible for further fusion with a vast amount of external knowledge. In this paper, we call for an alternative paradigm for the OK-VQA task, which transforms the image into plain text, so that we can enable knowledge passage retrieval, and generative question-answering in the natural language space. This paradigm takes advantage of the sheer volume of gigantic knowledge bases and the richness of pre-trained language models. A Transform-Retrieve-Generate framework (TRiG) framework is proposed, which can be plug-and-played with alternative image-to-text models and textual knowledge bases. Experimental results show that our TRiG framework outperforms all state-of-the-art supervised methods by at least 11.1% absolute margin. | ['Prem Natarajan', 'Ying Nian Wu', 'Aishwarya Reganti', 'Govind Thattai', 'Qing Ping', 'Feng Gao'] | 2022-01-01 | null | null | null | cvpr-2022-1 | ['generative-question-answering', 'passage-retrieval'] | ['natural-language-processing', 'natural-language-processing'] | [ 1.41339734e-01 3.79182130e-01 -1.17412619e-02 -1.04400188e-01
-1.28331232e+00 -9.40535784e-01 8.86889815e-01 -1.13861568e-01
-3.26512307e-01 6.95199072e-01 2.10300148e-01 -4.79492515e-01
2.05577329e-01 -9.87727940e-01 -8.97052944e-01 -5.44107080e-01
7.26683557e-01 5.78508556e-01 4.37948525e-01 -3.07701319e-01
-2.11574897e-01 -1.28136396e-01 -1.39263630e+00 6.01827443e-01
8.70571375e-01 9.69106376e-01 6.04037881e-01 6.69214547e-01
-6.96135044e-01 1.31901002e+00 -5.21675348e-01 -9.25097823e-01
4.04277779e-02 -7.81604886e-01 -1.13294601e+00 2.06464201e-01
5.61517358e-01 -4.77422804e-01 -4.23473030e-01 9.27627921e-01
3.08731556e-01 -3.06523126e-02 6.77331626e-01 -1.13934159e+00
-1.29479921e+00 4.25763994e-01 -2.79564023e-01 -2.53821183e-02
6.81238234e-01 2.75659382e-01 1.01558781e+00 -9.84909296e-01
9.46972489e-01 1.28099966e+00 7.56898597e-02 3.85031581e-01
-8.46033931e-01 -1.23776764e-01 8.06244686e-02 7.05956817e-01
-1.31413364e+00 -2.06756920e-01 7.23199904e-01 -3.79898846e-01
7.86009431e-01 2.86795825e-01 7.46918738e-01 1.06974101e+00
-1.92715824e-01 1.12124896e+00 1.13880086e+00 -6.87777817e-01
1.34185806e-01 4.36471581e-01 -1.05617002e-01 7.77618349e-01
-8.04104470e-03 -2.72426367e-01 -5.39648056e-01 8.99120197e-02
7.24632204e-01 -1.67417377e-01 -3.70585918e-01 -4.24560606e-01
-1.22630489e+00 8.67479205e-01 5.55765629e-01 2.47755617e-01
-5.32500088e-01 1.36548370e-01 7.99030587e-02 3.57931077e-01
1.69082940e-01 3.53440762e-01 -6.04643226e-02 9.32640880e-02
-8.55082393e-01 3.21462899e-01 9.20491099e-01 1.06362677e+00
9.88539696e-01 -1.90711692e-01 -2.92724252e-01 4.31419909e-01
3.57832134e-01 7.06470549e-01 3.51873636e-01 -1.07532156e+00
6.70211673e-01 6.33233607e-01 6.35421276e-02 -6.52247608e-01
2.37712160e-01 -1.40937507e-01 -5.94498098e-01 -3.80028300e-02
4.39716846e-01 3.67729776e-02 -1.06982613e+00 1.44159710e+00
6.18593037e-01 -1.67381376e-01 5.31189919e-01 9.25974250e-01
1.16830790e+00 9.60705876e-01 4.93411422e-02 -1.11845076e-01
1.81027699e+00 -1.27326095e+00 -8.55039358e-01 -4.77427870e-01
2.97516823e-01 -7.16998756e-01 1.46741271e+00 -2.60821786e-02
-1.15754545e+00 -4.80518758e-01 -8.88961136e-01 -6.70603335e-01
-7.14796484e-01 5.07221976e-03 2.22889990e-01 3.54684472e-01
-1.23750031e+00 -2.29201153e-01 -2.90904582e-01 -3.59044671e-01
3.35713923e-01 -3.12883705e-01 -2.52426744e-01 -5.73028207e-01
-1.30341208e+00 9.26451325e-01 7.39060879e-01 8.79133958e-03
-9.77374792e-01 -5.23055613e-01 -8.75132263e-01 -5.25030047e-02
9.31817710e-01 -1.38526046e+00 1.28448272e+00 -1.17726839e+00
-1.44694471e+00 7.84188271e-01 -2.27460220e-01 -3.34304094e-01
7.05412030e-01 -3.31434518e-01 -1.47723988e-01 7.22443223e-01
2.01156333e-01 7.38926351e-01 1.05227041e+00 -1.48857796e+00
-3.28009099e-01 -4.91873980e-01 6.89296961e-01 5.04885972e-01
-1.40723437e-01 -2.65891284e-01 -1.01969349e+00 -4.38615799e-01
-2.67483920e-01 -7.47577369e-01 2.13236324e-02 -9.10247117e-02
-2.20280394e-01 -1.94289088e-01 8.05133104e-01 -1.04369354e+00
1.00333405e+00 -1.93611097e+00 2.53601730e-01 -1.90755446e-02
1.89134300e-01 2.19102636e-01 -2.39454389e-01 7.09226251e-01
2.85931855e-01 -8.55410546e-02 -1.32812129e-03 -2.07219169e-01
1.20137133e-01 4.42120910e-01 -7.21479952e-01 -1.53541595e-01
1.80724099e-01 1.63761485e+00 -9.46326911e-01 -8.92045021e-01
2.56749224e-02 4.87058908e-01 -3.05576950e-01 3.27983737e-01
-7.92586923e-01 2.73396581e-01 -7.13853717e-01 5.30693293e-01
3.45804423e-01 -7.39879310e-01 2.09063720e-02 -3.21610421e-01
2.20658913e-01 -2.23893076e-02 -9.15313601e-01 1.91998553e+00
-4.28561002e-01 4.52260345e-01 -1.42794296e-01 -6.74759686e-01
6.21090174e-01 4.69754726e-01 -3.79854515e-02 -8.45473051e-01
5.63416816e-02 9.47933123e-02 -5.26846826e-01 -8.10939610e-01
5.58186769e-01 -1.31821305e-01 -1.19495973e-01 4.53155637e-01
3.73991430e-01 -4.05544847e-01 4.40702409e-01 7.94738352e-01
7.75368154e-01 3.55300248e-01 2.20107466e-01 2.44254634e-01
4.79476362e-01 2.76358336e-01 -1.38206139e-01 9.03738201e-01
1.62208170e-01 4.54555303e-01 3.81273687e-01 -1.31497413e-01
-9.41256821e-01 -1.21701431e+00 4.96257454e-01 9.98379409e-01
2.27470636e-01 -4.34725493e-01 -8.87459457e-01 -6.71363950e-01
-2.94269443e-01 9.83138800e-01 -5.21879733e-01 1.34787336e-02
-2.58132339e-01 -1.78133935e-01 4.89595860e-01 4.13937241e-01
7.46452570e-01 -1.16373956e+00 -5.14135242e-01 -2.56017204e-02
-7.85794497e-01 -1.47591197e+00 -3.50547850e-01 -4.23153341e-01
-4.79780883e-01 -9.45714772e-01 -1.06776226e+00 -8.07376146e-01
4.60043550e-01 3.67964268e-01 1.42301881e+00 -8.03426132e-02
-2.30368823e-01 1.09861338e+00 -7.52093613e-01 -3.60533953e-01
-6.17052555e-01 -1.72226161e-01 -5.65324008e-01 3.36371601e-01
2.00491473e-01 -4.34265882e-01 -5.76637626e-01 -1.79876611e-01
-1.41099811e+00 4.27505553e-01 8.33260417e-01 7.05940902e-01
8.17062020e-01 -2.36929312e-01 6.72702193e-01 -5.67133307e-01
5.22505522e-01 -3.97713363e-01 -4.05698150e-01 8.87961507e-01
-1.60879821e-01 2.92990506e-01 3.83573651e-01 -4.31949586e-01
-1.43016779e+00 7.77923539e-02 -7.76559522e-04 -6.57664835e-01
7.05277920e-02 6.77624226e-01 -4.70366210e-01 2.56335229e-01
5.67007601e-01 7.06311941e-01 -9.15082470e-02 -2.90137738e-01
1.34582889e+00 4.56815660e-01 8.95549715e-01 -4.06721652e-01
8.75312030e-01 6.53194308e-01 -2.70767272e-01 -7.89444685e-01
-1.09841979e+00 -5.03053546e-01 -5.97939730e-01 -3.36513817e-01
1.24841714e+00 -1.06176734e+00 -4.12197500e-01 2.52229005e-01
-1.36242115e+00 -2.57457078e-01 -5.61150908e-01 1.51520312e-01
-7.32289910e-01 6.65361583e-01 -3.76599193e-01 -7.04902947e-01
-4.30135131e-01 -8.70870471e-01 1.04871154e+00 3.58283222e-01
2.67425716e-01 -9.16903198e-01 -3.78473774e-02 1.06756222e+00
2.65939593e-01 2.47534364e-02 9.74461436e-01 -5.70983469e-01
-1.17416596e+00 -1.25906736e-01 -4.64789867e-01 3.79872501e-01
-1.44219548e-01 -5.45187235e-01 -9.06110287e-01 1.10767365e-01
2.26082891e-01 -8.08762252e-01 1.03723133e+00 -5.58954962e-02
9.14519429e-01 -5.07821441e-01 5.02968323e-04 1.07712150e-01
1.47434187e+00 -1.05720751e-01 9.50021923e-01 2.80216604e-01
7.89084136e-01 6.54674113e-01 3.77171010e-01 1.32863864e-01
1.21530914e+00 5.21125197e-01 4.40618515e-01 -9.41535234e-02
-5.83370566e-01 -4.89472359e-01 2.61020154e-01 9.16427255e-01
5.47709502e-02 -3.31907541e-01 -8.62747431e-01 8.95412266e-01
-2.02347755e+00 -1.04121673e+00 1.24537967e-01 1.72780812e+00
1.19934452e+00 -3.94469202e-01 -1.36835605e-01 -3.10117394e-01
3.36444199e-01 1.15971617e-01 -4.68776107e-01 9.36730113e-03
-2.25361839e-01 -1.63753629e-01 1.21006399e-01 5.50794303e-01
-7.53076971e-01 1.04509068e+00 5.48712349e+00 1.07800376e+00
-7.06995010e-01 2.45314568e-01 3.14214259e-01 2.87851930e-01
-6.08260453e-01 3.03447157e-01 -6.00706697e-01 2.04514805e-02
7.45226145e-01 -2.90290177e-01 4.87698197e-01 6.79601014e-01
-3.48223060e-01 -3.94787073e-01 -9.60036457e-01 1.08221900e+00
6.05918765e-01 -1.41508424e+00 5.53454101e-01 -1.81109577e-01
7.03079462e-01 -1.61834210e-01 -2.28354055e-02 3.88657659e-01
3.45505774e-01 -7.46128142e-01 8.80359113e-01 8.90654504e-01
6.79464400e-01 -3.37190539e-01 6.01072788e-01 5.84159434e-01
-1.05027485e+00 9.03916508e-02 -1.94801658e-01 2.34849274e-01
5.44387579e-01 5.33718407e-01 -8.27055454e-01 1.07655346e+00
7.03864217e-01 2.28859425e-01 -7.56645262e-01 5.99348783e-01
-4.14319396e-01 3.82906467e-01 -2.45442599e-01 3.46457437e-02
2.47326627e-01 -1.95459854e-02 2.74331927e-01 8.79287779e-01
2.93369919e-01 2.36606121e-01 1.53178334e-01 9.94583547e-01
-2.02194020e-01 3.23181868e-01 -4.91452903e-01 -3.27841341e-01
7.78621361e-02 1.15393007e+00 -4.91887242e-01 -8.83558750e-01
-7.06244230e-01 1.33932734e+00 2.76044279e-01 6.29066050e-01
-7.02538669e-01 -2.96046376e-01 -2.92441308e-01 -4.69173454e-02
6.01836205e-01 -1.53745532e-01 2.50736743e-01 -1.55768192e+00
4.12177622e-01 -8.54170561e-01 4.89209890e-01 -1.52756870e+00
-1.36737883e+00 6.34141207e-01 2.09410399e-01 -8.13599706e-01
-5.73747933e-01 -3.70635331e-01 -1.43160984e-01 7.14898169e-01
-1.95572746e+00 -1.73027456e+00 -4.88204092e-01 1.07139993e+00
6.06906414e-01 7.95248896e-02 7.63974071e-01 7.18761533e-02
6.91815093e-02 1.57421306e-01 -6.30676141e-03 1.92491964e-01
5.97819269e-01 -1.21900713e+00 1.89593539e-01 7.55289435e-01
6.83676422e-01 1.95741966e-01 4.96916533e-01 -6.23611689e-01
-1.79916823e+00 -1.07648635e+00 9.21095014e-01 -8.34291697e-01
8.05708349e-01 -2.85758048e-01 -9.96232569e-01 7.19874144e-01
5.96677840e-01 -1.61698431e-01 6.98396206e-01 -3.90131861e-01
-8.05096269e-01 -1.75712369e-02 -6.77712977e-01 8.64510715e-01
7.30424166e-01 -7.95579553e-01 -1.18397856e+00 4.35018331e-01
1.04181826e+00 -3.03227454e-01 -6.39075100e-01 9.23023671e-02
3.76525313e-01 -7.67623842e-01 1.09579539e+00 -3.92500907e-01
5.81076980e-01 -3.99391055e-01 -3.91419530e-01 -9.85802710e-01
1.15244977e-01 -5.21279216e-01 -2.72340536e-01 1.31393135e+00
4.14256781e-01 -2.39415348e-01 3.50099266e-01 5.19955039e-01
7.05713034e-02 -3.10116738e-01 -1.05028665e+00 -5.95441222e-01
1.15497718e-02 -4.73118514e-01 4.89172637e-01 6.13138258e-01
-2.95011066e-02 8.37378144e-01 -3.92443687e-01 1.89682506e-02
5.27816176e-01 4.38945293e-01 9.22651172e-01 -7.01423049e-01
-4.98579443e-01 -1.09895535e-01 -1.31726906e-01 -1.30187750e+00
1.38783837e-02 -9.33079541e-01 -5.29465377e-02 -2.20560312e+00
3.36971223e-01 2.96289623e-01 8.74506980e-02 5.81932008e-01
-4.20275897e-01 2.40676507e-01 6.84291482e-01 2.92821199e-01
-1.12236273e+00 8.54894638e-01 1.56100047e+00 -2.42830843e-01
6.40421584e-02 -6.12554252e-01 -8.11036587e-01 5.80875397e-01
3.95032197e-01 -2.16259137e-02 -7.23342896e-01 -6.71875596e-01
4.33669537e-01 2.74607927e-01 8.32981169e-01 -6.85138166e-01
3.91050607e-01 -1.10541172e-01 2.66363293e-01 -6.36015832e-01
6.38928771e-01 -8.78270090e-01 -4.42961231e-02 -8.44549462e-02
-2.16028377e-01 -3.60175595e-02 8.65751281e-02 7.77860820e-01
-4.18593794e-01 -8.58656764e-02 2.18232065e-01 -4.27998394e-01
-9.02502418e-01 5.02803139e-02 -1.96291849e-01 3.72292608e-01
9.65219200e-01 -6.82857782e-02 -5.63226700e-01 -8.04971635e-01
-6.29025877e-01 2.64209330e-01 2.45479330e-01 5.58472037e-01
6.73940599e-01 -1.37296379e+00 -7.83663630e-01 -1.06476545e-01
4.45630997e-01 -1.13963582e-01 7.39203453e-01 5.10666549e-01
-3.10725778e-01 6.02562606e-01 3.89935859e-02 -5.10974169e-01
-9.79021013e-01 9.51302409e-01 5.06910533e-02 -3.55235547e-01
-6.78748310e-01 5.72142839e-01 5.38499713e-01 -2.06487134e-01
-6.95822686e-02 -1.29953936e-01 -7.54780993e-02 1.62880301e-01
7.88996160e-01 -9.01910514e-02 -2.40998462e-01 -7.26555049e-01
6.21518940e-02 5.48893690e-01 4.97753248e-02 -4.94499415e-01
9.69541609e-01 -5.42038679e-01 -2.25275323e-01 4.41905856e-01
9.66196239e-01 -2.18612224e-01 -1.05202854e+00 -6.97086632e-01
-3.74505520e-01 -3.51486534e-01 -7.02333599e-02 -1.07324195e+00
-7.38527954e-01 9.70274687e-01 1.84634686e-01 3.81545097e-01
1.17241132e+00 6.87207699e-01 7.34939277e-01 6.00760400e-01
2.67767668e-01 -9.49148476e-01 6.47458673e-01 3.87004942e-01
1.21827722e+00 -1.15582490e+00 -1.24753244e-01 -3.53244185e-01
-1.07563269e+00 8.77860785e-01 4.21635151e-01 2.35434949e-01
4.17704701e-01 -3.29574645e-01 3.35256517e-01 -3.23634595e-01
-8.26108277e-01 -5.03590703e-01 4.33992982e-01 6.34526491e-01
-2.65767187e-01 -1.21146619e-01 1.07744515e-01 7.17585444e-01
-2.50884324e-01 2.29206681e-01 4.26558346e-01 9.00020897e-01
-4.14324075e-01 -9.62092280e-01 -4.48713481e-01 1.08455613e-01
-2.74765193e-01 -3.48701984e-01 -3.81059974e-01 7.27932930e-01
-1.72614921e-02 9.13115025e-01 -2.28477657e-01 4.37746532e-02
2.61050493e-01 5.18816292e-01 5.76311946e-01 -3.20707947e-01
-1.28677025e-01 5.46426959e-02 2.32960153e-02 -3.84215385e-01
-6.36660576e-01 -2.44676486e-01 -1.07483697e+00 1.45030782e-01
-3.02757859e-01 6.84049726e-02 6.36746407e-01 1.17866719e+00
4.61059570e-01 4.65512186e-01 5.99199068e-03 -4.74967659e-01
-4.23080146e-01 -7.55045116e-01 -3.44092309e-01 3.88223499e-01
2.50661075e-01 -2.90386856e-01 -1.63473278e-01 6.92279518e-01] | [10.895867347717285, 1.5566951036453247] |
4b25282e-154d-4bcd-89b5-445cd2dc3bdc | harnessing-expressive-capacity-of-machine | 2111.14998 | null | https://arxiv.org/abs/2111.14998v1 | https://arxiv.org/pdf/2111.14998v1.pdf | Harnessing expressive capacity of Machine Learning modeling to represent complex coupling of Earth's auroral space weather regimes | We develop multiple Deep Learning (DL) models that advance the state-of-the-art predictions of the global auroral particle precipitation. We use observations from low Earth orbiting spacecraft of the electron energy flux to develop a model that improves global nowcasts (predictions at the time of observation) of the accelerated particles. Multiple Machine Learning (ML) modeling approaches are compared, including a novel multi-task model, models with tail- and distribution-based loss functions, and a spatio-temporally sparse 2D-convolutional model. We detail the data preparation process as well as the model development that will be illustrative for many similar time series global regression problems in space weather and across domains. Our ML improvements are three-fold: 1) loss function engineering; 2) multi-task learning; and 3) transforming the task from time series prediction to spatio-temporal prediction. Notably, the ML models improve prediction of the extreme events, historically obstinate to accurate specification and indicate that increased expressive capacity provided by ML innovation can address grand challenges in the science of space weather. | ['Ryan M. McGranaghan', 'Jack Ziegler'] | 2021-11-29 | null | null | null | null | ['time-series-prediction'] | ['time-series'] | [-3.46754372e-01 -3.48692715e-01 -1.15371041e-01 -6.77844286e-01
-1.01453018e+00 -3.88051063e-01 8.96095693e-01 9.18195322e-02
-2.82724321e-01 1.17171419e+00 2.60611802e-01 -8.76976490e-01
-2.16892511e-01 -6.77271485e-01 -7.74119675e-01 -9.22318339e-01
-6.95254147e-01 6.64876521e-01 -5.02377450e-01 -4.81438041e-01
-5.59446216e-02 6.53793573e-01 -1.33579230e+00 6.77129328e-02
7.86978304e-01 1.25104296e+00 1.25571741e-02 8.43786359e-01
-3.28542218e-02 1.12829769e+00 -1.66071832e-01 3.56966883e-01
1.58526927e-01 -1.14308380e-01 -4.27932888e-01 -5.63234270e-01
3.40169996e-01 -2.39961445e-02 8.35502055e-03 3.22096974e-01
4.13062304e-01 4.24892396e-01 7.01092541e-01 -1.09377611e+00
-2.85764664e-01 -9.23585147e-03 -5.69819689e-01 5.23963332e-01
-4.56601173e-01 5.57491928e-02 1.14838815e+00 -6.44862235e-01
2.68615514e-01 1.09457004e+00 1.47676051e+00 1.93106100e-01
-1.04963374e+00 -5.53847909e-01 1.90094590e-01 -1.24757335e-01
-8.30517590e-01 -2.80080497e-01 3.82908493e-01 -8.48141134e-01
1.29298866e+00 5.66760838e-01 6.77325845e-01 1.00163352e+00
7.43995249e-01 6.97204471e-01 1.15339029e+00 -3.13601503e-03
1.30768135e-01 -6.77755922e-02 4.45752665e-02 3.72300357e-01
-1.10298812e-01 7.88838923e-01 -3.99111003e-01 -5.10262251e-01
3.17857534e-01 -7.04866126e-02 -6.51318207e-02 9.34560820e-02
-7.59686291e-01 1.20227587e+00 3.69126439e-01 1.15223169e-01
-4.48671341e-01 2.08029613e-01 4.04536605e-01 4.83386606e-01
1.57331967e+00 7.12731004e-01 -1.37129760e+00 1.92715842e-02
-1.54270983e+00 1.05163813e+00 9.96876419e-01 3.30529183e-01
6.50687575e-01 5.85856497e-01 -2.76580602e-01 6.74228430e-01
4.05270159e-01 1.01702893e+00 2.18818933e-01 -6.44846737e-01
3.31663042e-01 -2.02369258e-01 5.97876310e-01 -7.22740054e-01
-1.18116140e+00 -8.60111594e-01 -8.46626759e-01 3.37439388e-01
1.27805859e-01 -7.41329312e-01 -8.21768343e-01 1.73555195e+00
2.63205945e-01 4.33353812e-01 4.18702960e-02 8.29128087e-01
6.88290179e-01 1.27435648e+00 3.02915931e-01 -1.86237991e-01
1.05075908e+00 -1.01973569e+00 -6.33265316e-01 -7.25451231e-01
8.89627874e-01 -4.17919934e-01 5.85055232e-01 -1.02191493e-02
-1.02603042e+00 -4.44840968e-01 -8.37091327e-01 -1.06235094e-01
-6.91902339e-01 1.93854511e-01 8.88681531e-01 1.04313344e-02
-1.14995539e+00 8.46525490e-01 -1.17415333e+00 -1.77974313e-01
1.53888226e-01 -3.43842059e-03 1.21972926e-01 5.67573071e-01
-1.48992610e+00 1.17459345e+00 -2.36469179e-01 1.36311054e-01
-1.08575237e+00 -1.45578671e+00 -8.62760484e-01 1.63246796e-01
-6.04969382e-01 -8.32397163e-01 1.55424380e+00 -9.59219515e-01
-1.22927582e+00 7.68841505e-01 -3.26339424e-01 -1.07960927e+00
4.38112885e-01 -4.04812992e-01 -6.09869659e-01 -5.01962543e-01
1.43287078e-01 3.27780515e-01 9.06312048e-01 -8.68449807e-01
-9.53620255e-01 -1.32143751e-01 -6.24553382e-01 1.14445835e-01
9.08472389e-02 9.45256278e-02 3.76874208e-01 -6.23183012e-01
-4.44536358e-01 -9.25981700e-01 -5.51846743e-01 1.44885615e-01
1.36095896e-01 -2.12936059e-01 8.44284415e-01 -1.17927015e+00
9.33659196e-01 -2.06428933e+00 -7.33933896e-02 -5.40688261e-02
3.44118103e-02 1.12185366e-01 -1.81681037e-01 2.92551428e-01
-2.83892423e-01 3.78811322e-02 -5.22376895e-01 -9.74316120e-01
-5.70323644e-03 9.20691416e-02 -1.16511214e+00 7.16294527e-01
2.83703625e-01 8.94233644e-01 -6.81853950e-01 2.00621083e-01
2.27027968e-01 4.37102854e-01 -2.66804010e-01 3.85948211e-01
-7.73571074e-01 7.79029191e-01 -3.42121422e-01 3.76333773e-01
8.25980246e-01 -4.09235537e-01 -1.87076688e-01 2.48935759e-01
-7.03045726e-01 6.55018091e-01 -3.43874693e-01 1.23124492e+00
-7.20433831e-01 7.27959514e-01 2.62726068e-01 -8.51834834e-01
9.39876795e-01 2.75165707e-01 7.06487954e-01 -1.00996232e+00
-3.60518306e-01 2.41785124e-01 -4.01563644e-01 -4.55204159e-01
5.86890221e-01 -6.68935001e-01 3.88588309e-02 4.14507568e-01
-2.07388252e-01 -4.74937379e-01 -2.70297855e-01 -2.82223612e-01
6.03175402e-01 4.13115829e-01 1.18646517e-01 -7.91107476e-01
-1.17584601e-01 2.28072494e-01 7.65257955e-01 8.30790520e-01
5.70035949e-02 4.63003069e-01 3.91458154e-01 -1.23124540e+00
-1.32743657e+00 -8.73969197e-01 -6.65442944e-01 1.23797357e+00
-5.88177025e-01 -2.57097602e-01 2.25646257e-01 -1.33662775e-01
6.08157873e-01 8.85930419e-01 -9.20625925e-01 1.99734598e-01
-4.17204022e-01 -1.78183818e+00 5.90322137e-01 5.57274699e-01
-3.65090333e-02 -9.11235154e-01 -4.68881011e-01 3.58263731e-01
-7.72908926e-02 -7.40746200e-01 1.50514990e-01 5.87154567e-01
-8.64166677e-01 -5.52010477e-01 -4.87951070e-01 -3.04009020e-01
-1.84417844e-01 -3.35708737e-01 1.58835208e+00 -3.66130859e-01
-3.01057339e-01 4.42981794e-02 4.50042374e-02 -9.47293103e-01
-2.83359978e-02 -2.40712729e-03 1.88886032e-01 -2.85894185e-01
2.98413306e-01 -5.82854569e-01 -4.43313450e-01 -2.60201424e-01
-4.82425243e-01 1.86715350e-02 1.20274261e-01 7.29534209e-01
5.99362254e-01 -4.44460273e-01 7.92347491e-01 -6.93476498e-01
4.99729097e-01 -1.25863206e+00 -1.07329702e+00 6.58450201e-02
-5.18966973e-01 8.32264274e-02 6.38218105e-01 1.66934446e-01
-1.16431761e+00 -3.14409882e-01 -2.37350956e-01 -2.52213895e-01
2.47887224e-01 9.42230344e-01 9.05399501e-01 -7.01445937e-02
6.99222147e-01 2.09025428e-01 -8.41061547e-02 -6.64053679e-01
2.45748475e-01 2.35926315e-01 2.82287568e-01 -5.84788740e-01
4.50157285e-01 6.24583960e-01 3.01729709e-01 -8.93301725e-01
-1.25187647e+00 -2.16488108e-01 -4.31259274e-02 1.60081312e-02
1.01817906e+00 -1.58149672e+00 -3.10696274e-01 5.87433636e-01
-1.00860906e+00 -1.06530058e+00 -3.60318571e-01 5.46372533e-01
-4.98215824e-01 -1.66429639e-01 -8.25664878e-01 -1.03827310e+00
-6.80741251e-01 -5.46011925e-01 1.30230319e+00 2.28497125e-02
-5.02688289e-02 -1.65884781e+00 8.41643870e-01 -3.31701279e-01
9.55221832e-01 5.94488621e-01 9.71451879e-01 -4.20545995e-01
-1.80375233e-01 1.00613534e-01 -2.09985852e-01 2.28642926e-01
-5.01938276e-02 6.96230009e-02 -1.21977878e+00 -5.41447163e-01
4.84712422e-01 -5.05874515e-01 1.19538832e+00 1.28823578e+00
1.29073811e+00 -1.82039320e-01 -4.72945362e-01 1.35642362e+00
1.16356027e+00 -1.00938432e-01 3.54882568e-01 5.94562888e-01
3.92500013e-01 2.86395639e-01 4.89601552e-01 8.73876691e-01
4.77769226e-01 2.45830253e-01 3.47176641e-01 -5.78723133e-01
3.48308176e-01 5.74159175e-02 2.55304605e-01 6.07440948e-01
2.35282984e-02 -3.51699889e-01 -1.11909235e+00 7.94224679e-01
-1.92218733e+00 -9.33693171e-01 -3.95432860e-01 1.91751087e+00
7.24171817e-01 -2.93831825e-01 -2.34154806e-01 -6.74413502e-01
-3.68179865e-02 9.89068151e-01 -7.06653237e-01 -5.90306938e-01
-5.16401887e-01 4.08310682e-01 9.54451323e-01 7.98911452e-01
-1.73853981e+00 7.10325181e-01 7.61548519e+00 3.34846288e-01
-1.50628746e+00 2.30231136e-01 9.84569013e-01 -3.54851395e-01
-4.25297379e-01 -1.94982290e-01 -1.10170710e+00 4.24741387e-01
1.46938646e+00 -8.43016133e-02 3.56447577e-01 7.11188197e-01
6.10246837e-01 3.71879458e-01 -7.84392118e-01 5.78183949e-01
-2.10461959e-01 -1.63038981e+00 -3.07540208e-01 -1.48184672e-01
1.05663753e+00 1.06591368e+00 2.41022155e-01 5.59639454e-01
5.78204453e-01 -1.22309601e+00 5.48507988e-01 1.13907743e+00
7.09973097e-01 -4.98390079e-01 5.65383434e-01 2.55699545e-01
-1.08638811e+00 -1.46827668e-01 -4.60173875e-01 -4.91340756e-01
3.29265803e-01 1.15097356e+00 -4.74713743e-01 5.60390532e-01
1.11626136e+00 1.40130532e+00 -1.03891931e-01 8.99020016e-01
1.58413842e-01 8.82175505e-01 -7.09047496e-01 3.55875283e-01
6.48250282e-01 -2.87153423e-01 7.18343318e-01 1.28048885e+00
6.70082033e-01 -6.60570189e-02 9.97610576e-03 8.75280976e-01
6.50010109e-02 -2.38535717e-01 -6.86430097e-01 9.57845896e-03
6.45587891e-02 1.10724127e+00 3.26246232e-01 -9.67474803e-02
-5.61858475e-01 3.23455960e-01 4.21435028e-01 5.53762257e-01
-8.57346416e-01 3.88868861e-02 1.34491587e+00 -3.40246744e-02
4.27748561e-01 -3.69574457e-01 -5.97129047e-01 -1.05127728e+00
-2.92345911e-01 -5.81222951e-01 3.33193719e-01 -8.80313337e-01
-1.70222759e+00 5.35358250e-01 -3.10966939e-01 -1.14040351e+00
-2.44595796e-01 -6.70427799e-01 -9.22012269e-01 1.43610501e+00
-2.17897773e+00 -1.06306839e+00 8.97260606e-02 8.30870196e-02
2.46972695e-01 -3.20227653e-01 8.99118841e-01 3.88105392e-01
-2.17950299e-01 6.25226938e-04 9.59528506e-01 -5.12351215e-01
8.06546926e-01 -1.31668699e+00 1.07303417e+00 6.29744589e-01
-4.50797111e-01 1.16365917e-01 7.73701847e-01 -6.83595717e-01
-1.19625461e+00 -1.63335967e+00 1.40948045e+00 -6.70664430e-01
9.66303647e-01 -2.97620714e-01 -9.12625313e-01 1.18993914e+00
-1.26895666e-01 2.12504879e-01 5.65737128e-01 4.98465449e-01
-1.19254896e-02 -3.11360419e-01 -9.32011247e-01 9.82271880e-02
1.89824894e-01 -6.48240626e-01 -4.09009725e-01 1.01893342e+00
6.12051725e-01 -5.83029628e-01 -7.53394842e-01 8.01206231e-01
4.22673970e-01 -5.12789905e-01 8.90944600e-01 -1.20714307e+00
5.82537413e-01 -1.21210687e-01 -8.63564759e-02 -1.57640278e+00
-5.82584798e-01 -4.98237222e-01 -3.24030936e-01 5.14221251e-01
6.98430121e-01 -6.53746307e-01 3.23524177e-01 4.88640517e-01
-5.24655819e-01 -7.30134904e-01 -1.05273449e+00 -5.78758001e-01
6.54555380e-01 -5.75654030e-01 6.03954792e-01 1.20433068e+00
-3.70193064e-01 -2.50737630e-02 -9.65635598e-01 4.93621498e-01
4.05429780e-01 5.42663097e-01 3.34538966e-01 -1.60510564e+00
-2.36226827e-01 -2.19103813e-01 3.06713402e-01 -1.01644349e+00
2.60411471e-01 -7.80948102e-01 1.63145632e-01 -1.28174520e+00
-2.73191065e-01 -7.87109256e-01 -4.60971296e-01 5.54500699e-01
-7.75141194e-02 -2.00185478e-01 -1.61444783e-01 2.03380466e-01
-1.48095816e-01 9.26706135e-01 8.70186746e-01 5.67218580e-04
-1.09854907e-01 2.02239394e-01 -1.78993702e-01 5.05278528e-01
8.30261946e-01 -6.47073865e-01 -4.06139344e-02 -8.45709145e-01
6.72566354e-01 3.62713069e-01 5.03082216e-01 -1.00921714e+00
-1.03535920e-01 -2.94099987e-01 5.78107953e-01 -6.32640541e-01
4.56710637e-01 -3.71860713e-01 6.24204427e-02 2.66592115e-01
-2.72692740e-01 1.60308808e-01 9.26223040e-01 4.65237141e-01
-2.25978777e-01 4.04743880e-01 9.99754608e-01 -1.72343627e-01
-5.49633205e-01 6.90479517e-01 -4.75650370e-01 6.43102229e-02
5.94807684e-01 8.47395480e-01 -5.62373161e-01 -4.80242819e-01
-8.22955489e-01 7.42644548e-01 1.80679932e-02 3.52688164e-01
-4.94456626e-02 -1.03617096e+00 -1.22662926e+00 4.20465082e-01
-1.76572099e-01 -1.64877087e-01 3.81767064e-01 8.60283315e-01
-5.80972672e-01 6.62086546e-01 -4.12151637e-03 -5.23415029e-01
-5.66788852e-01 5.82948327e-02 1.02845943e+00 -5.87961137e-01
-7.36079156e-01 1.09286916e+00 4.00848687e-01 -9.59578574e-01
-1.76210836e-01 -8.17112386e-01 -6.74790218e-02 7.22064972e-02
6.47269428e-01 1.24128915e-01 2.56940067e-01 -6.35972321e-02
-1.96114436e-01 3.11804354e-01 2.60069668e-01 4.27255221e-02
1.86393726e+00 2.99351010e-02 -1.95727810e-01 8.77396524e-01
1.18680036e+00 -4.21760559e-01 -1.63552439e+00 -1.02611363e-01
-1.69818580e-01 1.69221178e-01 5.04009306e-01 -1.08118033e+00
-9.39418972e-01 1.01867437e+00 6.37075603e-01 1.60853490e-01
8.19872618e-01 -3.09811324e-01 1.00457489e+00 4.35758680e-01
-1.85288057e-01 -9.31086183e-01 -7.72533178e-01 1.11323571e+00
9.56757367e-01 -1.30931485e+00 1.87924832e-01 6.61967158e-01
-5.80806017e-01 9.40189183e-01 3.88870001e-01 -1.37409702e-01
1.29542828e+00 5.21046758e-01 1.42798051e-01 -4.36811328e-01
-1.16865838e+00 1.46393985e-01 3.15652996e-01 1.01738967e-01
4.56765711e-01 2.97396481e-01 -8.90960768e-02 6.92295671e-01
-2.64576644e-01 1.23646334e-01 -7.81407878e-02 7.35435426e-01
-4.94706720e-01 -5.84274173e-01 -2.12587625e-01 8.51186633e-01
-3.87909919e-01 -5.70295095e-01 3.15587819e-01 4.93416756e-01
-1.64582476e-01 6.20748520e-01 5.26347160e-01 1.44169137e-01
-1.49114355e-01 2.88297415e-01 -1.69692367e-01 -2.43289083e-01
-5.06813467e-01 -3.46137196e-01 2.58156478e-01 -4.60505992e-01
-3.69592100e-01 -9.34360802e-01 -7.67902613e-01 -4.21985000e-01
1.66217044e-01 3.62323910e-01 8.96022797e-01 1.27602839e+00
3.66837174e-01 5.69812953e-01 7.32489526e-01 -1.35182703e+00
-5.34026027e-01 -1.06012881e+00 -9.98920798e-01 -4.29885909e-02
1.26792991e+00 -3.19141626e-01 -8.67107749e-01 -1.81072399e-01] | [6.589149475097656, 2.9500465393066406] |
307cabdd-173c-451d-ba20-119b7468542c | estimating-parkinsonism-severity-in-natural | 2105.03464 | null | https://arxiv.org/abs/2105.03464v3 | https://arxiv.org/pdf/2105.03464v3.pdf | Estimating Parkinsonism Severity in Natural Gait Videos of Older Adults with Dementia | Drug-induced parkinsonism affects many older adults with dementia, often causing gait disturbances. New advances in vision-based human pose-estimation have opened possibilities for frequent and unobtrusive analysis of gait in residential settings. This work leverages novel spatial-temporal graph convolutional network (ST-GCN) architectures and training procedures to predict clinical scores of parkinsonism in gait from video of individuals with dementia. We propose a two-stage training approach consisting of a self-supervised pretraining stage that encourages the ST-GCN model to learn about gait patterns before predicting clinical scores in the finetuning stage. The proposed ST-GCN models are evaluated on joint trajectories extracted from video and are compared against traditional (ordinal, linear, random forest) regression models and temporal convolutional network baselines. Three 2D human pose-estimation libraries (OpenPose, Detectron, AlphaPose) and the Microsoft Kinect (2D and 3D) are used to extract joint trajectories of 4787 natural walking bouts from 53 older adults with dementia. A subset of 399 walks from 14 participants is annotated with scores of parkinsonism severity on the gait criteria of the Unified Parkinson's Disease Rating Scale (UPDRS) and the Simpson-Angus Scale (SAS). Our results demonstrate that ST-GCN models operating on 3D joint trajectories extracted from the Kinect consistently outperform all other models and feature sets. Prediction of parkinsonism scores in natural walking bouts of unseen participants remains a challenging task, with the best models achieving macro-averaged F1-scores of 0.53 +/- 0.03 and 0.40 +/- 0.02 for UPDRS-gait and SAS-gait, respectively. Pre-trained model and demo code for this work is available: https://github.com/TaatiTeam/stgcn_parkinsonism_prediction. | ['Babak Taati', 'Andrea Iaboni', 'Sina Mehdizadeh', 'Andrea Sabo'] | 2021-05-07 | null | null | null | null | ['2d-human-pose-estimation'] | ['computer-vision'] | [-2.33458921e-01 -3.95983793e-02 -3.41459751e-01 -2.64192134e-01
-7.78276145e-01 -1.00209396e-02 8.14907476e-02 -3.62670541e-01
-9.35329854e-01 1.06915820e+00 5.87994635e-01 -3.76642980e-02
-1.21932238e-01 -6.17131174e-01 -2.85925359e-01 -3.47914666e-01
-7.81060874e-01 8.01610231e-01 4.41965818e-01 -2.87554741e-01
-2.69090027e-01 3.35931063e-01 -1.21360505e+00 2.49803707e-01
7.25650012e-01 6.04429781e-01 1.52759597e-01 1.06472409e+00
7.56090879e-01 5.29902875e-01 -1.09700829e-01 -1.93770483e-01
8.03769454e-02 -8.94286409e-02 -4.91143465e-01 -1.84948429e-01
3.62294197e-01 -1.11150205e+00 -8.67424667e-01 5.90563357e-01
1.12165606e+00 -2.48284027e-01 6.71018422e-01 -1.21137881e+00
-1.67964578e-01 3.54248956e-02 -4.19104695e-01 6.78917408e-01
4.70180601e-01 7.41052091e-01 8.59453917e-01 -6.25235379e-01
4.94164377e-01 1.16278827e+00 1.42663252e+00 8.11314821e-01
-1.10096288e+00 -4.26605701e-01 -8.25288892e-02 7.28611469e-01
-9.55344856e-01 -2.33351827e-01 2.31359482e-01 -7.04834461e-01
1.35206366e+00 -6.52807727e-02 1.45727885e+00 1.81941080e+00
4.69438761e-01 6.75140381e-01 6.94783032e-01 -2.43104156e-02
4.27787632e-01 -1.02269244e+00 3.55239749e-01 7.83830166e-01
6.72135174e-01 1.34003505e-01 -5.69566727e-01 -4.01701123e-01
9.30632770e-01 1.23660006e-01 -4.00288194e-01 -2.74367720e-01
-1.25684857e+00 8.06102753e-01 6.56274378e-01 -1.54834211e-01
-7.34330833e-01 4.91935164e-01 6.19620860e-01 -2.97670215e-02
1.55786395e-01 -4.27295685e-01 -6.00794554e-01 -6.91567242e-01
-8.10143173e-01 8.78996193e-01 6.04886115e-01 7.82043278e-01
-8.88738632e-02 -2.67916113e-01 -2.07807481e-01 6.69590414e-01
8.60039413e-01 8.55891228e-01 8.55878174e-01 -9.55070674e-01
8.55323493e-01 6.68284953e-01 2.32274055e-01 -5.44641197e-01
-1.04931974e+00 -9.91449580e-02 -5.48526108e-01 4.87534165e-01
6.19921625e-01 -3.73402834e-01 -1.10246193e+00 1.44399464e+00
1.28195465e-01 -4.28736895e-01 -3.19156915e-01 9.58735049e-01
6.58947647e-01 -2.74894446e-01 3.50775808e-01 3.45408320e-02
1.55391753e+00 -5.60444295e-01 -4.29718405e-01 -7.60204136e-01
8.52237165e-01 9.11672041e-02 1.07712972e+00 3.10163081e-01
-7.38932908e-01 -1.77119330e-01 -9.02080357e-01 -7.61318132e-02
-1.40735635e-03 5.03235757e-01 3.43468904e-01 7.79169083e-01
-1.17922723e+00 8.78012598e-01 -1.74018252e+00 -7.36800611e-01
9.62442696e-01 7.65245676e-01 -4.47212309e-01 -4.27796654e-02
-1.07963634e+00 1.06835544e+00 2.45500639e-01 2.29251817e-01
-6.98171377e-01 -4.70322579e-01 -6.14265203e-01 -7.30459511e-01
-4.31921631e-01 -1.25194228e+00 1.34646606e+00 -9.38528255e-02
-1.09358013e+00 9.91979539e-01 -2.32018903e-02 -7.90439844e-01
1.41925204e+00 -9.51605141e-01 -1.53918371e-01 2.58334130e-01
5.92794716e-01 6.40226901e-01 5.61950743e-01 -3.63118976e-01
-6.27894580e-01 -1.08163130e+00 -4.42493051e-01 3.38765949e-01
-2.35564169e-02 -2.03364134e-01 -5.72479554e-02 -4.70090061e-01
1.13298304e-01 -1.12560797e+00 -3.74031395e-01 5.84697783e-01
-4.68769699e-01 1.43571123e-02 7.15379298e-01 -1.29642236e+00
1.05370903e+00 -1.47893751e+00 1.17709711e-02 -1.26320213e-01
7.90911853e-01 1.97511554e-01 5.15623987e-01 4.56707776e-02
-4.49771471e-02 -3.12682956e-01 -2.78936297e-01 -4.49918509e-01
-1.01930171e-01 3.23439896e-01 4.95025009e-01 7.42890835e-01
-2.17288211e-01 1.08729112e+00 -9.25830364e-01 -2.88332462e-01
2.97112882e-01 5.28775990e-01 -6.42810047e-01 -8.93168449e-02
5.07535739e-03 5.40330350e-01 -4.66728210e-01 8.09945583e-01
1.57874092e-01 -2.33703539e-01 1.23685114e-01 -3.16842318e-01
1.24110937e-01 1.13611199e-01 -5.78725278e-01 1.51151621e+00
1.71150640e-01 4.42404747e-01 -3.00988942e-01 -6.34767711e-01
4.88840938e-01 2.21540570e-01 8.09323788e-01 -4.94256645e-01
3.22684288e-01 5.04943311e-01 1.71222135e-01 -1.03228891e+00
-1.96734369e-01 1.83935110e-02 1.19412459e-01 1.56747401e-01
-1.68382570e-01 6.97340965e-01 2.10425019e-01 -2.69279093e-01
1.79252720e+00 4.39690799e-01 4.91159886e-01 -5.81720471e-03
5.47568165e-02 2.20358208e-01 4.53184932e-01 5.44253826e-01
-1.02999115e+00 8.13415229e-01 3.31714749e-01 -6.87666237e-01
-1.32568705e+00 -1.44790936e+00 1.44078985e-01 6.52217805e-01
-5.27807415e-01 -4.86056834e-01 -7.28363276e-01 -4.68420416e-01
3.09872717e-01 1.90640137e-01 -7.23163605e-01 -1.99424431e-01
-9.92604434e-01 -1.03105462e+00 9.60298300e-01 1.32440090e+00
8.97896230e-01 -8.49960089e-01 -1.07948172e+00 2.98956126e-01
-5.14123976e-01 -1.10636270e+00 -2.46246174e-01 1.12755030e-01
-1.46554720e+00 -1.35927355e+00 -1.22441912e+00 -6.93905175e-01
2.14689478e-01 -2.98034608e-01 6.47284031e-01 -6.56339049e-01
-3.26522201e-01 7.27809906e-01 -2.91689783e-01 3.51597145e-02
5.22824042e-02 5.58365174e-02 4.76570040e-01 -5.62257767e-01
7.17768192e-01 -1.22860360e+00 -1.03697073e+00 1.92873642e-01
1.86122954e-01 -8.16169456e-02 6.49609864e-01 4.24577802e-01
5.60797751e-01 -7.90829659e-01 2.73508012e-01 8.84542838e-02
6.88261926e-01 -3.78304958e-01 8.40206519e-02 -2.23822549e-01
-4.66902494e-01 -4.32832800e-02 1.86696365e-01 -5.18917322e-01
-3.02118987e-01 4.46004778e-01 -2.94077009e-01 -5.05313575e-01
-2.39874646e-01 2.41930842e-01 -1.77402303e-01 2.76470393e-01
1.11491585e+00 1.39968693e-01 3.62232596e-01 -4.71313596e-01
1.63583532e-01 6.79762304e-01 9.64331925e-01 -3.08058888e-01
4.02367830e-01 6.47858322e-01 -1.17948530e-02 -8.51935148e-01
-1.94711849e-01 -5.79633772e-01 -1.14784706e+00 -4.91538286e-01
1.25557315e+00 -1.21437132e+00 -5.99688828e-01 9.42470551e-01
-1.02206278e+00 -7.07521379e-01 -8.00538659e-02 8.74155223e-01
-1.08701837e+00 4.83291000e-01 -4.86370593e-01 -6.07042134e-01
-7.37238586e-01 -1.00183165e+00 1.13524640e+00 -5.53083532e-02
-8.89958203e-01 -8.65157783e-01 3.03762585e-01 4.79672283e-01
1.62984133e-01 8.89955938e-01 5.04926980e-01 -4.99716431e-01
6.41972646e-02 -4.02077883e-01 -9.59543586e-02 -4.89086695e-02
2.11277440e-01 -5.95368207e-01 -5.41294038e-01 -2.60257572e-01
-1.66644305e-01 -2.54100591e-01 6.22397661e-01 8.94992530e-01
1.42773703e-01 -2.83104450e-01 -5.40632665e-01 3.74484450e-01
1.02640641e+00 1.05654828e-01 9.69230711e-01 1.21367896e+00
1.09168684e+00 9.55142677e-02 1.24641396e-01 7.16374218e-01
6.88405514e-01 7.67162025e-01 5.41516781e-01 5.87328255e-01
-2.06050754e-01 6.15203194e-02 8.20610702e-01 2.13477805e-01
-1.04786718e+00 4.47891474e-01 -1.26541150e+00 4.53613967e-01
-2.19580007e+00 -9.01304543e-01 -6.44739807e-01 2.01789045e+00
5.61136782e-01 2.44763449e-01 1.03792393e+00 4.27210212e-01
7.79118240e-01 -1.29621401e-01 -8.89501631e-01 3.34899336e-01
2.52240211e-01 1.41043082e-01 8.82377863e-01 1.92486286e-01
-1.34817159e+00 4.48958635e-01 5.32770777e+00 1.40904216e-02
-6.82483196e-01 1.72854945e-01 8.37037340e-02 -4.81154591e-01
6.05489969e-01 -4.59311873e-01 -8.02903473e-01 5.94523489e-01
1.18965876e+00 -9.21754912e-02 1.37673810e-01 1.08452308e+00
9.96479452e-01 -1.02374792e-01 -1.09611893e+00 1.16282117e+00
-3.49574804e-01 -5.87904215e-01 -2.22136870e-01 4.10768658e-01
-8.65716040e-02 7.74802685e-01 -3.04384381e-01 1.75980195e-01
2.86300153e-01 -9.32469428e-01 7.53163338e-01 6.96258843e-01
9.54460919e-01 -4.01236296e-01 5.67430019e-01 1.41877517e-01
-1.41781199e+00 -2.97087640e-01 1.75584957e-01 -1.38455093e-01
4.34681177e-01 1.83988944e-01 -8.79720271e-01 1.63245559e-01
1.15434134e+00 1.16863501e+00 -7.26574719e-01 1.25948334e+00
-4.36242193e-01 5.38343489e-01 -5.16521990e-01 6.06161058e-02
2.44318485e-01 2.05197021e-01 8.12328041e-01 9.25820649e-01
4.55536216e-01 -2.48510595e-02 -8.64547864e-02 6.67427242e-01
4.62124944e-01 -3.61389071e-01 -5.24232924e-01 1.51337817e-01
1.02312565e-01 6.53394759e-01 -5.34277976e-01 1.12384893e-01
-4.15973186e-01 1.12722337e+00 2.17068810e-02 1.12512983e-01
-8.99956524e-01 -4.31077965e-02 6.48449659e-01 6.97205722e-01
1.66755393e-01 -5.44787169e-01 -6.19172871e-01 -1.01255214e+00
4.03723598e-01 -7.46392548e-01 4.71754521e-01 -8.04110289e-01
-1.23543799e+00 1.69308677e-01 1.08822145e-01 -1.65074670e+00
-5.57508230e-01 -1.04799187e+00 -6.26877725e-01 7.08815396e-01
-7.05910385e-01 -1.09375191e+00 -7.11537361e-01 9.31356013e-01
4.55396742e-01 -2.10749462e-01 8.52535725e-01 2.81421900e-01
-4.52011198e-01 2.11882457e-01 -3.15440260e-02 5.13404250e-01
3.62454623e-01 -1.25638640e+00 7.51864731e-01 5.29353440e-01
-7.03683376e-01 5.12691438e-01 6.98801756e-01 -1.26902688e+00
-1.02679777e+00 -1.31461895e+00 8.02015126e-01 -6.54588699e-01
9.09786999e-01 8.48413706e-02 -4.34093148e-01 9.04536486e-01
-5.87934792e-01 -5.14776520e-02 4.62589115e-01 -3.62015843e-01
-2.39600595e-02 2.90887505e-01 -1.27036607e+00 7.33214140e-01
1.65479314e+00 -2.70356864e-01 -1.02878606e+00 4.83452439e-01
9.17885005e-02 -4.14838880e-01 -8.89775276e-01 4.53115135e-01
1.21492100e+00 -8.91283929e-01 1.24756444e+00 -5.73808134e-01
4.60831553e-01 -1.97538491e-02 -4.09829468e-01 -8.40756893e-01
-3.32390338e-01 -2.35691532e-01 -5.67624331e-01 9.09272656e-02
2.91502941e-02 -5.03495753e-01 1.27579820e+00 7.27342069e-01
-2.16440275e-01 -8.32034588e-01 -1.30007076e+00 -1.04870129e+00
1.87972281e-02 -6.42267525e-01 -2.31310487e-01 2.48771682e-01
8.42572227e-02 2.54010648e-01 -2.44471028e-01 2.29206085e-01
1.02895367e+00 -9.53592956e-01 4.74701911e-01 -1.51628697e+00
-2.30530575e-01 -3.44727397e-01 -1.15646899e+00 -4.42355871e-01
-3.95794541e-01 -6.75568104e-01 -2.51156360e-01 -2.27280974e+00
-6.45124316e-02 3.26891184e-01 2.67173767e-01 7.09024131e-01
9.77248549e-02 -8.16996209e-04 -1.65714785e-01 6.80676639e-01
-1.61976859e-01 6.02126181e-01 9.42233145e-01 -1.57998934e-01
-5.84111750e-01 3.61349851e-01 -1.02424085e-01 9.91477489e-01
1.14068675e+00 -3.06444079e-01 -4.75009173e-01 -1.30128801e-01
-1.47069111e-01 -2.61083364e-01 1.13201392e+00 -1.64653051e+00
-2.57893298e-02 3.81760478e-01 8.39722574e-01 -8.31106842e-01
4.69659090e-01 -3.74468178e-01 9.59208906e-02 1.27436388e+00
3.30200762e-01 2.32681751e-01 6.44660220e-02 7.92554557e-01
6.10746801e-01 1.95665792e-01 7.54726231e-01 -5.39131701e-01
-7.09498107e-01 4.67256159e-01 -7.26364553e-01 1.86330035e-01
8.03729236e-01 -8.70707810e-01 -3.49219233e-01 -2.56231874e-01
-1.57347488e+00 6.82971105e-02 1.56146467e-01 1.63589284e-01
7.10094690e-01 -1.56913328e+00 -5.44147193e-01 -3.23938698e-01
1.88084602e-01 -5.83831854e-02 6.15961142e-02 1.28250611e+00
-8.83742809e-01 3.18643659e-01 -7.06359565e-01 -6.73063934e-01
-1.23532093e+00 -3.36024046e-01 7.01905727e-01 -4.72277731e-01
-1.38947773e+00 4.16366190e-01 -5.59207976e-01 -1.69374391e-01
3.49687725e-01 -5.85565567e-01 -2.83595234e-01 8.60417485e-02
4.58432287e-01 9.71788704e-01 2.14099661e-01 -7.64293909e-01
-7.50319064e-01 3.83248955e-01 -2.32541189e-01 -3.22830200e-01
1.69710302e+00 -2.88980696e-02 4.17067140e-01 2.45959267e-01
9.82033968e-01 -8.44254315e-01 -1.46921742e+00 2.90300339e-01
1.71885207e-01 2.02651456e-01 -1.92512512e-01 -7.44046867e-01
-6.97936356e-01 5.82542717e-01 1.48025131e+00 -3.63906145e-01
6.39620245e-01 -1.04479678e-01 1.06350768e+00 6.40722096e-01
6.71906292e-01 -1.00022638e+00 1.86337411e-01 4.48836565e-01
1.01295173e+00 -9.42452371e-01 5.95658310e-02 5.78534901e-02
-4.93779182e-01 1.34451771e+00 5.33602178e-01 -4.61005926e-01
5.32527030e-01 1.26804439e-02 -9.40744877e-02 -3.32313478e-01
-5.15432619e-02 -2.61085659e-01 4.56563123e-02 1.30356348e+00
2.16334388e-01 2.73636371e-01 -4.88392442e-01 1.02589309e+00
-5.21662295e-01 5.98420501e-01 3.24690223e-01 1.37917769e+00
-6.54954314e-01 -7.62919128e-01 -4.27421242e-01 9.42395508e-01
-1.61997691e-01 1.33048788e-01 -3.90690804e-01 1.13282073e+00
9.03607979e-02 5.04129589e-01 -3.40961784e-01 -5.54461718e-01
7.11753190e-01 2.97705948e-01 7.54521370e-01 -4.24692720e-01
-2.25278020e-01 -1.71809375e-01 5.49804747e-01 -8.67796719e-01
-3.95405382e-01 -1.17087853e+00 -1.40175641e+00 -1.91346124e-01
5.23691297e-01 -7.46957004e-01 1.88626155e-01 9.92699444e-01
1.20138049e-01 4.18265104e-01 -3.73725653e-01 -1.41333926e+00
-4.60791796e-01 -1.20470870e+00 -6.41505778e-01 -7.86726177e-02
3.87129873e-01 -1.04905581e+00 -3.32741529e-01 2.76319474e-01] | [7.103670597076416, 0.3224536180496216] |
f7f53dc4-7a6d-4fca-9765-7f8a1b766720 | actively-learning-a-bayesian-matrix-fusion | 2306.05331 | null | https://arxiv.org/abs/2306.05331v1 | https://arxiv.org/pdf/2306.05331v1.pdf | Actively learning a Bayesian matrix fusion model with deep side information | High-dimensional deep neural network representations of images and concepts can be aligned to predict human annotations of diverse stimuli. However, such alignment requires the costly collection of behavioral responses, such that, in practice, the deep-feature spaces are only ever sparsely sampled. Here, we propose an active learning approach to adaptively sampling experimental stimuli to efficiently learn a Bayesian matrix factorization model with deep side information. We observe a significant efficiency gain over a passive baseline. Furthermore, with a sequential batched sampling strategy, the algorithm is applicable not only to small datasets collected from traditional laboratory experiments but also to settings where large-scale crowdsourced data collection is needed to accurately align the high-dimensional deep feature representations derived from pre-trained networks. | ['Jordan W. Suchow', 'Yangyang Yu'] | 2023-06-08 | null | null | null | null | ['active-learning', 'active-learning'] | ['methodology', 'natural-language-processing'] | [ 1.82686344e-01 -2.15315685e-01 -3.33608299e-01 -7.53376186e-01
-8.79299045e-01 -6.47243917e-01 3.70509148e-01 7.15360194e-02
-9.87482607e-01 6.85706615e-01 3.37098986e-01 3.19446534e-01
1.36067076e-02 -2.99189985e-01 -9.15749490e-01 -4.85632032e-01
1.73062339e-01 6.80096149e-01 -1.48000851e-01 4.64207917e-01
3.94271426e-02 3.88585687e-01 -1.50895166e+00 2.19721481e-01
4.38981205e-01 1.03811550e+00 2.27036983e-01 6.76255107e-01
1.19922742e-01 6.46265149e-01 -2.73055196e-01 -3.05441290e-01
5.75992227e-01 -6.11990839e-02 -4.28689361e-01 2.13186637e-01
8.34432721e-01 -9.92530465e-01 -4.38025445e-01 8.80773842e-01
5.16170800e-01 3.04141521e-01 6.86660409e-01 -1.32802784e+00
-7.31833398e-01 2.40763947e-01 -4.82844919e-01 2.66976506e-01
2.39387468e-01 5.04738867e-01 8.34857345e-01 -8.31323922e-01
3.79675001e-01 1.13468599e+00 3.41312140e-01 7.47110069e-01
-1.83604753e+00 -8.96480024e-01 5.20251334e-01 8.93142894e-02
-1.18574083e+00 -7.02523768e-01 5.74886799e-01 -7.84412205e-01
6.60834193e-01 -2.59520382e-01 8.75585556e-01 1.80248940e+00
-2.11432695e-01 7.54216492e-01 9.47429657e-01 8.12509656e-02
6.74396574e-01 1.19715214e-01 7.54984766e-02 5.36777496e-01
5.18910229e-01 -3.83506827e-02 -9.70533013e-01 -7.03203976e-01
9.74283338e-01 4.95952785e-01 4.75047575e-03 -7.21546531e-01
-1.21700478e+00 8.01100552e-01 2.75197566e-01 -1.28829926e-01
-6.06108129e-01 3.89845878e-01 3.03595930e-01 -7.39609599e-02
4.75644112e-01 4.66045707e-01 -5.42863905e-01 -1.07329972e-01
-7.93797970e-01 4.28353280e-01 4.43175793e-01 1.03932798e+00
9.99541223e-01 7.65322009e-03 -2.73739517e-01 5.88721454e-01
2.71464527e-01 6.30818188e-01 5.42604923e-01 -1.32944620e+00
3.05236101e-01 5.08522570e-01 7.95414329e-01 -1.06505084e+00
-2.96679556e-01 -1.23109361e-02 -6.56174719e-01 -1.73065811e-02
5.71817517e-01 -3.17718148e-01 -6.29226983e-01 1.92483830e+00
4.24021959e-01 1.48619935e-01 -4.80695665e-01 1.00785506e+00
3.27534109e-01 1.88106790e-01 4.14765745e-01 -1.05272539e-01
1.05951238e+00 -4.92207021e-01 -4.72261548e-01 -4.91877437e-01
3.29997957e-01 -3.97253275e-01 1.26672912e+00 4.39769834e-01
-9.05974090e-01 -5.40469646e-01 -8.22067440e-01 -1.36595458e-01
2.66517829e-02 5.47640920e-02 1.08335102e+00 3.23556602e-01
-7.90199101e-01 3.60353947e-01 -1.01232147e+00 -1.89449236e-01
1.12773204e+00 5.66811025e-01 -5.36388457e-01 -1.56177476e-01
-6.50988758e-01 3.41282636e-01 6.17700256e-02 9.19436663e-02
-1.58708656e+00 -7.46369958e-01 -7.21944392e-01 8.50711092e-02
3.57139796e-01 -6.64144933e-01 1.39122033e+00 -1.19432247e+00
-1.46828163e+00 8.02150190e-01 -4.84212130e-01 -5.66482306e-01
3.94818574e-01 -4.93818223e-01 1.82494178e-01 3.09617549e-01
-6.89660944e-03 1.03117657e+00 8.62070322e-01 -9.00391459e-01
-1.73781171e-01 -5.72265565e-01 1.18196204e-01 2.28692785e-01
-7.24978268e-01 -4.58095260e-02 3.71630788e-02 -2.54562229e-01
-1.82463929e-01 -9.38771665e-01 -6.53316796e-01 6.62009835e-01
-2.52792519e-02 -1.41649798e-01 3.42902422e-01 -1.74198553e-01
5.64536333e-01 -2.16815591e+00 1.39946807e-02 -1.34784319e-02
7.23838627e-01 2.00945020e-01 -3.52183819e-01 1.08200647e-01
2.24664390e-01 -1.46461844e-01 5.18654659e-02 -3.99652332e-01
1.92054823e-01 1.06162988e-01 -5.85564077e-01 6.76961899e-01
1.46680832e-01 1.06069982e+00 -1.06606460e+00 -3.42228025e-01
-2.71039568e-02 2.56304979e-01 -8.41789722e-01 6.66531086e-01
-4.53338772e-01 6.03547573e-01 -4.36848640e-01 3.82124335e-01
4.65457201e-01 -5.67316532e-01 3.21068048e-01 -3.11805867e-02
2.75580615e-01 8.66519660e-02 -9.35182929e-01 1.89320886e+00
-3.51697564e-01 6.75680280e-01 1.23966159e-02 -1.07638597e+00
6.42538071e-01 3.27160388e-01 6.32738590e-01 -6.00489974e-01
2.81370610e-01 -8.85575861e-02 -1.78241491e-01 -5.01094878e-01
4.06009614e-01 2.73809165e-01 -2.14369539e-02 7.32168376e-01
2.77512401e-01 1.99745387e-01 -3.47622409e-02 2.92342246e-01
9.91568446e-01 1.09421566e-01 3.42540860e-01 -1.93054721e-01
-2.58870542e-01 -2.21252330e-02 7.24464655e-01 7.86140084e-01
-4.39527750e-01 3.88776779e-01 5.41752875e-01 -7.61000037e-01
-1.19330931e+00 -1.13719964e+00 7.30945915e-02 1.41952741e+00
-1.00369751e-01 -1.78006962e-01 -8.32170963e-01 -6.86960220e-01
9.60385129e-02 2.66975313e-01 -8.37152302e-01 -1.57774895e-01
-3.04961473e-01 -6.43016338e-01 2.42184639e-01 7.11607516e-01
8.62508342e-02 -4.96567130e-01 -6.11588597e-01 9.05249491e-02
1.73070580e-01 -1.23197556e+00 -4.44507211e-01 2.54497200e-01
-6.60358667e-01 -1.09352350e+00 -5.47116041e-01 -5.91105878e-01
1.03074718e+00 6.13012433e-01 1.10934663e+00 -1.32205084e-01
-3.70134383e-01 6.61925614e-01 6.34806529e-02 -5.38094819e-01
3.35658729e-01 -2.85118729e-01 3.88651490e-01 1.33669853e-01
8.52081180e-01 -7.33424604e-01 -1.09112346e+00 3.59668344e-01
-9.62866247e-01 -1.29213423e-01 4.33043599e-01 7.06433058e-01
7.64499426e-01 -6.73392177e-01 5.71175814e-01 -9.27619457e-01
4.52665687e-01 -5.00208139e-01 -1.09551728e+00 2.76713893e-02
-8.24068189e-02 -1.43725663e-01 6.91884577e-01 -1.07325566e+00
-8.31242561e-01 5.25137424e-01 3.50274742e-01 -5.89802504e-01
-4.73222792e-01 -4.17222865e-02 -2.54464537e-01 7.82709271e-02
1.05669677e+00 4.21439894e-02 -2.11817160e-01 -2.48578832e-01
4.80688483e-01 6.96156681e-01 1.53385803e-01 -7.18883812e-01
6.33945882e-01 6.30493760e-01 -2.59768009e-01 -4.82719481e-01
-1.09173679e+00 -2.62900263e-01 -8.17221761e-01 1.56198873e-03
7.64247656e-01 -1.36550295e+00 -9.04686332e-01 3.24544221e-01
-1.15007448e+00 -7.97267437e-01 -3.11046273e-01 8.72688651e-01
-8.10060978e-01 4.20726910e-02 -4.74583119e-01 -6.55440152e-01
-7.46001862e-03 -1.14556766e+00 1.11582124e+00 1.38704211e-01
-5.77140570e-01 -6.67087138e-01 2.85352319e-01 4.57732022e-01
3.49729538e-01 -1.22437365e-01 7.14127719e-01 -1.02898550e+00
-7.97342002e-01 -2.15388060e-01 -7.41494000e-02 2.55219072e-01
2.00762209e-02 -2.03544378e-01 -1.19972861e+00 -3.48711491e-01
9.67085063e-02 -1.21919954e+00 4.88883197e-01 3.08479369e-01
1.64482939e+00 -4.95142162e-01 -1.37677774e-01 3.28505039e-01
1.05230570e+00 -1.38382822e-01 2.39556909e-01 -4.18772131e-01
4.52109098e-01 4.00238931e-01 3.51095766e-01 6.86959505e-01
2.64106452e-01 5.39377987e-01 2.14122772e-01 1.64078712e-01
3.49740267e-01 -4.57257807e-01 1.71905756e-01 4.53925729e-01
5.13835371e-01 -1.53440759e-01 -7.78260112e-01 5.69946468e-01
-1.85328209e+00 -9.34047461e-01 4.64967579e-01 2.52753425e+00
1.01895761e+00 -5.38731106e-02 3.82440627e-01 -3.83717954e-01
6.71492517e-01 -4.22442444e-02 -1.08118677e+00 5.63410204e-03
1.12492077e-01 8.68572667e-02 3.14100623e-01 1.31691813e-01
-1.04094446e+00 7.04914093e-01 7.30181265e+00 5.23800373e-01
-1.20057166e+00 1.91070870e-01 8.22343946e-01 -7.85163105e-01
-3.08486462e-01 -1.16858438e-01 -8.65327775e-01 4.45680052e-01
8.44480157e-01 -1.41690060e-01 6.59748018e-01 1.15070164e+00
2.02434629e-01 -1.85298607e-01 -1.66556680e+00 1.28581691e+00
-1.04734916e-02 -1.17300630e+00 -2.03737207e-02 2.27139950e-01
8.43618393e-01 2.14590237e-01 2.33572721e-01 5.76338656e-02
8.55223000e-01 -1.03067696e+00 4.54148233e-01 5.37202299e-01
6.02662146e-01 -4.44544941e-01 1.13551961e-02 5.58866978e-01
-4.96684790e-01 -1.95732415e-01 -8.89583886e-01 -2.74490416e-01
-1.74610868e-01 7.61551738e-01 -9.33601856e-01 -5.24043858e-01
4.60588872e-01 4.59743619e-01 -3.46814483e-01 1.01594353e+00
-1.22690849e-01 7.26731300e-01 -3.07638109e-01 -2.44445205e-01
1.84569638e-02 1.24516688e-01 1.62924767e-01 7.38846004e-01
9.27705094e-02 -3.56773734e-02 3.04020733e-01 1.08535624e+00
-4.54095691e-01 -1.13099650e-01 -7.73212790e-01 -4.68199432e-01
7.58832216e-01 1.45965302e+00 -5.53837895e-01 -2.87085265e-01
-3.83574903e-01 7.98088670e-01 9.16422844e-01 5.44351459e-01
-6.97772205e-01 3.37023847e-02 9.30796504e-01 1.81387976e-01
3.56986105e-01 -5.37696481e-01 -4.06710096e-02 -1.28110039e+00
-1.23730458e-01 -8.23485315e-01 -8.68872404e-02 -6.76791131e-01
-1.68753433e+00 3.09449732e-01 8.37413967e-02 -1.05572331e+00
-3.97508293e-01 -5.71493983e-01 -2.83179671e-01 7.34907210e-01
-9.87931430e-01 -6.60537720e-01 -2.06549406e-01 7.62597978e-01
3.51075858e-01 -1.63893119e-01 1.14445150e+00 1.99454591e-01
-6.14806235e-01 6.74155056e-01 1.07787259e-01 3.31105709e-01
8.29627991e-01 -9.14582670e-01 2.73865670e-01 5.51601231e-01
4.63556081e-01 1.05565655e+00 4.86458659e-01 -4.57086354e-01
-1.47847593e+00 -1.12758517e+00 3.05152476e-01 -5.00889957e-01
6.13864183e-01 -1.17060339e+00 -7.20815420e-01 8.24299216e-01
8.53340477e-02 5.67054510e-01 1.42327356e+00 3.46992701e-01
-7.40888298e-01 -3.04063588e-01 -1.20693159e+00 6.17889464e-01
1.09072483e+00 -9.11865830e-01 -2.74638712e-01 6.74674213e-01
7.01027155e-01 -2.03219786e-01 -7.60101199e-01 -9.06577706e-02
7.03278124e-01 -5.20003855e-01 6.78152800e-01 -1.23725057e+00
3.00782472e-01 -4.39046770e-02 -4.11361784e-01 -1.11325014e+00
-4.81378317e-01 -5.01069248e-01 -2.13984236e-01 1.02989328e+00
2.77287513e-01 -4.70231444e-01 1.11469591e+00 1.14055955e+00
4.63266492e-01 -5.53455889e-01 -8.07286561e-01 -4.17836338e-01
-1.57408267e-01 -3.09168786e-01 4.22800243e-01 7.10973680e-01
-1.94419116e-01 4.19266731e-01 -3.04218888e-01 -1.44207224e-01
7.34786808e-01 4.36813273e-02 1.14057314e+00 -1.24312031e+00
-5.06081283e-01 9.44376811e-02 -5.01204610e-01 -1.31885457e+00
5.19749463e-01 -4.83189523e-01 8.73960480e-02 -1.02484512e+00
6.88071251e-01 -3.30771297e-01 -3.05317312e-01 3.69317204e-01
-1.85248807e-01 2.92945176e-01 -8.99686739e-02 1.07058950e-01
-1.14663172e+00 6.68841600e-01 1.06044340e+00 3.85462120e-02
1.39273137e-01 -1.66580752e-01 -1.03077781e+00 8.21459591e-01
6.02639973e-01 -7.99754381e-01 -8.36262047e-01 -9.14760649e-01
5.62658191e-01 -1.60131022e-01 6.64414465e-01 -7.57246792e-01
3.13309073e-01 -5.56644619e-01 9.61468935e-01 -1.84751868e-01
5.96619248e-01 -9.00447726e-01 -1.99224845e-01 5.37637323e-02
-9.76732969e-01 -2.54707426e-01 5.73339267e-03 1.03569031e+00
2.18296126e-01 -8.09578672e-02 7.48512924e-01 -4.21623111e-01
-2.28289023e-01 8.09014857e-01 -6.15567148e-01 3.04098010e-01
9.11279678e-01 -4.71459702e-02 -2.96015620e-01 -3.35893631e-01
-4.47172463e-01 -7.59898871e-02 6.13902092e-01 1.62699699e-01
5.27741849e-01 -1.41081023e+00 -2.99974203e-01 2.32075468e-01
7.11802170e-02 3.76839370e-01 3.27806115e-01 6.44567907e-01
9.83788352e-03 1.94822282e-01 -3.88619572e-01 -7.45792508e-01
-8.38908732e-01 6.82305157e-01 5.30062355e-02 5.83591349e-02
-1.38106361e-01 8.58363569e-01 6.19052589e-01 -2.27916896e-01
4.14920092e-01 -4.10271883e-02 1.54775545e-01 -3.43108140e-02
8.75665247e-01 4.19512279e-02 -8.42841193e-02 -2.72031963e-01
-6.88730553e-02 -1.09174587e-01 -3.81912589e-01 -1.38505384e-01
1.53284371e+00 -5.87497130e-02 1.23313412e-01 6.50981903e-01
1.31839728e+00 -4.36119400e-02 -1.88655305e+00 -6.51754379e-01
-2.58399546e-01 -6.92303956e-01 -1.59316529e-02 -3.32274973e-01
-1.04093301e+00 9.64444518e-01 6.26587331e-01 -4.10779685e-01
7.85507321e-01 -1.93215534e-01 6.06455266e-01 1.06029630e+00
4.44522351e-01 -1.03848195e+00 5.04057407e-01 1.79059312e-01
6.21295452e-01 -1.26577437e+00 -3.31138098e-03 1.98474631e-01
-6.75417304e-01 7.68645406e-01 8.33905816e-01 -5.54444373e-01
5.33592165e-01 2.61942655e-01 -1.39328733e-01 -6.57355338e-02
-1.14849973e+00 2.24992976e-01 3.72462347e-02 7.66085982e-01
2.46283725e-01 1.05037726e-01 3.27448159e-01 8.77652347e-01
1.33265555e-01 3.07304680e-01 3.57704937e-01 6.47641480e-01
-4.65043426e-01 -9.24221098e-01 -6.46219179e-02 7.04555094e-01
-2.74460942e-01 3.08566890e-03 -5.15081704e-01 2.28682801e-01
-2.16354057e-02 6.90898359e-01 3.92313391e-01 -1.72564030e-01
-1.21990610e-02 -3.44134010e-02 7.27594197e-01 -8.05217087e-01
-2.90477008e-01 -1.09381855e-01 -2.87815988e-01 -9.02700424e-01
-3.92195970e-01 -7.26704001e-01 -8.05234313e-01 -6.87578097e-02
-9.33662895e-03 -5.89348488e-02 5.26180685e-01 8.82833242e-01
6.55353069e-01 -2.02407882e-01 7.56878972e-01 -9.91738260e-01
-6.71996236e-01 -8.47271502e-01 -4.82868999e-01 5.86045861e-01
1.20627023e-01 -8.11389565e-01 -2.62088686e-01 7.60537907e-02] | [9.97717571258545, 2.3417396545410156] |
11f80348-fc61-48d1-9b5f-c2a67e68a591 | ym2413-mdb-a-multi-instrumental-fm-video-game | 2211.07131 | null | https://arxiv.org/abs/2211.07131v1 | https://arxiv.org/pdf/2211.07131v1.pdf | YM2413-MDB: A Multi-Instrumental FM Video Game Music Dataset with Emotion Annotations | Existing multi-instrumental datasets tend to be biased toward pop and classical music. In addition, they generally lack high-level annotations such as emotion tags. In this paper, we propose YM2413-MDB, an 80s FM video game music dataset with multi-label emotion annotations. It includes 669 audio and MIDI files of music from Sega and MSX PC games in the 80s using YM2413, a programmable sound generator based on FM. The collected game music is arranged with a subset of 15 monophonic instruments and one drum instrument. They were converted from binary commands of the YM2413 sound chip. Each song was labeled with 19 emotion tags by two annotators and validated by three verifiers to obtain refined tags. We provide the baseline models and results for emotion recognition and emotion-conditioned symbolic music generation using YM2413-MDB. | ['Juhan Nam', 'Taegyun Kwon', 'JongIk Jeon', 'Seolhee Lee', 'Yoonjin Chung', 'Eunjin Choi'] | 2022-11-14 | null | null | null | null | ['music-generation', 'music-generation'] | ['audio', 'music'] | [ 1.56626254e-01 -2.51511514e-01 1.53642260e-02 -5.99422567e-02
-1.09845138e+00 -1.14183772e+00 -4.47416957e-03 -3.45636815e-01
1.75477862e-01 3.92904282e-01 5.81948385e-02 2.98356622e-01
-3.14157337e-01 -3.11675668e-01 -3.25039476e-01 -5.10053337e-01
-8.30109790e-02 5.87783992e-01 -1.60488546e-01 -2.02444300e-01
1.01707228e-01 -1.71341449e-01 -1.71283269e+00 6.34182990e-01
2.59880632e-01 1.41866541e+00 1.84059054e-01 1.20136714e+00
2.13525593e-01 9.95244443e-01 -1.04535544e+00 -4.22424287e-01
2.31574923e-01 -6.89728618e-01 -6.25483155e-01 -4.22944725e-01
6.82689622e-02 1.30197942e-01 4.20856364e-02 1.02896881e+00
9.84252751e-01 3.20072412e-01 6.66560173e-01 -1.55135500e+00
-3.87545414e-02 1.59978461e+00 -2.13807598e-01 -4.67170089e-01
4.97415990e-01 -3.40281516e-01 1.42358351e+00 -7.35019863e-01
6.52346075e-01 8.84818316e-01 1.05061638e+00 5.12657464e-01
-1.09184813e+00 -1.20770848e+00 -7.53956914e-01 1.05035685e-01
-1.72844362e+00 -5.17276883e-01 8.46095443e-01 -6.41723871e-01
6.84351683e-01 6.83568835e-01 9.06782687e-01 1.34773207e+00
-2.94816285e-01 6.87142253e-01 1.01878953e+00 -4.10423100e-01
3.21894377e-01 -3.39433938e-01 -9.89368185e-02 3.68844450e-01
-5.48377872e-01 -8.74072164e-02 -1.13055634e+00 -4.40757126e-01
5.28524935e-01 -6.53851688e-01 -1.10241622e-01 3.09452385e-01
-1.23179448e+00 5.71538985e-01 -4.14177060e-01 2.61271000e-01
-2.30843782e-01 3.97475183e-01 7.08581984e-01 3.87035698e-01
1.69962466e-01 8.77542317e-01 -6.55617476e-01 -1.13064623e+00
-1.19691479e+00 2.92717874e-01 8.22704732e-01 1.26139402e+00
3.67726684e-01 4.63192135e-01 -2.35872328e-01 1.33068776e+00
2.55933404e-01 5.83955526e-01 7.60589004e-01 -1.52826917e+00
9.07135829e-02 1.00582885e-02 1.72215309e-02 -9.72924232e-01
-4.18720871e-01 -5.06053090e-01 -5.23517907e-01 1.39513031e-01
1.01912856e-01 -2.10029289e-01 -3.18466365e-01 1.87163281e+00
-2.98343539e-01 4.72699046e-01 -1.08685724e-01 8.89694571e-01
1.04406035e+00 4.91792053e-01 -1.80107564e-01 -4.99784164e-02
1.22009826e+00 -9.67337012e-01 -6.68731451e-01 1.92457706e-01
5.37823558e-01 -1.07393456e+00 1.45166731e+00 1.10836005e+00
-1.00034857e+00 -8.98594499e-01 -9.73654449e-01 2.10683808e-01
8.02682489e-02 8.08640957e-01 4.19599175e-01 7.10195065e-01
-6.04224324e-01 8.52327228e-01 -2.82750785e-01 8.68472978e-02
1.00872107e-02 3.79350960e-01 -2.33745761e-02 7.06161439e-01
-1.20703471e+00 2.47459766e-02 3.91458482e-01 -2.57331938e-01
-9.37271178e-01 -6.10897660e-01 -4.16575909e-01 -7.23603666e-02
4.59370278e-02 -3.38349015e-01 1.83350015e+00 -1.08177948e+00
-2.11917448e+00 1.13286972e+00 4.58790570e-01 -1.29565835e-01
9.15742591e-02 -1.65255189e-01 -8.89657795e-01 1.44985095e-01
2.37211019e-01 4.45147902e-01 9.65642989e-01 -9.69179094e-01
-8.06564093e-01 5.61852790e-02 -2.40139499e-01 4.01287451e-02
-1.00161523e-01 4.13124382e-01 -6.10709071e-01 -1.30489767e+00
9.53831710e-03 -1.44529700e+00 3.17533135e-01 -8.83295417e-01
-9.08366382e-01 1.46765247e-01 2.98748881e-01 -5.48287392e-01
1.59715700e+00 -2.53627086e+00 7.43317530e-02 4.40224320e-01
-1.20619297e-01 -9.23162624e-02 -2.58662730e-01 1.19966477e-01
-2.55873621e-01 2.03621294e-02 2.90537238e-01 -5.21805167e-01
6.46285355e-01 1.62075963e-02 -5.75264037e-01 -2.35344116e-02
-3.05067062e-01 6.73125446e-01 -9.18878734e-01 -5.14271557e-01
-1.35567933e-01 -3.59718502e-03 -6.27086103e-01 1.16493076e-01
-2.38515824e-01 4.56470519e-01 -1.55132458e-01 1.00464010e+00
1.10984229e-01 3.54803711e-01 6.47724345e-02 -2.85045207e-01
-1.07298784e-01 4.81320292e-01 -1.39825189e+00 2.24960923e+00
-2.08795518e-01 5.64422965e-01 7.83096179e-02 -3.59357446e-01
1.23085940e+00 6.73051953e-01 7.15767264e-01 -1.95396379e-01
5.21263957e-01 3.70003819e-01 -3.70468907e-02 -5.21006346e-01
1.08191109e+00 -4.15757418e-01 -9.13086295e-01 4.31519061e-01
2.86264956e-01 -6.37463987e-01 1.79676935e-01 -1.68861106e-01
1.18901253e+00 2.79077977e-01 1.78355843e-01 2.54291128e-02
2.77447492e-01 7.62697458e-02 8.59531760e-01 8.38910401e-01
4.19179834e-02 9.48440969e-01 2.96195865e-01 3.36531579e-01
-6.58285618e-01 -1.04130745e+00 1.84803948e-01 1.67953742e+00
-3.35473597e-01 -1.32600403e+00 -5.80687404e-01 -4.35708985e-02
-3.33287865e-01 5.38574159e-01 -2.86930114e-01 -1.86415523e-01
-1.61818132e-01 -1.80323705e-01 1.71639073e+00 3.54972035e-01
7.82303810e-02 -1.36256242e+00 -2.58516610e-01 5.13390720e-01
-6.74520910e-01 -8.51556897e-01 -4.35389251e-01 4.49487180e-01
-3.31131846e-01 -1.07756114e+00 -3.70771796e-01 -8.87296796e-01
-4.68613476e-01 -4.68436301e-01 1.31784403e+00 -5.75224578e-01
-2.37086162e-01 3.07496578e-01 -9.23302591e-01 -6.79794669e-01
-5.55678964e-01 4.09829170e-01 3.82079571e-01 -1.50913820e-01
1.14267049e-02 -1.03379714e+00 8.69848579e-03 3.66228253e-01
-4.18938071e-01 1.11461341e-01 1.02453180e-01 6.47515297e-01
9.12946284e-01 2.51175106e-01 5.38820088e-01 -5.33449829e-01
5.98804176e-01 -1.84294462e-01 -6.22212067e-02 -3.12732160e-01
4.64609824e-02 -5.65887392e-01 5.93553603e-01 -7.73150742e-01
-5.32088637e-01 1.43772066e-01 -2.97861814e-01 -6.64162636e-01
-2.32256986e-02 5.91371059e-01 -3.66970032e-01 2.65316457e-01
6.82591021e-01 -1.56804785e-01 -6.64054990e-01 -5.86146355e-01
3.00009668e-01 1.20896208e+00 1.28706014e+00 -1.18769491e+00
3.00177008e-01 -2.49693781e-01 -2.09554285e-01 -7.51784027e-01
-6.64461195e-01 -3.71680915e-01 -1.43521234e-01 -6.84689522e-01
6.65228844e-01 -9.84076977e-01 -1.02886713e+00 8.57114494e-01
-7.37661183e-01 -5.91620445e-01 -5.14622331e-01 6.51770532e-01
-1.18792021e+00 -7.62436762e-02 -1.02399278e+00 -9.00864422e-01
-4.86881584e-01 -6.53277159e-01 1.21261942e+00 -2.50016917e-02
-1.06764162e+00 -1.57884136e-01 2.53354967e-01 3.51015478e-01
-1.58883885e-01 2.76161432e-01 7.23153889e-01 -5.73569357e-01
3.75735790e-01 -1.60189018e-01 5.31242847e-01 4.50874120e-01
9.36063230e-02 2.97453910e-01 -1.30253375e+00 2.10744873e-01
-1.23792596e-01 -8.32065165e-01 3.56392145e-01 2.07091704e-01
1.12579381e+00 2.80406773e-01 1.59276083e-01 6.20696485e-01
7.85498917e-01 5.09760678e-01 4.55537260e-01 2.12302819e-01
5.32331765e-01 1.60306036e-01 1.11376405e+00 9.67518926e-01
-8.04787502e-02 9.03348446e-01 1.15289189e-01 4.88734245e-01
-1.13820359e-01 -3.68897408e-01 7.78234661e-01 1.69885218e+00
-4.59393978e-01 -3.36354554e-01 -7.48169541e-01 2.61207283e-01
-1.65085125e+00 -9.72562671e-01 -2.78928220e-01 1.78009975e+00
1.15272784e+00 -3.94383781e-02 3.14348876e-01 9.53173101e-01
7.59490609e-01 -1.19362727e-01 -1.79007381e-01 -3.14442515e-01
-5.38038850e-01 8.17831457e-01 -9.84629840e-02 7.10769668e-02
-1.15574574e+00 9.96459961e-01 6.21482325e+00 1.34148169e+00
-9.67523694e-01 1.68927863e-01 2.97432463e-03 -6.85720265e-01
1.74030773e-02 -3.00472349e-01 -4.82867151e-01 7.37328827e-01
1.00205946e+00 9.59002823e-02 7.80630767e-01 9.64424253e-01
1.19431354e-01 2.44438857e-01 -8.75841022e-01 1.57191038e+00
-6.45861179e-02 -9.47430491e-01 -5.33812582e-01 -2.58798301e-01
7.50859082e-01 -2.17363704e-02 1.71310216e-01 7.04847634e-01
1.44268289e-01 -8.52584243e-01 1.72321427e+00 7.03344107e-01
1.41370869e+00 -8.22069228e-01 4.01183099e-01 1.23280145e-01
-1.39906251e+00 -4.07461822e-02 1.65437818e-01 -2.71483660e-01
8.47446248e-02 2.34757870e-01 -5.51922023e-01 4.93432760e-01
8.02432239e-01 8.54671359e-01 -3.08535516e-01 6.85759425e-01
-2.10102618e-01 1.31633997e+00 -3.32874864e-01 1.89425081e-01
-1.63526610e-01 -2.29951933e-01 6.14925206e-01 1.25343347e+00
9.35583174e-01 -4.92257848e-02 9.66877639e-02 6.67786062e-01
5.00059091e-02 2.13715672e-01 -1.11734964e-01 -5.73402345e-01
6.03162706e-01 1.36188102e+00 -7.63310254e-01 -3.76407981e-01
1.81661770e-01 1.07845390e+00 -3.67446482e-01 1.92225091e-02
-1.27443779e+00 -6.25798225e-01 8.55717301e-01 -2.66549051e-01
2.03581482e-01 -5.44042587e-02 -8.16937163e-02 -9.39300537e-01
-6.00734890e-01 -1.34273362e+00 2.33368427e-01 -1.36675179e+00
-1.35329652e+00 7.90316761e-01 -5.96110582e-01 -1.62304103e+00
-6.31299138e-01 -5.59358895e-01 -2.96188980e-01 5.31813383e-01
-4.08553809e-01 -1.12863767e+00 -3.19880843e-01 6.51736021e-01
1.64219990e-01 -5.49302578e-01 1.56023574e+00 6.97623670e-01
-3.10770273e-01 6.83306932e-01 3.14211100e-02 1.78001523e-01
1.22780812e+00 -1.18204582e+00 -9.56941396e-02 8.10989441e-05
6.44057572e-01 1.89910978e-01 7.88189113e-01 -4.88587946e-01
-9.66557622e-01 -1.19741476e+00 6.05352581e-01 -1.98612705e-01
7.07336485e-01 -5.54140508e-01 -1.92635015e-01 6.23697102e-01
-1.97464287e-01 -3.85294735e-01 1.27903092e+00 2.24429980e-01
-2.85704583e-01 3.13811153e-02 -5.87089479e-01 4.29273367e-01
1.16305673e+00 -1.13142812e+00 -5.77214777e-01 -2.17706949e-01
3.47847223e-01 -6.30390525e-01 -1.10564482e+00 3.07169914e-01
9.94527102e-01 -8.17303717e-01 8.11731279e-01 -3.94618928e-01
4.23570603e-01 -5.86481690e-01 -7.79361665e-01 -1.35468686e+00
-3.92961472e-01 -1.06526983e+00 -1.68512501e-02 1.87500763e+00
3.42389196e-01 2.03116268e-01 5.96113443e-01 -2.78283924e-01
-5.75285852e-01 7.85410553e-02 -7.35397100e-01 -9.53216374e-01
-4.10761833e-01 -1.41063392e+00 6.87943935e-01 1.20361555e+00
4.73654896e-01 4.50787246e-01 -7.54167795e-01 -1.77821755e-01
1.06393188e-01 5.00850379e-01 8.20769429e-01 -1.29723454e+00
-9.62060869e-01 -5.92969179e-01 -5.42581141e-01 -5.29362798e-01
4.13412333e-01 -1.18521726e+00 3.32649738e-01 -5.46089828e-01
-5.38568497e-02 -6.64286852e-01 -6.13969445e-01 6.87543690e-01
3.34354848e-01 1.11065686e+00 2.99296468e-01 6.21521771e-02
-8.18899333e-01 3.65601003e-01 6.65205061e-01 -3.14825505e-01
-2.21133053e-01 2.43301332e-01 -4.24306601e-01 1.06301534e+00
1.00797176e+00 -7.88240850e-01 -2.12114751e-01 1.20282806e-01
8.08014274e-01 1.76173612e-01 1.29440844e-01 -1.40175724e+00
4.57135402e-02 1.48590088e-01 7.28411134e-03 -5.26516676e-01
8.17677915e-01 -4.33348626e-01 8.13161016e-01 -2.54651960e-02
-4.26033497e-01 -2.61627495e-01 3.00941467e-01 6.90868124e-02
-5.02265573e-01 -3.97228658e-01 3.63492280e-01 1.90006569e-01
-6.37241662e-01 -7.90227801e-02 -7.92864680e-01 3.32154095e-01
4.66999084e-01 1.24373706e-02 9.71093401e-02 -4.34002578e-01
-1.19100094e+00 -5.01811206e-01 3.00533921e-01 4.71854270e-01
5.74063174e-02 -1.86477029e+00 -4.29122657e-01 3.89571451e-02
3.21725011e-01 -5.91859639e-01 2.85758227e-01 5.06419003e-01
-2.40468398e-01 2.26903204e-02 -3.25292706e-01 -2.57973582e-01
-1.55958295e+00 -5.95043004e-02 9.61786062e-02 1.10564396e-01
-2.14260936e-01 9.69065547e-01 -3.57608378e-01 -5.39546192e-01
3.16432327e-01 -5.45922279e-01 -1.72833353e-02 3.87660354e-01
3.13119024e-01 5.39691806e-01 -9.16218981e-02 -7.74770379e-01
-3.73502076e-01 5.40023148e-01 1.04288125e+00 -6.20973110e-01
1.14507103e+00 1.31667599e-01 -1.59841701e-01 1.25314069e+00
7.87529230e-01 7.09588528e-01 -5.59958637e-01 3.69970739e-01
-9.23690125e-02 -1.21539153e-01 -1.63989022e-01 -1.06877029e+00
-8.19841802e-01 5.53831279e-01 2.92104304e-01 1.47294208e-01
1.27558601e+00 -3.82026583e-02 8.55716825e-01 3.53414267e-01
4.57673281e-01 -1.49216533e+00 8.00489038e-02 8.80989969e-01
8.04797947e-01 -5.18779635e-01 -7.09102988e-01 -1.60041228e-01
-8.76802385e-01 9.74233210e-01 2.52244413e-01 1.11574493e-01
4.92565989e-01 6.58686697e-01 3.41180563e-01 -1.13899678e-01
-5.63988209e-01 -1.01672590e-01 2.66210645e-01 4.95632201e-01
5.39752305e-01 5.46557784e-01 7.63755441e-02 2.16849351e+00
-1.31171846e+00 1.65866598e-01 3.03805232e-01 5.53157985e-01
-4.73399013e-02 -1.18544376e+00 -4.99268293e-01 2.82948822e-01
-7.32128024e-01 -3.02232891e-01 -6.75213218e-01 4.12347645e-01
9.54392254e-01 1.22812474e+00 -2.75401007e-02 -1.35190129e+00
2.28894413e-01 4.69812959e-01 5.79521120e-01 -5.78765512e-01
-1.03771472e+00 5.95021069e-01 5.05121171e-01 -3.73896509e-01
-3.75459135e-01 -8.12598348e-01 -1.45013082e+00 -1.67756677e-02
-2.91415989e-01 3.91690552e-01 5.18522799e-01 4.60803211e-01
2.55380601e-01 9.11005795e-01 6.46921515e-01 -9.61250186e-01
-1.06349774e-01 -1.31218660e+00 -1.55748439e+00 4.13606882e-01
-3.77528131e-01 -4.83607024e-01 -1.99860007e-01 3.87634367e-01] | [15.927909851074219, 5.307121753692627] |
d30fb973-136b-42a5-9725-87589c976cc7 | grown-up-a-graph-representation-of-a-webpage | 2208.02252 | null | https://arxiv.org/abs/2208.02252v2 | https://arxiv.org/pdf/2208.02252v2.pdf | GROWN+UP: A Graph Representation Of a Webpage Network Utilizing Pre-training | Large pre-trained neural networks are ubiquitous and critical to the success of many downstream tasks in natural language processing and computer vision. However, within the field of web information retrieval, there is a stark contrast in the lack of similarly flexible and powerful pre-trained models that can properly parse webpages. Consequently, we believe that common machine learning tasks like content extraction and information mining from webpages have low-hanging gains that yet remain untapped. We aim to close the gap by introducing an agnostic deep graph neural network feature extractor that can ingest webpage structures, pre-train self-supervised on massive unlabeled data, and fine-tune to arbitrary tasks on webpages effectually. Finally, we show that our pre-trained model achieves state-of-the-art results using multiple datasets on two very different benchmarks: webpage boilerplate removal and genre classification, thus lending support to its potential application in diverse downstream tasks. | ['Huijuan Wang', 'Benedict Yeoh'] | 2022-08-03 | null | null | null | null | ['webpage-object-detection', 'genre-classification'] | ['computer-vision', 'computer-vision'] | [ 3.87572914e-01 1.29337072e-01 -3.05470169e-01 -2.24261090e-01
-1.03043568e+00 -1.02050102e+00 7.35051751e-01 4.33385998e-01
-4.82832372e-01 2.39197597e-01 3.62143040e-01 -7.42943525e-01
-1.66677102e-01 -8.61676872e-01 -8.87552142e-01 -2.46825412e-01
-3.70104313e-01 2.51382947e-01 1.97597519e-01 -3.82161766e-01
1.08385585e-01 2.45598868e-01 -1.40019834e+00 4.27264899e-01
5.83041430e-01 1.03670704e+00 -1.64606452e-01 9.61532116e-01
-2.76808739e-01 6.25260830e-01 -1.14521571e-01 -7.49984205e-01
4.22430128e-01 -2.50356272e-02 -1.04481316e+00 2.49402955e-01
6.73790216e-01 -2.78174490e-01 -5.14711142e-01 1.04684019e+00
2.63662487e-01 -9.49524641e-02 6.83966696e-01 -7.91940451e-01
-6.75587356e-01 7.13320613e-01 -6.62453651e-01 4.91446584e-01
2.16399834e-01 1.31189242e-01 1.65110099e+00 -6.38600171e-01
7.55847633e-01 9.73706245e-01 8.55289519e-01 2.14180723e-01
-1.16825855e+00 -7.24268183e-02 4.00445282e-01 1.18084848e-01
-7.71578848e-01 -4.73360032e-01 6.63338184e-01 -2.50442147e-01
1.07253075e+00 2.69269913e-01 2.22182631e-01 1.57724535e+00
8.65962654e-02 9.14409041e-01 8.52215469e-01 -5.40646017e-01
2.36192849e-02 -2.35826075e-01 5.77187777e-01 1.06570125e+00
3.98856491e-01 -4.26774889e-01 -4.75256979e-01 -7.43025914e-02
2.60512948e-01 -1.33684576e-01 -1.03833020e-01 -3.08343858e-01
-9.10327375e-01 9.95492220e-01 3.47972274e-01 1.42981261e-01
-1.68915421e-01 1.01156998e-02 7.61434555e-01 4.41464543e-01
5.62741399e-01 4.67362642e-01 -7.22459316e-01 3.49656120e-02
-7.62634993e-01 1.56518832e-01 1.22298026e+00 8.24462473e-01
5.29613316e-01 -1.69033170e-01 1.40045598e-01 7.72552013e-01
1.86094597e-01 1.35089800e-01 5.27529359e-01 -6.71752632e-01
7.34730124e-01 7.02014387e-01 -4.59592789e-01 -9.03419435e-01
-3.76854271e-01 -4.54014271e-01 -7.20486641e-01 1.91913292e-01
6.28599584e-01 -2.03095794e-01 -1.02787745e+00 1.46591592e+00
2.61823326e-01 -4.63412642e-01 -1.93981916e-01 7.73080647e-01
8.72323155e-01 4.27272052e-01 1.37063250e-01 2.07460448e-01
1.55559123e+00 -1.00898147e+00 -1.33123875e-01 -6.37201071e-01
5.34109235e-01 -7.14881837e-01 1.35132968e+00 6.23422623e-01
-9.19852972e-01 -1.97428241e-01 -1.14831746e+00 -4.96774226e-01
-7.91314423e-01 -3.44900817e-01 9.94181693e-01 4.77367252e-01
-9.40771043e-01 6.97404385e-01 -6.38534248e-01 -6.35596156e-01
5.21290720e-01 1.78344801e-01 -4.49435592e-01 -6.50291964e-02
-1.09932029e+00 5.31166852e-01 3.47553849e-01 -1.97299406e-01
-7.18843937e-01 -4.60839093e-01 -6.93634033e-01 4.38062817e-01
6.83193862e-01 -7.78070211e-01 1.27289689e+00 -1.03064859e+00
-1.22896993e+00 1.16774726e+00 2.83855408e-01 -6.22802913e-01
3.73322010e-01 -3.10412228e-01 -2.79812992e-01 1.35444134e-01
2.56218389e-02 2.39010587e-01 9.46547210e-01 -1.06325042e+00
-8.02296877e-01 -5.83364010e-01 8.00839365e-02 1.70009315e-01
-9.76920366e-01 -1.28401905e-01 -7.92264342e-01 -4.94055212e-01
-1.79241449e-01 -6.36970997e-01 -2.23509714e-01 -2.72990584e-01
-7.34606385e-01 -3.74395221e-01 6.17546380e-01 -9.90756691e-01
1.03505623e+00 -1.91234231e+00 -7.68145025e-02 2.92033911e-01
7.00518370e-01 2.79792070e-01 -6.38769746e-01 5.02561808e-01
-5.19608967e-02 5.18780708e-01 2.73033138e-02 -5.61529137e-02
2.57747889e-01 -1.30221009e-01 -2.97296226e-01 1.98091403e-01
3.97390336e-01 1.06473482e+00 -9.21040952e-01 -3.94237727e-01
-1.12454146e-01 3.00405800e-01 -4.23686087e-01 3.96484211e-02
-5.58870494e-01 -3.08399320e-01 -5.77452958e-01 5.95144272e-01
3.01115125e-01 -7.87256718e-01 5.08755922e-01 -5.69455102e-02
1.13923989e-01 7.27515936e-01 -8.99555385e-01 1.60724461e+00
-3.51794094e-01 8.09119225e-01 3.18791270e-01 -1.05768847e+00
3.71234119e-01 -2.45168377e-02 5.96050203e-01 -9.38808143e-01
1.84173629e-01 1.50162391e-02 -9.62588117e-02 -5.97006738e-01
3.16140413e-01 5.56544960e-01 -1.36202782e-01 5.31642139e-01
3.12926352e-01 3.54621857e-01 6.81244373e-01 4.37283695e-01
1.75022018e+00 1.15857504e-01 1.30959347e-01 -2.82590926e-01
3.23984802e-01 1.20761909e-01 6.12169690e-02 8.23186159e-01
-5.05475886e-02 6.03023946e-01 7.26167679e-01 -6.37649894e-01
-1.27221131e+00 -1.18342972e+00 1.67878613e-01 1.85264719e+00
-1.93560734e-01 -6.16048813e-01 -8.99570763e-01 -1.22186840e+00
4.69618514e-02 2.30173528e-01 -4.98349756e-01 1.59187876e-02
-6.32301092e-01 -9.80794668e-01 6.06349051e-01 3.56488615e-01
3.11259717e-01 -8.47426176e-01 5.14558330e-02 2.75950849e-01
-1.71066776e-01 -1.37186766e+00 -3.19935352e-01 6.04935646e-01
-7.24067450e-01 -1.32224941e+00 -2.85971105e-01 -1.07956696e+00
3.56081694e-01 5.54124117e-01 1.70753527e+00 1.15518264e-01
-4.31320220e-01 4.81215239e-01 -3.65930706e-01 -1.90399528e-01
-5.42163730e-01 9.17895913e-01 -4.25172657e-01 -3.24424535e-01
7.09249198e-01 -7.71916628e-01 -6.79361701e-01 -2.44760692e-01
-1.06193590e+00 -1.69859007e-01 7.71303654e-01 6.49518251e-01
2.09970653e-01 1.05479941e-01 4.68064725e-01 -1.22431850e+00
1.07104421e+00 -6.42259181e-01 -5.68470538e-01 1.14463441e-01
-9.59124863e-01 2.84989864e-01 9.18211937e-01 -1.64787635e-01
-8.52860749e-01 -1.52629644e-01 -2.43829280e-01 1.77674249e-01
-3.16284508e-01 8.47379863e-01 -1.21828886e-02 7.89318010e-02
9.38725293e-01 1.19255334e-01 -5.73095828e-02 -6.80866659e-01
6.31893396e-01 5.30968130e-01 5.35903275e-01 -3.60249907e-01
9.39088464e-01 3.83548617e-01 3.13566327e-02 -8.77438426e-01
-1.15572989e+00 -5.90956390e-01 -4.42921400e-01 2.09449843e-01
7.60818481e-01 -8.17483008e-01 -6.82530522e-01 3.14532250e-01
-9.05670226e-01 -3.91190797e-01 4.95531373e-02 -3.40666860e-01
-2.60193765e-01 6.57702088e-01 -1.08297336e+00 -2.35144898e-01
-7.67033815e-01 -5.89414477e-01 9.81895208e-01 1.35856569e-01
-9.53477323e-02 -1.26591408e+00 6.21810295e-02 6.58072770e-01
1.88816026e-01 1.28542781e-01 1.25373209e+00 -1.18084145e+00
-5.91916621e-01 -3.69153291e-01 -4.84339088e-01 3.78547996e-01
1.07388750e-01 6.97094128e-02 -9.50326562e-01 -1.09939046e-01
-4.08329010e-01 -6.34528100e-01 1.18923581e+00 9.15335342e-02
1.16486073e+00 -6.24764979e-01 -2.38165155e-01 5.57265222e-01
1.54074252e+00 -5.19759834e-01 3.14692318e-01 7.44745731e-01
7.69963205e-01 6.55146360e-01 -1.90461114e-01 1.74827814e-01
4.16630924e-01 8.55084434e-02 5.47516525e-01 -2.62437433e-01
-2.16756016e-01 -2.55274355e-01 3.03142428e-01 7.58626103e-01
1.82369038e-01 -6.22400820e-01 -9.03049648e-01 5.91705084e-01
-1.77260113e+00 -7.67656088e-01 -3.43629897e-01 1.86781251e+00
1.06929088e+00 7.36417174e-01 1.68641821e-01 -2.61152405e-02
4.76246238e-01 3.46988529e-01 -5.12954414e-01 -3.37896764e-01
-1.80699751e-02 2.58051455e-01 7.04572678e-01 3.11633736e-01
-1.54754925e+00 9.19179559e-01 6.38270426e+00 7.01762557e-01
-9.26814139e-01 -8.88128653e-02 8.57176781e-01 3.33183445e-03
-2.39326090e-01 -3.65650386e-01 -8.64326656e-01 2.92769313e-01
1.18143380e+00 4.62795421e-02 8.95461559e-01 9.51539278e-01
1.45957828e-01 2.90727109e-01 -1.25976968e+00 6.46840215e-01
1.02107361e-01 -1.44589484e+00 9.77779105e-02 1.83249563e-01
7.19690323e-01 7.02766955e-01 9.36348140e-02 4.85986680e-01
8.69278431e-01 -9.45069134e-01 2.05997422e-01 7.09490199e-03
3.60394657e-01 -5.61941385e-01 5.09852707e-01 2.44969234e-01
-8.71458769e-01 1.27958218e-02 -1.50087401e-01 1.35575876e-01
-2.36256003e-01 7.93341577e-01 -9.73211050e-01 3.17428678e-01
7.83653796e-01 4.28971589e-01 -1.03758025e+00 7.86679983e-01
-1.87946051e-01 8.08251441e-01 -3.23658437e-01 -2.27093056e-01
5.09274423e-01 -6.62382096e-02 4.73007292e-01 1.55732441e+00
-2.87005305e-01 -5.20261705e-01 2.54550517e-01 5.32344639e-01
-6.92647159e-01 2.11465880e-01 -6.56648219e-01 -6.05171740e-01
-1.11793447e-02 1.77709329e+00 -8.45129132e-01 -3.63263905e-01
-7.62356341e-01 9.04099762e-01 7.09967136e-01 5.02799928e-01
-5.94738126e-01 -5.40675998e-01 5.76048136e-01 2.83060282e-01
4.87365961e-01 -3.25556219e-01 -5.19365549e-01 -1.37278414e+00
1.53901607e-01 -1.16685379e+00 6.70573115e-01 -3.93240869e-01
-1.86589301e+00 4.58642483e-01 -6.26013517e-01 -6.12536967e-01
-5.83441615e-01 -1.22602272e+00 -4.69338894e-01 4.83902454e-01
-1.78058517e+00 -1.37916207e+00 -2.67450780e-01 5.61419249e-01
6.94623411e-01 -1.28647193e-01 7.24224448e-01 2.46614143e-01
-4.45023030e-01 5.17514706e-01 4.43194479e-01 7.09840417e-01
6.25335217e-01 -1.54369140e+00 8.68127644e-01 9.71713662e-01
5.02836466e-01 6.12277567e-01 5.71308970e-01 -4.47106421e-01
-1.75348687e+00 -1.15928996e+00 7.78629422e-01 -7.08208203e-01
1.14727569e+00 -7.90241778e-01 -8.51337731e-01 8.13499868e-01
4.47208375e-01 -1.75073937e-01 7.04215109e-01 7.53679335e-01
-8.73587668e-01 -7.16924369e-02 -5.60312510e-01 9.65825677e-01
1.15076435e+00 -5.45413852e-01 -4.20995057e-01 7.78290510e-01
6.71064794e-01 1.17782624e-02 -8.31941903e-01 1.22287259e-01
4.09234434e-01 -8.59114647e-01 1.01550257e+00 -1.06506717e+00
9.23273802e-01 4.93245035e-01 1.04193382e-01 -1.07088983e+00
-4.91048515e-01 -7.13851988e-01 -2.64700592e-01 1.29591918e+00
6.59992635e-01 -3.72746497e-01 1.21377110e+00 5.10351837e-01
-1.42744556e-02 -5.89242041e-01 -3.81476611e-01 -5.70733309e-01
2.44588062e-01 -3.39058518e-01 6.03260584e-02 8.48683178e-01
1.01502821e-01 9.92083371e-01 1.21190645e-01 -2.37834185e-01
7.99130023e-01 5.34377843e-02 7.67978609e-01 -1.54418790e+00
-3.76677513e-01 -8.44639421e-01 5.75737767e-02 -8.87637436e-01
2.10657984e-01 -1.13090467e+00 1.62325811e-03 -1.70860493e+00
4.85779196e-01 1.32893873e-02 -3.72648031e-01 5.74448347e-01
-2.92290241e-01 2.32454628e-01 -1.70432687e-01 5.55321574e-02
-9.68433917e-01 1.62908137e-01 9.81540322e-01 -4.08433825e-01
-2.98742042e-03 -1.44558758e-01 -1.16635931e+00 8.17956686e-01
7.25397646e-01 -4.03094053e-01 -3.30673099e-01 -5.65062702e-01
6.60942316e-01 -5.28711557e-01 2.26835996e-01 -6.04078174e-01
-5.23628145e-02 -4.09069732e-02 5.18429875e-01 -1.58162221e-01
-9.17808563e-02 -5.56957662e-01 -9.49077487e-01 1.56747520e-01
-5.33687770e-01 -3.16027761e-03 1.81456923e-01 6.80869162e-01
6.08196408e-02 -3.24587882e-01 4.21121776e-01 -4.83781725e-01
-8.76939416e-01 4.84698355e-01 -6.70415163e-01 5.08997381e-01
4.67286587e-01 1.63309127e-02 -5.39540946e-01 -3.17867994e-01
-4.91609365e-01 4.95024920e-02 2.15810642e-01 7.01507628e-01
2.66545653e-01 -9.33695197e-01 -7.18192995e-01 4.60797548e-02
8.59661698e-02 -1.06382243e-01 -2.55857438e-01 1.66500807e-01
-5.19473672e-01 5.53292811e-01 -1.56074408e-02 -4.02694613e-01
-1.05431914e+00 7.75682211e-01 -7.17029274e-02 -8.24948490e-01
-6.06264114e-01 6.40735805e-01 -4.31571677e-02 -5.37741840e-01
2.81867832e-01 -1.52796097e-02 -1.52296156e-01 -5.22895046e-02
4.22669709e-01 9.73861217e-02 5.74060798e-01 8.70869234e-02
-5.72367683e-02 -1.55231416e-01 -4.71208394e-01 -9.23152827e-03
1.62508965e+00 -1.24389440e-01 -5.98016605e-02 -1.51339611e-02
1.46023715e+00 1.89856589e-02 -1.21691990e+00 -2.84499913e-01
6.01560533e-01 3.98500562e-02 1.14644602e-01 -8.68598938e-01
-8.54987919e-01 6.89662158e-01 2.33376667e-01 7.95971453e-01
9.93631363e-01 -7.89545774e-02 1.05976796e+00 9.62780595e-01
1.58327334e-02 -1.33688140e+00 4.38039511e-01 7.98317373e-01
5.79552233e-01 -1.43450570e+00 3.21472958e-02 -4.11263764e-01
-1.22969903e-01 1.26918471e+00 5.58819771e-01 -3.87485623e-01
5.69650650e-01 4.81416911e-01 -8.51597264e-03 -2.76379824e-01
-9.81644094e-01 -3.22295666e-01 4.38267827e-01 6.34562850e-01
7.19528198e-01 -3.54834497e-01 -2.59926498e-01 4.97589350e-01
-1.57468408e-01 -1.79015785e-01 4.15918261e-01 8.73161495e-01
-5.10310352e-01 -1.08434749e+00 -4.13573645e-02 7.19522536e-01
-9.03793991e-01 -5.91450810e-01 -6.95821047e-01 9.07749772e-01
-4.18250799e-01 8.47506344e-01 -4.23190556e-02 -2.83599466e-01
-8.00730586e-02 3.81538391e-01 3.53375584e-01 -6.92432284e-01
-7.31962204e-01 1.80307135e-01 5.73882520e-01 -5.82460105e-01
2.76740462e-01 -3.98332417e-01 -9.38477576e-01 -3.71660203e-01
-1.24290287e-01 -8.48006681e-02 8.40002537e-01 8.72940361e-01
6.21954083e-01 5.92745781e-01 3.94947171e-01 -5.62160671e-01
-9.83624935e-01 -8.94400835e-01 -2.68227667e-01 8.97239625e-01
2.18918875e-01 1.34909496e-01 -3.02068949e-01 4.44516659e-01] | [9.924803733825684, 7.8775954246521] |
400da096-5923-4b6e-9014-ab389e197bbc | a-novel-approach-for-neuromorphic-vision-data | 2210.15362 | null | https://arxiv.org/abs/2210.15362v1 | https://arxiv.org/pdf/2210.15362v1.pdf | A Novel Approach for Neuromorphic Vision Data Compression based on Deep Belief Network | A neuromorphic camera is an image sensor that emulates the human eyes capturing only changes in local brightness levels. They are widely known as event cameras, silicon retinas or dynamic vision sensors (DVS). DVS records asynchronous per-pixel brightness changes, resulting in a stream of events that encode the brightness change's time, location, and polarity. DVS consumes little power and can capture a wider dynamic range with no motion blur and higher temporal resolution than conventional frame-based cameras. Although this method of event capture results in a lower bit rate than traditional video capture, it is further compressible. This paper proposes a novel deep learning-based compression scheme for event data. Using a deep belief network (DBN), the high dimensional event data is reduced into a latent representation and later encoded using an entropy-based coding technique. The proposed scheme is among the first to incorporate deep learning for event compression. It achieves a high compression ratio while maintaining good reconstruction quality outperforming state-of-the-art event data coders and other lossless benchmark techniques. | ['Abhipraay Nevatia', 'Mansi Sharma', 'Sally Khaidem'] | 2022-10-27 | null | null | null | null | ['data-compression'] | ['time-series'] | [ 6.45323992e-01 -4.74882245e-01 -8.68979916e-02 -2.98447222e-01
-2.75133908e-01 -1.95560575e-01 5.79587579e-01 2.38363832e-01
-6.82224691e-01 7.02875793e-01 5.92117831e-02 2.40498856e-01
1.79359615e-01 -9.03680265e-01 -9.48624134e-01 -8.53281796e-01
4.79422044e-03 -2.77839899e-01 6.55215144e-01 5.45140505e-01
3.66643816e-01 4.12636131e-01 -1.82428324e+00 3.60060155e-01
-1.83203649e-02 1.65597057e+00 5.07915735e-01 8.66892576e-01
7.06577599e-02 1.30618787e+00 -6.04133010e-01 -3.09659123e-01
4.09335680e-02 -3.79191726e-01 -2.13175580e-01 -2.94878781e-01
2.98154473e-01 -9.72068131e-01 -1.18402827e+00 1.02958000e+00
4.74620283e-01 -1.12949915e-01 2.62754261e-01 -1.04783916e+00
-6.71942413e-01 3.63480985e-01 -3.14657629e-01 5.79943955e-01
5.55483937e-01 1.88133627e-01 5.77875853e-01 -3.54348034e-01
6.95380688e-01 8.89528632e-01 5.76925278e-01 6.30668819e-01
-1.06288207e+00 -5.70223987e-01 -5.00073910e-01 6.58551276e-01
-1.00783467e+00 -6.60579979e-01 7.74658918e-01 1.71748754e-02
1.60222292e+00 5.54967485e-02 1.25309265e+00 1.20971751e+00
9.16168213e-01 7.88213789e-01 7.48425484e-01 -1.98559001e-01
8.73475790e-01 -6.54276669e-01 -1.31069466e-01 4.90644813e-01
1.54809773e-01 1.11675054e-01 -1.07471383e+00 9.65256616e-02
9.52257097e-01 6.76551342e-01 -4.54705775e-01 -1.04988158e-01
-1.18931568e+00 3.83004874e-01 3.33407700e-01 4.42550927e-02
-6.55070484e-01 9.20993209e-01 5.09078741e-01 3.89061421e-01
-2.58328170e-01 -1.51425168e-01 1.10864632e-01 -7.69287646e-01
-1.19054151e+00 -3.85156572e-02 8.11396658e-01 7.93266416e-01
3.07883859e-01 2.82091111e-01 -1.65421382e-01 3.89592946e-01
3.68238240e-01 5.37089944e-01 9.45815563e-01 -1.22218704e+00
9.60298255e-02 5.04740655e-01 -1.74942300e-01 -8.23263824e-01
-5.92427421e-03 2.24081710e-01 -8.56416345e-01 2.92951792e-01
-5.52178994e-02 1.52302444e-01 -9.46984828e-01 1.64613390e+00
-2.94813037e-01 4.61065412e-01 1.43875733e-01 6.41878545e-01
5.54874659e-01 1.22486627e+00 -1.09391987e-01 -4.57378626e-01
1.26940346e+00 -3.01574886e-01 -1.14923060e+00 -2.41722390e-01
-2.55821377e-01 -4.06513065e-01 4.29700077e-01 5.35155535e-01
-1.60656130e+00 -3.82893860e-01 -1.53204274e+00 -4.01299059e-01
-2.13429987e-01 -2.85615742e-01 6.35073304e-01 4.44031030e-01
-1.13933885e+00 6.65116370e-01 -1.32144260e+00 -2.00061783e-01
6.32026911e-01 4.99804407e-01 -1.59965515e-01 6.44188970e-02
-1.01651776e+00 6.11989021e-01 3.47205877e-01 -3.52182955e-01
-1.20578408e+00 -4.40086544e-01 -8.56320500e-01 3.37241083e-01
-3.20447147e-01 -4.05409724e-01 1.46603990e+00 -9.03367281e-01
-1.64461005e+00 8.12336564e-01 -3.48119467e-01 -1.11929643e+00
-4.17769589e-02 1.87904075e-01 -4.61585581e-01 7.34413803e-01
-5.92149377e-01 8.38219702e-01 1.03286326e+00 -4.11052853e-01
-4.47990805e-01 -3.13028723e-01 -2.29620159e-01 -1.29669815e-01
-5.29054940e-01 -3.89689691e-02 -3.48001152e-01 -3.89463127e-01
6.33439645e-02 -5.89343607e-01 3.33238870e-01 6.08524919e-01
2.53136128e-01 -9.10051912e-02 1.13485217e+00 -3.47879708e-01
1.25166047e+00 -2.27995944e+00 1.15776993e-02 -5.01896739e-01
1.80239573e-01 3.99577349e-01 1.32540658e-01 3.85663211e-01
2.14658976e-01 -4.88027871e-01 -2.19212487e-01 -3.08182120e-01
-1.65893003e-01 3.62309128e-01 -5.59457302e-01 4.60100770e-01
3.32725868e-02 9.49803412e-01 -8.25386405e-01 -4.18372154e-01
6.69806123e-01 8.10368478e-01 -3.69517803e-01 1.44161850e-01
-2.07164258e-01 -7.49264285e-02 2.03706641e-02 6.70275569e-01
4.42360789e-01 -2.90340781e-01 6.96857348e-02 -4.36825246e-01
-2.78385967e-01 3.51158530e-01 -8.10313702e-01 2.00895667e+00
-1.85285643e-01 1.37954557e+00 -3.85825247e-01 -8.33535314e-01
1.10623240e+00 3.57138991e-01 5.58429182e-01 -1.10885525e+00
3.38649124e-01 2.28365920e-02 -3.42638314e-01 -3.97799402e-01
7.10729897e-01 -4.13520299e-02 3.50840129e-02 1.47803217e-01
4.18896675e-01 2.12761238e-01 3.37117724e-02 5.21225482e-02
1.63207543e+00 -1.09322272e-01 2.31532440e-01 4.73755926e-01
-6.70832694e-02 -5.51145911e-01 5.84895790e-01 6.75857902e-01
-4.96777087e-01 5.94400883e-01 2.53173411e-01 -6.23181701e-01
-1.25206411e+00 -1.48595011e+00 -3.15199494e-01 3.51663828e-01
5.21398008e-01 -4.63824719e-01 -6.20804608e-01 4.23920453e-01
-2.27149889e-01 6.78140402e-01 -1.88860178e-01 -6.44715667e-01
-2.70730466e-01 -5.52008271e-01 7.60179996e-01 4.49364483e-01
9.64168787e-01 -1.27173805e+00 -1.64255679e+00 4.79404420e-01
-1.23013191e-01 -1.16708660e+00 -6.60856217e-02 4.73256618e-01
-1.16308343e+00 -6.71131253e-01 -5.59014201e-01 -1.01261866e+00
1.53844088e-01 -5.29179908e-02 8.11699569e-01 -8.29044878e-01
-6.30261242e-01 4.48632032e-01 -2.31823757e-01 -4.12106693e-01
-9.08730030e-02 -6.28164411e-01 -2.22925991e-01 -1.29470136e-02
8.07371140e-01 -9.49450612e-01 -1.06851840e+00 -4.01170224e-01
-1.25131881e+00 1.13110933e-02 5.45062959e-01 5.80473721e-01
1.08507109e+00 1.71712279e-01 1.76625088e-01 -2.23650783e-01
3.52083862e-01 -4.75568205e-01 -6.20385885e-01 2.50025857e-02
-3.53818893e-01 2.42389031e-02 7.76464224e-01 -6.76001787e-01
-9.75892127e-01 2.25942001e-01 9.22411457e-02 -4.95128274e-01
-1.86406150e-02 1.07095361e-01 2.36170441e-01 -2.48214692e-01
5.24105489e-01 7.73483038e-01 -1.53306216e-01 -1.72471821e-01
-1.65301800e-01 8.16649854e-01 9.63541150e-01 2.22875968e-01
-1.56245723e-01 7.67724693e-01 -1.78729240e-02 -7.01687694e-01
3.84137146e-02 -2.32668757e-01 -6.19360059e-02 -4.21651602e-01
7.11656928e-01 -1.03977716e+00 -9.02312934e-01 1.05381691e+00
-1.35384429e+00 -2.35701457e-01 -5.02013505e-01 5.80303550e-01
-8.68865669e-01 1.99846402e-01 -1.14899182e+00 -6.68476284e-01
-4.55760628e-01 -8.83705795e-01 9.61221457e-01 6.43059909e-01
1.05801016e-01 -6.64794803e-01 2.82979995e-01 -2.83290118e-01
4.98590112e-01 3.80468339e-01 5.00417531e-01 1.99553698e-01
-9.01223838e-01 -4.94015813e-01 -5.21610454e-02 2.55315661e-01
-7.60162808e-03 -1.73624516e-01 -9.00938094e-01 -2.04640821e-01
5.07122219e-01 -4.78575677e-01 1.11489701e+00 6.04062140e-01
1.55216873e+00 -3.47903907e-01 -9.11036879e-02 1.02007234e+00
1.93223226e+00 7.37699986e-01 1.29607522e+00 2.14775711e-01
2.09339708e-01 -4.11106229e-01 -1.73999161e-01 1.00034666e+00
2.89689600e-01 3.72906655e-01 6.45483613e-01 3.71813506e-01
-3.48148614e-01 -3.63969415e-01 7.62917399e-01 8.37499499e-01
2.45011538e-01 -6.16385162e-01 -4.69880491e-01 5.83991945e-01
-1.69670296e+00 -1.48617899e+00 3.31875563e-01 2.30989552e+00
1.17617285e+00 1.10039927e-01 -3.56753588e-01 2.88663745e-01
8.13262939e-01 4.00408864e-01 -1.14850688e+00 -7.56260455e-01
-2.82750607e-01 4.62183475e-01 8.47144723e-01 -2.48630539e-01
-7.70597816e-01 4.39893484e-01 6.47935486e+00 3.54126126e-01
-1.39196444e+00 -1.36447653e-01 2.38607004e-01 -6.70094967e-01
1.57766230e-02 -1.21587060e-01 -6.33844614e-01 1.06911516e+00
1.70692515e+00 -1.80945501e-01 7.02837825e-01 5.78005910e-01
6.22090995e-02 -4.22770709e-01 -1.17427742e+00 1.81314015e+00
2.55802780e-01 -1.58716214e+00 1.85625911e-01 -7.89194778e-02
5.56806386e-01 1.83788195e-01 1.16670035e-01 -2.28037715e-01
-1.33458927e-01 -8.35828722e-01 9.27039027e-01 9.27728355e-01
1.05500114e+00 -5.44220865e-01 4.54278678e-01 3.36348042e-02
-9.89901543e-01 -3.30842108e-01 -7.75658906e-01 -3.04994226e-01
3.83867651e-01 7.29212880e-01 -1.85099959e-01 -2.31265798e-01
1.04245055e+00 1.03425586e+00 -1.36846930e-01 1.27732277e+00
3.09989780e-01 6.96788549e-01 -3.56070668e-01 -2.77727425e-01
6.16376894e-03 3.00527185e-01 5.58512092e-01 1.15893495e+00
6.31054282e-01 2.17815578e-01 -5.40249586e-01 7.12257206e-01
-4.62332875e-01 -7.59596348e-01 -7.37630725e-01 -2.91264981e-01
7.94265091e-01 7.77127445e-01 -4.58182275e-01 -3.94050837e-01
-7.06547618e-01 1.53132200e+00 -5.03746644e-02 1.57266080e-01
-8.66751492e-01 -8.22392046e-01 3.74484152e-01 -6.40586838e-02
5.98922253e-01 -2.88196832e-01 1.07125761e-02 -1.06347954e+00
8.82910043e-02 -4.31326807e-01 5.70717594e-03 -1.09565473e+00
-7.21800864e-01 2.03236386e-01 -3.34987462e-01 -1.24873257e+00
-1.51613265e-01 -5.63245773e-01 -3.15696269e-01 2.34290943e-01
-1.57399189e+00 -5.55380106e-01 -4.73631561e-01 9.10036266e-01
4.21298474e-01 -1.59746021e-01 7.89844275e-01 3.20863843e-01
-2.69492626e-01 4.08387363e-01 4.22052115e-01 -4.01927903e-03
5.95967770e-01 -1.02342260e+00 2.09484920e-01 8.83629739e-01
1.16289377e-01 3.62426877e-01 4.88509923e-01 -4.70286906e-01
-2.16674113e+00 -1.08163822e+00 6.07114613e-01 2.80144840e-01
3.90785217e-01 -2.02095062e-01 -7.09908783e-01 5.47978878e-01
3.54103327e-01 2.15072632e-01 6.74762189e-01 -9.55069423e-01
-2.80436426e-01 -6.01221740e-01 -1.35270441e+00 3.15587819e-01
7.38686979e-01 -8.74017179e-01 -7.45900393e-01 -4.18434069e-02
4.56470311e-01 -3.64031434e-01 -8.35509241e-01 -1.32449698e-02
7.40920305e-01 -1.12566638e+00 8.00377548e-01 1.76234588e-01
6.39788628e-01 -2.11333215e-01 -3.46901268e-01 -6.21655345e-01
-3.41595352e-01 -7.40172625e-01 -9.33723271e-01 9.17100430e-01
-3.06608856e-01 -3.36623877e-01 5.90948939e-01 5.57742894e-01
6.30552322e-03 -3.72996390e-01 -1.33117962e+00 -7.80405939e-01
-7.03872144e-01 -3.30138385e-01 2.58717716e-01 2.82280922e-01
1.54305950e-01 -1.09419756e-01 -4.74564105e-01 -7.10543022e-02
8.71806145e-01 -4.64140475e-02 -2.01544479e-01 -9.37039614e-01
-1.91651881e-01 -1.26710773e-01 -1.24144030e+00 -1.20743191e+00
-2.65269578e-01 -7.03033566e-01 2.67831683e-01 -1.34292436e+00
4.93337661e-01 3.37649733e-01 -6.20039046e-01 3.58781219e-01
5.28183222e-01 4.18100387e-01 3.19170356e-02 2.76789665e-01
-9.91335213e-01 7.12232590e-01 6.95498049e-01 -1.25965372e-01
-9.51840654e-02 -7.13359058e-01 -5.41378185e-03 4.47976321e-01
7.28835821e-01 -4.48294938e-01 -4.35108304e-01 -7.22477734e-01
3.33162367e-01 3.85515213e-01 9.07005906e-01 -1.68822861e+00
9.95470762e-01 2.51712501e-01 7.89617062e-01 -5.59971809e-01
4.38808173e-01 -8.01899970e-01 5.01112103e-01 6.08301759e-01
-4.25136507e-01 1.44035563e-01 2.95062333e-01 9.46941137e-01
-4.12412286e-01 -2.62399673e-01 9.23957229e-01 -2.30285645e-01
-1.05774188e+00 3.58880848e-01 -8.23444605e-01 -2.45469823e-01
1.04375625e+00 -6.89833403e-01 -4.05646622e-01 -3.37229222e-01
-2.16348395e-01 -3.90537530e-01 6.87871993e-01 2.05487058e-01
1.18165612e+00 -1.38218975e+00 -1.32358065e-02 3.18602026e-01
-1.40481710e-01 -2.02450469e-01 3.45598578e-01 2.04783216e-01
-7.45611191e-01 3.89224350e-01 -6.73818290e-01 -6.85727179e-01
-9.58146334e-01 3.98828030e-01 1.00172319e-01 2.30401129e-01
-9.39115822e-01 7.23745167e-01 -4.19952452e-01 8.20550859e-01
5.17564714e-01 -3.21254969e-01 1.08483501e-01 -2.29619533e-01
1.00790560e+00 3.83008003e-01 -1.69713810e-01 -1.01344816e-01
-3.60693365e-01 4.29034889e-01 5.84061258e-02 -3.56427491e-01
1.45509529e+00 -2.21715122e-01 -9.32383388e-02 7.33216405e-01
1.41892207e+00 -7.11756885e-01 -1.79745197e+00 -5.21822274e-02
-3.87876332e-01 -3.24488968e-01 6.07240796e-01 -7.24035442e-01
-1.14452159e+00 9.86113012e-01 1.21304619e+00 6.41151667e-02
1.80046034e+00 -3.13734859e-01 1.61091876e+00 4.86444920e-01
3.60429674e-01 -1.30771863e+00 2.62251228e-01 3.22598785e-01
5.07037222e-01 -8.34883988e-01 -2.49406070e-01 4.98862773e-01
-1.96208969e-01 1.46743608e+00 9.53106359e-02 -3.49962711e-01
4.73932654e-01 6.82351112e-01 -7.30464518e-01 -2.98101101e-02
-1.08588290e+00 1.74039140e-01 -3.87535483e-01 5.37075877e-01
1.45681828e-01 -1.24017626e-01 -1.13092624e-01 2.29818553e-01
-1.98670495e-02 8.18825483e-01 6.22599661e-01 1.10256195e+00
-6.35980070e-01 -6.92236543e-01 -1.04908962e-02 6.32905722e-01
-6.11134052e-01 -1.32545531e-01 5.36135100e-02 -8.13992247e-02
1.12968281e-01 7.69482195e-01 7.58032322e-01 -4.32539970e-01
-4.02385220e-02 7.52872080e-02 8.62721503e-01 2.87325457e-02
-2.63332110e-02 -3.05920660e-01 -5.50641060e-01 -9.13339794e-01
-5.21055162e-01 -6.66886568e-01 -1.43898654e+00 -4.26575661e-01
1.70940623e-01 -5.63261926e-01 9.73534226e-01 4.62756366e-01
5.34560680e-01 6.04761362e-01 5.21717072e-01 -5.61482489e-01
-1.31362036e-01 -3.81897599e-01 -5.24864256e-01 4.74011391e-01
7.27578044e-01 -2.11987302e-01 -3.56223822e-01 7.17769682e-01] | [8.740516662597656, -1.1224652528762817] |
68842705-e7f2-490b-b616-0b0035871219 | unification-based-reconstruction-of | 2004.00061 | null | https://arxiv.org/abs/2004.00061v2 | https://arxiv.org/pdf/2004.00061v2.pdf | Unification-based Reconstruction of Multi-hop Explanations for Science Questions | This paper presents a novel framework for reconstructing multi-hop explanations in science Question Answering (QA). While existing approaches for multi-hop reasoning build explanations considering each question in isolation, we propose a method to leverage explanatory patterns emerging in a corpus of scientific explanations. Specifically, the framework ranks a set of atomic facts by integrating lexical relevance with the notion of unification power, estimated analysing explanations for similar questions in the corpus. An extensive evaluation is performed on the Worldtree corpus, integrating k-NN clustering and Information Retrieval (IR) techniques. We present the following conclusions: (1) The proposed method achieves results competitive with Transformers, yet being orders of magnitude faster, a feature that makes it scalable to large explanatory corpora (2) The unification-based mechanism has a key role in reducing semantic drift, contributing to the reconstruction of many hops explanations (6 or more facts) and the ranking of complex inference facts (+12.0 Mean Average Precision) (3) Crucially, the constructed explanations can support downstream QA models, improving the accuracy of BERT by up to 10% overall. | ['André Freitas', 'Mokanarangan Thayaparan', 'Marco Valentino'] | 2020-03-31 | null | https://aclanthology.org/2021.eacl-main.15 | https://aclanthology.org/2021.eacl-main.15.pdf | eacl-2021-2 | ['science-question-answering'] | ['miscellaneous'] | [ 3.78241912e-02 1.13856983e+00 -3.70206356e-01 -3.52251351e-01
-1.15955472e+00 -7.55925179e-01 1.00184035e+00 6.99105144e-01
-5.28686568e-02 9.59504187e-01 8.10606122e-01 -8.53332520e-01
-1.03750312e+00 -7.07409263e-01 -9.23174143e-01 -3.52476984e-01
2.64285564e-01 1.13753235e+00 5.21604419e-01 -3.07793736e-01
8.99980903e-01 2.04589307e-01 -1.83440363e+00 6.63094759e-01
1.50096262e+00 5.90175867e-01 1.06279761e-01 7.09910274e-01
-5.39262414e-01 1.07760417e+00 -4.34244782e-01 -1.04017520e+00
-4.18044567e-01 -4.86041188e-01 -1.64706421e+00 -5.89928508e-01
5.14429927e-01 3.19069475e-01 1.15991741e-01 6.45874023e-01
8.38596821e-02 7.77820647e-02 7.54425466e-01 -1.02410662e+00
-7.40459323e-01 1.22505260e+00 2.55801648e-01 3.88682485e-01
6.46665335e-01 -3.59080195e-01 1.49391246e+00 -6.16296709e-01
9.05456483e-01 1.34042168e+00 3.74124616e-01 5.33604264e-01
-1.06860888e+00 -2.14003995e-01 2.57296443e-01 8.38275194e-01
-7.07197964e-01 -2.58248806e-01 3.99551958e-01 -2.89010704e-01
1.21358883e+00 7.26728082e-01 4.64515924e-01 7.88912416e-01
-1.76433416e-03 4.59251851e-01 8.57078612e-01 -6.63563848e-01
3.59439701e-01 2.67255872e-01 5.86722732e-01 5.59829414e-01
4.84664857e-01 -4.28965032e-01 -9.08030570e-01 -4.69543666e-01
2.06238747e-01 -4.72754806e-01 -3.27369988e-01 -7.94395357e-02
-1.22864830e+00 9.56624806e-01 5.91003001e-01 4.93453026e-01
-4.89973903e-01 1.64546706e-02 2.44564638e-01 9.70880091e-02
2.98912287e-01 1.04870605e+00 -8.19213986e-01 -1.76005647e-01
-6.90466583e-01 5.96595109e-01 9.52795923e-01 8.69711459e-01
5.44680536e-01 -5.80316901e-01 -7.89449215e-02 5.53429842e-01
2.24879608e-01 1.11424245e-01 3.83622915e-01 -1.46648872e+00
4.65999246e-01 8.94354761e-01 9.62800253e-03 -8.93174946e-01
-3.92400235e-01 -4.50102717e-01 -3.73664826e-01 -5.56971967e-01
5.82451046e-01 3.28790247e-01 -4.72086638e-01 1.58285487e+00
5.12143016e-01 5.26666082e-02 3.51232678e-01 7.81298339e-01
1.25933599e+00 4.49909896e-01 2.48392850e-01 -2.06239283e-01
1.73390698e+00 -1.02468419e+00 -7.26494908e-01 -1.89558104e-01
4.45349395e-01 -5.82836211e-01 9.58920300e-01 5.46463192e-01
-1.14018452e+00 -5.08997977e-01 -6.95494771e-01 -6.10420406e-01
-3.89684469e-01 -1.64954588e-01 1.02924883e+00 3.33878726e-01
-1.19974208e+00 5.21990716e-01 -2.07141474e-01 -3.41131806e-01
1.21530928e-01 2.48066559e-01 -1.00118592e-01 -2.16900200e-01
-1.27896607e+00 1.05227911e+00 3.88888329e-01 -4.04042184e-01
-3.10764998e-01 -1.03551745e+00 -5.38191080e-01 5.56292832e-01
5.09279549e-01 -1.20193326e+00 8.81769121e-01 -4.69588101e-01
-1.18936300e+00 6.02364540e-01 -7.91314840e-01 -5.39405704e-01
2.99419969e-01 -3.13379288e-01 -3.92614752e-01 5.81940472e-01
6.31114841e-01 6.98901296e-01 5.55769615e-02 -1.40633202e+00
-5.64447105e-01 -5.04344225e-01 3.94479841e-01 7.72533342e-02
-1.41010173e-02 -2.95727760e-01 -4.10826743e-01 -9.97745097e-02
5.34841001e-01 -4.69137222e-01 -1.05533339e-01 -5.47186792e-01
-4.48437482e-01 -9.16710675e-01 3.18071358e-02 -6.11868203e-01
1.10289454e+00 -1.63590956e+00 5.04662156e-01 1.96589142e-01
5.44347227e-01 -3.61989021e-01 1.56401321e-01 5.07326305e-01
-6.02725968e-02 2.33996734e-01 -2.59431481e-01 -5.18264398e-02
5.41917086e-02 5.79888225e-01 -8.36586952e-01 -2.11728439e-01
1.37016430e-01 1.18743312e+00 -1.10995460e+00 -5.68846524e-01
6.69976277e-03 1.83483005e-01 -7.49428153e-01 -1.55340478e-01
-6.30346119e-01 3.62776130e-01 -5.97543895e-01 4.79461819e-01
2.09266663e-01 -5.99232793e-01 4.24882770e-01 1.24455877e-01
-7.49471411e-02 1.07387900e+00 -7.80942619e-01 1.68995023e+00
-1.93666622e-01 4.64551479e-01 -6.08782709e-01 -1.03380859e+00
8.08959067e-01 3.92707020e-01 8.74324664e-02 -5.60889602e-01
-2.38849461e-01 4.56770301e-01 -6.86156154e-02 -6.18974626e-01
4.87432867e-01 -9.80668589e-02 1.81818172e-01 3.76588494e-01
1.24283716e-01 -1.64276779e-01 4.39882308e-01 8.84070754e-01
1.17458868e+00 3.66085465e-03 1.82986751e-01 -7.08473444e-01
7.74399996e-01 5.98954022e-01 3.03665578e-01 8.63266587e-01
3.11124355e-01 3.02374631e-01 9.50303435e-01 -6.49304748e-01
-8.23827565e-01 -8.72153997e-01 -2.68611073e-01 7.98890293e-01
2.76760042e-01 -7.70068884e-01 -6.46205544e-01 -7.48222113e-01
-1.95068046e-02 1.61453283e+00 -7.40958631e-01 1.02762610e-01
-4.13449734e-01 -3.23826641e-01 5.56698203e-01 3.32243115e-01
2.31040820e-01 -9.38618541e-01 -6.28268182e-01 2.53531467e-02
-9.44534123e-01 -9.62924480e-01 4.98093963e-01 2.19099194e-01
-1.25303519e+00 -1.42545676e+00 -1.20123595e-01 -2.07717434e-01
4.34936583e-01 1.91735491e-01 1.69337416e+00 5.66139936e-01
1.41288638e-01 3.26473951e-01 -3.91262174e-01 -4.10076827e-01
-3.94243896e-01 1.71581686e-01 -1.52451292e-01 -7.71245956e-01
5.74732661e-01 -4.85479087e-01 -4.99801785e-01 5.73961772e-02
-7.53345311e-01 -9.24515445e-03 3.46269816e-01 7.59802580e-01
4.74338114e-01 -1.97013482e-01 7.75775671e-01 -1.31373811e+00
6.12600327e-01 -7.25420237e-01 -3.58057439e-01 5.36399066e-01
-1.27765334e+00 4.76589501e-01 5.16708434e-01 5.96283227e-02
-1.38522327e+00 -5.08497477e-01 -2.23349899e-01 3.76311302e-01
-3.27675223e-01 4.70343709e-01 3.02177807e-03 4.64397579e-01
9.65121388e-01 -7.82539025e-02 -3.94271106e-01 -4.03984100e-01
9.84077394e-01 2.33693436e-01 7.06369817e-01 -8.28200400e-01
5.86692214e-01 3.94137502e-01 1.03678502e-01 -3.54789019e-01
-1.23146343e+00 -6.08756363e-01 -5.92070997e-01 -9.33310986e-02
8.08452606e-01 -6.51341081e-01 -1.08671284e+00 -7.93753207e-01
-1.45177424e+00 3.46348345e-01 -3.24377149e-01 5.53690314e-01
-5.11029482e-01 5.39946020e-01 -4.80012864e-01 -6.35185540e-01
-2.66278177e-01 -8.87241900e-01 9.47879732e-01 3.15705925e-01
-7.70163476e-01 -1.04560792e+00 1.43479779e-01 1.21068525e+00
5.56826666e-02 -1.13369653e-03 1.68308032e+00 -9.21380341e-01
-7.88201630e-01 2.96527773e-01 -1.11204214e-01 -5.63510001e-01
-4.32068527e-01 -1.58842146e-01 -1.08432782e+00 6.65129781e-01
9.00339987e-03 -1.94885060e-01 9.52620685e-01 3.44977885e-01
9.70450997e-01 -4.23606634e-01 -4.23512667e-01 -1.08831063e-01
1.26416683e+00 -1.69973001e-01 7.51431286e-01 6.87129080e-01
3.35853815e-01 1.00190377e+00 5.00894725e-01 -1.26021821e-02
8.42781305e-01 3.06949854e-01 5.33592224e-01 3.34748983e-01
-1.55591995e-01 -3.83472621e-01 -1.32692277e-01 7.70660460e-01
-3.98234308e-01 -1.60200298e-01 -9.10051167e-01 7.92118490e-01
-2.08581591e+00 -1.10338116e+00 -1.00263727e+00 1.80940044e+00
6.16821170e-01 4.17509340e-02 -2.12984011e-01 2.08347857e-01
2.39554778e-01 -4.25016403e-01 -4.22087103e-01 -5.71363270e-01
-3.73478025e-01 3.14630568e-01 6.52713887e-03 8.67498934e-01
-2.66032249e-01 9.37393367e-01 6.67874432e+00 6.09553695e-01
-2.53102750e-01 1.37804747e-01 5.96754551e-01 1.97467119e-01
-1.24400306e+00 4.91871268e-01 -4.72944826e-01 7.19505399e-02
1.14010763e+00 -1.11111671e-01 3.06541473e-01 7.17858732e-01
-1.77538499e-01 -3.81179452e-01 -1.01720083e+00 2.55893171e-01
3.53426561e-02 -1.87310874e+00 4.95433956e-01 -2.44548135e-02
6.55719936e-01 -2.27407575e-01 -3.07962000e-01 2.65325662e-02
5.15577316e-01 -8.58108222e-01 7.98358500e-01 5.73595345e-01
3.56676131e-01 -7.59154081e-01 9.09412622e-01 4.81924921e-01
-6.15638793e-01 -2.55563170e-01 -5.89024723e-01 -1.99413553e-01
1.58292845e-01 8.44643354e-01 -8.05315435e-01 1.27628374e+00
7.83834338e-01 4.21908051e-01 -6.47516906e-01 6.17037416e-01
-8.52416992e-01 8.43578935e-01 -6.52360469e-02 -6.74227923e-02
1.76205888e-01 -1.05664045e-01 3.68366510e-01 9.47795212e-01
3.74422580e-01 6.10895634e-01 -7.65277445e-01 9.99230683e-01
-1.11321822e-01 3.13715152e-02 -2.65607804e-01 1.36148646e-01
5.24795413e-01 1.06952918e+00 -8.63491654e-01 -7.75527358e-01
-1.48643879e-02 8.22501779e-01 5.01319170e-01 4.63615954e-02
-6.78964317e-01 6.97888387e-03 3.29036236e-01 -1.69485703e-01
9.44446027e-02 3.75940770e-01 -8.81573558e-01 -1.10952437e+00
-3.48209403e-02 -7.49554873e-01 8.70051026e-01 -1.06995606e+00
-1.18930566e+00 6.74079716e-01 5.00235930e-02 -4.64553535e-01
-4.47498918e-01 -4.22982782e-01 -4.56034601e-01 6.99647486e-01
-1.58153629e+00 -8.64456296e-01 -8.28604475e-02 2.39498794e-01
5.94993055e-01 9.79526192e-02 1.01285160e+00 4.83490713e-02
-4.61179130e-02 7.02283811e-03 -1.52668029e-01 -6.50129378e-01
6.04444802e-01 -1.67240357e+00 2.43267208e-01 6.74723029e-01
5.31556427e-01 1.18557370e+00 1.03220594e+00 -7.00547576e-01
-1.21364629e+00 -3.76904279e-01 1.89733028e+00 -1.09099877e+00
5.54136097e-01 1.30748212e-01 -1.12580466e+00 4.46333170e-01
3.73270333e-01 -8.08360219e-01 8.56590569e-01 8.11685026e-01
-6.16042197e-01 2.73532361e-01 -1.06117582e+00 5.21751463e-01
1.13240588e+00 -3.56173009e-01 -1.28549600e+00 6.49919450e-01
9.86463845e-01 -1.56003207e-01 -7.82006741e-01 2.80981898e-01
3.23212385e-01 -1.22513545e+00 1.10935342e+00 -8.31267357e-01
1.17875266e+00 -4.29928839e-01 1.31942891e-03 -9.88635123e-01
-5.27450442e-01 -2.86004633e-01 -1.96018502e-01 1.08332229e+00
8.30982447e-01 -5.05785167e-01 5.31088233e-01 9.45039690e-01
-2.62535512e-01 -6.08077109e-01 -1.13179827e+00 -1.56027466e-01
1.50248289e-01 -4.60532635e-01 7.75656044e-01 1.02560782e+00
4.36844349e-01 7.76484311e-01 1.08669393e-01 3.42669517e-01
5.76689601e-01 3.98829818e-01 4.85046536e-01 -1.63572562e+00
-2.44929865e-01 -4.10015196e-01 7.94480816e-02 -1.02628338e+00
1.72722623e-01 -1.02321517e+00 -1.51838064e-01 -2.03146791e+00
3.11890692e-01 -2.86002100e-01 -6.71752319e-02 2.08345890e-01
-6.23414457e-01 -9.49713439e-02 -1.11210145e-01 3.26166481e-01
-7.97324598e-01 4.48858887e-01 1.04208267e+00 2.76717305e-01
4.13494676e-01 -3.98520589e-01 -1.30648780e+00 6.62554324e-01
4.98147607e-01 -4.90434945e-01 -4.84269708e-01 -4.71556455e-01
6.63145959e-01 3.46976370e-02 3.94472301e-01 -5.33468127e-01
5.16516805e-01 2.39113793e-02 2.58346409e-01 -4.66965318e-01
1.48974150e-01 -5.88382483e-01 2.33054429e-01 5.90619862e-01
-6.84388936e-01 1.11379616e-01 1.11765847e-01 8.04359555e-01
-1.31584346e-01 -3.22329700e-01 2.40681320e-02 -1.80121243e-01
-5.06822288e-01 -6.12964571e-01 -1.90384910e-01 9.46986377e-02
4.45327282e-01 5.28796129e-02 -7.88137496e-01 -3.22138518e-01
-4.25985008e-01 2.50963867e-01 3.45581956e-02 4.22873378e-01
5.10029674e-01 -9.09001410e-01 -6.39384151e-01 -4.16916937e-01
2.20928237e-01 -8.75415206e-02 2.76219875e-01 6.85362518e-01
-3.57629210e-01 1.11287546e+00 2.00954691e-01 -4.51189995e-01
-1.13366890e+00 5.96184492e-01 -1.54417440e-01 -3.20956677e-01
-7.06793725e-01 9.26499784e-01 -9.39820614e-03 -6.04584098e-01
-1.39520830e-02 -2.85207510e-01 -5.89043796e-01 -9.46763083e-02
4.90245134e-01 5.51769674e-01 8.48305523e-02 -2.43631706e-01
-4.08797055e-01 3.85338545e-01 7.04874173e-02 -9.36746150e-02
1.49376607e+00 -1.44992188e-01 -5.79991281e-01 4.61869448e-01
3.02224964e-01 2.63122320e-01 -4.56468046e-01 -1.77165627e-01
5.94491661e-01 -2.18291834e-01 -3.27969313e-01 -1.33987021e+00
-3.13281506e-01 7.51240432e-01 -1.70253232e-01 4.26726460e-01
6.81742132e-01 7.41389155e-01 4.79886770e-01 5.22360802e-01
1.32325828e-01 -7.46476710e-01 -1.22359537e-01 3.83614510e-01
8.79275441e-01 -1.00445712e+00 -1.06300705e-03 -8.11738431e-01
-7.28867471e-01 1.21465158e+00 2.72123992e-01 4.43652987e-01
-9.95180830e-02 -3.25524658e-01 1.18261697e-02 -9.51624393e-01
-1.06995189e+00 -6.77063093e-02 5.90103984e-01 9.75052193e-02
5.99005640e-01 6.45885393e-02 -6.97014332e-01 9.22658086e-01
-5.00756323e-01 -2.65508235e-01 4.22233850e-01 1.63634896e-01
-6.85446680e-01 -1.16047728e+00 -3.82854611e-01 3.54589552e-01
-3.40677977e-01 -3.44039083e-01 -8.97151887e-01 6.43971741e-01
9.99334157e-02 1.53807521e+00 -1.77493379e-01 -8.76800250e-03
2.10848883e-01 4.65351909e-01 1.97061881e-01 -4.03055608e-01
-8.21352184e-01 -3.07394624e-01 3.35345507e-01 -5.32374024e-01
-5.75469077e-01 -4.88544792e-01 -1.54608238e+00 -4.29163456e-01
-2.86107540e-01 9.85539913e-01 6.43026352e-01 1.46482992e+00
6.91470623e-01 7.08477616e-01 -1.76197410e-01 1.32097527e-01
-2.23799869e-01 -8.04082274e-01 -4.93331328e-02 3.74924958e-01
-1.07454084e-01 -7.25060701e-01 -4.68592793e-01 8.10671151e-02] | [10.919465065002441, 7.878127574920654] |
fe64d4ea-c3fa-4beb-9068-e6c32615c0ee | clinical-temporal-relation-extraction-with | 2012.08790 | null | https://arxiv.org/abs/2012.08790v1 | https://arxiv.org/pdf/2012.08790v1.pdf | Clinical Temporal Relation Extraction with Probabilistic Soft Logic Regularization and Global Inference | There has been a steady need in the medical community to precisely extract the temporal relations between clinical events. In particular, temporal information can facilitate a variety of downstream applications such as case report retrieval and medical question answering. Existing methods either require expensive feature engineering or are incapable of modeling the global relational dependencies among the events. In this paper, we propose a novel method, Clinical Temporal ReLation Exaction with Probabilistic Soft Logic Regularization and Global Inference (CTRL-PG) to tackle the problem at the document level. Extensive experiments on two benchmark datasets, I2B2-2012 and TB-Dense, demonstrate that CTRL-PG significantly outperforms baseline methods for temporal relation extraction. | ['Wei Wang', 'Peipei Ping', 'Yizhou Sun', 'Kai-Wei Chang', 'J. Harry Caufield', 'Rujun Han', 'Yu Yan', 'Yichao Zhou'] | 2020-12-16 | null | null | null | null | ['temporal-relation-extraction'] | ['natural-language-processing'] | [ 4.21334803e-02 -2.17195172e-02 -7.02242732e-01 -5.67903936e-01
-1.14955521e+00 -3.09731811e-01 5.50684035e-01 8.92059982e-01
-1.02284372e-01 9.48016763e-01 5.01893520e-01 -5.53888500e-01
-7.41812289e-01 -6.19734585e-01 -3.95400107e-01 -5.25821269e-01
-5.16621649e-01 5.13911545e-01 2.54585445e-01 1.16783284e-01
-3.46251100e-01 4.26919222e-01 -5.42038381e-01 6.16500139e-01
5.01884937e-01 1.02246606e+00 -3.97585958e-01 3.79846841e-01
-2.10196082e-03 1.39055693e+00 -3.00836235e-01 -4.92933273e-01
-2.23211408e-01 -4.53190804e-01 -1.08855093e+00 -4.17307764e-01
-4.59035546e-01 -7.90268704e-02 -4.01171923e-01 6.69933677e-01
2.84990728e-01 1.53539971e-01 5.43917894e-01 -9.77957487e-01
-1.28830925e-01 7.76548088e-01 -5.18451691e-01 5.73501348e-01
3.84947449e-01 -1.13594852e-01 1.15947664e+00 -4.55392897e-01
9.96402025e-01 1.28759813e+00 7.52240896e-01 1.77135780e-01
-1.19316399e+00 -5.01295149e-01 2.43361145e-01 3.53835344e-01
-1.47032666e+00 -2.25125328e-01 4.02607441e-01 -6.35222256e-01
1.18626404e+00 3.63690555e-01 2.69354880e-01 1.06336820e+00
6.14411175e-01 7.62428701e-01 9.49063241e-01 -1.23247929e-01
4.75846492e-02 -1.98700503e-01 5.03529727e-01 6.63760245e-01
-7.03002606e-03 8.62713009e-02 -7.71278262e-01 -5.51478922e-01
4.70699102e-01 2.19524384e-01 -2.80374289e-01 4.37095702e-01
-1.19108486e+00 7.46009529e-01 2.47374937e-01 3.71498674e-01
-5.12856960e-01 1.76083893e-01 8.52177501e-01 8.94411132e-02
5.90096414e-01 2.58410811e-01 -7.65158355e-01 -3.44914347e-02
-7.12029934e-01 1.70209721e-01 7.13826120e-01 8.19393098e-01
-2.21638614e-03 -8.00265551e-01 -6.04673505e-01 6.37047529e-01
3.94069821e-01 3.04205026e-02 2.39936128e-01 -5.89854121e-01
4.33049381e-01 6.75865293e-01 3.39639150e-02 -1.19362330e+00
-5.63794911e-01 -8.31450000e-02 -9.59740639e-01 -7.87913978e-01
2.60051608e-01 -7.38028362e-02 -9.07553673e-01 1.64954734e+00
4.77274358e-01 2.93057233e-01 8.48114938e-02 5.68588376e-01
1.10999322e+00 7.66445756e-01 4.27614808e-01 -5.54255605e-01
1.65513349e+00 -5.49441040e-01 -1.24394798e+00 -1.50120929e-01
8.19685578e-01 -5.62659800e-01 3.75452042e-01 1.71804056e-01
-8.80967736e-01 2.51359910e-01 -7.29750693e-01 -1.48141205e-01
-3.57696027e-01 -1.19721837e-01 9.57940519e-01 -8.82412866e-02
-5.19788563e-01 4.88230348e-01 -1.38144445e+00 -2.16326997e-01
4.87641990e-01 3.16025734e-01 -3.95079762e-01 -6.13049977e-02
-1.64699316e+00 9.42252696e-01 5.35557985e-01 3.62886429e-01
-4.12354529e-01 -7.64608204e-01 -8.55313361e-01 2.39170175e-02
5.41013122e-01 -6.17252827e-01 1.14468205e+00 1.00634411e-01
-9.58775640e-01 7.54177034e-01 -3.99007022e-01 -5.16029775e-01
3.63351017e-01 -2.10436046e-01 -7.04247594e-01 1.02429762e-01
3.97790298e-02 1.07071497e-01 2.27309749e-01 -4.53063130e-01
-4.92356807e-01 -3.72782171e-01 -3.50322455e-01 -3.22234213e-01
-6.36675805e-02 3.49188417e-01 -7.83948779e-01 -7.84663320e-01
3.88904638e-03 -7.50774205e-01 -5.57209969e-01 -1.45814285e-01
-7.30302334e-01 -7.08378732e-01 5.83231568e-01 -7.53535390e-01
1.69763434e+00 -2.03605700e+00 2.02672295e-02 3.06660831e-01
2.19152704e-01 1.86800390e-01 3.26477289e-01 4.81814861e-01
-3.09871823e-01 1.96541086e-01 -1.24809407e-01 -1.48351133e-01
-3.32352400e-01 3.90828043e-01 -3.90240222e-01 3.93569887e-01
3.82127911e-01 1.26725841e+00 -1.11206043e+00 -1.04188776e+00
-2.24557564e-01 4.83866900e-01 -4.10525113e-01 9.20350999e-02
-3.64454895e-01 3.39034468e-01 -1.01808465e+00 7.42308617e-01
6.43109977e-02 -9.23685193e-01 5.02710581e-01 -3.32166970e-01
3.07190448e-01 7.41549134e-01 -7.03573883e-01 1.52063370e+00
-7.05147460e-02 3.42296928e-01 -3.65564108e-01 -8.94250929e-01
5.49010456e-01 6.88463449e-01 9.38038647e-01 -4.32517260e-01
1.45229638e-01 -8.94484222e-02 -2.72553921e-01 -8.32910240e-01
-5.86488396e-02 -4.68799502e-01 -3.44934255e-01 1.89794108e-01
-2.25606680e-01 2.63591111e-01 2.20855355e-01 3.72781962e-01
1.41532636e+00 3.75816189e-02 7.65429735e-01 1.17607880e-02
5.16054571e-01 8.66532177e-02 1.13839054e+00 4.96415079e-01
-2.58946002e-01 2.30512992e-01 9.62934494e-01 -4.70832497e-01
-2.00876981e-01 -9.57713544e-01 -4.89337295e-01 5.48584402e-01
-1.78226367e-01 -1.03628576e+00 6.80528730e-02 -1.01823843e+00
2.33274177e-02 5.05587041e-01 -7.17888832e-01 -2.06378192e-01
-7.02756643e-01 -1.02144957e+00 5.79340875e-01 7.38644421e-01
-7.61792213e-02 -9.15691853e-01 -2.29948089e-01 5.17696619e-01
-5.87173641e-01 -1.43085384e+00 -5.73996782e-01 2.86292315e-01
-8.97464097e-01 -1.15317643e+00 -5.50541505e-02 -4.89736140e-01
4.71260250e-01 -4.52922344e-01 1.01345384e+00 -1.89282537e-01
-5.91774285e-01 -7.58873075e-02 -1.95055962e-01 -2.65484273e-01
-2.16042906e-01 -1.25356227e-01 -3.38427067e-01 -1.66449994e-01
6.86771035e-01 -4.26846951e-01 -5.53088963e-01 1.30892292e-01
-7.97502756e-01 -4.41391245e-02 4.46753472e-01 9.14759278e-01
9.08806324e-01 7.18310922e-02 7.37011790e-01 -1.28962195e+00
5.16872883e-01 -7.67124534e-01 -4.49385941e-01 5.25074959e-01
-5.98491669e-01 2.93566585e-01 1.34218588e-01 -2.65935272e-01
-1.14245200e+00 -8.88791755e-02 -6.09865040e-02 8.84600799e-04
1.16284899e-01 1.26337230e+00 2.32013445e-02 6.87885284e-01
5.12467325e-01 -2.29273006e-01 -4.36256170e-01 -3.58072549e-01
5.60478084e-02 4.69525814e-01 4.22316492e-01 -4.31596845e-01
3.65306228e-01 5.01716554e-01 3.22462320e-01 -2.57191747e-01
-1.39167666e+00 -7.10749030e-01 -2.12570354e-01 2.74453372e-01
1.10382545e+00 -8.55324209e-01 -8.03205669e-01 5.84272258e-02
-1.26781130e+00 -1.95013896e-01 -1.18906036e-01 6.87377691e-01
-2.83595353e-01 1.71841279e-01 -1.07913184e+00 -5.15659273e-01
-4.79741126e-01 -1.12856519e+00 1.13705075e+00 -1.00329868e-01
-6.61143363e-01 -1.20270169e+00 1.57565489e-01 2.93399781e-01
-3.26068923e-02 6.43619299e-01 1.32602227e+00 -6.94997847e-01
-6.45211518e-01 -3.85293245e-01 -3.19583774e-01 -2.37814739e-01
5.11311233e-01 -1.26010031e-01 -5.21323085e-01 1.31366268e-01
-1.55905545e-01 -2.67913312e-01 8.32311809e-01 5.07718265e-01
1.16035557e+00 -4.19518411e-01 -9.40912306e-01 6.36999011e-01
1.15761745e+00 3.50036830e-01 4.47398871e-01 -1.38670400e-01
6.10765696e-01 5.11927426e-01 8.47136199e-01 4.64098543e-01
4.57350701e-01 6.63170576e-01 -1.16453625e-01 7.22531751e-02
9.19611380e-02 -2.04830706e-01 -9.00510326e-02 6.86166108e-01
-4.05543521e-02 -2.30788514e-01 -1.13621998e+00 7.07236826e-01
-2.30219340e+00 -6.37110770e-01 -1.94854647e-01 1.76913524e+00
1.58115470e+00 2.46066079e-01 -2.98945695e-01 -2.31541395e-01
4.72948283e-01 -5.67459501e-02 -2.86499262e-01 -2.39873126e-01
9.35611352e-02 7.11375400e-02 2.81293899e-01 5.04750371e-01
-1.21472836e+00 7.92161644e-01 6.52224302e+00 6.70057297e-01
-8.96197438e-01 1.69547692e-01 9.16813672e-01 -9.21506509e-02
-3.97248529e-02 3.61697711e-02 -1.03202105e+00 2.48054221e-01
1.13099766e+00 -3.08750778e-01 -9.14009735e-02 3.25026870e-01
3.64503205e-01 -4.54960577e-02 -1.34844923e+00 9.12015498e-01
-3.81697208e-01 -1.54154265e+00 -2.13839024e-01 -1.02902297e-02
6.24750316e-01 -1.62500456e-01 -2.86454529e-01 1.87451184e-01
4.02898520e-01 -1.10226810e+00 3.28739733e-02 8.98498118e-01
8.51900101e-01 -5.40122032e-01 9.32726443e-01 1.79937154e-01
-1.17851901e+00 2.30893955e-01 2.96529800e-01 4.15173650e-01
5.88728726e-01 1.02627075e+00 -1.16815555e+00 9.43152428e-01
7.79023409e-01 9.67317998e-01 -2.41074666e-01 7.80272484e-01
-4.17597592e-01 9.09671962e-01 -4.52751994e-01 1.02762863e-01
1.01319961e-01 1.82346508e-01 3.89777243e-01 1.46380830e+00
-2.64802743e-02 6.82955980e-01 1.76658571e-01 6.08267486e-01
-1.23425655e-01 1.44517526e-01 -4.44220781e-01 -5.81911385e-01
2.30507031e-01 1.04105115e+00 -7.83219099e-01 -3.76240909e-01
-4.00236756e-01 4.65587825e-01 2.55276293e-01 2.86137700e-01
-1.05058837e+00 -1.84287280e-01 3.83362889e-01 -1.14145555e-01
2.33179167e-01 -8.51880237e-02 -1.77449122e-01 -1.07742119e+00
-7.77051598e-02 -7.17810750e-01 1.03756404e+00 -4.15428370e-01
-1.48396802e+00 6.26879692e-01 8.81083086e-02 -8.92078221e-01
-4.46554303e-01 -3.44788104e-01 -5.39841466e-02 8.35051000e-01
-1.27011037e+00 -1.05308223e+00 1.83822587e-02 6.59223735e-01
1.13051616e-01 3.68042141e-01 9.45475757e-01 4.82615560e-01
-7.12751627e-01 5.35402954e-01 -3.12834680e-01 2.71998525e-01
9.58445311e-01 -1.04736042e+00 1.23402171e-01 5.86365879e-01
-1.90986946e-01 1.10649037e+00 5.38160086e-01 -9.31495011e-01
-1.21221602e+00 -1.36796844e+00 1.52276564e+00 -6.28773928e-01
8.20439219e-01 -1.49312824e-01 -9.98004675e-01 1.04082036e+00
-1.23885423e-01 2.70958662e-01 9.74615753e-01 3.86734843e-01
-3.73312086e-01 -1.45610675e-01 -1.00630045e+00 4.44658965e-01
8.19819212e-01 -6.46430135e-01 -7.03039706e-01 8.35709393e-01
9.65664625e-01 -4.30190384e-01 -1.38654470e+00 7.24887490e-01
1.96705818e-01 -2.68552423e-01 1.11222076e+00 -1.13071501e+00
5.76853752e-01 -2.72253454e-01 5.81121892e-02 -6.68253005e-01
-1.93466544e-01 -8.36424887e-01 -4.74957108e-01 1.15184474e+00
7.05917716e-01 -6.20431662e-01 3.18203211e-01 6.39931858e-01
2.20834851e-01 -1.18228984e+00 -8.84320498e-01 -5.42446494e-01
-3.37457985e-01 -5.43544233e-01 3.19196284e-01 1.29321527e+00
5.23871601e-01 4.78044271e-01 -2.08629221e-01 2.30426222e-01
2.27166414e-01 3.84645939e-01 4.50126268e-03 -1.12248838e+00
-6.08164668e-01 -2.75321245e-01 -3.86138409e-01 -7.12849915e-01
2.16673017e-02 -9.12213326e-01 1.88108072e-01 -1.78661180e+00
4.61686760e-01 -5.30991614e-01 -5.99667728e-01 9.75361168e-01
-5.32177985e-01 -1.52205229e-01 -6.82626069e-01 8.63814354e-02
-7.21789598e-01 2.48717457e-01 9.54971731e-01 -1.43808916e-01
-3.26070130e-01 6.22909144e-02 -5.39943278e-01 7.24865973e-01
4.31535870e-01 -8.07263017e-01 -3.97124648e-01 -5.78493588e-02
6.78467333e-01 7.03723848e-01 2.07195505e-01 -2.43074566e-01
4.90342289e-01 -3.85089338e-01 7.15555772e-02 -8.38984966e-01
1.70039356e-01 -5.89073598e-01 1.00423545e-01 4.43611979e-01
-5.25450766e-01 -4.83779842e-03 2.92955369e-01 8.91496360e-01
-4.86029744e-01 1.45748034e-01 4.10646766e-01 1.71645552e-01
-3.43219548e-01 4.33887362e-01 -1.44524649e-01 2.82634854e-01
9.74305511e-01 6.29508734e-01 -2.89620459e-01 -6.63257316e-02
-1.03991735e+00 5.90650976e-01 -3.90431315e-01 3.67256612e-01
5.74970126e-01 -1.18672371e+00 -7.36812592e-01 -3.73705626e-01
1.97809711e-01 2.23865554e-01 2.42102101e-01 1.37634361e+00
-3.38658780e-01 9.18642938e-01 6.09713495e-01 -5.97370684e-01
-1.39672112e+00 8.36028099e-01 1.56943589e-01 -1.00158429e+00
-8.25141430e-01 1.05962956e+00 1.96925312e-01 -5.95913082e-02
1.81326479e-01 -5.08989513e-01 -5.07204458e-02 1.22428097e-01
5.30622900e-01 -9.21746120e-02 1.60351157e-01 -1.22592203e-01
-8.76135886e-01 8.14564303e-02 -5.03367186e-01 8.32850859e-02
1.57184803e+00 1.73978254e-01 -4.70413923e-01 4.53579158e-01
1.18604505e+00 -1.12110727e-01 -7.26200879e-01 -4.43601996e-01
6.72769248e-01 -5.07927425e-02 1.00465439e-01 -9.88347411e-01
-8.43930423e-01 5.15057087e-01 8.38421285e-02 1.36131763e-01
1.20612645e+00 5.48822761e-01 8.47838223e-01 4.84107852e-01
1.55288666e-01 -6.03079140e-01 -2.07496718e-01 3.88335973e-01
7.28669524e-01 -1.01954520e+00 3.81681234e-01 -7.84533143e-01
-5.23695946e-01 7.95721591e-01 2.10198000e-01 2.75754899e-01
9.86511588e-01 6.12259984e-01 -8.79921392e-02 -5.30932784e-01
-1.26302981e+00 4.32070121e-02 6.52608275e-01 -4.00988013e-02
8.30445290e-01 1.02508388e-01 -5.25675297e-01 9.31716502e-01
3.37686032e-01 3.59214336e-01 3.60787351e-04 1.14026618e+00
2.16140971e-01 -1.31456399e+00 -1.03320599e-01 5.78375041e-01
-9.84843135e-01 -2.90332258e-01 -2.28709966e-01 5.96575797e-01
-3.03095579e-02 8.82730782e-01 -4.28402692e-01 -1.58271685e-01
1.66562930e-01 2.19703361e-01 2.87066877e-01 -6.44857347e-01
-5.12432337e-01 3.83263409e-01 5.33920288e-01 -7.55837083e-01
-5.05633831e-01 -9.35538769e-01 -1.71049750e+00 2.68983871e-01
-3.00580263e-01 2.32128590e-01 2.11580053e-01 1.16496754e+00
4.92712587e-01 9.98261511e-01 1.04408987e-01 4.06937361e-01
-3.37552965e-01 -6.81911409e-01 -2.44956225e-01 3.81476671e-01
3.51094246e-01 -5.20532846e-01 -3.76123637e-02 2.66781837e-01] | [8.649724006652832, 8.929250717163086] |
0606ff36-0363-41e2-8a79-ff21f00fff21 | automating-feature-engineering | null | null | http://workshops.inf.ed.ac.uk/nips2016-ai4datasci/papers/NIPS2016-AI4DataSci_paper_13.pdf | http://workshops.inf.ed.ac.uk/nips2016-ai4datasci/papers/NIPS2016-AI4DataSci_paper_13.pdf | Automating Feature Engineering | Feature Engineering is the task of transforming the feature space in a given learning problem to improve the performance of a trained model. It is a crucial but time intensive and skillful process, involving a data scientist or a domain expert. It is often the key determinant of the time and cost required to build an effective learner. In this paper, we discuss our system for performing feature engineering in an automated manner using a combination of exploratory and learning techniques. We also mention our larger charter of an automated data science pipeline. | ['Horst Samulowitz', 'Deepak Turaga', 'Elias Khalil', 'Udayan Khurana', 'Fatemeh Nargesian'] | 2016-01-01 | automating-feature-engineering-1 | https://pdfs.semanticscholar.org/959e/dbe6a4720d49d87d9b64440cd394f934815d.pdf | https://pdfs.semanticscholar.org/959e/dbe6a4720d49d87d9b64440cd394f934815d.pdf | nips-2016-2016-1 | ['automated-feature-engineering'] | ['methodology'] | [ 1.95028752e-01 1.16351554e-02 8.87654126e-02 -7.24427164e-01
-7.58245051e-01 -9.01024103e-01 5.76652169e-01 5.25219560e-01
-2.74981022e-01 5.57787299e-01 3.38920504e-02 -5.79867482e-01
-4.59278405e-01 -6.39508605e-01 -8.94131958e-01 -3.00838500e-01
2.45546132e-01 5.12232304e-01 3.50393471e-03 -1.64954662e-02
7.92912722e-01 8.63055289e-01 -1.90782857e+00 1.14307217e-01
9.82121527e-01 7.04701722e-01 -2.47910712e-02 5.73576331e-01
-3.91059041e-01 8.43678355e-01 -4.58653092e-01 -1.44105807e-01
2.55553365e-01 -3.83933812e-01 -8.03184807e-01 -5.08246087e-02
4.78213251e-01 2.50581622e-01 2.98546672e-01 7.91440248e-01
1.84308931e-01 2.61691719e-01 4.68966365e-01 -1.32332015e+00
-2.05157027e-01 4.06069398e-01 -1.50124356e-01 3.79728168e-01
4.40768838e-01 1.97650045e-01 7.70290315e-01 -1.18499911e+00
6.51153982e-01 9.34530199e-01 5.41292727e-01 1.53964341e-01
-1.21757543e+00 -7.75681496e-01 1.51157333e-02 2.27961674e-01
-9.42309082e-01 -6.97223783e-01 7.49102235e-01 -9.66123879e-01
9.04135466e-01 1.50833920e-01 1.07734942e+00 3.82631510e-01
2.65996277e-01 5.58580995e-01 8.60364199e-01 -6.18232131e-01
4.31014121e-01 4.81658995e-01 4.01826859e-01 7.42043436e-01
2.71546602e-01 1.24520928e-01 -8.73720169e-01 -1.30720511e-01
5.21806479e-01 3.18166107e-01 1.24594294e-01 -5.93803287e-01
-7.64496148e-01 9.29120183e-01 2.50560105e-01 2.60571301e-01
-3.47518325e-01 3.40339839e-02 7.46840090e-02 7.80675769e-01
3.26546401e-01 1.33436382e+00 -6.23857498e-01 -6.76773369e-01
-9.98375356e-01 2.76882857e-01 7.49170661e-01 8.89476180e-01
8.98932874e-01 -1.84101030e-01 1.80043787e-01 1.92885220e-01
4.24000740e-01 -1.27174199e-01 3.40905041e-01 -7.67346978e-01
-1.39784604e-01 1.45918739e+00 -2.25759577e-02 -5.32320619e-01
-3.32642108e-01 -5.00203252e-01 4.03739475e-02 4.34521079e-01
2.78724700e-01 -1.87666968e-01 -8.84084284e-01 9.69106555e-01
6.73710465e-01 -1.90111890e-01 -7.92681500e-02 3.72336090e-01
9.71721232e-01 2.79103696e-01 2.29013056e-01 -2.09760517e-01
9.38233316e-01 -6.53586388e-01 -4.68628675e-01 -2.94127792e-01
8.36455882e-01 -8.63201439e-01 1.16974914e+00 6.65705740e-01
-1.10970998e+00 -4.69242096e-01 -8.72553229e-01 -3.35733354e-01
-6.21146262e-01 -1.10528402e-01 8.63095045e-01 3.89743060e-01
-6.70484543e-01 8.51931274e-01 -7.74579585e-01 -3.85450870e-01
6.19796455e-01 3.62425953e-01 -5.55924416e-01 -1.77730247e-01
-5.90251505e-01 1.01045370e+00 3.68690223e-01 -3.80524933e-01
-5.59198260e-01 -1.44298911e+00 -7.10729182e-01 3.73694561e-02
3.96932811e-01 -4.42995667e-01 1.36820054e+00 -6.27331495e-01
-9.36950922e-01 4.44297433e-01 -6.75627068e-02 -1.52072012e-01
5.31118333e-01 -6.56840444e-01 -9.85904485e-02 -4.68506962e-01
-6.33638501e-02 1.31592736e-01 8.27283859e-01 -7.45131195e-01
-9.10893977e-01 -4.88234192e-01 -4.14900571e-01 1.76009282e-01
-1.97290391e-01 1.14619188e-01 -3.00603379e-02 -2.65362769e-01
1.15173191e-01 -8.22067142e-01 -2.74473250e-01 -1.35644764e-01
-2.22718716e-02 -5.54528117e-01 1.15416360e+00 -4.86834109e-01
1.35262823e+00 -2.28193188e+00 1.05963141e-01 4.97659743e-01
3.99875343e-01 2.22219244e-01 4.06540900e-01 5.52950799e-01
-1.70682967e-01 2.60388255e-01 8.38704631e-02 1.90509290e-01
-1.39535233e-01 -3.32615256e-01 -1.64127037e-01 1.79351881e-01
4.92819101e-01 8.18980455e-01 -1.07476234e+00 -3.43966156e-01
2.63484985e-01 4.89526272e-01 -6.29920244e-01 5.92003226e-01
-2.49919191e-01 5.21039844e-01 -5.64221621e-01 6.22127414e-01
-1.50781916e-02 -3.04526865e-01 -6.69057071e-02 -4.09938730e-02
-6.44916534e-01 4.16685790e-01 -1.24924493e+00 1.49083304e+00
-3.73808414e-01 9.09630656e-01 -3.76811475e-01 -5.81580997e-01
1.16663468e+00 2.32951660e-02 3.97032320e-01 -4.01567161e-01
-1.51058417e-02 2.35814422e-01 2.17587464e-02 -6.32758081e-01
3.51054549e-01 -4.44051474e-02 2.26518810e-01 9.19034660e-01
3.23354602e-01 -3.07204545e-01 3.80074143e-01 2.19529402e-02
1.41976416e+00 3.17061752e-01 6.43148422e-01 -3.10828954e-01
1.19698167e-01 5.38899779e-01 3.70454103e-01 6.10298753e-01
4.76333573e-02 1.67230621e-01 3.38964522e-01 -8.52875352e-01
-1.04704297e+00 -5.43367326e-01 -2.94885188e-01 1.35757303e+00
-7.69396901e-01 -5.31301141e-01 -3.35700840e-01 -8.80728304e-01
3.60010892e-01 9.04650629e-01 -7.51909852e-01 -4.56225038e-01
-2.85328060e-01 -1.41946629e-01 -1.80269688e-01 4.27558541e-01
2.47529179e-01 -1.15684652e+00 -9.44613993e-01 2.11822107e-01
4.91257221e-01 -4.45574939e-01 -1.45624995e-01 6.07445478e-01
-8.82964492e-01 -1.25840294e+00 1.91562116e-01 -8.03491294e-01
8.24250460e-01 1.36034831e-01 1.28610551e+00 2.95885980e-01
-6.72249913e-01 2.20238656e-01 -4.77830380e-01 -1.13254392e+00
-2.69924700e-01 1.84849814e-01 -2.23035872e-01 -4.30186123e-01
9.79371786e-01 -3.78824860e-01 -4.13049042e-01 8.58678147e-02
-7.93729544e-01 1.84395418e-01 6.89428329e-01 5.56385159e-01
6.87010646e-01 2.72756368e-01 4.74453777e-01 -1.14532399e+00
8.45984340e-01 -4.20624286e-01 -6.50200248e-01 6.33941412e-01
-1.06201625e+00 3.33392799e-01 4.01421338e-01 -2.66314059e-01
-7.56999731e-01 8.37507665e-01 1.30210981e-01 -1.35939330e-01
-2.97368199e-01 7.30107427e-01 -2.65721023e-01 -3.64467651e-01
9.73932147e-01 1.05876580e-01 2.99276579e-02 -8.26496243e-01
2.57850051e-01 5.54629564e-01 -5.11447750e-02 -3.34650040e-01
9.61395144e-01 -5.27297072e-02 1.12131171e-01 -7.76133239e-01
-8.14742923e-01 -4.92145061e-01 -1.17357767e+00 -2.92715341e-01
3.47013265e-01 -5.34609199e-01 -4.42837209e-01 3.85048203e-02
-5.20830870e-01 -2.64895260e-01 -6.96792543e-01 3.32131594e-01
-2.41849154e-01 -4.98510510e-01 2.95635045e-01 -3.53563040e-01
-4.04749274e-01 -1.04875839e+00 5.03282130e-01 7.59594381e-01
-4.67503846e-01 -1.03146541e+00 3.36717457e-01 2.40322486e-01
4.23439384e-01 3.69086266e-01 9.07414973e-01 -1.17678905e+00
-3.33535641e-01 -5.36951661e-01 -1.57749318e-02 -1.96824837e-02
4.07940716e-01 5.84318876e-01 -9.67551887e-01 -2.41731867e-01
-9.63619277e-02 -2.90260613e-01 4.11354333e-01 1.47245795e-01
1.09146249e+00 -2.09455460e-01 -1.98721319e-01 5.34181893e-01
1.15989578e+00 3.50527316e-01 -4.58394119e-04 6.25534952e-01
5.66925108e-01 7.45601416e-01 1.07618165e+00 5.22997200e-01
3.60620581e-02 -3.95671725e-02 -1.58855364e-01 -2.59273853e-02
1.92127064e-01 -4.39932317e-01 -9.71884727e-02 5.32650389e-02
1.29913867e-01 5.65149903e-01 -1.46006072e+00 5.15988410e-01
-1.79480314e+00 -7.85334945e-01 -1.43013662e-02 2.15943432e+00
9.67765510e-01 1.85889795e-01 1.99222073e-01 2.48890147e-01
8.78338441e-02 -5.87900400e-01 -8.25750947e-01 -5.07475615e-01
6.13098800e-01 3.19351196e-01 1.31185055e-01 2.88844109e-01
-9.50631082e-01 9.71468389e-01 6.88322926e+00 6.27633557e-02
-1.01153207e+00 -4.85163182e-01 1.96931690e-01 -3.38620186e-01
-4.04952049e-01 1.70023125e-02 -8.69671226e-01 1.30096644e-01
1.22628915e+00 -7.72235572e-01 3.82509708e-01 1.28206527e+00
1.90053135e-01 -1.51258484e-01 -1.59181476e+00 7.12024987e-01
-1.58262521e-01 -1.45390046e+00 -1.97226435e-01 1.03154130e-01
4.07992065e-01 -2.03498930e-01 5.25777563e-02 3.81214350e-01
7.36925840e-01 -1.21653819e+00 2.70017654e-01 6.67998433e-01
4.35968906e-01 -1.12858272e+00 3.71758729e-01 3.00489068e-01
-8.96500528e-01 -1.62765115e-01 6.58262968e-02 -1.22179404e-01
-3.40058863e-01 6.18449152e-01 -1.31126869e+00 -4.30104472e-02
8.36085081e-01 7.69269526e-01 -1.07881689e+00 1.49643326e+00
-2.86515266e-01 6.13076270e-01 -1.50484532e-01 -6.93810508e-02
-1.07729539e-01 2.91663408e-02 2.24569097e-01 1.03087163e+00
2.04854757e-01 8.54474381e-02 -4.32925150e-02 5.74165344e-01
4.09433320e-02 1.86753571e-01 -8.24224234e-01 -6.02591157e-01
6.79614127e-01 1.51185787e+00 -4.40782696e-01 -7.79561549e-02
-3.59050244e-01 1.53423443e-01 4.06920791e-01 -1.06264697e-02
-1.10444121e-01 -6.59762383e-01 6.57718539e-01 3.71434391e-01
1.32583052e-01 -2.00939029e-01 -5.55667222e-01 -6.39046669e-01
-8.23110044e-02 -1.09308636e+00 4.21588242e-01 -3.43935758e-01
-1.12900877e+00 1.68349668e-01 -1.37162581e-01 -1.12707639e+00
-3.61699611e-01 -3.15535098e-01 -7.48155713e-01 9.14204180e-01
-1.21951699e+00 -8.69664133e-01 -6.49791718e-01 5.22075891e-01
3.47335547e-01 -3.83266807e-01 6.76295042e-01 -9.01100785e-02
-6.27497673e-01 2.27799565e-01 2.30829976e-02 -1.97221056e-01
6.98629677e-01 -1.42331958e+00 5.52880287e-01 8.19020987e-01
8.68675858e-02 1.11731958e+00 9.41866875e-01 -8.93595457e-01
-1.72451472e+00 -1.03822780e+00 1.14072061e+00 -6.66221917e-01
6.50874019e-01 -2.41103947e-01 -1.10722375e+00 8.12514067e-01
-1.14502601e-01 -1.16710886e-01 1.37630785e+00 5.33143282e-01
-6.39477819e-02 -3.62136886e-02 -1.39017284e+00 2.43067741e-01
6.81578994e-01 -3.49504322e-01 -9.02541995e-01 3.16792309e-01
6.48501694e-01 -3.10520530e-01 -1.07085752e+00 1.74139604e-01
6.31521821e-01 -7.57794738e-01 5.68682492e-01 -1.08906806e+00
3.39593679e-01 -2.73259968e-01 2.88369894e-01 -1.42161298e+00
-4.21849608e-01 -6.16521716e-01 -2.69997299e-01 1.18588161e+00
5.22759378e-01 -2.93352067e-01 9.69366074e-01 1.44475067e+00
-3.22157852e-02 -6.49488866e-01 -4.20212567e-01 -3.97958398e-01
-1.11264758e-01 -2.88371712e-01 8.46477926e-01 1.08625257e+00
2.21771464e-01 3.92495006e-01 3.06127667e-01 -2.70630151e-01
6.39626443e-01 7.27508739e-02 1.01644135e+00 -1.88273644e+00
1.20904192e-01 -3.15491170e-01 -2.94170082e-01 -1.82491362e-01
-4.44478691e-01 -7.03790665e-01 1.39453597e-02 -1.67451334e+00
1.06880888e-01 -4.58027095e-01 -2.32342869e-01 5.87713122e-01
-2.59476662e-01 -2.83030063e-01 -6.51310897e-03 2.24856451e-01
-5.37808239e-01 3.67754288e-02 9.61134315e-01 3.95304710e-01
-7.58897781e-01 2.84044087e-01 -1.16262341e+00 7.14918852e-01
9.34372246e-01 -7.48515189e-01 -6.54754877e-01 1.28299568e-03
6.48937821e-01 -6.46058440e-01 1.19859621e-01 -9.03539538e-01
5.34753680e-01 -5.58787882e-01 9.52625871e-01 -4.65863258e-01
-1.56780630e-01 -1.05398154e+00 2.97067434e-01 4.50957984e-01
-5.48377275e-01 3.93119156e-01 3.14155608e-01 1.80390283e-01
-9.07966718e-02 -3.65854383e-01 5.92223287e-01 -2.86246806e-01
-8.87515426e-01 1.23066917e-01 -3.36947516e-02 3.51681262e-02
1.49117661e+00 -6.01166226e-02 -2.73528308e-01 5.58785200e-02
-7.28438914e-01 3.66757452e-01 5.25277495e-01 3.59528363e-01
7.28768706e-01 -9.98437762e-01 -4.09998417e-01 7.47744858e-01
2.04771042e-01 2.52497882e-01 -2.73934156e-01 4.44602191e-01
-3.33714902e-01 2.54347980e-01 -3.73657107e-01 -3.35412621e-01
-1.55845082e+00 2.45137379e-01 1.57155141e-01 2.62893513e-02
-6.27106965e-01 9.13490891e-01 -4.50900823e-01 -3.85982662e-01
4.33813035e-01 -1.37203950e-02 -5.15426099e-01 2.88358092e-01
9.48693216e-01 5.41532576e-01 3.09436440e-01 -1.05732650e-01
-3.90399933e-01 2.35187501e-01 -4.58943874e-01 -1.02183418e-02
1.91985452e+00 3.46642524e-01 9.39901322e-02 6.33158743e-01
9.63154852e-01 -3.49886745e-01 -1.32451189e+00 -2.96608090e-01
4.36515987e-01 -7.03037679e-01 3.96322608e-01 -9.90302026e-01
-4.24361438e-01 7.23626554e-01 6.91566169e-01 2.98693687e-01
9.19620812e-01 1.56064913e-01 -1.40271941e-02 5.64370573e-01
1.31918252e-01 -1.37151980e+00 3.05446945e-02 2.95932233e-01
1.02413344e+00 -1.61766505e+00 3.74123752e-01 -4.78076302e-02
-6.15718782e-01 1.25542665e+00 8.54378462e-01 1.50295511e-01
1.01205492e+00 2.27593347e-01 1.89799309e-01 -6.89919412e-01
-9.10914063e-01 -3.73265222e-02 7.04856753e-01 4.50638115e-01
8.97147596e-01 -2.66984999e-01 -8.16561505e-02 3.07378858e-01
-5.59556365e-01 2.61515439e-01 2.60970891e-01 1.63089490e+00
-9.05391991e-01 -1.18415964e+00 -1.23383991e-01 9.43507731e-01
-1.33797929e-01 4.59965691e-02 -9.49613988e-01 7.59357572e-01
-6.13392629e-02 8.57713819e-01 1.28644388e-02 -5.47574162e-01
5.99827111e-01 5.64038396e-01 2.12343246e-01 -1.15269804e+00
-1.08752048e+00 -1.39955983e-01 -1.14600353e-01 -5.86974144e-01
1.45631451e-02 -1.03456962e+00 -1.27950013e+00 -3.79346550e-01
2.04969235e-02 3.39240193e-01 1.07061172e+00 9.45464492e-01
5.52145362e-01 4.50085521e-01 6.65400863e-01 -3.80239606e-01
-3.37768108e-01 -8.56027007e-01 -2.87539482e-01 2.85471082e-01
4.24093038e-01 -6.08858705e-01 -3.62179697e-01 2.67212316e-02] | [8.367745399475098, 4.674412250518799] |
65f80190-c0a8-4d36-8f3d-ba85374c201c | vox-populi-vox-diy-benchmark-dataset-for | 2107.01091 | null | https://arxiv.org/abs/2107.01091v2 | https://arxiv.org/pdf/2107.01091v2.pdf | CrowdSpeech and VoxDIY: Benchmark Datasets for Crowdsourced Audio Transcription | Domain-specific data is the crux of the successful transfer of machine learning systems from benchmarks to real life. In simple problems such as image classification, crowdsourcing has become one of the standard tools for cheap and time-efficient data collection: thanks in large part to advances in research on aggregation methods. However, the applicability of crowdsourcing to more complex tasks (e.g., speech recognition) remains limited due to the lack of principled aggregation methods for these modalities. The main obstacle towards designing aggregation methods for more advanced applications is the absence of training data, and in this work, we focus on bridging this gap in speech recognition. For this, we collect and release CrowdSpeech -- the first publicly available large-scale dataset of crowdsourced audio transcriptions. Evaluation of existing and novel aggregation methods on our data shows room for improvement, suggesting that our work may entail the design of better algorithms. At a higher level, we also contribute to the more general challenge of developing the methodology for reliable data collection via crowdsourcing. In that, we design a principled pipeline for constructing datasets of crowdsourced audio transcriptions in any novel domain. We show its applicability on an under-resourced language by constructing VoxDIY -- a counterpart of CrowdSpeech for the Russian language. We also release the code that allows a full replication of our data collection pipeline and share various insights on best practices of data collection via crowdsourcing. | ['Dmitry Ustalov', 'Ivan Stelmakh', 'Nikita Pavlichenko'] | 2021-07-02 | null | null | null | null | ['crowdsourced-text-aggregation'] | ['natural-language-processing'] | [-1.61443055e-02 1.57379471e-02 2.11952522e-01 -5.04145622e-01
-1.48804760e+00 -9.08749878e-01 6.44117951e-01 1.70387670e-01
-5.88735521e-01 8.23571563e-01 5.67109644e-01 -1.50517821e-01
2.46686935e-01 -3.84017080e-01 -6.63095653e-01 -5.54135859e-01
2.71402776e-01 7.25672722e-01 3.22810829e-01 -5.07859588e-01
-6.86278641e-02 4.07902986e-01 -1.76163256e+00 5.72202504e-01
4.72654492e-01 9.24340546e-01 -1.31725371e-01 7.35998213e-01
-1.87142372e-01 7.61946499e-01 -8.99373651e-01 -6.79057479e-01
2.38646120e-01 -4.73843753e-01 -1.00729537e+00 2.64020383e-01
5.34785688e-01 -2.49615952e-01 6.15082942e-02 5.01598835e-01
9.95236456e-01 1.17175668e-01 2.53376454e-01 -1.55980766e+00
-6.35074198e-01 4.82950270e-01 -4.56797518e-02 -3.44078802e-02
8.31212282e-01 2.46110201e-01 1.03745222e+00 -1.09483027e+00
5.55482507e-01 1.05818856e+00 7.57820308e-01 7.08710372e-01
-9.96603251e-01 -3.52443665e-01 -2.37849414e-01 -9.85386148e-02
-1.38577235e+00 -9.98073995e-01 3.15341860e-01 -6.50211334e-01
9.75555420e-01 4.34284300e-01 4.61665541e-01 1.22519290e+00
-5.89214087e-01 1.12536848e+00 1.21540952e+00 -5.72580576e-01
4.00431156e-01 3.00790012e-01 -3.48254740e-01 3.44996542e-01
-6.75950423e-02 -4.02589738e-01 -1.01096034e+00 -5.17658234e-01
3.67032528e-01 -4.44661528e-01 -1.71589449e-01 -7.65411779e-02
-1.30483365e+00 6.59029067e-01 -1.40516281e-01 3.11035722e-01
-6.65949211e-02 -6.73730969e-02 5.00063181e-01 4.01588440e-01
8.21132004e-01 3.82239074e-01 -3.94658118e-01 -5.74513078e-01
-1.14657342e+00 6.40306532e-01 1.15732348e+00 1.17982447e+00
7.12311447e-01 -4.77877446e-02 -1.36718005e-01 9.86475348e-01
1.73493907e-01 4.99518275e-01 4.95459378e-01 -1.07409322e+00
8.01716864e-01 5.93370736e-01 4.57594901e-01 -5.08740604e-01
-2.15855807e-01 3.44753563e-01 -3.07526290e-01 1.62897348e-01
8.70396793e-01 -2.40886256e-01 -3.50090444e-01 1.37432516e+00
6.16091490e-01 -1.06351078e-01 -5.77314496e-02 1.19091642e+00
1.14532578e+00 1.77955702e-01 -2.93541625e-02 1.27483681e-01
1.39643061e+00 -7.42067099e-01 -5.06714702e-01 -4.26590107e-02
8.64340663e-01 -9.48246837e-01 1.33456481e+00 4.44333673e-01
-1.16040373e+00 -2.52294898e-01 -5.71245253e-01 -4.65633690e-01
-5.75526237e-01 1.20063566e-01 5.49661219e-01 9.43635702e-01
-1.12247360e+00 3.85051489e-01 -4.80197817e-01 -7.06881166e-01
5.99480152e-01 3.73574287e-01 -7.73541570e-01 -1.03159063e-02
-1.15035355e+00 6.75904334e-01 -1.14717059e-01 -4.27058637e-02
-8.11733603e-01 -5.05964160e-01 -7.84864485e-01 -5.49109161e-01
4.02164549e-01 -3.27057034e-01 1.66810691e+00 -1.11342275e+00
-1.47462344e+00 1.23463631e+00 -2.67073840e-01 -4.16160792e-01
7.52909184e-01 -1.22862175e-01 -2.40142077e-01 2.22864989e-02
3.65299761e-01 7.00863302e-01 6.49577856e-01 -9.78786349e-01
-7.21625507e-01 -1.67186350e-01 1.40332893e-01 1.00329764e-01
-4.61482167e-01 6.61655545e-01 -2.47354761e-01 -6.07304931e-01
-4.56685364e-01 -1.07616687e+00 -7.64803886e-02 -6.70716539e-02
-1.76025420e-01 -4.37623352e-01 4.80878472e-01 -5.70948243e-01
8.14038396e-01 -2.19121456e+00 -3.73250954e-02 -2.68451512e-01
3.94301891e-01 4.42982018e-01 -1.69101253e-01 8.75211954e-01
3.29035729e-01 2.84130394e-01 -2.18208566e-01 -8.43793809e-01
1.46714002e-01 2.30553523e-01 -6.21606708e-01 5.32181323e-01
4.48087543e-01 9.14704680e-01 -1.18599141e+00 -5.88122010e-01
-9.13905799e-02 3.06243807e-01 -3.49742830e-01 2.22589180e-01
-1.81898609e-01 5.74553907e-01 -2.36219153e-01 7.61757195e-01
3.67907703e-01 7.39395842e-02 -2.36757427e-01 3.95947039e-01
-2.14658245e-01 3.83669168e-01 -1.34097576e+00 1.88814402e+00
-3.28475803e-01 8.72539639e-01 3.39153498e-01 -5.52035272e-01
8.48669112e-01 7.05317199e-01 6.10187054e-01 -3.67841274e-01
-8.38446245e-03 6.59979820e-01 -3.51939976e-01 -7.27596879e-01
7.43475676e-01 -1.68627605e-01 -2.14355513e-01 8.33099544e-01
2.38672361e-01 -5.88627040e-01 2.86889762e-01 1.75305441e-01
1.13690233e+00 1.75620034e-01 2.15486482e-01 -7.53827617e-02
2.40361914e-01 4.35974985e-01 2.31055960e-01 8.00841689e-01
-5.80512941e-01 1.06081867e+00 2.60192573e-01 -3.91614556e-01
-1.16785395e+00 -6.77402318e-01 -1.43277952e-02 1.56688070e+00
-4.66954619e-01 -4.32359576e-01 -9.82664704e-01 -5.99057376e-01
2.87939608e-02 1.24574006e-01 -4.12665814e-01 6.18726254e-01
-4.63253438e-01 -5.45936048e-01 1.30166030e+00 5.56319356e-01
2.89277345e-01 -1.03679204e+00 -3.79688799e-01 2.09198430e-01
-4.85637367e-01 -1.48197460e+00 -3.12339544e-01 -1.88504711e-01
-4.57567781e-01 -8.49091887e-01 -9.10085738e-01 -5.57527721e-01
1.63338602e-01 3.78025621e-01 1.26827776e+00 1.49262667e-01
-2.18424201e-01 7.32278645e-01 -7.03216553e-01 -1.00861871e+00
-5.36341310e-01 3.18204790e-01 1.65727839e-01 6.27405867e-02
8.26673329e-01 -2.61022866e-01 -4.08600032e-01 5.73134065e-01
-9.16752815e-01 -3.95041138e-01 4.87888046e-02 4.22080725e-01
2.14465186e-01 -5.85665166e-01 7.61474788e-01 -6.20616257e-01
1.07881665e+00 -5.14386415e-01 -4.45911199e-01 1.88117810e-02
1.48911417e-01 -4.18876439e-01 4.70030189e-01 -2.77257442e-01
-7.27190673e-01 3.18548352e-01 -2.99119473e-01 -4.00102027e-02
-5.66045403e-01 2.25102112e-01 -7.26051033e-02 -4.43878248e-02
1.10960722e+00 -6.75858036e-02 4.96505871e-02 -3.88324708e-01
4.59921151e-01 1.23726010e+00 2.42268175e-01 -7.45985508e-01
5.45181990e-01 8.04211259e-01 -2.77109355e-01 -1.15713608e+00
-6.46342874e-01 -7.75728405e-01 -7.57456243e-01 -1.21024184e-01
8.14653218e-01 -1.31057084e+00 -6.58098876e-01 4.20416504e-01
-1.27781940e+00 -6.49863601e-01 -5.75959802e-01 2.62344807e-01
-5.94693244e-01 3.53097767e-01 -4.16935593e-01 -1.14817035e+00
-1.15983963e-01 -1.14221609e+00 1.86804652e+00 -1.83603689e-01
-5.80612481e-01 -9.30820644e-01 3.16073447e-01 7.93622494e-01
2.69592434e-01 9.05912146e-02 -9.60227177e-02 -1.01208818e+00
-4.80753869e-01 -4.29693878e-01 3.25315520e-02 4.92593616e-01
-1.09573051e-01 9.70974192e-02 -1.61967123e+00 6.79129213e-02
-5.91678694e-02 -9.99220133e-01 5.45372844e-01 -1.39571905e-01
6.92393005e-01 -9.76257026e-02 -6.10121293e-03 5.94215281e-03
8.87249529e-01 -3.81453037e-01 6.02475345e-01 2.64413029e-01
6.13077104e-01 1.03974187e+00 5.91884315e-01 5.96766829e-01
6.58443928e-01 8.62127244e-01 -6.17131665e-02 -1.18447151e-02
-8.14580172e-02 -6.71407357e-02 3.76669556e-01 8.78921330e-01
-3.23602945e-01 -1.99695483e-01 -1.09316540e+00 9.80606675e-01
-1.91154230e+00 -1.09689295e+00 -4.92284268e-01 2.18147516e+00
1.05517185e+00 -5.00533700e-01 8.05039465e-01 2.57390499e-01
3.91973406e-01 -3.58943716e-02 1.36888400e-01 -3.02707344e-01
-3.60650241e-01 8.11345875e-02 1.13423295e-01 5.78229129e-01
-1.12845945e+00 8.65566969e-01 6.57166147e+00 9.13206577e-01
-8.54111671e-01 4.67078328e-01 2.61850268e-01 -3.54061753e-01
-1.95242569e-01 -8.66758376e-02 -8.78090084e-01 4.60572481e-01
9.65491831e-01 2.02148974e-01 6.60768688e-01 7.23856151e-01
3.65057766e-01 -2.90396303e-01 -1.34362531e+00 1.01495814e+00
1.43484756e-01 -1.20736170e+00 -4.05504435e-01 4.76690084e-02
6.50444150e-01 4.05382723e-01 -3.93818647e-01 1.27961442e-01
4.42981720e-01 -1.03760505e+00 1.06272042e+00 1.09505475e-01
7.35590041e-01 -1.17846489e-01 6.79512858e-01 4.79333460e-01
-9.44358945e-01 1.28293931e-01 -3.32576811e-01 -3.46093416e-01
2.50153154e-01 7.67616808e-01 -1.21682906e+00 2.52854347e-01
6.71338856e-01 2.29373887e-01 -5.75161159e-01 9.07296062e-01
-1.82189092e-01 6.87932372e-01 -4.45350289e-01 -3.78646433e-01
9.38724279e-02 1.29871994e-01 5.22593141e-01 1.55666721e+00
2.29676679e-01 -8.13813731e-02 2.34360665e-01 5.33660531e-01
-2.30261207e-01 3.12501639e-01 -1.00371504e+00 -9.92005616e-02
7.01061845e-01 1.32177997e+00 -5.08251905e-01 -2.29895264e-01
-5.54209769e-01 7.45075643e-01 5.49054980e-01 2.20361322e-01
-5.84665000e-01 -3.15973967e-01 6.39281094e-01 2.17991650e-01
-6.64689168e-02 -3.35965365e-01 -3.98169100e-01 -1.17619240e+00
4.42357957e-01 -1.39669776e+00 2.01945916e-01 -7.65558600e-01
-1.40732384e+00 5.34465134e-01 -7.83108100e-02 -1.21128309e+00
-4.88622606e-01 -5.87979734e-01 -3.20449829e-01 9.80969191e-01
-1.05826652e+00 -1.28113520e+00 -5.32416582e-01 7.15392172e-01
6.21497154e-01 -3.51090372e-01 8.72520149e-01 3.35560024e-01
-1.04096517e-01 5.00205457e-01 -1.23713471e-01 3.87105167e-01
1.12951064e+00 -1.21989369e+00 7.26067305e-01 5.66633582e-01
6.47184551e-01 4.47813123e-01 8.08674514e-01 -4.53903139e-01
-1.25032008e+00 -9.33500946e-01 1.25318313e+00 -1.41912389e+00
9.96149659e-01 -9.92434919e-01 -7.41873205e-01 6.11088693e-01
1.48132905e-01 6.20471723e-02 1.05489326e+00 6.69706613e-02
-2.84561962e-01 -2.73082545e-03 -1.09972882e+00 4.64256138e-01
1.09272623e+00 -1.10415530e+00 -4.22701806e-01 6.89095676e-01
6.36343360e-01 -4.83652771e-01 -9.57265794e-01 -1.26971483e-01
6.01898015e-01 -1.00457263e+00 5.96160948e-01 -5.59191346e-01
4.14707810e-01 -3.86460662e-01 -3.62529606e-01 -1.34869361e+00
3.07031423e-01 -1.07014263e+00 3.50116193e-01 1.70959997e+00
4.32601273e-01 -6.56833827e-01 7.38473237e-01 9.82551575e-01
-1.15782067e-01 -4.86610204e-01 -1.19133532e+00 -8.05500448e-01
1.65839508e-01 -9.18681800e-01 8.60966146e-01 8.69682431e-01
2.09383920e-01 1.84388772e-01 -3.45306754e-01 -7.51155391e-02
2.21555501e-01 -3.35499525e-01 1.50081539e+00 -1.07173920e+00
-3.20833772e-01 -1.42716050e-01 -5.12866259e-01 -6.79771245e-01
1.90149382e-01 -6.50101304e-01 2.57960290e-01 -1.44206452e+00
-8.06067288e-02 -5.10777950e-01 5.21454453e-01 5.11404693e-01
-7.56548643e-02 6.93619192e-01 2.70390242e-01 4.80406165e-01
-9.03647423e-01 1.81633189e-01 9.46639478e-01 6.86530694e-02
-1.83663309e-01 -1.23882666e-01 -7.43431091e-01 5.96462548e-01
7.07831025e-01 -5.90006053e-01 -2.94069558e-01 -8.04189265e-01
4.04328167e-01 -3.27855796e-01 4.92232978e-01 -9.04160798e-01
4.16241616e-01 1.62673332e-02 -1.74936935e-01 -1.34110255e-02
3.03022653e-01 -5.86411119e-01 -1.77545413e-01 -3.55602682e-01
-3.57536107e-01 2.01926772e-02 1.19791791e-01 3.11230361e-01
-3.80651534e-01 -3.16100717e-01 1.99350402e-01 -2.55209595e-01
-5.18025994e-01 3.83704901e-03 -4.50315267e-01 5.52632332e-01
8.48012328e-01 -2.19469666e-01 -5.21819890e-01 -5.49783528e-01
-4.79065031e-01 -7.00184926e-02 6.73957646e-01 4.51463550e-01
2.04094037e-01 -1.33093536e+00 -1.05875731e+00 1.11997269e-01
3.68050903e-01 1.32465079e-01 -1.63208947e-01 9.61753249e-01
-4.45622206e-01 2.39244714e-01 1.48229763e-01 -6.33930206e-01
-1.30604899e+00 2.19467580e-01 1.64084896e-01 1.20647691e-01
-1.29357204e-01 8.69979262e-01 -2.83743471e-01 -6.12827182e-01
2.17866644e-01 -2.31142253e-01 1.35571182e-01 3.08211863e-01
9.01344121e-01 5.99310815e-01 5.55791259e-01 -9.36293840e-01
-5.58935940e-01 1.46605343e-01 4.04621094e-01 -6.47961378e-01
1.18888557e+00 -8.73787180e-02 -3.23756099e-01 7.16100693e-01
9.30425286e-01 5.01267195e-01 -9.80461657e-01 -5.04328050e-02
5.52943461e-02 -5.05518556e-01 -5.63257098e-01 -5.46147048e-01
-4.32254136e-01 9.13075328e-01 2.47246206e-01 6.46001160e-01
8.02936673e-01 2.97467321e-01 6.87531829e-01 5.21933675e-01
7.67479181e-01 -1.31475031e+00 -2.44233422e-02 4.70642388e-01
1.03891182e+00 -1.60775161e+00 -2.07959954e-02 -2.09060758e-01
-9.30372477e-01 9.00577426e-01 2.26222992e-01 9.76939648e-02
3.14201325e-01 4.88942444e-01 4.36168581e-01 -1.96336508e-01
-5.69803119e-01 -4.86964554e-01 2.41068527e-01 9.95590031e-01
8.70867670e-01 9.85330790e-02 2.95514357e-03 6.36545837e-01
-2.87356675e-01 3.20097446e-01 5.91770351e-01 9.55779433e-01
-3.17212641e-01 -1.24023604e+00 -7.34076500e-01 1.57550916e-01
-5.71313918e-01 -6.74451739e-02 -1.06369627e+00 6.82806611e-01
3.25113952e-01 1.57953000e+00 -2.29649901e-01 -2.45175049e-01
4.96921390e-01 2.64746815e-01 3.84168833e-01 -8.66547644e-01
-9.44485307e-01 -8.78131092e-02 5.13317466e-01 -5.48282802e-01
-8.06334019e-01 -9.00177419e-01 -8.93380225e-01 -3.71065289e-01
-2.02857807e-01 1.15749232e-01 7.79283047e-01 1.19600141e+00
4.98570830e-01 -1.79510966e-01 2.03711987e-01 -1.19099057e+00
-4.67945009e-01 -8.82637739e-01 -4.80145842e-01 7.17659295e-01
3.82139504e-01 -3.73330504e-01 -2.83600956e-01 1.96524411e-01] | [9.832755088806152, 4.915410995483398] |
97ef07f8-d409-46ee-b6a1-1e8196ac6baf | moviemat-context-aware-movie-recommendation | 2204.13003 | null | https://arxiv.org/abs/2204.13003v1 | https://arxiv.org/pdf/2204.13003v1.pdf | MovieMat: Context-aware Movie Recommendation with Matrix Factorization by Matrix Fitting | Movie Recommender System is widely applied in commercial environments such as NetFlix and Tubi. Classic recommender models utilize technologies such as collaborative filtering, learning to rank, matrix factorization and deep learning models to achieve lower marketing expenses and higher revenues. However, audience of movies have different ratings of the same movie in different contexts. Important movie watching contexts include audience mood, location, weather, etc. Tobe able to take advantage of contextual information is of great benefit to recommender builders. However, popular techniques such as tensor factorization consumes an impractical amount of storage, which greatly reduces its feasibility in real world environment. In this paper, we take advantage of the MatMat framework, which factorizes matrices by matrix fitting to build a context-aware movie recommender system that is superior to classic matrix factorization and comparable in the fairness metric. | ['Hao Wang'] | 2022-04-27 | null | null | null | null | ['movie-recommendation'] | ['miscellaneous'] | [-0.7365543 -0.731469 -0.22899254 -0.6104998 -0.01350747 -0.72249717
0.14971216 0.11632806 -0.18079877 0.41910937 0.5526306 -0.48951477
-0.4528572 -0.81788987 -0.22220637 -0.27221477 -0.05297009 0.06852005
-0.05517368 -0.7475377 0.361671 0.3325703 -1.2994012 0.78090125
0.6981041 1.0013363 0.11518873 0.45293763 -0.2986231 0.87554747
-0.23731951 -0.8558794 0.58388984 0.0291619 -0.2602821 -0.11865993
0.36509588 -0.70035934 -0.58668464 0.7703604 0.26899973 0.71592766
0.37652823 -0.9648837 -1.03649 1.0168426 -0.5729078 0.4767898
0.37976348 -0.42478105 1.3211017 -0.8267913 0.24097124 1.093812
0.45492586 0.10871357 -0.85841584 -0.6452118 0.658123 0.39509323
-1.0127621 -0.22044201 0.4368228 -0.40031415 0.5345878 0.6557109
0.7091063 0.52530366 0.20648438 0.7346223 1.0714519 0.05530317
0.01156021 0.29474366 0.2714182 0.15124154 -0.23056369 -0.10954823
-0.42397553 -0.27741814 0.69696957 1.124687 -0.03580178 0.02908096
-0.9510968 0.9004723 0.46729648 0.20256218 -0.40829635 -0.12218247
0.2190949 0.7815392 0.30482733 0.32854253 -0.3813214 -0.10676205
-1.0349813 0.36181945 0.76525456 0.7419647 0.61854476 0.0762418
0.09019708 0.8244666 0.6534058 0.42725784 0.5746629 -1.1110644
0.10700785 0.5121841 0.1179807 -1.4782469 -0.2652174 -0.5496755
-0.8521514 -0.11559488 0.24175522 -0.34459364 -0.31255892 1.056884
0.3438466 0.2834678 -0.33033273 1.3953894 0.941476 0.6265567
-0.24065922 -0.37186256 1.1003404 -0.92476374 -0.66863346 0.26013657
0.36929688 -1.2069467 0.7474967 0.91328734 -0.91163254 -0.6683837
-0.74728835 0.03562429 -0.11370908 0.13175608 1.3739368 0.81671697
-0.82903314 0.9275947 -0.31074917 -0.12317538 0.14586115 0.61788976
-0.2534576 -0.15977436 -1.1693903 0.44363347 -0.267403 0.16322084
-0.8324092 -0.65196157 -0.03780795 0.07480883 0.3942374 -0.5447572
0.99521524 -1.0618576 -1.647749 0.02266097 0.26508895 -0.21417953
0.12723638 -0.54075503 -0.7872626 -0.46043766 -0.46922895 -0.14695501
0.71765596 -0.65372366 -0.8426708 -0.5222136 0.9725236 0.2971945
-0.6098245 0.32528493 -0.35948285 -0.4922398 0.09566213 -0.9289519
-0.67096 -0.35166562 -0.10458261 -0.00659226 0.40498593 -0.5064778
1.5104234 -2.1577604 0.05985307 0.35653004 0.6027011 0.10137542
0.13028714 0.72278315 0.23158675 -0.13130684 0.952437 0.02882252
-0.18669096 0.11990673 -0.28881872 0.34314936 -0.8706606 0.28584546
-0.63144237 -0.20237626 0.20236354 0.26975062 -0.968471 0.24951926
0.0586382 0.20608035 -0.6408858 0.42171085 0.6780489 -0.16488743
0.38813832 -0.35206288 -0.21186464 0.08876648 -1.48399 1.6704211
-0.672962 0.24466784 0.28008822 -0.8645134 0.96512276 0.24898395
0.61054057 -0.3017177 0.4709671 -0.2502344 0.28808635 -0.4643028
0.906213 -0.02758073 0.29852563 0.71909684 -0.17881913 0.76737833
0.24582915 0.8454436 0.7960755 -0.10520711 -0.12541685 -0.14207445
0.46586516 -0.46144474 0.65237784 0.3925498 0.1519054 0.37166002
0.01777383 -0.65764195 -0.6287429 -0.74596334 0.18503815 1.7366678
0.08684167 -0.946076 -0.10264592 -0.48057884 0.26789832 0.21314831
-0.42117584 0.14719072 -0.15435198 -0.6140045 -0.27164897 0.43156877
-0.03833965 -0.31689346 0.21630402 0.2468768 -0.02385994 -0.6470842
-0.6472994 -0.36446005 -1.0705391 -0.7333546 -0.3373231 -0.14802234
0.3653597 0.99307233 1.0304389 0.29291597 0.10937098 0.12025056
-0.83513385 0.03332561 0.18766226 -0.22033292 0.5225953 0.34909862
0.3541719 -0.911734 -0.96538186 0.65136445 -0.6987805 -0.25698692
0.35177338 0.5246828 0.30030227 0.3355885 0.44066957 -1.3758646
0.86410874 -0.87395483 -0.23834082 0.14382914 -0.8900423 -0.46650216
0.9166607 -0.49836898 -1.0249602 -0.29340792 0.06189668 -0.34769568
0.22802979 0.78610563 -0.1346656 0.06921977 0.724798 -0.2748129
-0.2751137 -0.8818809 0.7588153 0.8435335 -0.08014661 -0.39150497
0.47316405 0.2854191 -0.27872694 -0.2716327 -0.8889335 -0.7292905
-0.3918435 -0.46200073 0.23526089 -0.9768397 -0.9755138 -0.2653751
-0.61615986 0.32625973 0.13845052 0.9323784 0.06822366 0.46130824
-0.7238778 -0.760732 -0.4069814 -1.0498309 0.04524913 0.47090185
0.13778828 -0.8084846 0.07132702 0.8288624 0.59153897 -0.26104808
0.51295406 -0.629148 -0.5861218 -0.30920705 -0.2227193 0.34412548
0.05919162 0.2796724 -0.5849618 -0.4853881 -0.31228724 0.07065272
0.52031225 0.34231266 1.0194347 -0.31624913 -0.07072807 0.5311792
1.3848385 0.32255548 0.4224388 -0.08936415 0.8698923 0.44773716
0.7864175 0.93345356 0.60829747 0.61511326 0.3153718 0.14900835
0.26698127 -0.08429907 0.38381153 1.245625 -0.58033144 -0.164099
-0.23150001 0.16701429 -2.271269 -0.9844686 -0.2100542 2.3504956
0.30350283 -0.0212987 0.5342314 -0.0552101 0.47179437 -0.09409948
-0.4024067 -0.65375394 0.04004347 -0.18770537 0.447925 0.13313226
-0.74476004 0.68375975 6.162766 0.7855337 -1.0571996 0.3235535
0.6068919 -0.66863346 -0.5943945 0.25441122 -0.5944803 0.5221229
0.988821 -0.20135373 0.9519588 0.9986975 0.46744183 0.1668456
-0.7750035 1.1330516 0.10546508 -1.3698589 -0.16811636 0.36202842
0.86680686 -0.01261731 0.22905177 0.37397397 0.6285874 -0.7424829
0.1073229 0.8107505 0.14133331 -0.9558484 0.7263041 0.17091991
-0.8501243 -0.44356948 -1.0044912 -0.6377746 0.22413173 0.9111021
-0.28658515 0.4950114 0.8996787 0.9061776 -0.1705413 1.1070952
0.22174789 0.7212951 -0.18397085 -0.10451843 0.0393497 -0.79722893
0.17250538 0.6114416 0.4241012 0.57315695 0.54424393 0.07655933
-0.18981785 1.0462385 -0.2402992 -0.14857003 0.27223596 1.7725686
-0.48730248 -0.17200948 -0.6651122 0.6715642 0.16663203 0.13510223
-0.74098206 -0.02767699 0.88518125 0.46322253 0.21743518 -0.32363948
0.0436151 -1.4097961 -0.43211356 -1.1482764 0.3734831 -0.57948476
-1.3401557 0.58798146 -0.4977933 -1.2665555 0.22026007 -0.41780803
-0.705043 0.65007424 -1.0305246 -1.1273348 -0.14497983 0.9802927
0.36444107 -0.64536995 0.62551266 0.945551 -0.655452 0.76487297
0.80927825 -0.26782215 1.0269783 -1.1755676 -0.08762413 0.673093
0.42521954 1.2707552 0.63692796 -0.3346775 -1.8623574 -0.5136419
0.5108011 -0.3300828 0.7896636 -0.05460994 -0.39712292 0.38489506
0.11913379 -0.10568219 1.5732119 1.1884351 -0.49857232 -0.695806
-0.9768657 0.49228644 0.81978613 -0.27443898 -0.10658361 0.758333
0.51716983 -0.06491513 -1.4364775 -0.2606192 0.9763253 -1.2459154
0.93656087 -1.1932507 0.5626907 -0.41110003 -0.49708733 -1.2096279
-0.970969 -0.722114 -0.2228191 1.3638769 0.483807 -0.22244658
0.9632646 1.1161444 0.05707208 -0.83270437 -0.27464545 -0.22560309
-0.19662097 -0.37068588 1.0627584 0.99916023 0.32488614 0.55406743
-1.1425223 -0.0290148 0.22126712 0.6995397 1.0276334 -1.3935122
-0.88210166 -0.10215873 -0.3286026 -1.2739631 -0.438482 -0.78270394
-0.6659616 -1.4269905 0.16214156 -0.9889946 -0.97953314 0.06206822
-0.01598378 0.5042078 0.41877955 0.44496173 -0.98117065 0.11227851
1.4563084 0.11291561 -0.10673398 0.41415378 -1.2159709 0.4288319
0.6391726 -0.22886887 -0.7222017 -0.59186465 1.0298765 0.413977
-0.6334638 -0.5799282 0.31239334 -0.5694928 0.37819737 -0.38238695
0.34901112 -0.9348954 0.46018 -0.20312564 -0.30847457 0.34318775
-0.40207297 0.7126301 -0.07228904 -0.23881266 0.20589077 -0.1901147
-0.5098428 0.6734577 -0.38603356 -0.454213 0.726861 -0.02476529
-0.2687311 -0.8249818 -0.92703855 0.10574415 0.23877369 0.56394935
0.5069262 -1.2820392 -0.66959566 -0.19391227 -0.11503522 -0.79917014
0.95278984 0.8016222 -0.37036148 0.10162601 -0.2944682 0.09386513
-1.4078212 0.85115856 -0.227211 -0.11730899 -0.25595552 1.0477419
0.06341209 -0.31706464 0.04322749 -0.04922835 -0.73512816 0.16702326
0.9301741 0.56594837 -0.0443432 -0.62778133 -0.13023467 0.15790275
-0.53839886 0.26050925 1.31852 -0.46088275 -0.23962787 0.14758384
0.8413904 0.41340148 -0.7436952 -0.34856647 -0.63404524 -1.1687152
0.6117659 -0.7978697 -1.6902024 0.78073895 0.59123063 0.38255778
1.0398599 -0.6380247 1.1098883 0.43725353 0.4986901 -1.1759839
-0.19815457 0.23743984 0.35611174 -1.368639 0.5156387 -0.10347177
-0.8921413 1.1762297 0.38874286 -0.4458635 1.3797308 -0.02728341
0.27558595 -0.07445373 -0.91361666 0.04531025 0.3458845 0.3009419
0.88064456 0.45061266 -0.62509906 1.0754064 -0.21109927 -0.0316514
0.6669067 0.49816746 -0.5298652 -1.4467429 -0.11475802 0.9630064
-0.821881 -0.23834686 -0.01972685 -0.17551585 0.3568656 1.3710823
-0.21378313 -1.001593 0.1285164 -0.56106496 0.1499039 -0.5373432
-1.0708323 0.2999988 0.20734447 -0.6214665 -0.39465284 -0.6633566
-0.71657693 -0.94074464 -0.7297763 0.5337439 0.7790997 0.89259577
0.47333393 0.36689878 1.0447488 -0.3600255 -0.3251828 -0.8482934
-1.0750617 0.43401924 -0.3434481 -0.7576208 -0.04716654 -0.15504712] | [10.056098937988281, 5.761201858520508] |
c5d7af01-eed3-4365-b7f3-ce418e294598 | on-the-validation-of-pansharpening-methods | 2111.07625 | null | https://arxiv.org/abs/2111.07625v1 | https://arxiv.org/pdf/2111.07625v1.pdf | On the validation of pansharpening methods | Validation of the quality of pansharpening methods is a difficult task because the reference is not directly available. In the meantime, two main approaches have been established: validation in reduced resolution and original resolution. In the former approach it is still not clear how the data are to be processed to a lower resolution. Other open issues are related to the question which resolution and measures should be used. In the latter approach the main problem is how the appropriate measure should be selected. In the most comparison studies the results of both approaches do not correspond, that means in each case other methods are selected as the best ones. Thus, the developers of the new pansharpening methods still stand in the front of dilemma: how to perform a correct or appropriate comparison/evaluation/validation. It should be noted, that the third approach is possible, that is to perform the comparison of methods in a particular application with the usage of their ground truth. But this is not always possible, because usually developers are not working with applications. Moreover, it can be an additional computational load for a researcher in a particular application. In this paper some of the questions/problems raised above are approached/discussed. The following component substitution (CS) and high pass filtering (HPF) pansharpening methods with additive and multiplicative models and their enhancements such as haze correction, histogram matching, usage of spectral response functions (SRF), modulation transfer function (MTF) based lowpass filtering are investigated on remote sensing data of WorldView-2 and WorldView-4 sensors. | ['Gintautas Palubinskas'] | 2021-11-15 | null | null | null | null | ['pansharpening'] | ['computer-vision'] | [ 4.61218894e-01 -4.18241322e-01 3.39836061e-01 -1.02583557e-01
-4.63221848e-01 -3.87106419e-01 4.54200476e-01 6.07621133e-01
-5.88119209e-01 9.85368490e-01 -2.87550509e-01 -2.70480543e-01
-6.92514539e-01 -1.17516947e+00 -2.07615837e-01 -1.06949449e+00
2.08110839e-01 2.54242092e-01 6.39935315e-01 -5.27879894e-01
2.25862145e-01 7.07229197e-01 -1.96966684e+00 -6.26239702e-02
1.17096782e+00 7.10669160e-01 5.02402484e-01 5.73315799e-01
-1.64846838e-01 3.21664900e-01 -5.94410896e-01 3.40822227e-02
2.99337476e-01 -6.49804831e-01 -4.28811878e-01 3.81599739e-02
2.56969839e-01 -1.02972411e-01 5.54244697e-01 1.16249847e+00
4.14291441e-01 2.10294351e-01 6.35941744e-01 -7.90422797e-01
5.71733862e-02 1.60027951e-01 -5.43534279e-01 1.90964207e-01
2.21171692e-01 -1.87152207e-01 4.42679763e-01 -6.24253511e-01
3.25679868e-01 7.86528170e-01 5.46969950e-01 -7.09379911e-02
-1.57308888e+00 -4.40261453e-01 -2.32685640e-01 1.73203215e-01
-1.47437429e+00 -1.66054331e-02 4.60466355e-01 -6.33130074e-01
5.33733070e-01 4.88314092e-01 7.52026439e-01 5.12161553e-01
1.29611254e-01 -3.26894850e-01 1.70630121e+00 -6.23137832e-01
2.37327844e-01 4.37520623e-01 1.83101535e-01 -5.45996539e-02
7.40527451e-01 2.55491614e-01 -5.46125248e-02 2.28134939e-03
5.73326349e-01 -2.52425313e-01 -6.35002792e-01 -1.46555752e-01
-7.03910053e-01 8.48110259e-01 2.29719251e-01 9.37853396e-01
-5.55900812e-01 -3.56022179e-01 1.35892570e-01 4.37116534e-01
5.23983121e-01 4.92122382e-01 -2.54434109e-01 8.69308948e-04
-1.20650136e+00 5.55170715e-01 6.01659596e-01 3.07913125e-01
1.01057243e+00 1.18974738e-01 3.66918743e-01 6.84899032e-01
3.76195014e-01 6.88094974e-01 1.11529835e-01 -5.40535629e-01
4.60412055e-02 4.98177856e-01 2.96056777e-01 -1.19657540e+00
-3.76952678e-01 -3.37128848e-01 -7.28986681e-01 9.80965197e-01
5.91999173e-01 -2.06768706e-01 -5.43456614e-01 1.26524794e+00
2.95330226e-01 -1.17210485e-01 2.77989477e-01 9.51501489e-01
5.12265325e-01 9.98559356e-01 1.55804073e-02 -5.35823643e-01
1.34659171e+00 -3.03510666e-01 -1.03755188e+00 -3.84715721e-02
2.31465906e-01 -1.24985373e+00 7.97828436e-01 6.50406301e-01
-6.84310198e-01 -6.21016383e-01 -1.16325700e+00 3.73559535e-01
-9.03681099e-01 8.26997459e-02 1.40618443e-01 7.50240505e-01
-8.20884585e-01 9.41043735e-01 -6.66396201e-01 -6.65659904e-01
-4.89625335e-01 2.06897095e-01 -3.47406387e-01 2.45880440e-01
-1.14583552e+00 1.27254546e+00 4.66661155e-01 3.61167461e-01
5.38297519e-02 -6.04916930e-01 -4.27729249e-01 -3.75637505e-03
2.22214371e-01 -1.51582643e-01 7.71930754e-01 -1.17752671e+00
-1.40097678e+00 6.91731453e-01 2.52689242e-01 -3.51986080e-01
6.18299544e-01 -2.55994529e-01 -6.83466434e-01 -2.99635511e-02
3.93477380e-02 -5.57866544e-02 6.37817860e-01 -1.43454003e+00
-7.03136861e-01 -2.59529084e-01 -5.22258393e-02 6.15677126e-02
1.94324657e-01 2.48330999e-02 9.44838896e-02 -5.68229914e-01
2.70175219e-01 -7.23900080e-01 -3.72666568e-02 -1.59944341e-01
2.43232459e-01 2.10067898e-01 8.70985329e-01 -8.32750320e-01
1.24842048e+00 -2.06909227e+00 -5.36635518e-02 3.84024441e-01
-2.70511091e-01 5.38620114e-01 2.94614583e-01 6.53490663e-01
-1.45722359e-01 -8.44913051e-02 -3.70427161e-01 2.51535773e-01
-3.50549549e-01 2.19911411e-01 -2.27094010e-01 5.21113336e-01
-5.49973547e-02 -1.66558221e-01 -7.44743466e-01 -4.69532073e-01
7.34643757e-01 7.70884871e-01 -7.07931519e-02 6.16739914e-02
5.49133494e-02 3.88865978e-01 -3.37425172e-01 1.42892435e-01
1.25014210e+00 4.65977997e-01 1.05212338e-01 -3.79796237e-01
-8.07942748e-01 9.22095850e-02 -1.81100488e+00 9.03517246e-01
-4.30711150e-01 6.22206151e-01 2.33662128e-01 -8.90409231e-01
1.28462851e+00 5.24059176e-01 3.13090712e-01 -7.51942396e-01
-1.49442419e-01 8.56552660e-01 -3.17933824e-04 -5.71921766e-01
5.68777204e-01 -4.61064219e-01 7.20323622e-01 -2.15428583e-02
-2.98420608e-01 -4.42457616e-01 4.56433386e-01 -5.91571987e-01
2.98058003e-01 3.45635444e-01 4.43867713e-01 -5.40368855e-01
8.75958741e-01 1.80074632e-01 3.07033241e-01 4.67048436e-01
2.28798270e-01 6.55364394e-01 4.00950134e-01 -2.08722651e-01
-9.90789533e-01 -7.00996697e-01 -5.36513865e-01 5.55356562e-01
3.22487950e-02 -1.58893481e-01 -6.91352367e-01 1.87922288e-02
-3.19875628e-01 7.67485917e-01 -5.18241465e-01 2.08248079e-01
-3.60499114e-01 -9.44203794e-01 1.90595798e-02 -5.78233376e-02
5.28860688e-01 -8.17860723e-01 -1.24199915e+00 5.07555008e-01
3.55812311e-02 -7.41672099e-01 5.74121952e-01 1.90628931e-01
-1.23585093e+00 -1.11966705e+00 -7.53969729e-01 -1.38778463e-01
4.45411235e-01 5.25429845e-01 8.27146888e-01 1.13592230e-01
9.01548564e-02 -3.29492167e-02 -7.65830457e-01 -4.48864371e-01
-6.56146467e-01 -9.56590548e-02 -4.90473777e-01 -3.53336483e-02
2.68014371e-01 -6.12213552e-01 -5.06882131e-01 3.75361860e-01
-1.15588474e+00 -1.35862023e-01 4.99693125e-01 4.61221009e-01
4.55789626e-01 3.67222965e-01 1.51051894e-01 -9.71601427e-01
4.72219795e-01 -1.03434801e-01 -1.17442882e+00 2.11575702e-01
-8.19530070e-01 -1.99399620e-01 5.53632081e-01 -6.12698644e-02
-1.12850595e+00 -1.15236118e-01 -3.86883557e-01 1.88698769e-01
-5.12662828e-01 6.77148879e-01 -3.24270390e-02 -2.69035667e-01
9.94491518e-01 8.74579251e-02 7.85003528e-02 -7.50279665e-01
-1.98849455e-01 6.60521030e-01 3.17006469e-01 -3.15287024e-01
9.41331267e-01 3.53010476e-01 1.79519981e-01 -1.40284538e+00
-3.28831643e-01 -5.40261805e-01 -3.96130323e-01 -3.83561879e-01
9.42634583e-01 -5.23326874e-01 -2.65272826e-01 4.26963151e-01
-1.04371083e+00 -2.02884048e-01 -2.83275813e-01 6.90699816e-01
-1.98986128e-01 5.00427306e-01 3.46802138e-02 -1.09785426e+00
-1.48185417e-01 -1.26360810e+00 6.26665890e-01 4.17766064e-01
-5.51992767e-02 -7.39651918e-01 2.52369761e-01 1.22125484e-01
7.35472977e-01 5.52831233e-01 7.70674229e-01 2.83182156e-03
-2.09255084e-01 -2.67819285e-01 -1.21806093e-01 5.71923733e-01
3.52745712e-01 6.07900620e-01 -9.64509487e-01 -1.67341441e-01
2.93564051e-01 4.56830323e-01 6.00055397e-01 5.29550791e-01
4.97184873e-01 4.79368353e-03 4.05913703e-02 4.56291080e-01
2.08760667e+00 3.79951537e-01 1.05108953e+00 6.80313647e-01
1.07846335e-02 9.49738324e-01 1.00810063e+00 1.55785993e-01
-3.31821203e-01 8.91433954e-01 5.15537024e-01 -2.09929109e-01
7.42629766e-02 2.61880994e-01 2.00235754e-01 1.14972092e-01
-8.22927535e-01 -1.38102159e-01 -8.13211203e-01 3.32292199e-01
-1.62047136e+00 -1.00618613e+00 -9.47448134e-01 2.81751370e+00
2.80274928e-01 2.28469849e-01 4.04172577e-02 7.79261470e-01
6.86492026e-01 1.79230914e-01 3.61058712e-01 -4.57761347e-01
-2.08106443e-01 3.03028345e-01 6.72365010e-01 8.89831960e-01
-1.10387838e+00 1.55218065e-01 4.61098480e+00 5.84238112e-01
-1.53100586e+00 1.60809308e-01 -8.43380690e-02 4.96697664e-01
-8.48281607e-02 3.85922372e-01 -5.21857023e-01 4.63803053e-01
1.01420403e+00 -3.03215999e-02 2.80292779e-01 4.25501466e-01
6.77477002e-01 -8.69616866e-01 -2.30905607e-01 7.85098314e-01
-2.38254517e-01 -8.04452181e-01 -3.79320651e-01 6.02437332e-02
4.43844169e-01 -1.31493628e-01 -3.30145955e-01 -2.20524698e-01
-4.75830078e-01 -7.48323739e-01 6.20444775e-01 6.69065118e-01
2.93115526e-01 -6.42579317e-01 1.12379110e+00 1.54005006e-01
-1.13506746e+00 1.46000087e-01 -4.93945569e-01 2.17996873e-02
2.04119384e-01 6.83834553e-01 -2.29484722e-01 1.02955115e+00
7.41895139e-01 1.38191029e-01 -4.32006449e-01 1.45151150e+00
-2.54139781e-01 5.53092957e-01 -5.85627198e-01 6.51560053e-02
1.25097439e-01 -7.77366638e-01 6.20885551e-01 1.29316115e+00
7.24982262e-01 1.31593645e-01 -4.04510975e-01 9.35181320e-01
1.03133702e+00 6.10171974e-01 -5.19305110e-01 1.48124725e-01
1.40407205e-01 1.26875448e+00 -9.34990108e-01 -3.27124596e-02
-5.07321894e-01 4.91712689e-01 -6.65139735e-01 2.96412051e-01
-5.72465777e-01 -3.27733159e-01 3.91997546e-01 6.17111921e-01
1.50221273e-01 -1.36438608e-01 3.07481438e-02 -6.30677819e-01
-5.39075956e-02 -8.86634648e-01 3.17664027e-01 -8.45492899e-01
-7.59022713e-01 5.68225265e-01 4.72272307e-01 -1.35901356e+00
-5.17067611e-02 -7.32925832e-01 -3.57363075e-01 1.16511834e+00
-1.46753311e+00 -7.08701313e-01 -6.84611320e-01 1.35886610e-01
1.69488043e-01 3.54703099e-01 8.98854077e-01 6.68853343e-01
-1.55040920e-01 -3.41821522e-01 4.34641719e-01 -4.57890302e-01
5.94958127e-01 -1.23120952e+00 -4.61650312e-01 1.03034520e+00
-2.07248405e-01 2.70530254e-01 1.45556700e+00 -5.31417727e-01
-7.50299454e-01 -4.80027854e-01 1.06901062e+00 3.32115859e-01
5.43230355e-01 8.26656371e-02 -1.44524622e+00 1.08384170e-01
2.38591433e-01 -2.00572997e-01 4.11773682e-01 -1.79931760e-01
1.81228846e-01 -4.97603357e-01 -1.11102152e+00 2.57881701e-01
6.60699978e-02 -2.00786263e-01 -6.67978704e-01 1.24394953e-01
-4.81101498e-02 -2.08866253e-01 -1.00079632e+00 5.95479667e-01
4.79235440e-01 -1.60456848e+00 7.06123471e-01 2.36858934e-01
1.66009754e-01 -8.55075002e-01 -5.99267855e-02 -1.08013904e+00
-1.04090497e-02 -2.34447271e-01 7.22208321e-01 1.40113163e+00
3.33350778e-01 -9.52870786e-01 2.45709419e-01 2.95650750e-01
1.93845153e-01 -3.13042819e-01 -8.51557374e-01 -6.13645852e-01
-1.01483658e-01 -3.12671840e-01 4.72963452e-01 7.87191212e-01
-4.68337297e-01 -7.61330035e-03 -1.46070570e-01 5.14836192e-01
4.52912360e-01 2.71225214e-01 7.15794802e-01 -1.59268832e+00
-2.60156095e-01 -4.33639854e-01 -3.23447078e-01 -6.36145845e-02
-5.52839994e-01 -5.81656434e-02 -1.59742743e-01 -1.72574246e+00
-3.98557156e-01 -4.31673765e-01 -1.45409286e-01 4.43735858e-03
5.33343013e-03 5.19629717e-02 2.51624107e-01 1.72940865e-01
5.22333741e-01 9.82488692e-02 8.86136711e-01 1.19117081e-01
-3.79760981e-01 3.30726027e-01 -1.45496458e-01 6.54456258e-01
7.36817360e-01 -5.73758304e-01 -4.04023916e-01 -9.99262854e-02
6.44326925e-01 2.76204403e-02 4.94975448e-01 -1.30052340e+00
4.78182733e-02 -2.16628641e-01 4.18654224e-03 -6.29217029e-01
2.53488392e-01 -1.40511453e+00 9.48273361e-01 5.83176792e-01
1.59979224e-01 9.87564772e-02 2.78797448e-01 1.10919796e-01
-3.59537721e-01 -7.82706022e-01 1.15313411e+00 -1.46733582e-01
-8.01869631e-01 -2.87614793e-01 -5.41267037e-01 -5.88496387e-01
9.26579356e-01 -6.48925483e-01 -2.40109876e-01 -3.91403705e-01
-6.52695894e-01 -2.38976762e-01 6.43314540e-01 -8.61544907e-02
3.19157243e-01 -7.61775732e-01 -8.08173001e-01 -6.54099956e-02
1.16893396e-01 -8.04403871e-02 5.51745415e-01 1.19370615e+00
-1.11987746e+00 2.86427643e-02 -3.21345061e-01 -3.20604175e-01
-1.41208112e+00 5.11768043e-01 6.45574927e-01 -6.10890761e-02
-4.25045907e-01 1.32669166e-01 -2.02265695e-01 5.85836591e-03
-3.53182584e-01 -2.19630390e-01 -6.43348157e-01 5.23198962e-01
4.25668687e-01 6.95469439e-01 4.01046634e-01 -9.38873947e-01
-1.45761147e-01 9.88786757e-01 5.60097039e-01 -1.65669039e-01
1.48158383e+00 -1.51929945e-01 -4.68428254e-01 5.94058394e-01
7.06687391e-01 3.35866362e-01 -9.04151559e-01 2.29806453e-01
-3.56935300e-02 -6.43390477e-01 2.36744523e-01 -6.21488750e-01
-8.65349054e-01 9.12151814e-01 1.07264030e+00 8.40897799e-01
1.47265708e+00 -6.15370512e-01 -8.88405293e-02 2.74609998e-02
2.00844705e-01 -1.34864330e+00 -9.36990917e-01 3.90273444e-02
1.01362896e+00 -9.49806631e-01 4.04331833e-01 -6.52846158e-01
-1.12038873e-01 1.63208866e+00 1.32145733e-01 -3.18559147e-02
7.20691204e-01 2.00278267e-01 3.79474998e-01 -1.36859208e-01
-1.40179023e-01 -6.09944284e-01 8.85128677e-02 5.44516206e-01
7.62166619e-01 7.48558864e-02 -1.30421531e+00 -1.71952426e-01
-2.71728002e-02 1.29112571e-01 5.64675152e-01 7.39649773e-01
-6.86323404e-01 -1.22500908e+00 -1.11000979e+00 2.27565661e-01
-4.08700913e-01 2.05597028e-01 -7.11658448e-02 1.23380935e+00
6.01833165e-01 1.16196144e+00 -2.76653141e-01 -6.60783425e-02
7.03945279e-01 -6.32013083e-02 2.57152289e-01 -3.67532343e-01
-5.90113163e-01 3.80688846e-01 1.25073656e-01 -1.49804816e-01
-9.83507514e-01 -7.07786441e-01 -8.38691354e-01 -3.05223942e-01
-6.31320119e-01 6.06594145e-01 1.03612995e+00 7.15438008e-01
-2.97117651e-01 3.60599816e-01 3.39300483e-01 -8.03812265e-01
-3.29741836e-01 -8.55506122e-01 -9.57542002e-01 1.67469814e-01
3.17470640e-01 -8.34097922e-01 -7.04499662e-01 -1.31810367e-01] | [10.008824348449707, -2.0457656383514404] |
e188b92e-de74-499f-827d-bc08a63470ce | camliflow-bidirectional-camera-lidar-fusion | 2111.10502 | null | https://arxiv.org/abs/2111.10502v4 | https://arxiv.org/pdf/2111.10502v4.pdf | CamLiFlow: Bidirectional Camera-LiDAR Fusion for Joint Optical Flow and Scene Flow Estimation | In this paper, we study the problem of jointly estimating the optical flow and scene flow from synchronized 2D and 3D data. Previous methods either employ a complex pipeline that splits the joint task into independent stages, or fuse 2D and 3D information in an "early-fusion" or "late-fusion" manner. Such one-size-fits-all approaches suffer from a dilemma of failing to fully utilize the characteristic of each modality or to maximize the inter-modality complementarity. To address the problem, we propose a novel end-to-end framework, called CamLiFlow. It consists of 2D and 3D branches with multiple bidirectional connections between them in specific layers. Different from previous work, we apply a point-based 3D branch to better extract the geometric features and design a symmetric learnable operator to fuse dense image features and sparse point features. Experiments show that CamLiFlow achieves better performance with fewer parameters. Our method ranks 1st on the KITTI Scene Flow benchmark, outperforming the previous art with 1/7 parameters. Code is available at https://github.com/MCG-NJU/CamLiFlow. | ['Lijun Chen', 'Wenjie Li', 'Jia Liu', 'Yihui Xu', 'Tao Lu', 'Haisong Liu'] | 2021-11-20 | null | http://openaccess.thecvf.com//content/CVPR2022/html/Liu_CamLiFlow_Bidirectional_Camera-LiDAR_Fusion_for_Joint_Optical_Flow_and_Scene_CVPR_2022_paper.html | http://openaccess.thecvf.com//content/CVPR2022/papers/Liu_CamLiFlow_Bidirectional_Camera-LiDAR_Fusion_for_Joint_Optical_Flow_and_Scene_CVPR_2022_paper.pdf | cvpr-2022-1 | ['scene-flow-estimation'] | ['computer-vision'] | [-2.28211284e-01 -5.18098533e-01 -2.71213233e-01 -3.47957104e-01
-6.00956380e-01 -6.00545526e-01 4.72825587e-01 -2.06929550e-01
-2.84878492e-01 4.74289924e-01 3.75368804e-01 -1.37035355e-01
-5.77561446e-02 -4.65754807e-01 -4.69416410e-01 -5.06128728e-01
-2.29976885e-02 1.98338076e-01 3.93126249e-01 4.70103770e-02
4.41417068e-01 5.03957927e-01 -1.36924851e+00 2.15635881e-01
7.95184493e-01 1.02165663e+00 1.28311262e-01 8.87716830e-01
-1.59604788e-01 1.06642365e+00 5.85964210e-02 -1.45778388e-01
7.23648071e-01 -4.50184286e-01 -9.07236397e-01 9.98331681e-02
8.83393586e-01 -6.81289792e-01 -6.54386044e-01 7.18490064e-01
5.42546511e-01 2.22289458e-01 3.45625967e-01 -1.42775285e+00
-3.42834353e-01 1.03094421e-01 -1.01162398e+00 4.52494949e-01
4.16558892e-01 4.60563302e-01 1.02485311e+00 -1.10636854e+00
6.31010652e-01 1.31017780e+00 5.14062405e-01 2.66037017e-01
-1.33201420e+00 -5.51332533e-01 2.38943815e-01 2.00462088e-01
-1.29505002e+00 -6.83555067e-01 8.65564525e-01 -7.27050245e-01
9.16311502e-01 -1.67678639e-01 6.25698566e-01 7.97941446e-01
5.36386156e-03 9.04463112e-01 1.01888371e+00 -1.20496020e-01
-1.22416548e-01 -2.75325388e-01 3.84475775e-02 1.00112021e+00
1.35601178e-01 2.63281435e-01 -7.82003343e-01 -7.92924091e-02
1.04819465e+00 1.36510089e-01 -5.67340374e-01 -5.74276924e-01
-1.45455956e+00 6.85733318e-01 6.65312707e-01 1.14005215e-01
-2.34308571e-01 2.56186157e-01 1.92013368e-01 1.57910377e-01
5.96204102e-01 2.06138968e-01 -3.72589946e-01 -3.51988643e-01
-9.14526105e-01 3.33326310e-01 6.73089385e-01 8.65933955e-01
1.18510258e+00 -2.38616779e-01 -1.09888099e-01 5.70277333e-01
4.14847940e-01 3.98118287e-01 1.48397312e-01 -1.37287462e+00
5.99579453e-01 3.09935927e-01 9.17733833e-02 -9.91590738e-01
-4.12379920e-01 -2.47207031e-01 -6.52549505e-01 2.87405998e-01
5.43914497e-01 -1.22118659e-01 -9.14886713e-01 1.60696077e+00
4.91429776e-01 9.29131269e-01 -1.63436532e-01 1.28631759e+00
8.51572096e-01 7.04794228e-01 -1.34564057e-01 1.32984370e-01
9.31278646e-01 -1.48067975e+00 -3.78983587e-01 -2.75469512e-01
6.62532628e-01 -9.76094246e-01 8.01596224e-01 1.17022194e-01
-1.24204385e+00 -6.45090401e-01 -7.96744764e-01 -6.02378666e-01
-1.17229395e-01 -5.67854941e-02 6.98423505e-01 2.67185569e-01
-1.17012918e+00 4.68148887e-01 -1.08419168e+00 -1.93088666e-01
6.34893239e-01 1.32136315e-01 -5.35412908e-01 -3.58904213e-01
-7.34167218e-01 6.83286130e-01 1.17030693e-03 1.92305714e-01
-7.03043044e-01 -1.21899152e+00 -1.18994236e+00 -7.73106441e-02
2.81836569e-01 -1.21260405e+00 1.10115600e+00 -5.51487923e-01
-1.58276856e+00 8.31989586e-01 -5.68706036e-01 -1.01675212e-01
6.16726875e-01 -4.18529242e-01 2.07885310e-01 4.41667587e-01
2.22959131e-01 8.62838209e-01 8.02991629e-01 -1.18963730e+00
-7.64539659e-01 -2.33655378e-01 2.06549063e-01 1.58069074e-01
3.45829315e-02 -2.53118634e-01 -7.11730301e-01 -4.16641861e-01
1.74554288e-01 -9.95831788e-01 -2.72644430e-01 3.56493801e-01
-3.12396109e-01 -2.42702011e-02 8.63706708e-01 -3.21879536e-01
8.96363616e-01 -2.15710521e+00 2.08553284e-01 -2.96951681e-02
6.56752229e-01 2.86784142e-01 -2.68126011e-01 2.76213288e-01
1.01385295e-01 -1.19800277e-01 -2.11665109e-01 -9.77300584e-01
-1.48557052e-01 1.10088177e-01 -1.86237216e-01 7.69870758e-01
4.27145422e-01 9.53716755e-01 -1.17081714e+00 -5.93961477e-01
6.66910052e-01 7.04795301e-01 -8.69825959e-01 3.55766505e-01
2.51344025e-01 8.18700254e-01 -5.25541425e-01 6.59044921e-01
8.54172945e-01 -5.70410907e-01 -3.79022509e-01 -4.01178211e-01
-3.30020130e-01 3.93161446e-01 -1.21345556e+00 2.32882643e+00
-5.12837827e-01 7.62736320e-01 1.35736957e-01 -8.31861258e-01
6.68118000e-01 1.81295440e-01 8.97758365e-01 -5.08758068e-01
5.91500588e-02 4.69727606e-01 -2.04586715e-01 -6.78050220e-01
3.33744645e-01 1.24481320e-02 2.73812056e-01 2.11717382e-01
3.04243892e-01 -2.37089366e-01 2.74651498e-01 3.41664851e-01
1.13645887e+00 5.95868886e-01 1.43505827e-01 -1.30230591e-01
6.55454457e-01 -5.84653206e-02 7.73863316e-01 7.09650517e-01
-5.68756163e-01 1.08797395e+00 6.07743204e-01 -4.44432765e-01
-7.94873893e-01 -1.13700533e+00 -1.25744464e-02 5.20113528e-01
5.22297919e-01 -5.56132674e-01 -1.61628202e-01 -7.35641718e-01
2.91531593e-01 1.58955142e-01 -5.44965982e-01 1.26982465e-01
-7.07345366e-01 -4.06627536e-01 1.95351675e-01 3.98930967e-01
6.11575484e-01 -5.54784715e-01 -8.28230739e-01 1.01361096e-01
-4.76198465e-01 -1.55951226e+00 -7.55563140e-01 -2.23791867e-01
-7.80577838e-01 -1.11223114e+00 -6.81433737e-01 -4.62045580e-01
4.06128377e-01 8.76001060e-01 1.10271311e+00 1.33655056e-01
-3.10492218e-01 5.31786919e-01 -4.57756579e-01 1.64449820e-03
2.30137229e-01 1.36282355e-01 -2.70067096e-01 2.89631367e-01
1.12792462e-01 -8.55907738e-01 -9.70649004e-01 1.46336138e-01
-6.99497283e-01 2.31611446e-01 3.65271151e-01 6.93392754e-01
3.53144735e-01 -4.72442567e-01 2.51274168e-01 -5.02337515e-01
1.54342487e-01 -4.66346949e-01 -5.12402475e-01 -1.33661613e-01
-2.45085344e-01 2.67181452e-03 4.74630356e-01 -6.18384555e-02
-9.09469664e-01 3.68820846e-01 -2.68021114e-02 -1.05474710e+00
-1.69999361e-01 2.76236862e-01 1.53593393e-02 -2.53845006e-01
3.56220692e-01 -7.82842636e-02 1.20343223e-01 -5.03238082e-01
4.60712790e-01 3.40509623e-01 5.72517514e-01 -5.69016993e-01
8.06995988e-01 9.27115381e-01 1.98141515e-01 -7.52346992e-01
-1.09089923e+00 -8.93759310e-01 -8.72995496e-01 -3.85955393e-01
9.23707306e-01 -1.07323527e+00 -6.90379024e-01 6.17331982e-01
-1.27828813e+00 -3.76092613e-01 -2.36653343e-01 8.93185377e-01
-6.19035006e-01 4.55856979e-01 -6.58303857e-01 -4.60350692e-01
-9.86953750e-02 -1.40148664e+00 1.36008203e+00 3.25437456e-01
9.81262177e-02 -1.04922724e+00 2.79962122e-01 4.31939721e-01
5.39319038e-01 2.49812633e-01 1.40625939e-01 -1.31673008e-01
-8.29236686e-01 5.87899871e-02 -5.72035253e-01 2.61160910e-01
1.44473329e-01 3.87234949e-02 -1.10152733e+00 -3.04195195e-01
-1.50412917e-01 -4.00568098e-01 1.20194888e+00 4.92038429e-01
8.58544409e-01 1.57004580e-01 -1.25559017e-01 1.17999220e+00
1.57176602e+00 -1.93308756e-01 4.64621574e-01 1.11544833e-01
1.11459255e+00 5.94630539e-01 3.39537352e-01 4.64662284e-01
8.34917605e-01 6.02959991e-01 3.77743036e-01 -2.23822638e-01
-3.78995568e-01 -2.13595569e-01 3.70481133e-01 6.04029536e-01
-1.53522789e-01 -2.44307756e-01 -8.34676683e-01 5.99179149e-01
-2.07820678e+00 -8.88366044e-01 -1.83838382e-01 1.93135834e+00
4.17539567e-01 7.77089782e-03 1.63201496e-01 -1.56445324e-01
4.37141418e-01 4.15893555e-01 -5.32697022e-01 -2.13838019e-03
2.64934134e-02 1.06133960e-01 5.12449443e-01 9.13170516e-01
-1.18005610e+00 9.90748584e-01 5.06262970e+00 3.72550756e-01
-1.29138052e+00 6.74514621e-02 4.79501754e-01 -3.30485255e-01
-1.32079273e-01 4.33455914e-01 -7.69591868e-01 4.19878811e-01
4.58910078e-01 3.01131345e-02 2.02861249e-01 3.44151914e-01
4.03283924e-01 -2.85075426e-01 -1.12111163e+00 1.22616935e+00
5.32129072e-02 -1.54496586e+00 -1.52768567e-01 1.25062346e-01
7.25364149e-01 3.84099156e-01 -2.30652630e-01 -1.01335429e-01
1.85975701e-01 -7.60938585e-01 7.12765217e-01 7.57106662e-01
4.82368797e-01 -2.61233568e-01 4.76776868e-01 1.30000249e-01
-1.43294501e+00 1.01289503e-01 -4.71317917e-02 -2.18048394e-01
5.99162221e-01 7.39979386e-01 -1.95108518e-01 7.73570418e-01
8.50570142e-01 1.48462677e+00 -4.32940781e-01 1.45688021e+00
-5.95805123e-02 4.11272138e-01 -5.42272508e-01 4.99462932e-01
4.03505832e-01 -2.38798559e-01 7.20524371e-01 1.15770757e+00
3.94652873e-01 1.90781839e-02 4.97888058e-01 8.79067898e-01
1.02005810e-01 -2.37562567e-01 -5.66333175e-01 3.56949627e-01
3.49645853e-01 1.34931922e+00 -6.04885519e-01 -3.35427880e-01
-6.99476302e-01 9.61652160e-01 3.57753694e-01 4.21346366e-01
-7.11836040e-01 -3.61943126e-01 8.05957735e-01 1.48494735e-01
4.74800080e-01 -6.14473879e-01 -2.09874332e-01 -1.59046185e+00
1.56082198e-01 -2.06257224e-01 3.77846032e-01 -7.81840682e-01
-1.38891780e+00 4.42487866e-01 1.13047421e-01 -1.39921021e+00
-1.31266862e-01 -5.52472055e-01 -6.79227173e-01 9.80635405e-01
-2.10440469e+00 -1.22033536e+00 -6.33500993e-01 8.35320413e-01
3.76206160e-01 2.57809758e-01 2.43616655e-01 5.36527514e-01
-6.89808488e-01 2.20181510e-01 -3.71227056e-01 2.15838090e-01
9.44834650e-01 -1.19050992e+00 2.52786189e-01 1.09159482e+00
9.12445262e-02 2.86865741e-01 3.61110747e-01 -2.21354485e-01
-1.44687700e+00 -8.37954581e-01 1.09127700e+00 -5.46444535e-01
6.44531369e-01 -2.46940449e-01 -8.18688750e-01 6.70491755e-01
1.72515571e-01 7.09379137e-01 3.54091018e-01 -1.77479953e-01
-4.79184806e-01 -2.27060720e-01 -8.86142552e-01 3.82408798e-01
1.35175323e+00 -3.86858493e-01 -2.85949260e-01 9.91365314e-02
7.60114849e-01 -5.89308858e-01 -9.05067682e-01 3.98419350e-01
4.84187871e-01 -1.39690793e+00 1.10093558e+00 -2.54867822e-01
6.89853370e-01 -6.48859680e-01 -9.68693048e-02 -1.18671072e+00
-2.88297892e-01 -8.95129681e-01 -2.52050549e-01 9.78074431e-01
2.10230947e-01 -7.56823659e-01 6.22877955e-01 3.88209850e-01
-3.26563299e-01 -7.15338051e-01 -8.96125376e-01 -5.96279979e-01
-2.20797993e-02 -4.35847193e-01 3.95011932e-01 9.20899451e-01
-1.53212756e-01 4.54382271e-01 -4.21259552e-01 1.89357892e-01
7.43148148e-01 4.97005433e-01 1.03966963e+00 -9.67509508e-01
-2.17123210e-01 -6.54806852e-01 -5.83366096e-01 -1.64531183e+00
1.26328498e-01 -8.94215167e-01 -1.38848526e-02 -1.52231812e+00
9.68007278e-03 -4.47622538e-01 -1.44251660e-01 4.81143177e-01
-2.38705263e-01 3.07287842e-01 4.63926166e-01 3.67634326e-01
-6.10591114e-01 6.44738078e-01 1.56641543e+00 1.74986511e-01
-3.41249436e-01 -2.40088671e-01 -5.54227352e-01 4.77494717e-01
6.87514305e-01 -2.74626702e-01 -5.20432949e-01 -7.82334566e-01
-7.51845762e-02 2.41840735e-01 7.93200254e-01 -9.88498926e-01
4.41322595e-01 -1.35223970e-01 1.78784743e-01 -5.50958276e-01
3.96445215e-01 -6.76952183e-01 -3.28735143e-01 1.44210294e-01
-2.04912081e-01 1.56517386e-01 4.38876301e-02 4.98440146e-01
-3.17486048e-01 1.71665147e-01 8.65593970e-01 -6.75483942e-02
-7.52752900e-01 8.10250103e-01 4.98514585e-02 2.34012216e-01
8.83601546e-01 -1.07217610e-01 -4.22074795e-01 -3.18366885e-01
-4.31773752e-01 5.19006193e-01 4.64544624e-01 6.71544611e-01
5.91168821e-01 -1.16437161e+00 -9.24196124e-01 4.38336849e-01
8.25133622e-02 4.12767410e-01 4.24859405e-01 1.36829770e+00
-7.62239993e-01 3.16987753e-01 -1.98929638e-01 -1.06047511e+00
-7.84211397e-01 3.19576383e-01 5.34304857e-01 -2.32523069e-01
-8.86313736e-01 8.51850688e-01 2.92247057e-01 -3.75456572e-01
7.16866404e-02 -2.52546728e-01 7.30397329e-02 -1.35276720e-01
4.60813165e-01 3.69389772e-01 -1.10987976e-01 -8.01574826e-01
-4.34479177e-01 8.93904746e-01 3.48739396e-03 -1.50741875e-01
1.26128066e+00 -4.02377278e-01 -1.18335053e-01 3.66653651e-01
1.69083095e+00 -3.57282050e-02 -1.77608430e+00 -3.77802253e-01
-3.51220846e-01 -1.07205248e+00 2.63019592e-01 -2.86081165e-01
-1.51305890e+00 1.03465664e+00 3.04608941e-01 6.15989305e-02
1.23272896e+00 -2.02261899e-02 1.03068686e+00 -1.04966357e-01
8.37026164e-02 -4.97402698e-01 -1.07133025e-02 6.31453156e-01
6.66024625e-01 -1.43796945e+00 8.48987252e-02 -4.82507437e-01
-3.96695346e-01 1.06591451e+00 7.45519876e-01 -3.50248486e-01
1.01453757e+00 2.20818371e-01 1.08008482e-01 -1.97972566e-01
-8.62772167e-01 -4.78526205e-01 3.75188917e-01 3.98029596e-01
4.07272726e-01 -3.31935436e-01 -1.46536911e-02 -1.47071213e-01
6.12374432e-02 2.01740578e-01 3.27206045e-01 1.19434166e+00
-1.59258947e-01 -1.07305741e+00 -1.13974795e-01 1.86120868e-01
-1.32865861e-01 6.11268543e-02 -2.94683352e-02 7.35413492e-01
1.11473411e-01 9.33410227e-01 2.29520693e-01 -4.42170620e-01
3.92284244e-01 -3.33920360e-01 5.86739123e-01 -3.38816106e-01
-3.06173533e-01 2.53399998e-01 -1.56177416e-01 -1.15072596e+00
-8.92237723e-01 -8.21846902e-01 -1.08892787e+00 -5.03976762e-01
9.30681899e-02 -1.77513704e-01 3.78390431e-01 8.58787715e-01
7.16098368e-01 2.13641018e-01 8.70419025e-01 -1.40191412e+00
-9.73093063e-02 -6.47461355e-01 -2.39490956e-01 4.26841944e-01
9.18096066e-01 -8.31248641e-01 -5.66535711e-01 5.18196486e-02] | [8.664029121398926, -1.960135579109192] |
bb415987-801e-4956-a7bd-27f388ba6875 | ecg-classification-system-for-arrhythmia | 2303.03660 | null | https://arxiv.org/abs/2303.03660v1 | https://arxiv.org/pdf/2303.03660v1.pdf | ECG Classification System for Arrhythmia Detection Using Convolutional Neural Networks | Arrhythmia is just one of the many cardiovascular illnesses that have been extensively studied throughout the years. Using a multi-lead ECG data, this research describes a deep learning (DL) technique based on a convolutional neural network (CNN) algorithm to detect cardiovascular arrhythmia in patients. The suggested CNN model has six layers total, two convolution layers, two pooling layers, and two fully linked layers within a residual block, in addition to the input and output layers. In this study, the classification of the ECG signals into five groups, Left Bundle Branch Block (LBBB), Right Bundle Branch Block (RBBB), Atrial Premature Contraction (APC), Premature Ventricular Contraction (PVC), and Normal Beat is the main goal (N). Using the MIT-BIH arrhythmia dataset, we assessed the suggested technique. The findings show that our suggested strategy classified 15000 cases with an average accuracy of 98.2%. | ['Jaskaran Singh Walia', 'Aryan odugoudar'] | 2023-03-07 | null | null | null | null | ['arrhythmia-detection', 'ecg-classification'] | ['medical', 'medical'] | [-9.92323160e-02 -1.12093017e-01 1.51874796e-01 -1.14047222e-01
-2.08481066e-02 -4.79288548e-01 -1.83120683e-01 8.48412588e-02
-2.66865909e-01 9.26399827e-01 -8.73241350e-02 -7.30101407e-01
-4.79063690e-02 -6.10905766e-01 -7.34514818e-02 -5.64369261e-01
-3.62037361e-01 3.59792076e-02 -2.94413745e-01 2.69800723e-01
9.61967260e-02 8.39675963e-01 -7.78512299e-01 5.98667145e-01
7.00854301e-01 1.16989386e+00 -5.30517340e-01 1.05024636e+00
1.85258940e-01 8.51839304e-01 -1.06314862e+00 -7.76429698e-02
1.85679212e-01 -6.79057181e-01 -6.53156519e-01 -4.87024128e-01
-2.24158764e-01 -2.99748659e-01 -4.60067168e-02 4.75853920e-01
1.42374825e+00 -6.28164530e-01 3.38793665e-01 -8.32283556e-01
-2.10155129e-01 5.84515452e-01 -4.21088219e-01 8.58129084e-01
-2.43671209e-01 1.69623673e-01 3.55531931e-01 -8.83772969e-01
2.11456507e-01 5.00305116e-01 1.51753294e+00 2.74087220e-01
-8.93141806e-01 -8.52555394e-01 -7.07526803e-01 -1.28743544e-01
-1.63655055e+00 6.01298772e-02 7.66873658e-01 -6.23069584e-01
1.10445356e+00 1.47785932e-01 1.21547687e+00 6.78114176e-01
7.23680079e-01 3.50774080e-01 7.59552360e-01 -2.38368675e-01
-1.40365520e-02 -1.07960336e-01 4.44841057e-01 6.35867298e-01
5.85666120e-01 1.90421030e-01 -1.29752457e-01 -5.03648520e-01
1.13117063e+00 1.52462259e-01 -3.83825034e-01 2.61489838e-01
-1.24704099e+00 6.62592649e-01 2.80379206e-01 3.89386147e-01
-5.69874048e-01 9.16281864e-02 7.29849875e-01 3.86677265e-01
1.36126682e-01 7.28407085e-01 -7.45606065e-01 -1.68646976e-01
-9.26353633e-01 1.47947282e-01 8.04108799e-01 5.36451161e-01
3.25094223e-01 3.54420722e-01 -5.57443798e-01 7.19755411e-01
1.51863277e-01 5.72437756e-02 6.93986416e-01 -6.49573565e-01
3.77358019e-01 7.34664500e-01 -1.18865743e-01 -1.03452098e+00
-7.93797016e-01 -1.25928831e+00 -1.76898539e+00 1.06620625e-01
1.79732889e-01 -9.71397936e-01 -6.85247540e-01 1.28541422e+00
-2.61281103e-01 1.53677881e-01 2.24037483e-01 7.14507401e-01
1.17145240e+00 4.17749316e-01 -1.07991952e-03 -2.33718336e-01
1.43073881e+00 -6.41718209e-01 -7.15079069e-01 1.51050672e-01
8.16582859e-01 -5.24540722e-01 2.46919706e-01 6.32867157e-01
-1.02931798e+00 -8.33844304e-01 -1.21464610e+00 1.13631189e-01
-2.75377512e-01 7.08149433e-01 5.97300410e-01 7.90055037e-01
-1.15040839e+00 5.37688732e-01 -4.73310530e-01 -2.00821579e-01
7.26142168e-01 4.00345027e-01 -1.02216691e-01 4.76841450e-01
-1.38071573e+00 7.58284807e-01 3.83692324e-01 5.52381039e-01
-6.41682565e-01 -7.74115980e-01 -4.57708567e-01 1.70192257e-01
-4.99788016e-01 -1.03595281e+00 7.93068409e-01 -8.75950933e-01
-1.16927099e+00 9.54306424e-01 1.64164409e-01 -7.48102784e-01
4.31116104e-01 -2.81334966e-01 -5.06568670e-01 -4.46991017e-03
-1.63411751e-01 2.19536111e-01 5.25260866e-01 -7.22142339e-01
-5.70535958e-01 -4.57535207e-01 -2.54488349e-01 4.31337729e-02
1.05726749e-01 1.00779064e-01 6.66354597e-02 -1.03385520e+00
2.69522280e-01 -6.68660820e-01 -2.91528493e-01 -3.30721527e-01
-5.10184288e-01 -2.40110695e-01 6.93446994e-01 -9.42832172e-01
1.91023111e+00 -2.33892131e+00 -7.36383200e-02 4.03399527e-01
6.13176525e-01 5.82593381e-01 4.70477462e-01 2.38228470e-01
-5.49382448e-01 5.22341907e-01 -3.91686916e-01 1.61284789e-01
-6.62242591e-01 6.17892407e-02 9.35095698e-02 3.70274961e-01
1.80930004e-01 1.21884918e+00 -4.46953654e-01 -2.03180224e-01
-7.25906789e-02 3.59328985e-01 -1.50908068e-01 2.51145273e-01
7.36690521e-01 6.69934094e-01 -1.64288953e-01 6.58071876e-01
7.14707971e-01 -2.82465369e-01 8.63579884e-02 -3.74221981e-01
-1.04598522e-01 -1.44587249e-01 -1.03694618e+00 1.48124170e+00
7.49952197e-02 6.58443689e-01 -9.95235369e-02 -9.74348247e-01
1.14915884e+00 8.69906068e-01 5.51656365e-01 -2.69987106e-01
3.02462965e-01 3.67826879e-01 6.15019143e-01 -7.81818926e-01
-5.42702317e-01 -1.71814915e-02 3.06284159e-01 4.62781012e-01
4.94520068e-02 3.16970944e-01 -2.18060628e-01 -3.14055353e-01
1.32838988e+00 -2.80219793e-01 6.14628911e-01 -4.56827641e-01
4.99938101e-01 -3.59328181e-01 1.03624487e+00 8.67856026e-01
-5.24806499e-01 1.02482700e+00 8.45721602e-01 -1.46912992e+00
-8.93781364e-01 -8.52799535e-01 -1.41124815e-01 -2.42606789e-01
-4.11669016e-01 -3.18510741e-01 -6.82365716e-01 -3.22786063e-01
-1.95980996e-01 -4.77451906e-02 -5.03159881e-01 -2.09442928e-01
-8.60282183e-01 -1.18096185e+00 1.40421689e+00 8.27354789e-01
9.92717326e-01 -1.68525386e+00 -1.37909102e+00 5.25857389e-01
-4.02168632e-02 -5.18773496e-01 4.12358008e-02 4.40979719e-01
-1.16215849e+00 -1.17730808e+00 -1.02722526e+00 -9.22828794e-01
3.92619312e-01 -6.53031409e-01 1.37420666e+00 2.75249213e-01
-8.19262505e-01 -5.17899990e-01 -5.20249344e-02 -8.66763711e-01
-1.28611669e-01 1.79998443e-01 -7.40606710e-02 1.31156668e-01
4.33746159e-01 -5.19273281e-01 -9.77349222e-01 -1.40516043e-01
-3.13464075e-01 6.09883219e-02 8.76165867e-01 7.38250971e-01
4.36497301e-01 -1.79199219e-01 9.37317252e-01 -7.94002056e-01
1.06515646e+00 -4.24623549e-01 -6.22927658e-02 -4.75717448e-02
-8.56903076e-01 -7.72048056e-01 5.22984385e-01 -7.89285824e-02
-3.06636661e-01 8.57095942e-02 -2.79800147e-01 -4.97684926e-01
-4.43327069e-01 7.05736816e-01 1.51597694e-01 2.36895293e-01
8.19453657e-01 2.79664904e-01 -2.50630174e-02 -1.77320495e-01
-3.89317691e-01 8.99838150e-01 5.41193128e-01 -1.09990671e-01
2.97356874e-01 4.29440252e-02 -1.00933775e-01 -6.84512079e-01
-5.34272313e-01 -3.28862667e-01 -8.19197714e-01 -1.75361529e-01
1.24697053e+00 -9.58678722e-01 -5.98274648e-01 7.50569046e-01
-1.42765963e+00 -1.21959232e-01 -3.31737161e-01 6.55232191e-01
-1.55836314e-01 2.37118483e-01 -8.56178641e-01 -7.03940392e-01
-1.18138111e+00 -8.49115968e-01 4.84361708e-01 2.26841360e-01
-5.13943374e-01 -6.86965585e-01 1.33085519e-01 -1.86893523e-01
6.08296931e-01 8.60585213e-01 1.01565576e+00 -6.70112729e-01
-3.49276117e-04 -3.72385055e-01 -2.48621926e-01 8.96822214e-01
2.76117206e-01 -1.49187639e-01 -9.23646450e-01 -2.73910165e-01
1.52148038e-01 2.29130551e-01 5.72630405e-01 8.72040629e-01
1.44361806e+00 1.16003091e-02 -2.07198009e-01 1.04587400e+00
1.35247827e+00 9.08499420e-01 1.05581379e+00 5.01381606e-02
6.99026585e-01 -1.27175987e-01 -2.46382430e-01 3.98340434e-01
-8.24535452e-03 -1.10370591e-01 1.54971212e-01 -8.33854079e-01
8.50071907e-02 5.31977713e-01 -1.08139843e-01 7.46437848e-01
-5.85110664e-01 4.52959724e-02 -1.28661108e+00 4.63651657e-01
-1.59379327e+00 -9.06043947e-01 -5.07672846e-01 1.83128750e+00
6.61896884e-01 8.88570026e-02 -9.61955786e-02 6.79566205e-01
7.37432063e-01 -2.63136655e-01 -5.39008200e-01 -9.02480006e-01
-2.66740292e-01 7.64055073e-01 2.08681673e-01 -1.82112738e-01
-1.25432587e+00 1.55614436e-01 6.69472504e+00 -6.43358007e-02
-1.37187099e+00 -4.94630598e-02 1.07155788e+00 4.31816548e-01
5.43600023e-01 -3.60599488e-01 -3.49235743e-01 4.94895309e-01
8.93170476e-01 2.46385187e-01 -8.21967889e-03 4.71992999e-01
2.33931273e-01 4.31508154e-01 -9.29298699e-01 1.39321268e+00
-2.76254788e-02 -1.52329338e+00 -5.81788346e-02 -1.70322776e-01
3.13173562e-01 1.64793190e-02 -2.76937693e-01 2.41298199e-01
-7.21438110e-01 -1.32643580e+00 2.26756185e-01 7.91009545e-01
1.15283275e+00 -8.82753253e-01 1.49341631e+00 2.43490264e-01
-1.14147568e+00 -4.83798206e-01 -9.82228890e-02 -3.33439708e-01
-2.54698813e-01 8.81522834e-01 -6.89156532e-01 7.01848447e-01
1.19382560e+00 1.07710302e+00 -4.24463838e-01 1.22671354e+00
-3.67447771e-02 1.17326081e+00 -4.00641337e-02 3.24023247e-01
-1.20518787e-03 -1.21208737e-02 3.71827155e-01 1.19270360e+00
3.59361976e-01 3.70481163e-01 -3.37433033e-02 9.17990863e-01
-2.46653873e-02 8.42547938e-02 -6.27555907e-01 3.13225895e-01
1.20814368e-01 1.27360928e+00 -7.65073597e-01 -5.45971334e-01
-2.10791394e-01 6.95957005e-01 -2.12737948e-01 3.62417221e-01
-6.67426169e-01 -1.00303018e+00 9.96141061e-02 -6.63276995e-03
-6.87434003e-02 2.86141932e-01 -1.03724825e+00 -9.18572009e-01
4.42405492e-02 -9.76077080e-01 4.63486135e-01 -6.20459378e-01
-1.10514069e+00 7.41656482e-01 -6.23478115e-01 -1.25146627e+00
1.81318700e-01 -4.59321231e-01 -9.72861111e-01 1.64875984e+00
-1.26896417e+00 -7.35770643e-01 -5.90183437e-01 4.41004992e-01
2.86869735e-01 -5.79272628e-01 1.26998997e+00 6.57934964e-01
-7.53382504e-01 4.44096923e-01 -5.14680862e-01 8.50163043e-01
4.54875380e-01 -1.33998954e+00 6.06847554e-02 8.10097873e-01
-3.90468866e-01 8.41813922e-01 -4.83964756e-02 -4.80587035e-01
-5.39353967e-01 -1.28753781e+00 1.22731638e+00 -1.48059681e-01
-2.70181000e-01 7.97443185e-03 -7.58314967e-01 4.51496094e-01
4.02093500e-01 2.57015795e-01 1.02158868e+00 -8.91631246e-02
1.88199982e-01 -2.41436809e-01 -1.04758215e+00 1.73772946e-01
4.74195004e-01 -1.67039588e-01 -5.69438159e-01 -1.78000882e-01
1.30957305e-01 -3.90690565e-01 -1.11522889e+00 9.69097733e-01
9.72312808e-01 -9.86053944e-01 8.44426751e-01 -6.48391843e-01
5.00311852e-01 -4.06593859e-01 4.53788877e-01 -1.03344131e+00
-6.10460937e-01 -6.89110816e-01 -4.47487831e-01 7.62018323e-01
4.94915634e-01 -7.38232434e-01 6.50400221e-01 -7.75735900e-02
-5.00563443e-01 -1.37168479e+00 -6.44004524e-01 -2.04965740e-01
-3.69259305e-02 8.41138791e-03 5.60221970e-01 1.15588737e+00
-4.21228647e-01 4.46343571e-01 -2.90566891e-01 1.06692731e-01
1.23394571e-01 -4.59799804e-02 3.65619034e-01 -1.74093914e+00
-7.14500621e-02 -3.71147364e-01 -5.16483426e-01 -1.56915903e-01
-6.60751820e-01 -9.22561824e-01 -4.16304976e-01 -1.68282425e+00
8.74188021e-02 -6.07074320e-01 -1.08850217e+00 5.93292832e-01
-9.87865552e-02 3.68326217e-01 -2.60336518e-01 2.02385470e-01
2.16224954e-01 -2.77272701e-01 1.06984794e+00 1.01628564e-02
-4.48274344e-01 4.72352982e-01 -6.50570452e-01 7.92403042e-01
1.28653288e+00 -3.19655985e-01 -2.09334105e-01 -3.35636407e-01
2.79285252e-01 2.37985104e-01 3.14612985e-01 -1.38207400e+00
9.42556262e-02 6.83939338e-01 1.16091144e+00 -8.78408074e-01
-2.66533256e-01 -4.76167679e-01 3.68599921e-01 1.22112989e+00
-2.72369206e-01 6.32157028e-01 2.60879874e-01 8.85780752e-02
-4.36373949e-01 2.18053773e-01 7.13211656e-01 -2.80589283e-01
-2.14239404e-01 5.70560157e-01 -6.96666062e-01 -1.62572473e-01
1.04535437e+00 -3.89897764e-01 1.74260795e-01 5.71417361e-02
-1.13081670e+00 2.08409540e-02 -5.30377388e-01 1.75030008e-01
9.48854864e-01 -1.29782879e+00 -1.00907564e+00 4.73939806e-01
9.61278901e-02 2.41749451e-01 2.78171301e-01 1.19250524e+00
-1.28627241e+00 4.14934635e-01 -5.69841921e-01 -7.72993267e-01
-1.15586066e+00 -2.19573695e-02 1.11607051e+00 -3.93020332e-01
-9.74137783e-01 7.93141603e-01 -2.82989115e-01 -2.15357885e-01
4.07362401e-01 -5.38904071e-01 -8.20066392e-01 1.47360722e-02
3.45331520e-01 4.10751015e-01 2.28698909e-01 -8.03304091e-02
-4.35428023e-01 5.01191497e-01 3.43492955e-01 4.84603673e-01
9.85840261e-01 3.43881130e-01 -4.47246581e-01 4.94753957e-01
1.01791358e+00 -4.22747552e-01 -4.43300128e-01 2.87334174e-01
-1.49049059e-01 3.32901299e-01 -1.63608734e-02 -1.15931606e+00
-1.25931382e+00 1.10164666e+00 1.22531593e+00 3.40795845e-01
1.26639271e+00 -6.09564185e-01 8.20999205e-01 2.71166086e-01
7.94052407e-02 -7.75291622e-01 -3.58913243e-01 2.81327009e-01
7.58038998e-01 -6.68072104e-01 -2.71895289e-01 -4.81413677e-02
-3.39257568e-01 1.57090771e+00 3.91777992e-01 -4.22590584e-01
1.34982085e+00 3.64895582e-01 4.96612459e-01 -4.97357249e-01
-5.32157958e-01 3.82163882e-01 -2.36035064e-02 3.64459723e-01
7.53407001e-01 1.14712909e-01 -6.55921817e-01 9.47479963e-01
5.16510627e-04 5.91404557e-01 4.35085267e-01 9.92236674e-01
-1.00222714e-01 -6.12498999e-01 5.38644521e-03 1.03139532e+00
-1.14800608e+00 -3.06169719e-01 -3.23114812e-01 4.80537534e-01
7.78189361e-01 6.57035470e-01 9.27440748e-02 -2.95796186e-01
3.47048789e-01 2.41245538e-01 3.85544971e-02 -7.76206374e-01
-1.29756820e+00 5.71295470e-02 -2.69630313e-01 -3.19584668e-01
-3.02254349e-01 -1.98914319e-01 -1.17242789e+00 2.87279934e-01
-3.61129604e-02 1.48245499e-01 3.02575231e-01 6.56677485e-01
6.42494917e-01 1.12982261e+00 4.47907746e-01 -1.85173199e-01
-1.36490732e-01 -1.31514132e+00 -7.40366578e-01 5.67083098e-02
5.62521458e-01 -6.83593303e-02 -1.94993421e-01 2.30917022e-01] | [14.329151153564453, 3.2804625034332275] |
42df1ec3-2487-4e19-8d34-deb46c0247a8 | modeling-the-graphotactics-of-low-resource | 2210.14409 | null | https://arxiv.org/abs/2210.14409v1 | https://arxiv.org/pdf/2210.14409v1.pdf | Modeling the Graphotactics of Low-Resource Languages Using Sequential GANs | Generative Adversarial Networks (GANs) have been shown to aid in the creation of artificial data in situations where large amounts of real data are difficult to come by. This issue is especially salient in the computational linguistics space, where researchers are often tasked with modeling the complex morphologic and grammatical processes of low-resource languages. This paper will discuss the implementation and testing of a GAN that attempts to model and reproduce the graphotactics of a language using only 100 example strings. These artificial, yet graphotactically compliant, strings are meant to aid in modeling the morphological inflection of low-resource languages. | ['Isaac Wasserman'] | 2022-10-26 | null | null | null | null | ['morphological-inflection'] | ['natural-language-processing'] | [ 2.39470258e-01 3.81781042e-01 3.10129672e-01 -4.76438671e-01
-5.52208006e-01 -7.30903864e-01 5.47505498e-01 2.94088037e-03
-5.70931613e-01 7.30758727e-01 4.09636736e-01 -6.82551026e-01
3.87921453e-01 -8.46718788e-01 -6.97204411e-01 -2.72435993e-01
2.50015974e-01 1.03480434e+00 -5.94515204e-01 -5.50887942e-01
-1.10450247e-02 3.63392740e-01 -1.27415812e+00 -6.99785054e-02
1.20664155e+00 -8.18452910e-02 3.00854117e-01 6.82451010e-01
-2.12440506e-01 4.98878360e-01 -9.06181276e-01 -8.76107574e-01
1.12209514e-01 -5.04773855e-01 -5.86707175e-01 -1.17124915e-01
5.61017513e-01 7.15689808e-02 -5.36203161e-02 1.16470861e+00
3.44633311e-01 1.47899449e-01 8.00251842e-01 -8.71024072e-01
-1.00254047e+00 1.10506070e+00 -1.04512669e-01 3.14391971e-01
4.57624286e-01 3.76700610e-01 9.33783352e-01 -6.62333608e-01
8.11264753e-01 1.58823156e+00 6.17212772e-01 7.00192809e-01
-1.26517594e+00 -5.72120309e-01 -1.47674412e-01 -2.90061831e-01
-1.17047668e+00 -3.47545117e-01 8.85540426e-01 -3.06302547e-01
1.07305956e+00 8.91988203e-02 7.95258522e-01 1.36759698e+00
3.12747844e-02 6.66697621e-01 1.02635980e+00 -6.67884588e-01
-2.30696611e-03 -1.12052159e-02 2.65941191e-02 6.13609374e-01
4.52447772e-01 1.72651097e-01 -1.70588598e-01 4.45743091e-02
6.25936270e-01 -4.58258778e-01 1.66891918e-01 4.56989855e-01
-9.44045544e-01 1.08296609e+00 2.59932667e-01 7.34993041e-01
-4.36964929e-01 8.70168582e-02 2.74649143e-01 1.69353157e-01
4.61847723e-01 9.60704982e-01 -1.37723029e-01 -3.48435551e-01
-8.50835979e-01 3.99857759e-01 7.09004581e-01 9.78050947e-01
3.78006846e-01 8.52062345e-01 2.07424045e-01 9.76776242e-01
2.17467755e-01 7.49736845e-01 8.94239187e-01 -7.53886878e-01
3.66946131e-01 4.86287624e-01 -1.35028616e-01 -8.91946197e-01
-3.92753899e-01 -1.59558371e-01 -6.76790059e-01 -8.88490081e-02
5.58711290e-01 -2.72094727e-01 -1.18312693e+00 2.06860805e+00
1.35865346e-01 -2.66502678e-01 3.33197892e-01 4.57447737e-01
8.48230660e-01 7.08228290e-01 4.98204917e-01 1.24218628e-01
1.29903555e+00 -5.24224997e-01 -6.10159457e-01 -8.86736810e-01
6.36198938e-01 -8.03937495e-01 1.68413627e+00 -9.62386932e-03
-1.46162236e+00 -7.27178454e-01 -1.04401183e+00 -3.06316614e-01
-6.01564527e-01 -3.26700866e-01 1.09664404e+00 1.00069213e+00
-1.03805995e+00 2.88478345e-01 -8.62074912e-01 -3.64995390e-01
4.56061393e-01 1.30745023e-01 -3.78664166e-01 -9.73327365e-03
-1.35701418e+00 1.17935872e+00 6.16999626e-01 1.60330147e-01
-3.82656962e-01 -5.33454537e-01 -1.13707793e+00 -4.64101415e-03
-6.50481954e-02 -6.15886629e-01 1.16375494e+00 -1.06323063e+00
-1.04482293e+00 1.18511629e+00 -3.10537014e-02 -5.31677067e-01
3.36557686e-01 1.48759559e-01 -6.15959048e-01 -3.15479517e-01
-3.80714890e-03 7.87178040e-01 6.22938991e-01 -1.12296879e+00
-4.09967124e-01 -2.71643788e-01 -4.37193848e-02 3.06816787e-01
-1.61715433e-01 2.84330368e-01 3.45714204e-02 -1.19273520e+00
-6.38705418e-02 -9.89234328e-01 -2.21062422e-01 -7.81744897e-01
-3.95119518e-01 -2.70598531e-01 3.25605541e-01 -9.72356081e-01
8.14514041e-01 -1.95737779e+00 -1.83566175e-02 2.29979783e-01
-8.59599039e-02 3.53338003e-01 -1.12889297e-01 2.24231914e-01
1.79177523e-01 6.37873232e-01 -3.17748904e-01 -3.14923972e-01
1.86716974e-01 5.01042247e-01 -2.45664582e-01 -4.20710351e-03
2.96460420e-01 1.22466099e+00 -8.86119783e-01 -4.13848519e-01
-8.27000737e-02 3.44511390e-01 -6.20939910e-01 2.49377698e-01
-2.98075140e-01 5.31009257e-01 -1.42647699e-02 6.38719320e-01
2.31430396e-01 1.88102052e-01 2.96912760e-01 3.16439390e-01
1.86426938e-01 4.86441642e-01 -7.47923672e-01 1.43092895e+00
-7.33658075e-01 7.49949813e-01 -2.75325388e-01 -5.68396628e-01
9.65942502e-01 6.23543859e-02 -1.38968065e-01 -5.73546350e-01
2.50460744e-01 3.82521600e-01 7.92824507e-01 -3.68568629e-01
9.43230510e-01 -5.52076876e-01 -4.50904280e-01 7.40522563e-01
-4.39139120e-02 -9.78154123e-01 4.94253367e-01 2.21102666e-02
8.67220759e-01 3.45912054e-02 1.88192874e-01 -3.49731445e-01
1.17975831e-01 2.48202041e-01 5.94457924e-01 6.45727217e-01
5.97555144e-03 5.12421191e-01 1.81296080e-01 -5.20004392e-01
-1.56311333e+00 -1.02563143e+00 -4.00398523e-02 8.23671162e-01
-4.89842653e-01 -1.24292681e-03 -9.97222662e-01 -2.73998111e-01
-2.13064775e-01 1.38499236e+00 -6.79702997e-01 -3.41197878e-01
-8.95637989e-01 -8.60928774e-01 9.58717465e-01 5.50314069e-01
1.11710504e-01 -1.73963952e+00 -5.17718911e-01 4.48976308e-01
-1.16274051e-01 -1.14999998e+00 -3.04042011e-01 -1.14252664e-01
-6.17964089e-01 -6.31027400e-01 -3.66262794e-01 -1.18219101e+00
8.44478428e-01 -4.73881751e-01 1.39723110e+00 2.52742916e-01
-2.01556906e-01 -1.56064443e-02 -2.86557794e-01 -1.01190460e+00
-1.27996457e+00 2.07516551e-01 1.23573475e-01 -4.05993670e-01
6.38340414e-01 -5.01250565e-01 1.60599023e-01 -9.35011432e-02
-1.21437275e+00 1.20835923e-01 1.45939544e-01 1.07702279e+00
4.91335332e-01 1.20133519e-01 4.41983193e-01 -1.62625909e+00
9.24742520e-01 -5.90034187e-01 -3.70239079e-01 2.86296487e-01
-3.27936083e-01 -8.20741206e-02 8.80891621e-01 -6.65855467e-01
-1.04438901e+00 -2.72257209e-01 -4.88930821e-01 1.48763061e-01
-4.51700866e-01 5.78993857e-01 -1.30435288e-01 1.94653064e-01
7.92309821e-01 4.73764762e-02 -1.64165437e-01 -2.08770126e-01
3.25283676e-01 6.97843313e-01 8.36345434e-01 -8.15756798e-01
9.16295290e-01 5.79916872e-02 -2.68531680e-01 -9.87672925e-01
-4.74807292e-01 4.30629909e-01 -6.25662565e-01 2.99156811e-02
6.70335472e-01 -6.61361456e-01 -1.12012602e-01 6.71571970e-01
-9.79920208e-01 -6.00168824e-01 -7.82016695e-01 3.29168409e-01
-6.68635070e-01 -1.08245149e-01 -7.04425514e-01 -5.54374218e-01
-4.42368418e-01 -1.08179581e+00 6.44433081e-01 4.56038654e-01
-7.60556221e-01 -1.33508873e+00 7.86493793e-02 2.61884063e-01
4.57206279e-01 3.68961215e-01 1.52154720e+00 -9.96337116e-01
8.04975554e-02 -3.77003700e-01 3.55328470e-01 2.46837169e-01
1.32877961e-01 2.92127490e-01 -9.37420070e-01 -1.00144714e-01
8.26325491e-02 -4.25234586e-01 -5.34558780e-02 3.55962276e-01
7.58780301e-01 -3.43963206e-01 1.82019651e-01 6.41715288e-01
1.22006595e+00 4.45547312e-01 6.35708153e-01 4.06815857e-02
8.51582527e-01 7.15519249e-01 2.18084231e-01 -6.53534457e-02
6.14130795e-01 2.40434200e-01 -1.63268387e-01 -2.00209320e-01
-2.84150243e-01 -7.96260834e-01 1.40963197e-01 1.30406892e+00
1.63643271e-01 -5.57064891e-01 -1.45647514e+00 7.74882913e-01
-1.18081665e+00 -8.43207896e-01 -3.82060930e-02 1.73611724e+00
1.06878042e+00 3.68403822e-01 2.01288871e-02 5.36573716e-02
8.45696092e-01 2.98720952e-02 -5.03082275e-01 -1.08073652e+00
-5.48145771e-01 5.42465985e-01 5.61568588e-02 6.25309825e-01
-7.07149506e-01 1.60531795e+00 7.30251884e+00 5.27980983e-01
-8.12447786e-01 -2.42970814e-03 8.06380630e-01 2.99643129e-01
-7.74790406e-01 -1.48489727e-02 -6.47585928e-01 7.11392879e-01
1.31646252e+00 -3.73375654e-01 6.44149780e-01 4.62521136e-01
6.67048246e-02 -9.83563215e-02 -1.07945299e+00 5.92771828e-01
1.81395739e-01 -9.57767308e-01 2.15832978e-01 -1.46399541e-02
8.52740049e-01 -1.38007089e-01 2.79215962e-01 3.79080713e-01
1.04936159e+00 -1.51343703e+00 8.71988058e-01 2.94249713e-01
9.43851531e-01 -8.03520799e-01 6.37575626e-01 3.61315370e-01
-5.84471405e-01 1.18742190e-01 -5.29853702e-01 -1.68885082e-01
2.85565525e-01 1.33107573e-01 -1.10867572e+00 -2.14082748e-01
4.35074866e-01 1.16828077e-01 -7.14116037e-01 6.69521093e-01
-6.03824496e-01 8.95647705e-01 -5.01783609e-01 -3.39159250e-01
3.27333987e-01 -2.52905309e-01 5.10493517e-01 1.20712328e+00
3.49644244e-01 1.08666234e-01 -1.32159010e-01 1.05061054e+00
-3.05513948e-01 3.83085489e-01 -1.00793850e+00 -5.80085278e-01
6.24892414e-01 8.51864040e-01 -8.72168601e-01 -2.94305772e-01
-1.72339827e-01 6.50027514e-01 5.26476920e-01 9.14935321e-02
-5.37899852e-01 -3.10929477e-01 4.34677243e-01 4.77442890e-02
-1.30624650e-03 -5.24495423e-01 -6.89505517e-01 -9.01150107e-01
-2.06259742e-01 -1.41870522e+00 1.94325343e-01 -8.89512300e-01
-1.43648827e+00 8.47745538e-01 -9.69699323e-02 -5.58343291e-01
-8.19247246e-01 -4.81782377e-01 -6.49412572e-01 1.18448615e+00
-7.50731647e-01 -1.51202703e+00 1.75615679e-02 3.06914657e-01
7.26413012e-01 -4.87546086e-01 1.03367627e+00 1.46351820e-02
-4.88793343e-01 8.43361616e-01 -3.47955041e-02 3.35346788e-01
2.27100402e-01 -1.43457711e+00 1.28827703e+00 1.08569324e+00
6.58903539e-01 9.03822601e-01 8.51592243e-01 -9.36897457e-01
-1.09971893e+00 -7.56064236e-01 1.23702800e+00 -5.46261549e-01
7.66240478e-01 -5.92710674e-01 -1.02032244e+00 9.57546055e-01
2.22932190e-01 -2.83015549e-01 8.58844995e-01 8.43877494e-02
-1.60957247e-01 3.57587546e-01 -1.45347345e+00 1.02171099e+00
1.01557302e+00 -6.96111500e-01 -9.59283173e-01 4.16517019e-01
5.78952730e-01 -6.58597469e-01 -8.15306365e-01 7.19083101e-02
3.35196435e-01 -3.16508472e-01 6.30526423e-01 -1.14842904e+00
5.91801882e-01 5.79431318e-02 -4.24848348e-02 -1.89022541e+00
-2.70822406e-01 -7.85908937e-01 3.20466995e-01 1.48404098e+00
6.28432870e-01 -6.21170163e-01 7.58254468e-01 9.20662701e-01
-1.77710637e-01 -2.23519355e-01 -6.70786560e-01 -5.99776864e-01
4.76145595e-01 -4.27980781e-01 1.00308526e+00 1.04993999e+00
-2.92054415e-01 3.50271374e-01 -2.13575121e-02 -2.14219898e-01
4.23338056e-01 -4.93628234e-02 5.82730174e-01 -1.08055329e+00
-2.78986514e-01 -4.88595337e-01 -5.35970092e-01 -2.17102826e-01
5.91051817e-01 -1.01517546e+00 9.93705392e-02 -1.19908738e+00
-3.34668964e-01 -8.30514252e-01 9.75738168e-02 4.65044796e-01
-4.12297994e-01 4.08589900e-01 3.22981566e-01 5.30964602e-03
3.78087401e-01 2.57523775e-01 1.05394375e+00 -1.07060954e-01
-1.17353283e-01 -1.19323045e-01 -1.04679799e+00 9.93077457e-01
1.04884875e+00 -6.12729192e-01 -3.62119585e-01 -7.51310170e-01
4.55489814e-01 -2.94250786e-01 -1.32887483e-01 -9.06794667e-01
1.42626446e-02 -2.26475641e-01 3.99169385e-01 -1.24358326e-01
1.63656697e-01 -5.74156106e-01 3.76667202e-01 4.61708784e-01
-3.46175432e-01 8.72841418e-01 3.89699459e-01 -4.49651778e-02
-1.94415808e-01 -6.09962344e-01 8.66392374e-01 -4.69632000e-01
-6.26048505e-01 9.43860412e-02 -4.58354503e-01 7.38537133e-01
8.46456409e-01 -2.37486005e-01 -2.86589652e-01 -4.83124584e-01
-5.87236881e-01 -1.65067241e-01 7.96718895e-01 5.15254736e-01
2.97854245e-01 -1.28533781e+00 -9.45843935e-01 5.98930001e-01
-1.88877761e-01 5.17487489e-02 1.07928608e-02 -1.18824348e-01
-1.04729879e+00 8.40769708e-02 -5.92734575e-01 3.90593484e-02
-7.44882405e-01 4.33102369e-01 2.16819763e-01 -1.42348260e-01
-4.68012214e-01 1.07565987e+00 -1.52002633e-01 -5.47423303e-01
-1.25212669e-01 -6.32590726e-02 -1.92179352e-01 -1.28553644e-01
7.46529475e-02 1.12953700e-01 -1.30475452e-03 -9.07966435e-01
4.45294529e-02 1.13723524e-01 -1.37766629e-01 -3.42374444e-01
1.41433239e+00 3.05971295e-01 -5.49926385e-02 6.61578059e-01
7.24277079e-01 3.79523128e-01 -8.61311615e-01 -6.73354715e-02
-1.92323588e-02 -2.28361472e-01 -3.54022473e-01 -7.51299739e-01
-7.89342284e-01 8.78704250e-01 2.53680438e-01 3.10579389e-01
6.80362165e-01 -3.30958009e-01 1.03850639e+00 2.79289752e-01
2.93027818e-01 -1.29225278e+00 -3.69151622e-01 8.64905298e-01
7.49435425e-01 -9.97244596e-01 -2.96628267e-01 -1.26731917e-01
-7.47102857e-01 7.77610421e-01 8.21526051e-01 -2.76770949e-01
3.89829576e-01 5.43116570e-01 2.51313686e-01 -2.38516424e-02
-4.02516782e-01 8.16597790e-03 6.49995580e-02 8.43389988e-01
6.20344341e-01 4.45000619e-01 -3.65351826e-01 3.61713231e-01
-1.30538130e+00 -6.42965376e-01 7.70350575e-01 7.80094504e-01
7.63974804e-03 -1.27455485e+00 -3.10434848e-01 5.55184305e-01
-6.84297681e-01 -5.65236092e-01 -4.94833201e-01 9.51050937e-01
1.88608170e-01 7.39571154e-01 4.33069021e-01 -1.92755908e-01
8.57315809e-02 4.03511912e-01 5.01045346e-01 -9.56702352e-01
-8.27639699e-01 -2.45476857e-01 3.39997262e-01 1.15264915e-01
-2.88887825e-02 -8.59844029e-01 -1.17840874e+00 -4.73931730e-01
1.97504699e-01 1.04897425e-01 8.38358879e-01 7.91464508e-01
-4.04452235e-02 3.64164442e-01 3.53390485e-01 -5.58665037e-01
-4.05864775e-01 -1.16883564e+00 -6.41645372e-01 7.85345614e-01
-1.57374486e-01 -1.69707820e-01 -2.27434844e-01 1.25813335e-01] | [10.777045249938965, 9.721062660217285] |
59043c23-96d2-4801-858f-b02ed16a0806 | nubo-a-transparent-python-package-for | 2305.06709 | null | https://arxiv.org/abs/2305.06709v1 | https://arxiv.org/pdf/2305.06709v1.pdf | NUBO: A Transparent Python Package for Bayesian Optimisation | NUBO, short for Newcastle University Bayesian Optimisation, is a Bayesian optimisation framework for the optimisation of expensive-to-evaluate black-box functions, such as physical experiments and computer simulators. Bayesian optimisation is a cost-efficient optimisation strategy that uses surrogate modelling via Gaussian processes to represent an objective function and acquisition functions to guide the selection of candidate points to approximate the global optimum of the objective function. NUBO itself focuses on transparency and user experience to make Bayesian optimisation easily accessible to researchers from all disciplines. Clean and understandable code, precise references, and thorough documentation ensure transparency, while user experience is ensured by a modular and flexible design, easy-to-write syntax, and careful selection of Bayesian optimisation algorithms. NUBO allows users to tailor Bayesian optimisation to their specific problem by writing the optimisation loop themselves using the provided building blocks. It supports sequential single-point, parallel multi-point, and asynchronous optimisation of bounded, constrained, and/or mixed (discrete and continuous) parameter input spaces. Only algorithms and methods that are extensively tested and validated to perform well are included in NUBO. This ensures that the package remains compact and does not overwhelm the user with an unnecessarily large number of options. The package is written in Python but does not require expert knowledge of Python to optimise your simulators and experiments. NUBO is distributed as open-source software under the BSD 3-Clause licence. | ['Richard D. Whalley', 'Kevin Wilson', 'Mike Diessner'] | 2023-05-11 | null | null | null | null | ['bayesian-optimisation'] | ['methodology'] | [-1.70263007e-01 -2.07580268e-01 2.05921590e-01 -1.65074855e-01
-7.34389484e-01 -4.97739494e-01 6.05472088e-01 2.44812250e-01
-4.06727850e-01 7.39639878e-01 -1.75733343e-01 -7.17982411e-01
-7.23837197e-01 -5.73315859e-01 -3.63793522e-01 -9.52979505e-01
-1.01324223e-01 6.88554227e-01 1.75800413e-01 6.71533123e-02
4.17217523e-01 6.83520734e-01 -1.42580104e+00 -5.59682012e-01
5.87835491e-01 6.71264410e-01 3.32887292e-01 1.10540879e+00
3.60709906e-01 1.13140203e-01 -4.69492972e-01 -2.70030666e-02
2.00110137e-01 -2.91372269e-01 -4.33314741e-01 -5.51366270e-01
-5.44356644e-01 4.07311730e-02 2.15849355e-01 7.11749673e-01
9.85041261e-01 4.21435326e-01 6.65548980e-01 -9.64110017e-01
-7.91130662e-02 -7.47435018e-02 -1.36177167e-01 1.82812512e-01
4.08317447e-01 8.10318768e-01 5.17295301e-01 -5.44677854e-01
8.47529843e-02 1.25283647e+00 6.92635179e-01 1.09962180e-01
-1.57089019e+00 -5.15697837e-01 -1.87104076e-01 -2.92881310e-01
-1.39379752e+00 -5.16514778e-01 1.44930631e-01 -5.73325038e-01
1.15289700e+00 6.76886737e-01 8.05814266e-01 8.44556093e-01
6.40562892e-01 1.41896099e-01 9.72176552e-01 -3.70545983e-01
6.12896919e-01 -1.93172414e-02 -1.62623703e-01 2.42947087e-01
1.40654653e-01 7.64967561e-01 -2.28740975e-01 -6.22982860e-01
7.02490330e-01 -3.40645641e-01 -1.84361115e-01 -5.46661496e-01
-1.21253335e+00 9.37487662e-01 -7.82082900e-02 -2.31587887e-01
-4.40859497e-01 3.67785364e-01 2.19973460e-01 -3.16957124e-02
1.81322396e-01 5.99073231e-01 -7.05996037e-01 -7.36767232e-01
-6.45390928e-01 5.72216928e-01 1.00371945e+00 6.71300948e-01
3.05902541e-01 2.84879631e-03 6.08067289e-02 8.06909204e-01
1.08956432e+00 7.42025912e-01 1.45975843e-01 -1.09226120e+00
-1.03770554e-01 4.59474772e-02 3.60504001e-01 -6.02948368e-01
-5.07677734e-01 -6.44192398e-01 -5.13318121e-01 7.15038300e-01
2.72727758e-01 -2.48598814e-01 -7.10130453e-01 1.10847509e+00
7.12109447e-01 -3.75925779e-01 -5.44510148e-02 8.80885422e-01
6.09255314e-01 9.25138116e-01 2.30794400e-01 -2.54203349e-01
1.21580970e+00 -3.18256825e-01 -3.51722479e-01 -2.87073284e-01
3.45472991e-01 -1.14959157e+00 8.74080002e-01 5.74901819e-01
-1.17568791e+00 -2.52839297e-01 -1.10432863e+00 3.35258335e-01
-2.12524682e-01 -4.47553277e-01 5.31219184e-01 1.03975809e+00
-6.82546616e-01 7.65864134e-01 -1.04010844e+00 4.19147946e-02
1.15141630e-01 6.45982087e-01 2.16591224e-01 9.62224007e-02
-1.10937989e+00 1.22282076e+00 5.13271034e-01 2.46248856e-01
-6.56393409e-01 -8.51112247e-01 -9.29665327e-01 -2.05255181e-01
8.00966248e-02 -8.89673054e-01 1.39128637e+00 -5.10193288e-01
-1.91566396e+00 2.47494236e-01 -5.15975803e-02 -2.95957760e-03
4.52697366e-01 -1.46945402e-01 -3.44912410e-01 -3.11695844e-01
-8.10370743e-02 2.95808643e-01 6.73949301e-01 -7.59533048e-01
-3.01081985e-01 -5.29963449e-02 -3.79366100e-01 3.36425304e-01
6.69549406e-01 5.35920680e-01 -3.40288997e-01 -3.99218678e-01
-1.29315197e-01 -7.83704817e-01 -5.83931804e-01 -1.85291022e-01
5.14484607e-02 4.25521433e-02 5.88114917e-01 -6.21822834e-01
1.00379729e+00 -1.96632338e+00 -7.36070871e-02 5.77465832e-01
-2.45335877e-01 1.72698200e-02 4.09791708e-01 5.00888884e-01
-1.36784583e-01 1.29379303e-04 -3.68101358e-01 1.05235666e-01
3.80968601e-01 1.40576810e-01 2.03034654e-01 8.62154603e-01
2.25197971e-02 6.63333654e-01 -1.13970971e+00 -2.50463694e-01
6.64961219e-01 5.55990458e-01 -1.96479693e-01 2.94933856e-01
-8.11153054e-02 7.83179998e-01 -3.54623556e-01 5.46011329e-01
5.88855863e-01 9.76148248e-02 -1.36480257e-01 4.31453943e-01
-4.28934902e-01 4.41310406e-01 -1.84941316e+00 1.25891590e+00
-6.77724659e-01 4.02144402e-01 4.17235106e-01 -6.13774240e-01
9.55448925e-01 4.01473582e-01 1.38138622e-01 -4.43385243e-01
7.42551014e-02 3.86606067e-01 1.25061110e-01 -3.00539345e-01
1.65019765e-01 -3.38440329e-01 -9.44825858e-02 7.55305111e-01
-2.16821715e-01 -8.96458745e-01 1.94225967e-01 -2.48698965e-01
7.97584236e-01 5.89863598e-01 5.39310336e-01 -5.42623818e-01
1.60086244e-01 -2.81410098e-01 3.90295953e-01 6.99531615e-01
2.06596240e-01 2.21526861e-01 1.93622902e-01 -9.28015709e-02
-1.01270080e+00 -1.08082366e+00 -7.49746084e-01 1.05267596e+00
-1.52991757e-01 -1.60786271e-01 -2.28464514e-01 8.36498663e-02
3.80167030e-02 1.01953661e+00 -9.30085853e-02 -2.51708124e-02
-2.65513480e-01 -1.01618242e+00 1.66690685e-02 1.92708582e-01
3.47008333e-02 -9.15980399e-01 -9.30966675e-01 4.56711829e-01
3.71591657e-01 -3.16430300e-01 -2.19906822e-01 4.74782735e-01
-9.50761318e-01 -6.67692780e-01 -5.38705170e-01 -2.95210302e-01
3.78182352e-01 -3.56654227e-01 9.05424416e-01 -2.11099640e-01
-2.48784333e-01 4.49664205e-01 -3.24986205e-02 -6.72404885e-01
-6.27138138e-01 -2.93054402e-01 -1.50282800e-01 -5.58538496e-01
1.10895522e-01 -4.20770735e-01 -4.48823363e-01 6.69526279e-01
-5.96999466e-01 -1.07709296e-01 3.37275892e-01 7.59699881e-01
4.46454883e-01 2.59485990e-01 1.73492998e-01 -1.78211048e-01
7.84273982e-01 -4.29670423e-01 -1.12801516e+00 -7.51276910e-02
-7.89501727e-01 2.59858251e-01 1.80058748e-01 -1.89184681e-01
-7.43381023e-01 7.64634535e-02 -2.08724633e-01 4.55736741e-02
-4.36090738e-01 6.98835433e-01 -3.20876062e-01 -1.81846023e-01
6.67822599e-01 -1.72758162e-01 1.57169193e-01 -4.89343464e-01
2.02363908e-01 8.43920052e-01 3.67536098e-01 -7.43142009e-01
4.41046208e-01 6.54114857e-02 1.80396229e-01 -8.70410681e-01
-1.33810833e-01 -5.33302605e-01 -4.81744498e-01 -3.83703768e-01
5.49979627e-01 -5.20897210e-01 -1.01509416e+00 4.19610381e-01
-7.33965576e-01 -5.32704115e-01 -2.19206125e-01 7.59730101e-01
-5.58101833e-01 3.01721781e-01 1.59881681e-01 -1.14266849e+00
-2.10391343e-01 -1.53364897e+00 1.01576209e+00 2.86188006e-01
-9.07138765e-01 -1.12860811e+00 1.69903934e-01 -1.29416957e-01
3.57925355e-01 4.45368797e-01 4.49416190e-01 -4.43755567e-01
-5.00903606e-01 -4.57470119e-01 2.41632909e-01 1.60685435e-01
-1.74778044e-01 5.44973552e-01 -8.49081159e-01 -3.61092806e-01
1.67416736e-01 1.16107285e-01 2.67776430e-01 6.71281636e-01
8.41386974e-01 -1.80877090e-01 -3.82418782e-01 6.20984137e-01
1.05122578e+00 3.12751323e-01 4.30435926e-01 5.63911796e-01
2.11425364e-01 7.03516185e-01 5.77982545e-01 4.44433153e-01
8.60548206e-03 6.98708653e-01 3.21788967e-01 1.15432337e-01
5.94293833e-01 2.57525146e-01 2.60862440e-01 4.37353611e-01
6.58718571e-02 4.15013023e-02 -1.20794415e+00 4.00822908e-01
-1.73624909e+00 -7.46521533e-01 -6.17728829e-01 2.84397340e+00
8.48603666e-01 4.02608335e-01 2.11665705e-01 1.99816793e-01
6.61066532e-01 -2.63320267e-01 -2.48307779e-01 -1.04222989e+00
4.83600646e-01 4.81004268e-01 8.43819857e-01 7.79836893e-01
-9.07052338e-01 1.48547634e-01 6.64187479e+00 8.43892753e-01
-8.07225704e-01 1.56627923e-01 3.28095406e-01 -3.88965070e-01
-1.73140496e-01 2.70889670e-01 -5.60774684e-01 6.77540004e-01
1.58767450e+00 -2.00750932e-01 6.07500553e-01 5.35793722e-01
8.97498786e-01 -6.91083968e-01 -8.35753560e-01 6.13388240e-01
-7.03994095e-01 -1.15724373e+00 -9.50097859e-01 2.51242757e-01
3.81813496e-01 8.66495743e-02 -1.24736026e-01 1.47369251e-01
7.60149717e-01 -1.38927543e+00 8.24091375e-01 8.04004312e-01
4.45686013e-01 -1.11252725e+00 8.52361202e-01 4.57176119e-01
-8.20560634e-01 5.20960391e-02 -8.68803933e-02 -2.65553951e-01
5.76990604e-01 7.47959793e-01 -6.15517378e-01 5.48545420e-01
8.75810385e-01 7.55772144e-02 -1.08730905e-01 1.73114824e+00
-1.60317615e-01 8.12803090e-01 -9.00997996e-01 -4.12377357e-01
1.25548840e-01 -2.08941758e-01 7.91243076e-01 1.19330323e+00
7.61042312e-02 -5.95165603e-02 -9.55769420e-02 8.37409317e-01
1.02553415e+00 -4.37722309e-03 -1.26938328e-01 1.16567515e-01
3.89782608e-01 9.19376194e-01 -7.11506069e-01 1.82921976e-01
-9.38574448e-02 1.73406214e-01 -5.16564786e-01 3.27528954e-01
-6.76606536e-01 -6.59922004e-01 5.14405370e-01 3.12389843e-02
4.04669583e-01 -4.86837059e-01 -6.51220739e-01 -3.27916332e-02
-3.20616841e-01 -9.08668518e-01 3.63204151e-01 -9.17225003e-01
-8.95843208e-01 -1.23200333e-03 6.18577898e-01 -7.36232758e-01
-5.55269897e-01 -6.53520226e-01 -7.26942301e-01 1.57527089e+00
-8.96674156e-01 -5.76135635e-01 2.76254471e-02 -1.00175299e-01
6.24305718e-02 8.04906338e-02 9.12055433e-01 -1.74957544e-01
-4.09197599e-01 8.43308866e-02 6.98352993e-01 -5.15862226e-01
3.60661507e-01 -1.30156553e+00 5.48945367e-01 5.21711111e-01
-4.55788821e-01 9.05049860e-01 1.28849328e+00 -8.13435495e-01
-1.23453343e+00 -5.82414448e-01 5.74384928e-01 -6.24390244e-01
8.52126539e-01 -2.22562343e-01 -6.96096659e-01 1.01724289e-01
-1.60999313e-01 -3.89738798e-01 6.35172963e-01 1.23242147e-01
4.42307055e-01 1.13842800e-01 -1.06478465e+00 4.77050245e-01
1.18690692e-01 -1.89654440e-01 -5.95568657e-01 3.77753168e-01
2.11863458e-01 -7.88575947e-01 -1.16462600e+00 3.51218522e-01
6.43899560e-01 -6.16037726e-01 1.04730761e+00 -8.36728513e-02
-2.51244575e-01 -6.36495948e-01 2.63941139e-01 -1.13424397e+00
-1.04527317e-01 -1.36318338e+00 -7.99030736e-02 1.00695503e+00
4.83965486e-01 -9.61802900e-01 2.01347992e-01 8.53051662e-01
-1.74818665e-01 -7.60198951e-01 -1.22002745e+00 -7.76082993e-01
2.01167651e-02 -1.00926471e+00 7.74132013e-01 3.29417408e-01
-9.06632394e-02 4.34515812e-02 1.34003256e-02 3.60176653e-01
7.46029377e-01 -1.70986637e-01 8.47985625e-01 -1.22021854e+00
-6.66712284e-01 -6.98804855e-01 -3.73128593e-01 -7.43329406e-01
-4.25177902e-01 -5.47494113e-01 2.64948696e-01 -1.53792250e+00
-3.18198860e-01 -4.82353449e-01 2.50480682e-01 1.63061604e-01
-1.91287905e-01 -4.43447307e-02 -3.35463285e-01 1.27970666e-01
-5.07777147e-02 5.12814462e-01 8.40958178e-01 2.59003550e-01
-3.85831922e-01 7.20384359e-01 -5.39397418e-01 5.48699617e-01
7.71608591e-01 -7.34233439e-01 -2.18592510e-01 1.41868383e-01
4.18994874e-01 -1.68336242e-01 7.91421473e-01 -7.01710284e-01
3.46201994e-02 -3.94476593e-01 3.76379877e-01 -4.90466475e-01
1.99000448e-01 -7.76535511e-01 7.38823950e-01 3.56352746e-01
6.05571866e-02 8.95448104e-02 5.39541364e-01 4.26283300e-01
3.84703666e-01 -7.55214572e-01 8.39117765e-01 -9.58878919e-02
-2.57607579e-01 -2.21919775e-01 -6.54284120e-01 -9.30871665e-02
1.02280796e+00 -5.39919376e-01 2.61794776e-01 -5.53590417e-01
-9.08964694e-01 5.95694065e-01 6.86102271e-01 5.58895245e-02
1.56714633e-01 -9.42612588e-01 -6.41562045e-01 1.07465066e-01
-1.25268787e-01 3.39105666e-01 6.55565262e-02 9.54418957e-01
-7.38422692e-01 5.00501096e-01 3.01986814e-01 -8.13380778e-01
-1.02271581e+00 2.96092629e-01 5.49058437e-01 2.60468032e-02
-5.01900852e-01 8.10964942e-01 -4.41487581e-01 -7.57844031e-01
1.02997765e-01 -2.21823484e-01 2.91470289e-01 -3.19494009e-01
4.23742592e-01 6.87495887e-01 2.22719580e-01 -3.82131487e-01
-3.66452187e-01 2.88516223e-01 5.45140922e-01 -5.65531731e-01
1.41500151e+00 -2.55299419e-01 -9.30808708e-02 6.79419041e-01
8.03071797e-01 -2.19609980e-02 -1.46043146e+00 4.52267081e-01
-3.66342291e-02 -4.17541027e-01 6.05350316e-01 -9.66649532e-01
-1.87971100e-01 6.21700287e-01 6.10946536e-01 1.43025979e-01
6.67686224e-01 -2.34612584e-01 1.84868440e-01 -3.33102196e-02
1.17496826e-01 -1.14299273e+00 -6.13110483e-01 4.82023448e-01
9.86376286e-01 -7.53791988e-01 6.70670152e-01 8.75359103e-02
-5.38812995e-01 1.14227831e+00 8.70143399e-02 3.09206933e-01
8.36399198e-01 3.53797615e-01 -8.92368704e-02 -3.07512015e-01
-5.44727027e-01 2.91625839e-02 4.38169569e-01 7.79287994e-01
2.61527628e-01 4.45123687e-02 -1.86789617e-01 2.75658488e-01
-3.80003184e-01 -1.24040246e-01 1.99620575e-01 1.10175192e+00
-2.80446291e-01 -1.21060765e+00 -1.10227203e+00 4.23276156e-01
-3.46177399e-01 -1.48685768e-01 1.12516396e-01 7.87498057e-01
-2.52832204e-01 1.01060748e+00 -4.40028671e-04 2.56529212e-01
2.28345320e-01 3.14670205e-02 2.86654115e-01 -5.40103018e-01
-7.39912927e-01 3.82492870e-01 3.35512877e-01 -3.26715797e-01
-3.93949151e-02 -1.02631080e+00 -9.94916320e-01 -3.74711096e-01
-5.85585117e-01 4.88572925e-01 1.13989019e+00 1.01404667e+00
3.03704798e-01 4.86248344e-01 1.96429938e-01 -1.20248079e+00
-6.53635383e-01 -9.07261908e-01 -3.93454939e-01 -6.25194371e-01
8.23114663e-02 -8.05569530e-01 -4.87953693e-01 -4.06287938e-01] | [6.251134395599365, 3.760422945022583] |
b0dae793-7cdf-46d0-b22d-2092554bbcdb | zmbart-an-unsupervised-cross-lingual-transfer | 2106.01597 | null | https://arxiv.org/abs/2106.01597v1 | https://arxiv.org/pdf/2106.01597v1.pdf | ZmBART: An Unsupervised Cross-lingual Transfer Framework for Language Generation | Despite the recent advancement in NLP research, cross-lingual transfer for natural language generation is relatively understudied. In this work, we transfer supervision from high resource language (HRL) to multiple low-resource languages (LRLs) for natural language generation (NLG). We consider four NLG tasks (text summarization, question generation, news headline generation, and distractor generation) and three syntactically diverse languages, i.e., English, Hindi, and Japanese. We propose an unsupervised cross-lingual language generation framework (called ZmBART) that does not use any parallel or pseudo-parallel/back-translated data. In this framework, we further pre-train mBART sequence-to-sequence denoising auto-encoder model with an auxiliary task using monolingual data of three languages. The objective function of the auxiliary task is close to the target tasks which enriches the multi-lingual latent representation of mBART and provides good initialization for target tasks. Then, this model is fine-tuned with task-specific supervised English data and directly evaluated with low-resource languages in the Zero-shot setting. To overcome catastrophic forgetting and spurious correlation issues, we applied freezing model component and data argumentation approaches respectively. This simple modeling approach gave us promising results.We experimented with few-shot training (with 1000 supervised data points) which boosted the model performance further. We performed several ablations and cross-lingual transferability analyses to demonstrate the robustness of ZmBART. | ['Kumari Deepshikha', 'Yoshinobu Kano', 'Maunendra Sankar Desarkar', 'Kaushal Kumar Maurya'] | 2021-06-03 | null | https://aclanthology.org/2021.findings-acl.248 | https://aclanthology.org/2021.findings-acl.248.pdf | findings-acl-2021-8 | ['headline-generation', 'distractor-generation'] | ['natural-language-processing', 'natural-language-processing'] | [ 2.02344641e-01 1.72200024e-01 -2.46247500e-02 -2.12749317e-01
-1.50657892e+00 -4.63299990e-01 9.00549352e-01 -2.20416173e-01
-5.16336858e-01 1.35217953e+00 5.87500215e-01 -2.38717824e-01
4.28274810e-01 -6.82016611e-01 -7.99585104e-01 -5.89564145e-01
2.40235761e-01 6.51825190e-01 -8.63254145e-02 -5.72547019e-01
-5.80241857e-03 -1.44145429e-01 -1.25034726e+00 4.99844342e-01
1.49497223e+00 4.00212109e-01 5.02091706e-01 7.46593952e-01
-2.64071971e-01 8.35494876e-01 -8.23980749e-01 -6.41645610e-01
-6.04567900e-02 -9.22349632e-01 -8.89779270e-01 -3.98119353e-02
6.74312860e-02 -6.73785508e-02 5.13666868e-02 6.98634088e-01
1.00613487e+00 2.79110491e-01 6.08144522e-01 -9.53986406e-01
-1.02542305e+00 9.74679768e-01 -5.21399319e-01 8.28093290e-02
4.05740649e-01 3.91626775e-01 9.49566364e-01 -1.18414164e+00
9.94590402e-01 1.42210686e+00 5.26737809e-01 8.60853851e-01
-1.06023157e+00 -4.66170073e-01 -2.50805747e-02 -9.76484343e-02
-1.10051906e+00 -6.78087354e-01 5.04702032e-01 -1.24088846e-01
1.08077109e+00 -1.70994047e-02 2.10637242e-01 1.80461407e+00
2.81953573e-01 9.34396684e-01 1.04311001e+00 -6.11726701e-01
1.66886047e-01 3.32767487e-01 1.16892099e-01 4.35039520e-01
1.10794917e-01 -9.02902558e-02 -7.53713846e-01 -4.98901717e-02
2.75669128e-01 -6.62433922e-01 -3.71136278e-01 3.05647969e-01
-1.46150064e+00 9.98237789e-01 -1.91953406e-02 4.15097833e-01
-3.56669307e-01 -1.82611614e-01 5.89256406e-01 4.95897800e-01
8.82817805e-01 5.26620924e-01 -5.06917596e-01 -8.33869800e-02
-8.89870286e-01 1.92808121e-01 8.59813094e-01 1.23144948e+00
6.27342761e-01 4.94329005e-01 -8.08313251e-01 1.28656912e+00
-1.13595612e-01 6.12754703e-01 1.25320697e+00 -2.58381039e-01
9.66578424e-01 1.99024647e-01 -8.67785886e-03 -4.27104384e-01
-1.44019842e-01 -4.46609050e-01 -1.14771342e+00 -3.80246878e-01
1.32981345e-01 -7.73147881e-01 -9.26204920e-01 1.96817148e+00
1.70344204e-01 -5.61835952e-02 8.25526893e-01 6.63105965e-01
8.92885089e-01 1.17711377e+00 -2.27755830e-02 -4.02178705e-01
1.30881858e+00 -1.31640530e+00 -7.65607893e-01 -3.60217005e-01
8.06884289e-01 -9.68071520e-01 1.52685797e+00 1.40935302e-01
-1.07489681e+00 -7.65116155e-01 -9.12156582e-01 -3.03111255e-01
-3.41539621e-01 5.43137312e-01 2.42798492e-01 3.12448651e-01
-8.05032849e-01 4.62975562e-01 -5.64541817e-01 -4.80895698e-01
2.35559251e-02 -2.00931102e-01 -3.62525433e-01 -2.44353712e-01
-1.72496295e+00 9.77193415e-01 8.87824476e-01 7.09230155e-02
-1.02386498e+00 -5.42714953e-01 -9.10969436e-01 -2.16102824e-01
2.84730136e-01 -9.36151087e-01 1.15144575e+00 -8.73490691e-01
-1.71848726e+00 7.22040772e-01 -1.20193899e-01 -5.86372674e-01
5.23960829e-01 -4.14058596e-01 -3.97804171e-01 -2.52682090e-01
3.86898398e-01 6.39587820e-01 8.91447604e-01 -9.98765469e-01
-4.07830089e-01 2.90456135e-02 -3.96008730e-01 4.61192608e-01
-1.82172775e-01 6.92020506e-02 -3.17975730e-01 -9.72289324e-01
-4.02257144e-01 -7.61585176e-01 -1.58967152e-01 -9.83247817e-01
-5.28435290e-01 -2.45000109e-01 2.74089426e-01 -1.05555665e+00
1.13175178e+00 -1.82504404e+00 2.16372192e-01 -4.08287942e-01
-5.25894821e-01 3.30928326e-01 -6.74893796e-01 7.79946208e-01
-5.13634197e-02 1.47340938e-01 -3.43234062e-01 -6.35336578e-01
-1.58721238e-01 1.59797564e-01 -5.25632083e-01 -6.72398731e-02
5.24160147e-01 1.12170982e+00 -9.74332631e-01 -4.63180542e-01
-2.03477457e-01 3.31433147e-01 -4.34058189e-01 5.18948197e-01
-3.75511706e-01 5.21113694e-01 -2.98146516e-01 3.68070573e-01
3.50519747e-01 3.48706841e-02 -7.06705227e-02 -6.82564005e-02
-1.12898707e-01 5.22863567e-01 -7.58057714e-01 1.96395290e+00
-7.84199119e-01 1.96117952e-01 -4.62030292e-01 -7.05925226e-01
1.09526384e+00 5.66924453e-01 -4.50688228e-02 -6.23633623e-01
-2.12528482e-02 3.80727619e-01 -6.53928146e-02 -5.41906953e-01
7.96648085e-01 -3.27745706e-01 -3.79289955e-01 6.32080793e-01
6.20467186e-01 -1.87216729e-01 5.00627339e-01 2.81309038e-01
7.66946733e-01 4.67680514e-01 4.39189672e-01 -1.71872512e-01
4.67247039e-01 6.45903423e-02 3.78832728e-01 7.60451555e-01
3.37480634e-01 8.82125139e-01 3.36802036e-01 7.19578490e-02
-1.04831564e+00 -9.46426272e-01 2.68290758e-01 1.21287191e+00
-3.52852225e-01 -5.13531923e-01 -8.76596689e-01 -6.41054630e-01
-4.58199412e-01 1.33516526e+00 -3.86017680e-01 -3.70329767e-01
-5.93370020e-01 -1.30975258e+00 8.00834775e-01 9.54028964e-02
4.62385118e-01 -1.45936918e+00 -6.53445944e-02 3.76565665e-01
-6.87927544e-01 -1.16039050e+00 -6.81534350e-01 5.79042360e-03
-5.91287732e-01 -6.64394200e-01 -1.14071143e+00 -7.58503020e-01
4.63756680e-01 1.43871186e-02 1.28894818e+00 -5.42349637e-01
1.45698693e-02 6.72636107e-02 -6.51826560e-01 -3.18166703e-01
-1.00109780e+00 5.94854176e-01 8.50238055e-02 1.09307148e-01
1.23113953e-03 -3.43556583e-01 -7.26519674e-02 -1.00276299e-01
-8.82963896e-01 4.29100007e-01 7.78329790e-01 1.14401674e+00
4.68359888e-01 -4.02449727e-01 1.20035291e+00 -1.01048279e+00
1.34433401e+00 -5.79363644e-01 -3.28401476e-01 5.14038980e-01
-3.09343755e-01 4.47748661e-01 9.46576059e-01 -4.91100758e-01
-1.54598391e+00 -4.65050668e-01 -2.31206954e-01 -1.41918093e-01
3.74413095e-02 8.10423434e-01 -2.34964222e-01 6.99454963e-01
9.11290467e-01 5.87792039e-01 -1.61491528e-01 -5.06376565e-01
6.48688853e-01 8.00136387e-01 5.15967846e-01 -7.00516105e-01
6.68360710e-01 -2.31812015e-01 -6.09891176e-01 -8.55848312e-01
-1.04290569e+00 4.38673794e-02 -4.80903745e-01 8.48075598e-02
8.50430489e-01 -1.20464039e+00 4.72797081e-02 5.05898654e-01
-1.52992654e+00 -4.72954214e-01 -5.14123678e-01 4.94513363e-01
-5.28477430e-01 2.43476003e-01 -7.11659968e-01 -7.60365427e-01
-9.98612463e-01 -7.89726555e-01 1.22073448e+00 -7.05415159e-02
-1.94921762e-01 -1.03443277e+00 4.96856183e-01 3.17735642e-01
4.15586591e-01 1.72408968e-01 7.68721342e-01 -7.92934358e-01
-2.25383863e-01 -6.13977872e-02 1.02236517e-01 6.91764057e-01
1.75673991e-01 -2.80647218e-01 -8.61121893e-01 -3.56145203e-01
1.23480394e-01 -7.68595874e-01 8.91112685e-01 1.07945897e-01
5.68957210e-01 -5.58491290e-01 5.76072372e-02 3.84321392e-01
1.22132158e+00 -1.49551809e-01 6.32218301e-01 2.09945813e-02
7.32569456e-01 7.26973474e-01 7.14835167e-01 2.46092558e-01
4.71122921e-01 5.50913930e-01 -3.94726157e-01 -8.92297551e-02
-2.62840003e-01 -5.79090774e-01 1.00825691e+00 1.51248097e+00
-4.91046645e-02 -6.97755575e-01 -8.26547384e-01 5.85667312e-01
-1.80752480e+00 -9.33420718e-01 -5.71670048e-02 2.18688059e+00
1.37842858e+00 -3.17707881e-02 -2.10116237e-01 -5.28441906e-01
6.73841596e-01 1.76586702e-01 -3.57664496e-01 -4.30354387e-01
-5.14047086e-01 1.11907907e-01 1.35843202e-01 5.40252686e-01
-7.30558813e-01 1.48997021e+00 5.32470131e+00 1.14735079e+00
-1.03742695e+00 4.26267594e-01 8.19133341e-01 -1.22496532e-03
-3.85417581e-01 -9.96121243e-02 -1.15917253e+00 5.89690149e-01
1.21198046e+00 -6.22940600e-01 3.84576499e-01 7.08359301e-01
3.42039436e-01 3.14844325e-02 -9.90368009e-01 7.62991548e-01
4.33178067e-01 -1.05968606e+00 4.52131063e-01 -3.89485866e-01
1.04574037e+00 3.00113231e-01 -3.15724909e-01 8.30221713e-01
5.64666629e-01 -8.66965175e-01 5.41854143e-01 5.03904343e-01
9.66777921e-01 -7.00921714e-01 7.75247216e-01 8.25775027e-01
-8.60108316e-01 3.40367436e-01 -5.57600081e-01 2.62564510e-01
5.25393069e-01 6.23833895e-01 -1.01077652e+00 1.00768316e+00
1.67000577e-01 6.07559860e-01 -6.42423153e-01 3.19738954e-01
-3.72322172e-01 6.97251379e-01 -1.20598368e-01 1.01722062e-01
2.83290088e-01 -2.23258749e-01 5.44291556e-01 1.32549465e+00
6.48398876e-01 -2.64626026e-01 9.48351547e-02 8.42920601e-01
-2.33823314e-01 3.90917331e-01 -6.96857810e-01 -9.92079899e-02
1.99107155e-01 1.27296853e+00 -2.34613061e-01 -5.42382360e-01
-2.13302717e-01 1.37738073e+00 4.70875353e-01 5.38843989e-01
-7.34855354e-01 -3.62215787e-01 1.34299368e-01 -2.33406484e-01
-7.72620877e-03 -2.18924172e-02 1.08438730e-01 -1.73551059e+00
-1.11991659e-01 -1.23782611e+00 2.23303199e-01 -8.57167244e-01
-1.61363423e+00 1.04561830e+00 1.80348326e-02 -1.22059715e+00
-9.59753335e-01 -1.24295354e-01 -7.13241518e-01 1.31468320e+00
-1.46487093e+00 -1.33082867e+00 -3.51709574e-02 5.22191346e-01
1.00594354e+00 -5.65674305e-01 8.94249499e-01 2.11222827e-01
-6.30977869e-01 6.74919605e-01 1.20398812e-01 -9.40879993e-03
1.10915267e+00 -1.09419048e+00 7.80054748e-01 1.09344125e+00
4.03082445e-02 5.24307311e-01 5.28194368e-01 -7.68139780e-01
-1.06206357e+00 -1.53776181e+00 1.25657654e+00 -3.39748025e-01
6.55847013e-01 -6.47032022e-01 -8.73054504e-01 6.77723587e-01
6.57322824e-01 -4.49522555e-01 5.78885257e-01 -3.41201350e-02
2.43806373e-02 6.22547716e-02 -7.51222491e-01 8.53229880e-01
7.93269455e-01 -3.51559281e-01 -6.13048375e-01 4.42185640e-01
1.08584797e+00 -2.58520156e-01 -5.87320387e-01 2.22065657e-01
1.03585869e-01 -5.09762347e-01 6.70771003e-01 -7.38695920e-01
6.77691698e-01 -6.66277707e-02 1.05248196e-02 -1.88440180e+00
4.68737297e-02 -8.32445085e-01 3.57666239e-02 1.70501971e+00
6.37721062e-01 -5.91297984e-01 1.64791569e-01 1.86625104e-02
-4.28450942e-01 -5.36335051e-01 -7.32552707e-01 -7.69931257e-01
3.00774634e-01 -1.65386692e-01 3.56518954e-01 8.98806870e-01
-1.29446954e-01 1.33940232e+00 -9.40734267e-01 -3.01619947e-01
3.81490529e-01 4.10767272e-02 9.98988271e-01 -6.61513865e-01
-4.04093534e-01 -2.60738116e-02 4.38509464e-01 -8.38464439e-01
2.80241191e-01 -1.26609826e+00 3.38795990e-01 -1.47114563e+00
7.53696561e-02 -1.06533200e-01 8.74971040e-03 4.71326619e-01
-6.48899138e-01 -5.07253930e-02 1.18890546e-01 1.68560237e-01
-3.45488697e-01 1.08726335e+00 1.26493680e+00 1.42203033e-01
-3.40296686e-01 1.13361459e-02 -6.96547747e-01 2.71081984e-01
9.84440565e-01 -4.26096857e-01 -6.08944178e-01 -5.03513575e-01
4.99529243e-02 1.85623124e-01 -6.86396360e-02 -8.55080307e-01
-7.48540759e-02 4.26365621e-02 1.22305788e-01 -5.30899704e-01
8.79993588e-02 -1.25715975e-02 1.87790516e-04 1.73386037e-01
-5.59715629e-01 8.26896057e-02 1.89793378e-01 1.92818999e-01
-3.62810820e-01 -4.63278562e-01 5.81720471e-01 -3.12970221e-01
-3.35307658e-01 1.85952231e-01 -3.02952856e-01 5.47163665e-01
5.68712533e-01 2.71949828e-01 -4.09694731e-01 -5.06223917e-01
-4.13356245e-01 2.30462700e-01 1.45334601e-01 7.41859138e-01
4.75182235e-01 -1.51226723e+00 -1.50574243e+00 1.73329592e-01
2.35630617e-01 -6.06471561e-02 3.28697264e-01 6.95472062e-01
-2.22798839e-01 4.84587133e-01 -4.09135297e-02 -2.36860305e-01
-8.19190681e-01 4.35017467e-01 1.11600079e-01 -8.93946350e-01
-3.22028756e-01 5.86009800e-01 1.96851730e-01 -8.05253208e-01
-1.51546821e-01 -9.56936777e-02 -2.48008132e-01 3.32559794e-01
4.94687170e-01 2.49690354e-01 2.16354243e-02 -5.44139802e-01
2.50348859e-02 2.26851285e-01 -3.05272609e-01 -5.18796265e-01
1.11126447e+00 -1.48904160e-01 -6.51871040e-02 7.74730206e-01
1.15723598e+00 5.37581220e-02 -8.52027178e-01 -1.93885043e-01
6.37114346e-02 1.84700131e-01 -3.20277333e-01 -8.03963542e-01
-6.14206433e-01 9.16406035e-01 8.36013183e-02 -1.10052511e-01
8.50874126e-01 -1.78679287e-01 9.18747187e-01 4.12345916e-01
2.76991785e-01 -1.19914865e+00 1.96935222e-01 9.54653144e-01
1.30576599e+00 -1.29998410e+00 -2.47058958e-01 -2.94589531e-02
-1.06959021e+00 7.69805789e-01 6.60022736e-01 3.89978886e-02
7.10939839e-02 4.06272598e-02 6.11273870e-02 2.28139192e-01
-1.26910377e+00 -2.34619707e-01 3.63743693e-01 5.21441638e-01
7.58028209e-01 -1.56870082e-01 -5.34013689e-01 8.04468334e-01
-5.97585559e-01 1.76309660e-01 5.10055840e-01 4.69850421e-01
-1.92361340e-01 -1.18925536e+00 -2.57500768e-01 2.18592986e-01
-3.98859203e-01 -7.01028585e-01 -1.92249522e-01 6.68230474e-01
-4.17255834e-02 8.86689246e-01 -1.88210964e-01 -1.53751701e-01
1.91915080e-01 3.93148959e-01 1.51839450e-01 -9.47158277e-01
-7.31713355e-01 2.40678191e-01 3.76714796e-01 -9.74844694e-02
-1.33400530e-01 -4.91424471e-01 -9.29232180e-01 1.33021977e-02
-3.22708219e-01 3.86502534e-01 4.28688824e-01 8.51894736e-01
4.36250210e-01 6.76124930e-01 6.21852219e-01 -6.22403920e-01
-6.84622586e-01 -1.62612951e+00 -2.20968574e-01 3.82771105e-01
3.30358408e-02 4.16437909e-02 -2.99642324e-01 3.06703180e-01] | [11.911966323852539, 9.291004180908203] |
f6fddc64-dbdd-4966-bc7c-95ef32759bee | multi-task-self-supervised-time-series | 2303.01034 | null | https://arxiv.org/abs/2303.01034v1 | https://arxiv.org/pdf/2303.01034v1.pdf | Multi-Task Self-Supervised Time-Series Representation Learning | Time-series representation learning can extract representations from data with temporal dynamics and sparse labels. When labeled data are sparse but unlabeled data are abundant, contrastive learning, i.e., a framework to learn a latent space where similar samples are close to each other while dissimilar ones are far from each other, has shown outstanding performance. This strategy can encourage varied consistency of time-series representations depending on the positive pair selection and contrastive loss. We propose a new time-series representation learning method by combining the advantages of self-supervised tasks related to contextual, temporal, and transformation consistency. It allows the network to learn general representations for various downstream tasks and domains. Specifically, we first adopt data preprocessing to generate positive and negative pairs for each self-supervised task. The model then performs contextual, temporal, and transformation contrastive learning and is optimized jointly using their contrastive losses. We further investigate an uncertainty weighting approach to enable effective multi-task learning by considering the contribution of each consistency. We evaluate the proposed framework on three downstream tasks: time-series classification, forecasting, and anomaly detection. Experimental results show that our method not only outperforms the benchmark models on these downstream tasks, but also shows efficiency in cross-domain transfer learning. | ['Pilsung Kang', 'Heejeong Choi'] | 2023-03-02 | null | null | null | null | ['time-series-classification'] | ['time-series'] | [ 2.16691360e-01 -3.56934249e-01 -2.71057129e-01 -6.43800318e-01
-8.72579396e-01 -4.05971378e-01 7.24794745e-01 2.03095585e-01
-1.79159299e-01 5.63110590e-01 -1.41200930e-01 1.69470552e-02
-3.68838906e-01 -4.60291356e-01 -5.09642005e-01 -9.83231962e-01
-5.26261628e-01 4.25440431e-01 -6.17116019e-02 1.82328485e-02
7.51523972e-02 3.71411711e-01 -1.52944636e+00 1.81826189e-01
1.07493663e+00 1.53928494e+00 -1.83013394e-01 -4.47352380e-02
-1.19761497e-01 8.38638186e-01 -4.45804536e-01 5.80341592e-02
2.40641102e-01 -3.00926059e-01 -3.17269295e-01 6.51550386e-03
6.29080981e-02 9.51315239e-02 -1.64165869e-01 8.96973431e-01
3.65194261e-01 4.55458790e-01 1.09811282e+00 -1.73853302e+00
-8.51803422e-01 3.38861108e-01 -8.96246254e-01 5.04276931e-01
-1.75627455e-01 -4.40153033e-02 1.07083416e+00 -8.97000372e-01
1.53432235e-01 1.14143908e+00 5.43175519e-01 4.17293191e-01
-1.29320037e+00 -1.04246926e+00 7.03458011e-01 3.68290007e-01
-1.15930212e+00 -3.14617217e-01 1.36583412e+00 -6.61943138e-01
8.42288256e-01 1.80339441e-02 2.17901930e-01 1.38639498e+00
2.15047717e-01 9.06102300e-01 1.00681686e+00 -1.98767006e-01
4.60477859e-01 2.28442419e-02 1.83738157e-01 2.60141253e-01
-3.76294434e-01 2.27037176e-01 -5.54939091e-01 -3.08996558e-01
4.72218513e-01 4.03839737e-01 -9.39388499e-02 -6.52922332e-01
-1.22772515e+00 8.33036840e-01 3.55384141e-01 3.80893379e-01
-5.09163797e-01 -1.84521060e-02 8.18203032e-01 8.47019434e-01
1.00449228e+00 1.84442237e-01 -6.45114124e-01 9.49375853e-02
-6.75874114e-01 -1.03406198e-01 2.60827899e-01 8.07613790e-01
5.62596381e-01 4.00907069e-01 -4.22373205e-01 1.01609457e+00
1.91622034e-01 2.89026678e-01 1.03377581e+00 -5.10591865e-01
7.46951044e-01 4.04694289e-01 -4.11458947e-02 -9.76975858e-01
-2.70615965e-01 -5.09995162e-01 -1.17572284e+00 -3.73312645e-02
1.94253489e-01 -1.79623425e-01 -9.54730928e-01 2.06734681e+00
2.32835338e-01 6.53706491e-01 9.75581631e-02 6.63806200e-01
3.56634319e-01 9.78815496e-01 2.45495543e-01 -6.72136247e-01
8.15208912e-01 -7.24398255e-01 -8.61999393e-01 -2.35385254e-01
6.29136562e-01 -5.69334805e-01 1.01886237e+00 1.89617276e-01
-8.27616394e-01 -5.77834606e-01 -1.07761049e+00 2.68468499e-01
-3.37320566e-01 1.64710462e-01 4.05520588e-01 3.93767888e-03
-6.35470331e-01 7.61969209e-01 -1.09935713e+00 -2.09664945e-02
4.12777513e-01 9.89231020e-02 -2.80631065e-01 2.19157800e-01
-1.30521333e+00 5.41701555e-01 4.61884826e-01 4.77624945e-02
-6.87456429e-01 -7.15946913e-01 -9.67470229e-01 7.39684328e-02
5.51623330e-02 -1.86587363e-01 1.01707602e+00 -1.15318048e+00
-1.38221455e+00 7.09168851e-01 3.73031199e-02 -5.48329175e-01
5.16808569e-01 -1.89378589e-01 -8.44790220e-01 4.68298681e-02
2.94121712e-01 3.23107034e-01 1.32102668e+00 -8.35083187e-01
-4.82976347e-01 -4.34192926e-01 -6.14085555e-01 1.41764671e-01
-7.22213626e-01 -1.22216992e-01 3.06038726e-02 -1.15985596e+00
3.78535330e-01 -7.82258153e-01 -1.58099666e-01 1.95997879e-01
-4.67941724e-02 -5.71581781e-01 1.25640619e+00 -4.92541015e-01
1.14899349e+00 -2.53063989e+00 1.73845753e-01 4.76589203e-01
-1.07038744e-01 5.84932752e-02 -2.52326041e-01 3.20782334e-01
-4.84615624e-01 -2.94587731e-01 -3.63150060e-01 -3.59045088e-01
-3.86262909e-02 1.64541066e-01 -7.34870732e-01 5.05205810e-01
5.23790896e-01 6.16755843e-01 -9.41445172e-01 -2.53724843e-01
2.30413616e-01 3.00406098e-01 -1.73771665e-01 3.59891206e-01
-1.62923515e-01 7.38922060e-01 -6.84030890e-01 6.47388339e-01
4.79634345e-01 -5.73346257e-01 -7.06500262e-02 -9.47141126e-02
1.16037004e-01 1.94482431e-01 -9.69804227e-01 1.43971479e+00
-5.47618151e-01 4.15574849e-01 -3.73365998e-01 -1.65218627e+00
1.40149200e+00 4.96178538e-01 8.22434664e-01 -8.09985459e-01
-1.20548226e-01 4.07323599e-01 -2.03042492e-01 -4.02441144e-01
7.05889240e-03 -2.15369433e-01 -1.64176971e-01 4.99222428e-01
8.79687294e-02 5.21041267e-02 -1.34336591e-01 -1.66305020e-01
7.41948843e-01 2.29676500e-01 3.20948988e-01 -1.77275077e-01
4.92963701e-01 -4.60894823e-01 7.80338347e-01 3.08076888e-01
-4.25035775e-01 3.92761528e-01 4.35061127e-01 -6.10945046e-01
-8.76226842e-01 -1.14100397e+00 -1.39509603e-01 1.29684091e+00
9.01254360e-03 3.01055927e-02 9.75081995e-02 -8.21729600e-01
1.60368875e-01 7.45424688e-01 -7.28704393e-01 -5.10067344e-01
-4.43060338e-01 -8.31639409e-01 1.22059502e-01 7.93899417e-01
2.64433622e-01 -1.19974768e+00 -2.54594147e-01 3.40931654e-01
-2.38781616e-01 -7.28986442e-01 -5.46984613e-01 5.44631720e-01
-1.11682963e+00 -6.54790342e-01 -8.08735251e-01 -8.56363893e-01
5.82409024e-01 1.44383490e-01 8.82031679e-01 -4.42888826e-01
3.55447493e-02 4.34237570e-01 -3.66528183e-01 -3.10617089e-01
-1.01125184e-02 -1.13724664e-01 2.93777972e-01 5.10296881e-01
4.18941438e-01 -1.05135787e+00 -4.84840602e-01 4.46175307e-01
-9.00650740e-01 -5.05244017e-01 3.44656914e-01 1.14171088e+00
9.54162657e-01 -9.95567907e-03 1.11680567e+00 -5.91351807e-01
6.34927630e-01 -1.00918078e+00 -4.41059172e-01 3.51928890e-01
-7.44370580e-01 5.10881059e-02 8.06980371e-01 -8.94187212e-01
-1.04717731e+00 -1.20357759e-01 3.27531576e-01 -1.06347537e+00
1.83082130e-02 6.00485265e-01 1.68558925e-01 3.32200468e-01
6.34873509e-01 4.12509739e-01 1.30784318e-01 -3.29002529e-01
2.18406633e-01 4.44886088e-01 3.17837000e-01 -8.55792165e-01
8.60437274e-01 4.30588931e-01 -2.39688009e-01 -5.68661451e-01
-9.09979045e-01 -5.90485632e-01 -5.71683526e-01 -5.31255268e-02
4.88276631e-01 -1.11339891e+00 -1.57912359e-01 5.57274103e-01
-7.12689102e-01 -2.05618024e-01 -5.31755507e-01 7.38463938e-01
-7.12582350e-01 3.11766297e-01 -5.03753603e-01 -9.39679205e-01
-3.02584320e-01 -6.69050336e-01 1.02910018e+00 4.54510637e-02
-1.86872452e-01 -1.14785910e+00 5.84628284e-02 -1.86639473e-01
4.27417934e-01 4.22766656e-01 9.94032383e-01 -1.05679715e+00
-1.33082882e-01 -1.81355029e-01 -1.01606064e-01 5.32511294e-01
4.06202734e-01 -1.78728387e-01 -1.03614962e+00 -4.55390364e-01
2.17694357e-01 -5.73658705e-01 9.39183533e-01 4.36358988e-01
1.61115694e+00 -3.74033630e-01 -3.58951628e-01 5.80149353e-01
1.04236984e+00 4.15786356e-01 3.50398272e-01 3.11021566e-01
5.95697463e-01 8.80592823e-01 9.84778166e-01 6.27457917e-01
1.41218618e-01 5.00838399e-01 2.00315088e-01 1.11741640e-01
4.97403979e-01 -9.51665118e-02 5.17332494e-01 1.07140100e+00
1.15481727e-01 -1.01302147e-01 -8.30780864e-01 7.13627338e-01
-2.21425653e+00 -1.10051978e+00 2.89722323e-01 2.09378767e+00
7.31228232e-01 2.02018723e-01 3.16261984e-02 2.20869929e-01
7.75299668e-01 4.33280349e-01 -1.03239346e+00 3.37725170e-02
-1.12791695e-01 5.92159517e-02 -6.24935105e-02 -8.99681598e-02
-1.49466848e+00 5.90764582e-01 5.36912394e+00 9.55111623e-01
-1.37341583e+00 7.29362965e-02 8.97669256e-01 -1.16049893e-01
-4.25354719e-01 -3.52561116e-01 -2.96743393e-01 7.77639329e-01
7.94186711e-01 -3.88350815e-01 1.83471471e-01 8.05688798e-01
1.76169679e-01 6.37131572e-01 -1.44589925e+00 1.01997530e+00
-2.49258623e-01 -9.10277724e-01 4.46857251e-02 -3.08284938e-01
7.99729943e-01 -9.28416476e-03 5.35230458e-01 6.72484159e-01
1.54797751e-02 -7.74655640e-01 6.62918568e-01 4.88391042e-01
5.53419113e-01 -6.39974415e-01 6.32420599e-01 2.86947846e-01
-1.45714104e+00 -3.52850944e-01 -1.42477810e-01 1.03192672e-01
1.00968580e-03 6.35165632e-01 -2.69197881e-01 5.82753003e-01
7.12200642e-01 1.33259928e+00 -1.90658465e-01 8.05608273e-01
8.06545615e-02 6.15434527e-01 -2.10748151e-01 1.87136188e-01
2.60119826e-01 -5.22562683e-01 5.63503265e-01 8.04954410e-01
7.45972335e-01 -9.31230411e-02 4.43016261e-01 8.48270893e-01
1.57777473e-01 2.49417037e-01 -7.24987388e-01 -1.00021578e-01
7.31672943e-01 8.80388618e-01 -3.85713935e-01 -3.06089222e-01
-4.13333148e-01 1.00141132e+00 5.69958448e-01 6.27301991e-01
-7.76280761e-01 -2.17259929e-01 5.81224144e-01 -2.72074074e-01
3.16508740e-01 -9.67663229e-02 -9.08311270e-03 -1.24578321e+00
4.28929240e-01 -6.11991286e-01 1.08602595e+00 -4.95721161e-01
-2.14850163e+00 6.52428865e-01 7.21642971e-02 -1.96049201e+00
-5.77033103e-01 -3.24502856e-01 -9.18338776e-01 7.62966752e-01
-1.75987554e+00 -1.09444642e+00 9.58074108e-02 9.53331292e-01
5.76028168e-01 -5.50818145e-01 6.64816260e-01 3.20403963e-01
-6.28555775e-01 6.61690474e-01 4.29236859e-01 -5.17252013e-02
8.35497677e-01 -1.21482265e+00 8.17158893e-02 6.24073505e-01
-1.84090603e-02 5.13806164e-01 2.90457785e-01 -4.04495925e-01
-8.10477197e-01 -1.50557542e+00 6.67094946e-01 1.54523307e-03
9.44232106e-01 -1.06009983e-01 -1.43651581e+00 8.23886991e-01
-2.64790803e-01 4.35726702e-01 8.41759324e-01 3.11094075e-01
-6.71699047e-01 -4.48956907e-01 -1.06956255e+00 3.64977330e-01
8.77258897e-01 -8.21776986e-01 -7.08297133e-01 5.25696993e-01
6.52830780e-01 -1.57346234e-01 -8.80607963e-01 5.21666288e-01
2.79395461e-01 -6.42321527e-01 9.84606445e-01 -7.24305511e-01
2.88885176e-01 -1.85201362e-01 -1.35071605e-01 -1.60259593e+00
-4.44855094e-01 -5.30124903e-01 -4.97273594e-01 1.41399801e+00
2.13079080e-01 -9.21066701e-01 3.63703668e-01 4.26463842e-01
-1.57895669e-01 -6.53281152e-01 -1.03273749e+00 -1.24065375e+00
1.50244161e-01 -3.42987657e-01 4.84082788e-01 1.46404052e+00
3.38353887e-02 2.87238568e-01 -5.68753779e-01 2.70642668e-01
6.17212832e-01 4.78297621e-01 1.82879850e-01 -1.51813412e+00
-4.12830450e-02 -4.70997632e-01 -3.13377589e-01 -8.19331050e-01
4.97436494e-01 -1.03572857e+00 -6.98917806e-02 -8.21085870e-01
-1.23447567e-01 -6.98456943e-01 -1.08026493e+00 5.33074558e-01
-1.99810371e-01 -2.08421528e-01 -2.24912301e-01 7.04715371e-01
-5.20790696e-01 1.12275970e+00 8.82260084e-01 -2.48670608e-01
-4.89230931e-01 1.93927705e-01 -3.67955863e-01 5.57233512e-01
9.24838662e-01 -4.19583917e-01 -7.72248924e-01 -1.76655799e-01
-2.31737167e-01 1.67207539e-01 2.87229836e-01 -9.18753803e-01
2.89216563e-02 -2.15181991e-01 4.40618664e-01 -6.55662179e-01
3.45756948e-01 -1.04458129e+00 -1.21794529e-01 3.38481039e-01
-6.54538035e-01 7.93926939e-02 1.78229257e-01 1.09578466e+00
-5.96455336e-01 1.91153780e-01 8.19613278e-01 1.00977011e-01
-8.63104284e-01 7.37565815e-01 -4.46903370e-02 2.08422035e-01
1.24032271e+00 1.38627854e-03 -1.76120028e-02 -4.19889867e-01
-8.92457187e-01 5.04468739e-01 -1.22768447e-01 7.58747816e-01
6.18506908e-01 -1.94151258e+00 -7.07548261e-01 4.91984040e-01
6.18660688e-01 -2.39675179e-01 3.82653773e-01 6.69412673e-01
3.48023444e-01 1.62519947e-01 -5.07112503e-01 -9.89851594e-01
-7.43802726e-01 8.18396926e-01 2.86438406e-01 -3.55957687e-01
-6.67599201e-01 7.19348907e-01 5.10951996e-01 -5.07811427e-01
4.01022583e-01 -4.00357664e-01 -3.71905029e-01 3.81039351e-01
4.75858241e-01 2.33368114e-01 -1.45955496e-02 -4.41785872e-01
-3.94926131e-01 5.56414247e-01 -3.25407267e-01 9.34451744e-02
1.44181359e+00 -1.49086937e-01 1.63667127e-01 1.01243269e+00
1.44806647e+00 -5.65214574e-01 -1.45558691e+00 -7.52725780e-01
2.68262804e-01 -3.95676315e-01 -1.71609208e-01 -3.59102428e-01
-1.18080235e+00 8.82189393e-01 8.11215460e-01 4.89337236e-01
1.35614479e+00 -7.15555176e-02 6.02057159e-01 2.72080451e-01
2.22292617e-01 -1.25293422e+00 5.98087668e-01 4.62316245e-01
1.06894708e+00 -1.48478687e+00 -3.65534812e-01 -1.42727152e-01
-7.88547993e-01 9.60691988e-01 7.97846019e-01 -2.76896209e-01
8.64785969e-01 -1.31495282e-01 8.90372600e-03 -2.19977319e-01
-9.82587218e-01 1.37511164e-01 6.24329925e-01 6.69131637e-01
4.42824304e-01 4.38046008e-02 2.29067802e-02 6.12887084e-01
3.56488734e-01 -1.91801995e-01 -1.56275198e-01 8.19723427e-01
-8.08980688e-02 -8.20178151e-01 -1.45637795e-01 6.36725128e-01
-3.29486877e-01 2.05958948e-01 1.61530618e-02 5.88444531e-01
-2.03501612e-01 7.51404643e-01 3.24210852e-01 -2.20019624e-01
2.59663969e-01 3.30826670e-01 8.75723585e-02 -4.26955283e-01
-2.79184461e-01 1.57239750e-01 -2.19882399e-01 -4.89014596e-01
-4.84842181e-01 -7.32560873e-01 -1.09613657e+00 2.17055351e-01
-1.30674183e-01 1.95858702e-01 2.06531495e-01 9.18791473e-01
4.43763614e-01 6.29908144e-01 1.25889647e+00 -8.29409480e-01
-1.01553893e+00 -9.04603302e-01 -7.43395388e-01 7.12545395e-01
7.15782881e-01 -9.24563110e-01 -6.59190416e-01 2.04655342e-03] | [7.373469829559326, 2.9037482738494873] |
1b915ec3-deb9-4d31-a0b5-605d2ed3a834 | graph-to-sequence-neural-machine-translation | 2009.07489 | null | https://arxiv.org/abs/2009.07489v1 | https://arxiv.org/pdf/2009.07489v1.pdf | Graph-to-Sequence Neural Machine Translation | Neural machine translation (NMT) usually works in a seq2seq learning way by viewing either source or target sentence as a linear sequence of words, which can be regarded as a special case of graph, taking words in the sequence as nodes and relationships between words as edges. In the light of the current NMT models more or less capture graph information among the sequence in a latent way, we present a graph-to-sequence model facilitating explicit graph information capturing. In detail, we propose a graph-based SAN-based NMT model called Graph-Transformer by capturing information of subgraphs of different orders in every layers. Subgraphs are put into different groups according to their orders, and every group of subgraphs respectively reflect different levels of dependency between words. For fusing subgraph representations, we empirically explore three methods which weight different groups of subgraphs of different orders. Results of experiments on WMT14 English-German and IWSLT14 German-English show that our method can effectively boost the Transformer with an improvement of 1.1 BLEU points on WMT14 English-German dataset and 1.0 BLEU points on IWSLT14 German-English dataset. | ['Hai Zhao', 'Sufeng Duan', 'Rui Wang'] | 2020-09-16 | null | null | null | null | ['graph-to-sequence'] | ['natural-language-processing'] | [ 5.10323882e-01 2.35357881e-01 -4.72865254e-01 -3.29982370e-01
-4.39495236e-01 -6.91557646e-01 6.50475800e-01 7.50656277e-02
-3.41461718e-01 7.54041672e-01 6.04845464e-01 -7.00521946e-01
1.29472777e-01 -9.27694619e-01 -9.39319551e-01 -3.99628133e-01
2.40624286e-02 5.91643333e-01 -2.97037531e-02 -5.56362510e-01
-4.97907400e-02 1.70730148e-02 -7.02396572e-01 5.12893677e-01
1.02120209e+00 3.01027596e-01 3.66604090e-01 4.85368699e-01
-6.67218566e-01 5.55419445e-01 -5.05868256e-01 -7.58779287e-01
3.19973916e-01 -1.02885628e+00 -7.80449092e-01 7.48426467e-02
3.81921351e-01 2.65052438e-01 -5.75454295e-01 1.24883401e+00
4.20492738e-01 -1.87102750e-01 6.10906482e-01 -1.06990731e+00
-9.91087556e-01 1.36928570e+00 -8.46057296e-01 1.46222070e-01
2.83822805e-01 4.62786928e-02 1.43726850e+00 -7.78599560e-01
8.55506957e-01 1.40226018e+00 5.68835676e-01 5.75479507e-01
-1.29075480e+00 -4.63248968e-01 4.52423215e-01 3.78694743e-01
-1.14441693e+00 -4.78038378e-02 5.78367054e-01 -4.66223955e-02
1.37107289e+00 4.16670650e-01 7.92731583e-01 1.17353392e+00
6.21146858e-01 7.88292587e-01 1.00513315e+00 -4.17608231e-01
-1.20231591e-01 -3.02332014e-01 3.40047956e-01 8.39140415e-01
3.55410069e-01 -2.18148947e-01 -6.66688383e-01 1.61304787e-01
5.52928567e-01 -1.82735950e-01 -2.82626063e-01 -2.14119583e-01
-1.56426573e+00 8.97544146e-01 6.58679783e-01 5.91307521e-01
-2.16064304e-01 2.67599583e-01 5.15490294e-01 7.65444100e-01
6.60908818e-01 2.23498210e-01 -4.01926100e-01 1.51705518e-01
-6.51859105e-01 -1.65060535e-01 6.44979835e-01 1.38465929e+00
9.41834271e-01 1.50561601e-01 -5.33615768e-01 7.50816286e-01
1.85814589e-01 5.11089921e-01 5.72875082e-01 -1.84531808e-01
9.85613167e-01 8.48742425e-01 -5.68347275e-01 -8.65990818e-01
-2.08049700e-01 -8.08811605e-01 -1.16548991e+00 -5.91032028e-01
4.10949253e-02 -8.78152996e-02 -1.09298658e+00 2.06686258e+00
-9.90870297e-02 3.78686637e-01 2.75007430e-02 6.74041152e-01
8.79778624e-01 1.00019550e+00 -1.43148273e-01 -5.47496974e-01
1.44362772e+00 -1.25793171e+00 -8.80304754e-01 -4.73440945e-01
9.26396072e-01 -6.41753733e-01 1.20635605e+00 -1.77126098e-02
-1.02588809e+00 -5.04442632e-01 -8.14711332e-01 -1.04738958e-01
-4.13689107e-01 -2.42037296e-01 4.57841277e-01 3.57843578e-01
-1.42519248e+00 5.77946842e-01 -4.57555205e-01 -3.70961219e-01
4.56897393e-02 2.09343374e-01 -4.91429389e-01 -3.56909007e-01
-1.60661256e+00 9.25847113e-01 5.22728741e-01 2.01018199e-01
-6.55579805e-01 -5.48295915e-01 -1.12896466e+00 3.68765630e-02
3.28193396e-01 -1.05151618e+00 8.27722490e-01 -1.13107014e+00
-1.26899195e+00 8.94937754e-01 -4.26145494e-01 -4.91280079e-01
4.09523934e-01 6.16721623e-02 -5.31976581e-01 -1.08344868e-01
3.50259803e-02 5.81928730e-01 6.89288497e-01 -8.81696105e-01
-2.79965222e-01 -3.88785422e-01 -1.74756497e-02 3.54353756e-01
-3.19774359e-01 5.21523925e-03 -5.33001006e-01 -8.16581905e-01
3.92865278e-02 -1.09807491e+00 -2.59350300e-01 -5.80636978e-01
-6.64582729e-01 -3.02335829e-01 6.02225125e-01 -7.34561801e-01
1.58953953e+00 -1.64456880e+00 6.96941853e-01 -4.96388152e-02
4.93456900e-01 1.97447851e-01 -7.31094480e-01 8.70936334e-01
-1.91221699e-01 2.93823868e-01 -3.98148954e-01 -2.83050805e-01
1.32649066e-02 6.17679060e-01 1.91596411e-02 1.90871373e-01
4.09852386e-01 1.47291791e+00 -1.17215598e+00 -4.39115226e-01
-1.18862011e-01 3.02904695e-01 -3.35117400e-01 3.51649970e-02
-3.01266164e-01 2.72575587e-01 -4.31062758e-01 1.15951210e-01
7.08353698e-01 -3.06519091e-01 6.03850782e-01 -1.50586575e-01
5.60404472e-02 5.60925484e-01 -6.62190378e-01 1.92840111e+00
-5.26335776e-01 4.78253424e-01 -1.70792416e-01 -1.05685103e+00
8.23199749e-01 1.76932737e-01 -3.49574424e-02 -6.31950319e-01
1.52809853e-02 2.46511161e-01 2.84410059e-01 -2.84377903e-01
5.54847240e-01 -2.63251960e-01 -2.85981745e-01 3.72932196e-01
3.23713720e-01 9.99761224e-02 4.96983260e-01 3.42554480e-01
1.23467350e+00 9.72837210e-02 4.19866800e-01 -5.01735508e-01
3.97429705e-01 -9.66408104e-02 5.60775220e-01 5.01954079e-01
3.18829387e-01 4.67878431e-01 6.98063195e-01 -3.51627588e-01
-1.07717490e+00 -8.88515234e-01 4.52127665e-01 9.40064132e-01
-8.80419388e-02 -8.25332642e-01 -7.61214793e-01 -8.59079838e-01
-2.09187835e-01 8.62314343e-01 -6.25096917e-01 -4.10400003e-01
-9.80191886e-01 -8.38142097e-01 4.93411720e-01 3.51814181e-01
3.53167593e-01 -9.76429343e-01 4.85057160e-02 1.90706953e-01
-5.86830437e-01 -1.16387272e+00 -1.13438237e+00 2.70738155e-01
-9.09118414e-01 -7.11725831e-01 -7.90103614e-01 -1.14304709e+00
7.59359121e-01 4.98398364e-01 1.61805010e+00 6.40401095e-02
1.00153476e-01 -5.93158267e-02 -5.49411178e-01 -5.66171147e-02
-6.71624601e-01 4.39714283e-01 -8.17424506e-02 2.11548060e-01
2.47219801e-01 -6.64138377e-01 -3.29397827e-01 3.77064608e-02
-8.75291467e-01 5.28861701e-01 7.48979211e-01 8.43265057e-01
5.14116287e-01 -4.78900790e-01 2.69993544e-01 -1.10930872e+00
7.80328393e-01 -6.03420436e-01 -3.10935766e-01 5.41217923e-01
-6.01589441e-01 4.50599253e-01 8.39685917e-01 -3.79019588e-01
-6.33038759e-01 -3.43223840e-01 3.26360986e-02 -4.16044116e-01
3.57625186e-01 8.81322503e-01 -3.67279947e-01 2.97143459e-01
2.50191450e-01 6.71486139e-01 -7.68762082e-02 -3.59086245e-01
6.10742807e-01 4.13229853e-01 7.19128773e-02 -4.34111238e-01
8.71139050e-01 -2.21998706e-01 1.55939028e-01 -5.67056775e-01
-5.53918064e-01 -1.39182389e-01 -6.66460216e-01 -6.73941150e-02
8.92544925e-01 -8.43194425e-01 -2.71288753e-01 4.51287478e-01
-1.27289784e+00 -3.70876729e-01 -8.14427286e-02 2.65706599e-01
-2.55319804e-01 6.72610462e-01 -9.16625500e-01 -2.18293384e-01
-5.47822475e-01 -1.11395240e+00 1.05895483e+00 -1.69098347e-01
-1.64318994e-01 -1.28367305e+00 1.51340976e-01 4.43771817e-02
3.16845834e-01 1.36075124e-01 1.41877782e+00 -6.04862511e-01
-4.51095700e-01 2.26578359e-02 -1.52141407e-01 3.63969982e-01
3.41991514e-01 -3.44352126e-01 -1.96625605e-01 -6.09044790e-01
-2.29280323e-01 6.34299964e-02 9.78179455e-01 1.02316938e-01
6.72878087e-01 -5.76909900e-01 -2.93247908e-01 5.05885959e-01
1.59798467e+00 -2.20303401e-01 6.32311881e-01 -3.32096443e-02
1.07665277e+00 4.58872020e-01 9.58853364e-02 -1.76876590e-01
4.18897152e-01 6.77085102e-01 3.23250353e-01 -8.70032534e-02
-6.20770574e-01 -7.68282652e-01 8.18590403e-01 1.77182424e+00
-7.43142739e-02 -7.92350650e-01 -6.67406142e-01 3.56927603e-01
-1.80761504e+00 -8.03136170e-01 -4.90694165e-01 1.96072137e+00
8.89848471e-01 1.89942807e-01 -5.88218980e-02 -1.48875400e-01
9.97597039e-01 5.11766851e-01 -2.99982637e-01 -7.06367195e-01
-3.54349524e-01 1.97907940e-01 6.16871774e-01 7.01330483e-01
-3.48956227e-01 1.22280443e+00 6.02506828e+00 1.05880916e+00
-1.04158902e+00 9.70919356e-02 5.47897458e-01 2.66786128e-01
-8.87781799e-01 1.20699584e-01 -4.93217111e-01 6.11662447e-01
1.10815179e+00 -4.41046268e-01 6.80725992e-01 1.27905175e-01
1.39258057e-01 5.37077606e-01 -1.03965867e+00 7.86180258e-01
8.88308808e-02 -1.29877675e+00 8.32022786e-01 1.80926919e-01
7.11553216e-01 1.74066380e-01 -1.74499646e-01 4.71758395e-01
4.58392680e-01 -1.05569148e+00 6.63815379e-01 3.66008729e-01
9.35133517e-01 -6.71019971e-01 6.82325184e-01 4.83185530e-01
-1.50348544e+00 3.71750414e-01 -5.11984468e-01 -1.46272838e-01
1.47458792e-01 5.56138754e-01 -7.84567237e-01 1.28727591e+00
2.26328805e-01 1.02073956e+00 -3.34392518e-01 4.14668411e-01
-6.32512808e-01 9.15746570e-01 1.52891621e-01 -3.31652850e-01
6.27230704e-01 -7.07952917e-01 6.99381292e-01 1.57842493e+00
5.39460540e-01 -1.26050591e-01 1.47035003e-01 6.68063998e-01
-5.03364861e-01 3.49815488e-01 -9.10164475e-01 -1.12100914e-01
2.91465074e-01 1.10403609e+00 -6.23037219e-01 -4.22287583e-01
-5.68193018e-01 1.30203533e+00 5.69861412e-01 4.01154608e-01
-8.47480237e-01 -2.09806442e-01 5.31589389e-01 -5.28131786e-04
2.62637883e-01 -4.06809062e-01 2.27461997e-02 -1.37930727e+00
1.01268953e-02 -1.08189273e+00 2.99192846e-01 -7.30304301e-01
-1.34400225e+00 1.01965630e+00 -2.30977964e-02 -1.31301486e+00
-1.73955798e-01 -6.12945259e-01 -5.93059301e-01 9.75572884e-01
-1.46572030e+00 -1.19511962e+00 -9.77830589e-03 6.38635099e-01
8.18236887e-01 -1.13837510e-01 8.00177038e-01 9.43444744e-02
-6.14739716e-01 8.32560897e-01 8.78320411e-02 2.29168832e-01
3.85795385e-01 -1.27687609e+00 1.15412128e+00 1.05981088e+00
6.35483682e-01 7.80804813e-01 6.42950714e-01 -7.82368243e-01
-1.71523595e+00 -1.29590404e+00 1.53199542e+00 -2.14775041e-01
7.75205195e-01 -6.97672844e-01 -8.52026522e-01 8.90337944e-01
8.47248971e-01 -3.06550741e-01 5.09792924e-01 -1.32367406e-02
-5.72922826e-01 1.04672685e-01 -6.57503426e-01 9.14401472e-01
1.82960761e+00 -4.31406170e-01 -7.09449470e-01 4.18451726e-01
1.26487458e+00 -3.79671365e-01 -6.46994591e-01 3.82495046e-01
-3.94758247e-02 -5.89145780e-01 5.93354642e-01 -9.93761480e-01
5.01781940e-01 -2.54581302e-01 -6.16799295e-02 -1.94791174e+00
-6.66536629e-01 -9.75607157e-01 -7.58773694e-03 9.03898954e-01
6.98831975e-01 -8.50417316e-01 5.08902311e-01 -4.81195420e-01
-4.80454743e-01 -6.39757216e-01 -1.07928348e+00 -9.49482203e-01
2.36842975e-01 -1.58123210e-01 7.23558009e-01 1.15327883e+00
1.58380687e-01 1.11495984e+00 -5.47209263e-01 -2.02321097e-01
5.21257162e-01 2.08888188e-01 4.97966349e-01 -8.68720412e-01
-3.64583492e-01 -5.89602232e-01 -3.87154996e-01 -1.29033613e+00
3.45917463e-01 -1.68119562e+00 -1.60994798e-01 -2.00313902e+00
5.35866976e-01 2.27856919e-01 -3.75661433e-01 5.12367547e-01
-3.97071868e-01 1.94431409e-01 1.19889706e-01 1.51483836e-02
-3.02935392e-01 6.14903629e-01 1.54861057e+00 -4.66551811e-01
2.81645030e-01 -3.17420512e-01 -6.39649212e-01 1.84869841e-01
6.44773424e-01 -4.70205605e-01 -6.37107670e-01 -1.02905464e+00
4.00032192e-01 1.25578731e-01 -9.72383991e-02 -4.60961074e-01
6.70888796e-02 -1.18391104e-01 -1.32600531e-01 -4.09083903e-01
-4.26061964e-03 -6.00452900e-01 3.58985364e-01 7.33099282e-01
-5.22109389e-01 5.84717870e-01 7.67368898e-02 6.97865427e-01
-2.53992468e-01 1.46786841e-02 3.84186715e-01 -3.70525569e-01
-5.97541809e-01 4.92684841e-01 -3.46610665e-01 1.89403072e-01
6.85560822e-01 -1.90577909e-01 -2.88376004e-01 -3.80180895e-01
-4.67249542e-01 3.54288518e-01 2.08000079e-01 6.27623439e-01
6.63144350e-01 -1.66808498e+00 -1.33691907e+00 9.61281955e-02
2.50934809e-01 -4.08501685e-01 9.21344757e-02 8.90829086e-01
-2.53664762e-01 4.54848498e-01 -2.70492107e-01 -3.68434608e-01
-1.44932497e+00 6.57618880e-01 2.81070232e-01 -7.76150405e-01
-5.52218854e-01 8.74170542e-01 3.80397767e-01 -5.52944601e-01
-2.11451456e-01 -5.74432135e-01 -1.60087585e-01 -3.77081074e-02
1.25531748e-01 1.03624903e-01 1.20012648e-01 -8.48184526e-01
-2.77124703e-01 6.40705585e-01 -2.20404863e-01 1.30703330e-01
1.22061753e+00 -2.17810627e-02 -6.69954240e-01 3.72558266e-01
1.40137911e+00 1.15704700e-01 -5.98148286e-01 -5.54538548e-01
1.63773164e-01 -1.45705029e-01 -2.61618823e-01 -6.56754613e-01
-1.13864028e+00 9.00633812e-01 -4.01386693e-02 1.69162914e-01
1.08627701e+00 1.92262940e-02 1.02646279e+00 1.39566466e-01
6.45052731e-01 -6.14399135e-01 7.80416280e-02 7.57617712e-01
9.66231763e-01 -6.21371686e-01 -4.55401242e-01 -6.10730052e-01
-5.29351056e-01 1.14055419e+00 2.38787353e-01 -1.48424849e-01
1.15198597e-01 5.70342876e-02 -1.49429500e-01 -1.12120882e-02
-8.97280037e-01 -2.05297783e-01 5.37426829e-01 3.19387823e-01
6.05475545e-01 4.05536622e-01 -7.98454225e-01 4.79099184e-01
-1.94535762e-01 -3.32977176e-01 3.85622859e-01 5.81733465e-01
-3.42979074e-01 -1.56331325e+00 1.28816187e-01 3.51129383e-01
-2.76073486e-01 -6.26482964e-01 -7.07473218e-01 8.04363847e-01
-6.36028200e-02 8.92213523e-01 -1.38096347e-01 -7.62707353e-01
5.06078064e-01 7.67485276e-02 5.74662626e-01 -8.35340858e-01
-7.19995141e-01 5.87747432e-03 3.89682710e-01 -4.19332981e-01
-1.97565630e-01 -5.14206052e-01 -1.18132448e+00 -4.33227837e-01
-8.28453526e-02 3.66970301e-01 2.72975117e-01 8.59528720e-01
4.39613134e-01 8.58239174e-01 6.92186534e-01 -3.33551705e-01
-5.75915217e-01 -1.29793322e+00 -4.24433261e-01 4.00343150e-01
1.59291685e-01 -1.26533657e-01 -2.26851434e-01 -2.28455693e-01] | [10.288275718688965, 8.35886001586914] |
11ac6698-fa79-4157-904f-3739af31b44a | jukebox-a-multilingual-singer-recognition | 2008.03507 | null | https://arxiv.org/abs/2008.03507v1 | https://arxiv.org/pdf/2008.03507v1.pdf | JukeBox: A Multilingual Singer Recognition Dataset | A text-independent speaker recognition system relies on successfully encoding speech factors such as vocal pitch, intensity, and timbre to achieve good performance. A majority of such systems are trained and evaluated using spoken voice or everyday conversational voice data. Spoken voice, however, exhibits a limited range of possible speaker dynamics, thus constraining the utility of the derived speaker recognition models. Singing voice, on the other hand, covers a broader range of vocal and ambient factors and can, therefore, be used to evaluate the robustness of a speaker recognition system. However, a majority of existing speaker recognition datasets only focus on the spoken voice. In comparison, there is a significant shortage of labeled singing voice data suitable for speaker recognition research. To address this issue, we assemble \textit{JukeBox} - a speaker recognition dataset with multilingual singing voice audio annotated with singer identity, gender, and language labels. We use the current state-of-the-art methods to demonstrate the difficulty of performing speaker recognition on singing voice using models trained on spoken voice alone. We also evaluate the effect of gender and language on speaker recognition performance, both in spoken and singing voice data. The complete \textit{JukeBox} dataset can be accessed at http://iprobe.cse.msu.edu/datasets/jukebox.html. | ['Arun Ross', 'Anurag Chowdhury', 'Austin Cozzo'] | 2020-08-08 | null | null | null | null | ['text-independent-speaker-recognition'] | ['speech'] | [ 2.23439299e-02 -3.42430443e-01 5.62769547e-03 -6.54295564e-01
-1.17214322e+00 -8.84591520e-01 5.26728213e-01 -4.05910611e-01
-1.42509788e-01 3.12644780e-01 3.05186093e-01 -3.04250747e-01
-4.30187723e-03 -1.10030279e-01 -3.95643324e-01 -7.28434265e-01
1.06207415e-01 1.94258496e-01 -1.40376896e-01 -2.32760042e-01
5.74664511e-02 3.32775652e-01 -1.95609283e+00 1.89533696e-01
3.68522137e-01 8.17845941e-01 1.57344818e-01 1.04283333e+00
-1.11794984e-03 3.51684749e-01 -9.64889765e-01 7.62247443e-02
-6.84192404e-02 -7.40960896e-01 -5.38234651e-01 6.96023181e-02
6.23512566e-01 -2.26879641e-01 -1.67975739e-01 7.25763202e-01
9.48950350e-01 1.12841628e-01 4.64926630e-01 -1.09077072e+00
-3.54317546e-01 8.66171658e-01 3.03116292e-01 3.87846202e-01
5.43843329e-01 2.49326304e-01 1.27051556e+00 -1.04797935e+00
2.77861446e-01 1.39370799e+00 4.76463974e-01 8.13325405e-01
-1.28026330e+00 -9.45450604e-01 -1.62767500e-01 -1.86103340e-02
-1.38863730e+00 -1.32674158e+00 9.18496668e-01 -5.54519951e-01
7.64048636e-01 6.95221961e-01 3.00766110e-01 1.15989459e+00
-5.06676137e-01 6.33934200e-01 1.07888556e+00 -6.08168423e-01
9.15089324e-02 3.27486813e-01 3.34280908e-01 2.83845216e-01
-5.69854200e-01 2.94550329e-01 -9.97758806e-01 -2.35622272e-01
4.21728224e-01 -6.72879636e-01 -5.38550019e-01 1.63703874e-01
-1.22937846e+00 6.28067374e-01 -1.51662320e-01 5.40861309e-01
-4.79019322e-02 -1.16519764e-01 3.66195053e-01 5.96083581e-01
4.87788647e-01 3.21044505e-01 -3.48187238e-01 -4.88147825e-01
-1.24914765e+00 1.24914825e-01 1.09726024e+00 5.27434647e-01
2.98864841e-01 5.38345039e-01 5.97259146e-04 1.43450022e+00
4.47187394e-01 7.15134740e-01 5.76890588e-01 -7.92233169e-01
1.64142355e-01 1.00874407e-02 -1.17647037e-01 -4.45343137e-01
-1.98722646e-01 -2.08987892e-01 -1.31220192e-01 1.16059229e-01
6.13568664e-01 -1.33870486e-02 -7.72239387e-01 1.77629340e+00
1.19065538e-01 -9.40729398e-03 2.20894128e-01 9.82741117e-01
1.20577967e+00 5.75545967e-01 -3.50571543e-01 -3.69794011e-01
1.29779398e+00 -8.02148640e-01 -9.32196975e-01 -6.53074682e-02
1.16308048e-01 -1.17335474e+00 1.49891222e+00 4.54659492e-01
-9.80201364e-01 -6.07708812e-01 -8.62898171e-01 1.75369829e-01
-1.77760869e-01 9.01343003e-02 1.84867010e-01 1.23711991e+00
-9.81678903e-01 2.56333828e-01 -6.21854126e-01 -2.64739275e-01
-1.33161962e-01 3.53144914e-01 -2.65200526e-01 2.09520712e-01
-1.22406948e+00 5.69349825e-01 -6.82921261e-02 -9.03756940e-04
-1.07995808e+00 -5.73796570e-01 -7.26756155e-01 -2.54045606e-01
9.64651033e-02 -1.34330280e-02 1.79116845e+00 -7.55636990e-01
-1.92821252e+00 7.72169709e-01 -4.99216974e-01 -9.94540527e-02
3.24680567e-01 4.54935916e-02 -9.32397068e-01 -3.99797037e-02
-2.67316967e-01 1.64004073e-01 1.11282694e+00 -1.12236881e+00
-2.83099025e-01 -4.69387800e-01 -6.53597593e-01 1.63260236e-01
-4.42635447e-01 6.07993722e-01 -2.56116331e-01 -4.58491057e-01
9.85967219e-02 -9.01378632e-01 5.33284128e-01 -5.70056200e-01
-3.62614095e-01 -2.44410515e-01 9.15146112e-01 -8.40642810e-01
1.30908883e+00 -2.52135491e+00 -2.10038945e-02 1.76147819e-02
-3.03702056e-01 2.28976071e-01 -8.62903446e-02 3.80012065e-01
1.64850075e-02 1.84941962e-01 -1.24696992e-01 -5.82333922e-01
1.59529179e-01 2.20580339e-01 -3.64665896e-01 4.17967409e-01
-3.79155874e-02 5.17557681e-01 -5.02641797e-01 -2.38486260e-01
2.40198418e-01 6.37999177e-01 -3.93596560e-01 3.26615632e-01
1.66948721e-01 6.94784164e-01 1.59379557e-01 7.72752702e-01
2.13314831e-01 4.98139918e-01 -1.62532389e-01 1.33943111e-01
-3.75878483e-01 8.79609585e-01 -1.28046775e+00 1.20137608e+00
-5.88551819e-01 7.33160317e-01 6.82208180e-01 -4.83264834e-01
1.18934751e+00 8.22970629e-01 1.55556291e-01 -2.10452959e-01
1.12028152e-01 6.78742111e-01 4.68483031e-01 -4.24229145e-01
5.21661520e-01 -4.44306761e-01 6.30708188e-02 3.01831812e-01
2.67239869e-01 -5.05760431e-01 1.59212556e-02 -3.41336042e-01
7.05426633e-01 -4.40622181e-01 -2.83076204e-02 -2.37656027e-01
5.69513798e-01 -4.09332275e-01 4.56650734e-01 5.06523192e-01
-4.34337854e-01 8.11823845e-01 6.41172379e-02 2.91426778e-01
-6.82402492e-01 -1.08850896e+00 -6.26915455e-01 1.62265694e+00
-5.33708513e-01 -3.13648045e-01 -8.27152491e-01 -9.46075395e-02
-1.01645477e-03 9.22722697e-01 -1.04636721e-01 1.35326162e-01
-4.62121755e-01 -4.41078186e-01 1.16796386e+00 2.72582412e-01
-3.39983217e-02 -1.28786755e+00 -2.72940565e-02 2.85774767e-01
-1.77955940e-01 -9.66134846e-01 -8.45315754e-01 1.87222555e-01
-4.61215883e-01 -7.11645126e-01 -6.60570681e-01 -7.20478475e-01
-6.61503375e-02 -4.64900862e-03 9.67888713e-01 -2.17816249e-01
-4.07668427e-02 6.11429095e-01 -2.27417380e-01 -5.79569519e-01
-1.03967631e+00 3.46800201e-02 5.99330842e-01 1.68020770e-01
2.50341177e-01 -5.30678928e-01 -1.99520662e-01 7.19203889e-01
-6.65959418e-01 -5.85243046e-01 1.07151093e-02 8.69793594e-01
2.92337805e-01 4.54592071e-02 1.15850115e+00 -4.52449858e-01
7.66299546e-01 -1.67057410e-01 -3.42991740e-01 -6.89805225e-02
-3.12994033e-01 -4.78728592e-01 4.95873541e-01 -6.90274239e-01
-9.43429410e-01 -2.53631193e-02 -5.17157018e-01 -3.74951303e-01
-5.54681599e-01 5.62402070e-01 -4.79454488e-01 2.37258390e-01
7.34090626e-01 4.27576840e-01 3.26252908e-01 -7.20580101e-01
1.74355030e-01 1.61760437e+00 9.71802652e-01 -3.62968504e-01
4.56440449e-01 -2.06378829e-02 -6.80551231e-01 -1.63058293e+00
-3.58613759e-01 -8.44940603e-01 -4.90527809e-01 -3.15156788e-01
3.99483860e-01 -8.56595695e-01 -6.40756905e-01 8.63328695e-01
-7.24613905e-01 -3.55652690e-01 -8.22550729e-02 7.45479882e-01
-5.23713231e-01 2.20596299e-01 -5.87412953e-01 -1.41519761e+00
-3.85354429e-01 -1.19204223e+00 9.37283754e-01 -2.32508138e-01
-6.59128010e-01 -8.22491348e-01 -4.31309007e-02 7.96442032e-01
5.12961745e-01 -2.87815630e-01 5.92616320e-01 -9.81199205e-01
5.86912297e-02 -1.81598932e-01 5.63533664e-01 6.88227654e-01
4.33376223e-01 2.73401320e-01 -1.60977912e+00 -4.18198466e-01
2.14228287e-01 -3.84466767e-01 6.23812437e-01 3.19090873e-01
5.34241796e-01 -2.16625467e-01 3.50449502e-01 2.72540331e-01
4.82628703e-01 1.28248230e-01 2.12927356e-01 -1.55055121e-01
5.36592901e-01 8.59530747e-01 3.42678696e-01 2.26738915e-01
1.76119685e-01 7.96165287e-01 2.53343396e-02 2.87368983e-01
-3.60572040e-01 -2.94017822e-01 8.64925146e-01 1.29228437e+00
2.40715519e-01 -2.76485533e-01 -1.10202742e+00 6.60518706e-01
-1.11500645e+00 -9.38020110e-01 -4.18200940e-02 2.36401367e+00
9.97968853e-01 -1.27025276e-01 4.65423197e-01 8.11967492e-01
8.59199584e-01 2.96682417e-01 -4.66564804e-01 -4.20744538e-01
-1.32572830e-01 1.34269847e-02 4.43007238e-02 6.98982298e-01
-9.34778988e-01 8.40851903e-01 6.38707829e+00 7.44272232e-01
-1.56235552e+00 -4.22179475e-02 1.86780289e-01 -3.16131860e-01
-2.38348380e-01 -3.13343942e-01 -9.70748127e-01 3.38583261e-01
1.38278854e+00 -5.25314510e-02 8.34938228e-01 6.06566608e-01
5.46064198e-01 2.62968510e-01 -1.32397830e+00 1.08848023e+00
1.42624334e-01 -6.69125974e-01 -3.04117322e-01 3.54383849e-02
1.63767919e-01 1.62046477e-01 2.44168893e-01 4.33616728e-01
-2.44660497e-01 -1.19328058e+00 1.06299293e+00 1.57525241e-01
1.04835010e+00 -6.05230331e-01 3.65458637e-01 5.61054826e-01
-1.01927662e+00 6.13205368e-03 2.60586321e-01 -6.27078563e-02
1.99059457e-01 2.39693955e-01 -1.19225264e+00 1.32790238e-01
6.98272586e-01 3.06099862e-01 -2.82573164e-01 6.61336362e-01
-6.26548827e-02 1.51545072e+00 -5.00040293e-01 1.88276656e-02
-2.85536796e-01 1.97288290e-01 9.03649747e-01 1.30314553e+00
4.40976530e-01 -9.57252085e-02 4.62420918e-02 7.55031586e-01
8.29247907e-02 4.02383476e-01 -4.14152950e-01 -6.19162858e-01
8.66545439e-01 1.00850689e+00 -3.04389626e-01 -1.26998667e-02
-3.05983782e-01 4.75425124e-01 -1.67112440e-01 3.56188595e-01
-2.36706570e-01 -1.71520725e-01 8.27638924e-01 1.58872992e-01
1.33573219e-01 -4.21588182e-01 -1.24124922e-01 -9.21571553e-01
-4.25980091e-02 -1.30445242e+00 1.71513170e-01 -3.37805986e-01
-1.30757797e+00 6.39288366e-01 -2.42295980e-01 -9.56666768e-01
-7.19101846e-01 -6.05407238e-01 -7.20904529e-01 1.17060804e+00
-1.36620128e+00 -1.04143465e+00 -4.46481369e-02 5.16018987e-01
8.03221464e-01 -5.42879641e-01 1.12487733e+00 3.17764521e-01
-6.36797547e-01 8.87879848e-01 1.08193159e-01 9.88831297e-02
9.95527089e-01 -1.18848312e+00 4.85092878e-01 3.15439224e-01
3.99467111e-01 6.83089316e-01 7.52986908e-01 -2.07634389e-01
-1.45591855e+00 -6.91905379e-01 9.67628777e-01 -5.09059966e-01
6.31061256e-01 -6.37160182e-01 -1.07149601e+00 3.85945141e-01
7.40140080e-02 -7.37873688e-02 1.17554367e+00 3.08574885e-01
-4.10917222e-01 -9.44834389e-03 -1.01373470e+00 4.06064183e-01
6.95984602e-01 -1.07387197e+00 -7.04383373e-01 -1.21587642e-01
4.67595875e-01 -2.59313017e-01 -1.07105267e+00 2.60752082e-01
7.40236878e-01 -1.01762867e+00 8.09278011e-01 -1.94551393e-01
-1.78469330e-01 -3.05880368e-01 -4.72308934e-01 -1.42671454e+00
1.15116946e-01 -7.80338764e-01 1.97805360e-01 1.86343539e+00
5.78362584e-01 -7.12990344e-01 3.25784296e-01 6.12852156e-01
-8.22573528e-02 -1.14441849e-01 -1.14879596e+00 -1.00745130e+00
2.28253230e-01 -7.36762941e-01 6.27900839e-01 9.57188427e-01
9.50644985e-02 4.99037534e-01 -3.08974087e-01 2.15240806e-01
3.52275074e-01 2.12139226e-02 8.61715019e-01 -1.14226151e+00
-3.92546952e-01 -3.78733456e-01 -2.02140540e-01 -7.35562682e-01
3.64771426e-01 -9.63479102e-01 3.66863489e-01 -9.35824513e-01
-2.89586663e-01 -3.31031621e-01 -2.40369886e-01 3.92105311e-01
-7.35137686e-02 2.13768780e-01 2.93221295e-01 3.99859905e-01
1.07440323e-01 4.73653495e-01 9.80072260e-01 5.29470518e-02
-5.28761208e-01 3.85556281e-01 -2.82508671e-01 6.98236465e-01
1.07047713e+00 -1.59285232e-01 -2.09021509e-01 -4.82775597e-03
-6.27986968e-01 3.52713555e-01 3.67455035e-01 -9.06536400e-01
1.55607805e-01 1.19014740e-01 -9.87870991e-02 -3.80515218e-01
6.89587533e-01 -3.99731636e-01 -3.68760824e-02 1.20969303e-01
-5.41130960e-01 -4.45888370e-01 2.90925771e-01 8.27214718e-02
-5.65795541e-01 -2.81777978e-01 6.68935955e-01 1.71898291e-01
-3.04608583e-01 -1.30977720e-01 -7.40008891e-01 -5.63638052e-03
2.08295673e-01 -2.05604464e-01 -1.14924088e-01 -6.30021393e-01
-6.97212577e-01 -1.51762992e-01 3.42879385e-01 7.50571728e-01
3.71981382e-01 -1.09020269e+00 -9.79683518e-01 6.57047212e-01
2.32342184e-01 -3.89541358e-01 2.04379514e-01 6.65169477e-01
5.67550212e-02 4.11751419e-01 1.49814606e-01 -7.20516205e-01
-1.82328725e+00 1.81113303e-01 4.54301804e-01 5.81380367e-01
-3.87740918e-02 8.00732970e-01 -1.06201127e-01 -1.02532995e+00
4.39867198e-01 -9.34620202e-02 -1.89066693e-01 3.47208619e-01
5.02257943e-01 4.33713555e-01 1.83709323e-01 -1.18851805e+00
-3.82907540e-01 1.71833232e-01 2.18338117e-01 -5.89366615e-01
1.01828492e+00 -2.94092029e-01 1.45870717e-02 1.08428752e+00
1.22370529e+00 5.71067214e-01 -7.58565903e-01 -1.92620605e-01
5.99483736e-02 -3.41403693e-01 3.44282806e-01 -7.74252832e-01
-5.59915066e-01 8.91527236e-01 5.81975043e-01 3.91864598e-01
8.44201207e-01 5.20809926e-02 5.49044907e-01 2.84537971e-01
9.90243927e-02 -9.37966526e-01 -2.01691777e-01 6.83401823e-01
1.13994133e+00 -1.40588975e+00 -5.92898607e-01 -3.93424839e-01
-6.66073263e-01 1.00441718e+00 2.44785577e-01 5.73936522e-01
6.15048170e-01 3.62637281e-01 8.42069268e-01 1.30690560e-01
-5.25445819e-01 -1.80624604e-01 5.16669333e-01 4.07041192e-01
9.87619758e-01 2.91986048e-01 -1.55698890e-02 8.39292765e-01
-1.11248851e+00 -3.38596523e-01 2.65574843e-01 5.99788606e-01
-4.49385345e-01 -1.17737365e+00 -9.01059687e-01 1.82760656e-01
-7.23146975e-01 -1.59961671e-01 -7.65723228e-01 2.45274603e-01
-2.33575940e-01 1.69595623e+00 -1.19302161e-01 -4.85104620e-01
4.04365361e-01 7.06322074e-01 1.28260612e-01 -6.90296888e-01
-8.22329640e-01 4.82747912e-01 3.51321220e-01 3.54345106e-02
-2.42538124e-01 -1.03125775e+00 -1.10431051e+00 -1.25419989e-01
-5.51144361e-01 2.79543936e-01 1.04157889e+00 5.79658329e-01
5.17307557e-02 5.35806775e-01 7.64843762e-01 -9.76533413e-01
-9.20773923e-01 -1.34671223e+00 -9.57832634e-01 2.76054561e-01
6.86982632e-01 -3.24673057e-01 -9.00642455e-01 3.05130128e-02] | [14.67196273803711, 6.248265743255615] |
b29c256f-a440-446a-b14a-a872cc0d980f | word-segmentation-by-separation-inference-for | null | null | https://aclanthology.org/2022.findings-acl.309 | https://aclanthology.org/2022.findings-acl.309.pdf | Word Segmentation by Separation Inference for East Asian Languages | Chinese Word Segmentation (CWS) intends to divide a raw sentence into words through sequence labeling. Thinking in reverse, CWS can also be viewed as a process of grouping a sequence of characters into a sequence of words. In such a way, CWS is reformed as a separation inference task in every adjacent character pair. Since every character is either connected or not connected to the others, the tagging schema is simplified as two tags “Connection” (C) or “NoConnection” (NC). Therefore, bigram is specially tailored for “C-NC” to model the separation state of every two consecutive characters. Our Separation Inference (SpIn) framework is evaluated on five public datasets, is demonstrated to work for machine learning and deep learning models, and outperforms state-of-the-art performance for CWS in all experiments. Performance boosts on Japanese Word Segmentation (JWS) and Korean Word Segmentation (KWS) further prove the framework is universal and effective for East Asian Languages. | ['Guokai Zheng', 'Ge Chen', 'Jizhe Zhou', 'Jingzhi Guo', 'Yu tong'] | null | null | null | null | findings-acl-2022-5 | ['chinese-word-segmentation'] | ['natural-language-processing'] | [ 2.89447159e-01 -7.02334419e-02 -4.67075437e-01 -4.48251247e-01
-4.99584436e-01 -7.58048713e-01 4.07152802e-01 2.00250968e-01
-7.10079253e-01 5.18650055e-01 2.22551927e-01 -1.02183664e+00
6.66025400e-01 -6.69235528e-01 -3.63678277e-01 -7.94023395e-01
3.12717766e-01 5.08527696e-01 5.95656455e-01 -9.27176774e-02
1.10070623e-01 2.82765150e-01 -5.48283696e-01 1.52555257e-01
1.04690778e+00 5.99894047e-01 3.94390196e-01 5.20948827e-01
-5.95564485e-01 2.62692779e-01 -6.34798706e-01 -5.71945429e-01
-1.91723295e-02 -5.78867853e-01 -1.36603880e+00 -1.19818687e-01
7.47652026e-03 -2.11368099e-01 2.07070857e-01 1.24814093e+00
7.12351203e-02 -6.99916184e-02 5.70326269e-01 -6.68116987e-01
-7.32089221e-01 1.16864955e+00 -7.60555446e-01 3.33735649e-03
-7.13654701e-03 -2.08640829e-01 1.41958261e+00 -7.03633726e-01
3.40879917e-01 1.08331716e+00 4.56259012e-01 7.68841922e-01
-1.10275018e+00 -4.82349843e-01 6.81876242e-01 -5.21728583e-02
-8.93035531e-01 1.07594997e-01 2.73226172e-01 -3.09811771e-01
1.02182686e+00 4.30553198e-01 6.81385636e-01 8.59074593e-01
1.53817415e-01 1.33755338e+00 8.60068798e-01 -7.56846488e-01
3.37266177e-01 -1.92296535e-01 1.18021286e+00 4.22072202e-01
3.14817905e-01 -6.46352947e-01 -1.09295487e-01 2.94210166e-01
1.52796000e-01 -6.73401207e-02 -9.65047479e-02 3.42719167e-01
-1.10016978e+00 1.02039480e+00 2.62532420e-02 4.59581196e-01
-4.41204570e-02 -1.30739868e-01 2.44119063e-01 -1.00966245e-01
4.24276590e-01 3.24189305e-01 -7.79920459e-01 -1.13797709e-01
-1.00886202e+00 -1.24505863e-01 9.15119708e-01 1.15954292e+00
6.41560674e-01 4.18101884e-02 -2.95816451e-01 7.43458629e-01
2.47446209e-01 2.67078161e-01 7.54108489e-01 -6.39762282e-01
3.84275526e-01 4.37449068e-01 -1.71125218e-01 -3.28221709e-01
-4.15945321e-01 -3.56323928e-01 -7.53786445e-01 -3.46342951e-01
5.34549057e-01 -6.49333596e-01 -1.30456305e+00 1.70707524e+00
8.86105001e-02 2.01869667e-01 3.51243950e-02 7.38127291e-01
8.04965973e-01 1.05481958e+00 3.18312883e-01 -2.61290669e-01
1.97720385e+00 -1.45785701e+00 -7.33849764e-01 -7.17482746e-01
6.12809360e-01 -7.26758301e-01 1.07848716e+00 3.65395129e-01
-1.06270385e+00 -4.33895439e-01 -9.28621411e-01 -2.02340633e-01
-6.51445150e-01 1.21000022e-01 6.37899935e-01 8.23345721e-01
-7.25583017e-01 3.30742061e-01 -1.09232402e+00 -1.59055024e-01
8.16599056e-02 2.20214546e-01 -4.27371860e-02 -1.46472631e-02
-1.45077050e+00 6.70935392e-01 6.24552727e-01 4.13025945e-01
-4.79857117e-01 -4.08481300e-01 -7.51468301e-01 1.09459899e-01
4.01273429e-01 -3.91678303e-01 1.27894688e+00 -1.04506457e+00
-1.53446662e+00 1.02176535e+00 -4.63970423e-01 -5.81057549e-01
1.71403021e-01 -5.10480344e-01 -6.82928324e-01 1.41018376e-01
1.72828555e-01 8.23341906e-01 7.44792581e-01 -1.06734192e+00
-7.92181432e-01 -2.22453073e-01 -8.32121149e-02 9.84239802e-02
-2.74311602e-01 5.33709288e-01 -9.98411357e-01 -6.99612916e-01
3.58884692e-01 -1.01522946e+00 -3.25583488e-01 -7.54438519e-01
-9.62907374e-01 -3.57181162e-01 5.26615977e-01 -1.04299355e+00
1.58136165e+00 -1.98182368e+00 9.34766307e-02 6.12850748e-02
2.48597972e-02 7.34316051e-01 -1.24269731e-01 3.90381724e-01
-1.59851089e-01 4.80726510e-01 -5.65223336e-01 -5.40024459e-01
-2.97369510e-02 5.59682310e-01 -2.78570563e-01 1.39544234e-01
3.88354957e-01 1.07300508e+00 -1.01324153e+00 -5.18643737e-01
-3.21409851e-01 2.21774867e-03 -3.65320802e-01 6.89227134e-02
-2.54080951e-01 1.60841584e-01 -3.71578604e-01 4.74683195e-01
9.18367743e-01 -5.31906523e-02 6.47738159e-01 7.15929866e-02
-3.01641107e-01 6.62111461e-01 -9.87522125e-01 1.43861568e+00
-1.97772071e-01 2.54916877e-01 -6.47987798e-02 -9.07226801e-01
7.92002797e-01 4.31977622e-02 -1.56293228e-01 -5.26397407e-01
1.47909429e-02 1.12797491e-01 2.85262316e-01 -2.45765448e-01
9.01842594e-01 2.23592259e-02 -5.13995469e-01 3.22149694e-01
-2.61573363e-02 2.07425803e-02 3.82537425e-01 3.23039144e-01
6.82049751e-01 6.40604645e-02 1.65187925e-01 -5.49195290e-01
6.80674374e-01 -8.52256566e-02 8.83435845e-01 6.36562109e-01
-1.51806012e-01 5.91714740e-01 8.78036141e-01 -5.93119897e-02
-6.75898731e-01 -1.23959982e+00 6.14986308e-02 1.18675923e+00
2.14227617e-01 -5.56717098e-01 -1.39668691e+00 -9.13555980e-01
-3.96397024e-01 1.06963825e+00 -5.55536509e-01 3.71760712e-03
-9.39325392e-01 -8.28382730e-01 7.59626627e-01 8.85502458e-01
4.35689598e-01 -1.11752748e+00 -3.33717972e-01 1.63297385e-01
-3.52045238e-01 -1.22482049e+00 -9.75331604e-01 6.02686942e-01
-6.40498161e-01 -8.15089822e-01 -8.02636325e-01 -1.47423005e+00
7.30869949e-01 3.22987914e-01 1.08247221e+00 1.35789275e-01
1.28982320e-01 -3.08674306e-01 -7.80423999e-01 -9.52094793e-02
-3.34564120e-01 5.83493233e-01 -2.30676994e-01 -5.92647456e-02
6.27257109e-01 -4.93611805e-02 -5.05134583e-01 1.31041393e-01
-1.03654516e+00 1.92063376e-01 5.32288551e-01 7.22761035e-01
3.73585165e-01 -1.57566488e-01 2.71814197e-01 -1.19851995e+00
5.96510589e-01 -3.37480068e-01 -4.64540392e-01 4.56624418e-01
-5.24688840e-01 8.57464075e-02 7.48854995e-01 -5.17023981e-01
-1.10398436e+00 -8.46487880e-02 -3.83885622e-01 4.10578966e-01
-3.76043588e-01 6.89411819e-01 -4.81017292e-01 8.05088282e-01
-1.91122696e-01 3.23568434e-01 -3.08675736e-01 -8.23379278e-01
5.23379564e-01 8.75100255e-01 6.67254806e-01 -7.00954497e-01
4.13630754e-01 2.06718892e-01 -4.16216522e-01 -9.14328456e-01
-8.56464028e-01 -5.48037112e-01 -1.08733535e+00 2.51514316e-01
1.47946918e+00 -6.97140515e-01 -4.96712893e-01 9.89943504e-01
-1.40970838e+00 -4.50396717e-01 4.76285890e-02 3.06539208e-01
3.36856395e-01 8.68198454e-01 -1.08203125e+00 -6.24383152e-01
-3.39882195e-01 -1.12527955e+00 7.98784614e-01 6.72153234e-01
-4.22122180e-01 -1.08148992e+00 -1.12224862e-01 3.77153218e-01
-8.66751671e-02 -1.41132608e-01 1.15766263e+00 -1.08059180e+00
-8.80896021e-03 -2.89902568e-01 -2.56606787e-01 8.04039598e-01
-6.33312901e-03 2.73125976e-01 -4.98881906e-01 -8.00081342e-02
-1.39992520e-01 -1.25383064e-02 1.14451993e+00 1.64111346e-01
8.61159205e-01 -2.95455158e-01 -2.24362314e-01 3.33932698e-01
1.18811977e+00 5.04819512e-01 8.12340975e-01 2.19337076e-01
8.01588774e-01 4.68550652e-01 4.74722654e-01 -7.64676481e-02
4.66326207e-01 2.56898791e-01 6.04921356e-02 -2.76880741e-01
4.29937989e-02 -1.98399156e-01 5.60290039e-01 1.29679859e+00
2.42447615e-01 -3.57664227e-01 -1.08504069e+00 5.63682914e-01
-1.49508119e+00 -4.56638187e-01 -6.75514817e-01 1.96175659e+00
1.18912232e+00 3.68306786e-01 -2.60124672e-02 -7.05937743e-02
9.79278207e-01 3.48461926e-01 -2.31466144e-01 -8.22110355e-01
-2.47274026e-01 6.28086150e-01 5.30113816e-01 6.82275534e-01
-1.02593970e+00 1.65820789e+00 5.92134953e+00 1.08963132e+00
-9.48729217e-01 4.45453711e-02 9.86244261e-01 6.44820511e-01
-4.26638544e-01 3.73537213e-01 -1.33433533e+00 8.28140438e-01
8.24138463e-01 4.92912203e-01 9.31679979e-02 4.59678799e-01
-5.24233542e-02 -4.54781860e-01 -8.65725935e-01 2.41450995e-01
-1.73292965e-01 -1.07286525e+00 1.75013557e-01 -2.28564724e-01
7.24357784e-01 -1.46420673e-01 -1.64010495e-01 3.76061678e-01
5.37722468e-01 -8.53380084e-01 9.68699038e-01 -5.34016527e-02
5.92260003e-01 -6.36869252e-01 8.69588077e-01 4.40788180e-01
-1.21719849e+00 4.53994513e-01 -1.46091551e-01 -7.93986395e-02
4.07750458e-01 3.88639927e-01 -3.46899897e-01 7.18241453e-01
4.49994773e-01 3.79722267e-01 -6.02119505e-01 5.45018196e-01
-8.48232567e-01 1.31316841e+00 -2.53279833e-03 -2.62081861e-01
9.14685428e-01 -7.96320140e-01 1.69129759e-01 1.87157786e+00
-1.64031625e-01 1.42664433e-01 1.65773973e-01 5.54608583e-01
8.70773271e-02 -8.43895376e-02 3.86266679e-01 -1.73780665e-01
5.17686069e-01 1.35736561e+00 -1.44594026e+00 -5.27645051e-01
-4.73240942e-01 1.25742924e+00 1.28600299e-01 4.64208424e-01
-9.42600727e-01 -6.45788610e-01 6.94726706e-01 -2.90918231e-01
6.51790679e-01 -4.78443742e-01 -4.27849799e-01 -9.94074643e-01
-2.03959376e-01 -7.36802936e-01 2.99391240e-01 -4.40589905e-01
-9.88541424e-01 6.89710259e-01 -1.91192567e-01 -8.75062287e-01
1.83125407e-01 -8.22010875e-01 -8.95686150e-01 1.02542043e+00
-1.65811992e+00 -1.06139779e+00 1.40335619e-01 2.42236078e-01
6.14463627e-01 2.34250605e-01 8.44520569e-01 4.16930430e-02
-1.13604379e+00 5.76645255e-01 1.55812100e-01 8.18722010e-01
4.26199526e-01 -1.44508326e+00 9.65537667e-01 1.42828953e+00
2.23837644e-01 8.82564247e-01 4.14129436e-01 -7.42527843e-01
-9.33546305e-01 -8.72155011e-01 1.30545008e+00 8.55964050e-02
7.93111742e-01 -6.93361223e-01 -9.69655335e-01 6.97786212e-01
4.77691382e-01 -3.47853422e-01 8.29739153e-01 8.48645791e-02
-5.28218985e-01 1.66350245e-01 -5.33212125e-01 9.32653248e-01
7.54908264e-01 -2.73833483e-01 -9.25733745e-01 -3.80874388e-02
1.27158928e+00 -2.87413508e-01 -3.67304385e-01 5.31049073e-02
4.21123058e-01 -8.81836414e-01 7.08444357e-01 -9.65932369e-01
3.81873459e-01 -2.75441140e-01 -1.17231332e-01 -1.06574094e+00
-3.50566417e-01 -8.63699973e-01 1.38850600e-01 1.51447427e+00
7.76586175e-01 -7.21814930e-01 7.09837019e-01 3.77705038e-01
-5.48247933e-01 -6.12408757e-01 -6.92946136e-01 -9.01098430e-01
5.99829495e-01 -5.35659134e-01 5.47653496e-01 9.76997972e-01
-3.77775840e-02 6.55973971e-01 -1.06154293e-01 1.42838702e-01
1.97334558e-01 1.14020936e-01 1.59092903e-01 -9.69783962e-01
-3.82921338e-01 -7.25347042e-01 2.77793169e-01 -1.70418489e+00
4.53728050e-01 -7.70236731e-01 3.26508284e-01 -1.54816449e+00
7.57358745e-02 -3.30463082e-01 -1.61170438e-01 5.81350029e-01
-6.58307076e-01 1.94098666e-01 3.13004255e-01 -7.67766908e-02
-5.92102826e-01 1.39831960e-01 1.05368233e+00 -1.18780918e-01
-1.02786198e-01 1.21507682e-01 -7.02186346e-01 8.90829802e-01
7.98543453e-01 -3.18344951e-01 9.97663885e-02 -6.05132639e-01
1.92752764e-01 -1.41702801e-01 -4.02840108e-01 -4.58502740e-01
1.90366685e-01 -1.87223673e-01 -4.94561568e-02 -7.63473690e-01
-1.86564140e-02 -5.49770832e-01 -3.67555439e-01 7.12595999e-01
-1.83922753e-01 2.65088789e-02 5.49760275e-03 7.96265975e-02
-1.27031684e-01 -6.81593120e-01 7.92208731e-01 -4.41457853e-02
-8.65541220e-01 1.38327898e-02 -6.49431884e-01 4.16824102e-01
8.98978889e-01 -3.52680773e-01 -2.21460342e-01 3.43369357e-02
-5.83567977e-01 4.84467685e-01 4.50894050e-02 4.33925807e-01
3.88409555e-01 -8.81864488e-01 -5.38276494e-01 2.00177759e-01
-2.48637110e-01 4.12076920e-01 5.69243403e-03 7.66413689e-01
-8.75711322e-01 4.25383985e-01 2.61028171e-01 -3.12175125e-01
-1.31343043e+00 4.01015908e-01 -1.33005083e-01 -4.51240540e-01
-3.67726237e-01 1.15244246e+00 7.58295646e-03 -2.87865311e-01
3.12341869e-01 -7.66429007e-01 -3.04770738e-01 4.77746040e-01
4.79231477e-01 3.36488456e-01 -1.65029809e-01 -5.81423700e-01
-4.64366317e-01 5.67459285e-01 -4.35381055e-01 -1.64950505e-01
1.10261679e+00 -1.47458419e-01 -6.76548660e-01 5.72548985e-01
1.17882311e+00 2.79537112e-01 -1.03735018e+00 -1.14342503e-01
5.71622670e-01 1.73901424e-01 -2.51873195e-01 -7.75274038e-01
-9.35664475e-01 8.99865389e-01 -1.11953847e-01 3.50584477e-01
8.82186830e-01 1.13334060e-01 1.34242785e+00 1.87431853e-02
-4.81391475e-02 -1.35550272e+00 -4.98806350e-02 1.10972118e+00
4.03036885e-02 -8.77769530e-01 -4.01770175e-01 -4.76426929e-01
-1.02375448e+00 1.21475542e+00 3.26549441e-01 -1.44578114e-01
7.14468420e-01 4.25631106e-01 1.15334906e-01 2.07681164e-01
-1.67789981e-01 -3.41255426e-01 2.06137270e-01 2.43930623e-01
5.04861712e-01 4.11493599e-01 -7.46927381e-01 1.04780674e+00
-1.34789482e-01 -6.22536302e-01 4.43422198e-01 7.64492214e-01
-5.15106380e-01 -1.48598826e+00 -5.22441678e-02 1.10683039e-01
-6.56635225e-01 -6.74987793e-01 -3.60896438e-01 6.97486937e-01
2.59725749e-01 1.14798391e+00 4.98856008e-01 -2.98232794e-01
-2.33264677e-02 3.10081333e-01 4.76279072e-02 -9.56213653e-01
-7.75263011e-01 3.63964975e-01 6.16140924e-02 -2.41004992e-02
-2.14849323e-01 -7.14244604e-01 -1.79611433e+00 -1.98720708e-01
-4.97194976e-01 5.21281719e-01 4.66749936e-01 1.18527353e+00
-2.59356916e-01 6.32786632e-01 5.81988633e-01 -3.11983645e-01
-2.49795750e-01 -8.31508577e-01 -7.38705635e-01 2.38464430e-01
2.13116854e-02 8.94021168e-02 -1.33141384e-01 1.48153022e-01] | [10.073678970336914, 10.089434623718262] |
8a4759ba-844e-42f0-986b-43ed850839ac | reflectance-capture-using-univariate-sampling | null | null | http://openaccess.thecvf.com/content_iccv_2017/html/Hui_Reflectance_Capture_Using_ICCV_2017_paper.html | http://openaccess.thecvf.com/content_ICCV_2017/papers/Hui_Reflectance_Capture_Using_ICCV_2017_paper.pdf | Reflectance Capture Using Univariate Sampling of BRDFs | We propose the use of a light-weight setup consisting of a collocated camera and light source --- commonly found on mobile devices --- to reconstruct surface normals and spatially-varying BRDFs of near-planar material samples. A collocated setup provides only a 1-D "univariate" sampling of the 4-D BRDF. We show that a univariate sampling is sufficient to estimate parameters of commonly used analytical BRDF models. Subsequently, we use a dictionary-based reflectance prior to derive a robust technique for per-pixel normal and BRDF estimation. We demonstrate real-world shape and capture, and its application to material editing and classification, using real data acquired using a mobile phone.
| ['Sunil Hadap', 'Joon-Young Lee', 'Jian Wang', 'Zhuo Hui', 'Kalyan Sunkavalli', 'Aswin C. Sankaranarayanan'] | 2017-10-01 | null | null | null | iccv-2017-10 | ['brdf-estimation'] | ['computer-vision'] | [ 7.63663232e-01 -3.23519796e-01 5.45203984e-01 -2.80511945e-01
-6.12757742e-01 -3.63768578e-01 2.69705415e-01 -2.25741267e-01
8.44370201e-02 5.53398669e-01 -3.11245918e-01 -2.52271950e-01
8.88049826e-02 -8.09799850e-01 -7.13350892e-01 -7.61319280e-01
3.43968332e-01 3.99068952e-01 3.00139636e-01 -2.84531284e-02
2.36426100e-01 1.03122878e+00 -1.78508186e+00 -4.59527969e-02
6.16785347e-01 1.02048278e+00 3.53938460e-01 8.75990868e-01
-8.70551020e-02 -4.69918586e-02 -3.99698019e-01 -2.73914367e-01
4.44558561e-01 -1.41606763e-01 -2.04278246e-01 4.28232551e-01
7.35620618e-01 -5.15523195e-01 4.71594632e-02 8.11476111e-01
3.12361270e-01 2.13754565e-01 1.06456411e+00 -5.10107815e-01
-6.39768958e-01 -8.18718314e-01 -1.02716374e+00 -3.76668960e-01
9.17847514e-01 -8.22430402e-02 6.34605825e-01 -9.31844532e-01
5.87184846e-01 1.14397311e+00 8.84619832e-01 3.98829043e-01
-1.51553226e+00 -1.22400060e-01 -3.10756475e-01 -3.48037213e-01
-1.36189938e+00 -7.58930445e-01 1.18754530e+00 -4.67417926e-01
7.55524337e-01 2.99830377e-01 8.64948332e-01 7.16185272e-01
2.44281426e-01 1.72868460e-01 1.40230179e+00 -8.43986809e-01
7.17358142e-02 5.89673175e-03 5.67088984e-02 7.26454616e-01
5.06660104e-01 5.14100753e-02 -2.17635974e-01 -5.07888377e-01
1.26655638e+00 -2.37343848e-01 -5.34089804e-01 -6.28825784e-01
-9.95115519e-01 1.72440067e-01 -1.12701282e-01 -6.56437054e-02
-4.65636045e-01 5.91648258e-02 -3.00593585e-01 3.02298456e-01
9.08853412e-01 2.08651111e-01 -1.82947323e-01 -7.31737614e-02
-6.40097857e-01 -6.35940582e-03 8.19463253e-01 1.17467499e+00
1.06269729e+00 2.61445362e-02 3.94569665e-01 1.11016905e+00
6.23732924e-01 1.38213861e+00 -4.68930125e-01 -1.37289298e+00
-5.89505099e-02 3.98378968e-02 7.84572661e-01 -9.03488815e-01
-1.32880688e-01 1.09284997e-01 -3.12528998e-01 4.64651346e-01
5.81240118e-01 9.97693688e-02 -8.04912686e-01 1.05904603e+00
7.24294782e-01 2.33271301e-01 -2.12450609e-01 6.98357761e-01
5.62799454e-01 3.79185140e-01 -7.88091302e-01 -5.30307472e-01
1.12492669e+00 -2.12872609e-01 -5.72606981e-01 3.72951984e-01
2.28003226e-02 -1.12951219e+00 1.45774615e+00 9.51132774e-01
-1.44615126e+00 -2.13149339e-01 -8.60863745e-01 -1.95903122e-01
2.23164126e-01 2.35974774e-01 9.74323973e-02 9.22534347e-01
-9.46406543e-01 2.74903923e-01 -6.48539424e-01 -3.31878006e-01
1.49255037e-01 -6.73693512e-03 -2.34704107e-01 -4.07094985e-01
-1.20017767e-01 6.93040729e-01 -6.88248038e-01 7.03701153e-02
-5.10818243e-01 -8.18661809e-01 -3.94003451e-01 -4.87212032e-01
3.11167934e-03 -7.50473261e-01 1.22400796e+00 -8.03651333e-01
-2.03813243e+00 1.11875570e+00 -6.08579993e-01 4.26707417e-01
1.17579684e-01 -1.94771409e-01 -4.46023285e-01 6.62756920e-01
-2.73871332e-01 -8.80718529e-02 1.35634398e+00 -1.96861827e+00
-1.87763065e-01 -2.87100285e-01 8.38604271e-02 9.51959491e-02
4.72886972e-02 2.14755759e-02 -1.66077256e-01 -3.65240693e-01
3.99969190e-01 -7.05523670e-01 1.56991839e-01 4.87818629e-01
-1.83917731e-01 4.73235548e-01 8.81790161e-01 -7.13547945e-01
5.93948603e-01 -2.12435865e+00 -1.20401867e-01 4.36082214e-01
5.22097088e-02 -4.26641181e-02 -6.75143152e-02 2.76763797e-01
-6.17601760e-02 -4.33461577e-01 -2.90648460e-01 -4.80089307e-01
-3.77055347e-01 2.32164547e-01 -1.88628569e-01 9.93376613e-01
-4.36221249e-02 3.27966571e-01 -6.25945151e-01 -1.10861324e-01
3.75268608e-01 8.08155298e-01 -3.68798077e-01 2.73855627e-01
-2.73520112e-01 5.25141120e-01 -7.25639462e-02 8.98660839e-01
1.26970625e+00 1.55053511e-01 -2.23602355e-01 -4.69035774e-01
-4.25949305e-01 -1.33113235e-01 -1.22069931e+00 1.59440386e+00
-8.95540535e-01 6.95019007e-01 9.05347824e-01 -6.16468549e-01
1.06826115e+00 1.65856872e-02 5.61845124e-01 -6.10312343e-01
6.40864391e-03 5.01685977e-01 -6.12660944e-01 -4.94545132e-01
4.45757091e-01 -4.08871949e-01 6.06570184e-01 3.91525477e-01
-4.85172421e-01 -6.25947654e-01 -4.65126783e-01 -2.57489771e-01
9.66389060e-01 2.32724175e-01 -4.95133549e-02 -4.39263344e-01
5.36797941e-01 -8.27542916e-02 1.12067237e-01 4.17389899e-01
3.43053937e-01 1.02276313e+00 1.06718145e-01 -1.57827854e-01
-1.11396992e+00 -1.22991896e+00 -5.91356218e-01 5.64335525e-01
3.62033784e-01 1.62108958e-01 -6.88213408e-01 -4.92277183e-02
2.45835423e-01 8.91215682e-01 -2.41210029e-01 3.48928988e-01
-5.12766719e-01 -4.79016393e-01 5.82822366e-03 1.84690766e-02
2.68722568e-02 -4.65098172e-01 -6.21510327e-01 -8.59431997e-02
1.26719460e-01 -9.59987581e-01 -2.04968870e-01 -2.82479465e-01
-7.68614471e-01 -1.29789877e+00 -1.01185441e+00 -4.74304199e-01
9.96727526e-01 8.49959373e-01 1.22387695e+00 -1.36282116e-01
-4.57509756e-01 1.59665048e+00 -2.50661492e-01 -2.50523478e-01
-3.84461939e-01 -7.47561991e-01 -8.92648250e-02 5.19262969e-01
3.91157158e-02 -6.56105995e-01 -6.77776694e-01 6.24915421e-01
-6.85246587e-01 4.12035100e-02 -6.79149176e-04 3.93587351e-01
7.73720443e-01 -2.48078927e-01 1.28277332e-01 -7.19020307e-01
3.88025016e-01 -1.86878890e-01 -9.42883730e-01 1.32941186e-01
-4.43275928e-01 -4.46749181e-01 2.76221901e-01 -4.97633517e-01
-1.50899374e+00 -2.39478990e-01 1.30732134e-02 -5.34304917e-01
-4.22799051e-01 1.91539768e-02 -2.21603081e-01 -4.97813970e-01
7.99704492e-01 -3.32701369e-03 1.71321183e-01 -9.13716972e-01
3.63854647e-01 8.13632727e-01 6.60610318e-01 -1.05396104e+00
8.19478810e-01 1.15884423e+00 4.75276411e-01 -1.76748490e+00
-3.86966586e-01 -7.80796289e-01 -4.23584789e-01 -4.63838607e-01
5.57766557e-01 -7.19676614e-01 -6.99275613e-01 8.42534900e-01
-1.51040936e+00 -6.68355942e-01 -5.94786584e-01 3.75074148e-01
-7.02903569e-01 4.93782252e-01 -3.36847425e-01 -1.13479757e+00
4.79721872e-04 -8.20310533e-01 1.53587842e+00 -2.06806902e-02
3.09086978e-01 -1.13750601e+00 4.14180785e-01 3.52753550e-01
3.10423344e-01 9.12074372e-02 7.88834691e-01 7.26886272e-01
-3.99617612e-01 1.67238429e-01 -5.34613967e-01 4.57783639e-01
4.39374208e-01 4.60818052e-01 -1.43654704e+00 -3.44765455e-01
2.38599539e-01 1.80213541e-01 3.66150171e-01 6.31040156e-01
1.06717288e+00 1.52548805e-01 -2.79297084e-01 7.84972191e-01
1.70056486e+00 7.24707469e-02 6.98433876e-01 -7.42439628e-02
9.48317409e-01 7.12222219e-01 6.57892227e-01 4.86698091e-01
3.27939838e-01 8.50348890e-01 4.81676549e-01 -1.71378776e-01
-5.56677699e-01 2.46328294e-01 2.33400375e-01 8.08020294e-01
-5.11368513e-01 -2.15266287e-01 -6.30346537e-01 4.31707740e-01
-1.20730591e+00 -6.91759765e-01 -8.90705466e-01 2.64375210e+00
6.35409653e-01 -4.37929332e-01 1.09954327e-01 -2.58261375e-02
7.45381057e-01 -2.97927231e-01 -3.78983080e-01 -3.79002094e-01
-2.09058419e-01 4.96092588e-01 6.51334524e-01 1.10224223e+00
-5.00006735e-01 3.59894961e-01 7.52032042e+00 5.95212579e-01
-1.10273325e+00 1.85940385e-01 1.64602194e-02 1.73491701e-01
-1.23744106e+00 -6.62914813e-02 -7.10655868e-01 2.21226215e-01
6.86147213e-01 3.83266717e-01 6.97592318e-01 4.12601531e-01
6.07576013e-01 -6.18839204e-01 -7.85681129e-01 1.34192693e+00
1.00690730e-01 -8.64248395e-01 -1.71841532e-01 1.61852583e-01
7.48607814e-01 -8.09731260e-02 -1.21346720e-01 -6.58647776e-01
2.95574460e-02 -5.98160088e-01 6.84159875e-01 1.01067448e+00
1.55134344e+00 -2.00208798e-01 -4.77275699e-02 3.30103114e-02
-1.05470884e+00 3.39043081e-01 -4.87785220e-01 1.91562653e-01
2.29355067e-01 1.19078445e+00 -5.11937201e-01 4.85833585e-01
5.21138310e-01 7.49018133e-01 -8.00811127e-02 1.04940724e+00
5.47914505e-02 3.88866752e-01 -7.37081110e-01 2.80362755e-01
-5.90950012e-01 -8.75465453e-01 8.66800129e-01 1.21929979e+00
7.06037641e-01 2.36518130e-01 -1.61711216e-01 8.34983647e-01
2.64266223e-01 -1.03297822e-01 -6.96205199e-01 5.56967974e-01
2.65045792e-01 1.19951749e+00 -5.98451257e-01 6.42468482e-02
-6.77441597e-01 8.58101428e-01 -1.30363733e-01 8.27929735e-01
-4.80755329e-01 -4.52095568e-01 7.77259231e-01 7.12309003e-01
3.83102089e-01 -6.40845180e-01 -4.52740490e-01 -1.25863826e+00
2.33770236e-01 -5.58363438e-01 -3.81598234e-01 -1.35399890e+00
-1.56620896e+00 -1.58638582e-01 3.55121046e-01 -1.13969886e+00
3.13509643e-01 -1.05521607e+00 -4.61112618e-01 1.27627957e+00
-1.78316402e+00 -1.14180326e+00 -5.13616145e-01 8.37447882e-01
1.33644894e-01 1.25455439e-01 8.70390475e-01 1.90400615e-01
-1.78749442e-01 -1.06770761e-01 5.93160689e-01 -6.19641483e-01
6.26435280e-01 -1.21731985e+00 1.67276472e-01 6.02635801e-01
-2.92906344e-01 5.94828784e-01 8.58648241e-01 -6.25381708e-01
-1.95854330e+00 -5.04462123e-01 2.11169824e-01 -3.66162956e-01
3.82239819e-01 -3.48524064e-01 -7.80458510e-01 3.88832986e-01
1.69745868e-03 2.47567058e-01 4.96604651e-01 -1.42479628e-01
-2.47176647e-01 -4.17290211e-01 -1.51991081e+00 3.79909724e-01
1.18247759e+00 -7.71657944e-01 -4.77548063e-01 2.17716351e-01
2.00195804e-01 -5.22779703e-01 -9.50473130e-01 2.07804844e-01
9.94460344e-01 -1.07201636e+00 1.12877023e+00 2.43356839e-01
-1.87630445e-01 -4.76292014e-01 -4.61424679e-01 -1.29175341e+00
-1.68258935e-01 -1.10254908e+00 -5.49774170e-02 9.76424038e-01
-6.82230070e-02 -9.99427259e-01 5.90918243e-01 6.18938625e-01
-3.52089137e-01 -3.74936938e-01 -8.11277807e-01 -5.80163240e-01
-3.53775144e-01 -6.24892712e-01 2.54058152e-01 8.14880908e-01
-6.53936148e-01 -2.34571788e-02 1.21749341e-01 3.34416062e-01
1.20638728e+00 1.34398684e-01 9.22309339e-01 -1.70695269e+00
-3.32003802e-01 1.68933328e-02 6.85448349e-02 -1.38148963e+00
5.37220351e-02 -3.17204922e-01 9.11811814e-02 -1.36581433e+00
-1.93508908e-01 -9.75410342e-01 4.19466048e-01 -4.17651795e-02
4.30518687e-01 2.67424375e-01 -4.99314725e-01 3.01937252e-01
1.01489499e-01 2.14438632e-01 1.32355511e+00 2.06390440e-01
-2.65789121e-01 1.27316207e-01 -3.38850081e-01 7.87229776e-01
3.17143768e-01 -2.87097037e-01 -6.52903438e-01 -6.22191846e-01
6.68352187e-01 -6.72168136e-02 6.48319244e-01 -6.04285777e-01
-2.76749194e-01 -2.84905583e-01 1.83854982e-01 -6.38682127e-01
7.90972590e-01 -1.24454510e+00 4.44111973e-01 -2.16747418e-01
1.58943728e-01 -3.02639693e-01 6.04229495e-02 6.64361358e-01
4.89819318e-01 -5.67680955e-01 9.02477026e-01 -1.02101527e-01
7.35635236e-02 3.10089160e-03 -2.41015494e-01 -1.23918638e-01
6.32728159e-01 -6.83111489e-01 -3.88816297e-01 -2.85314500e-01
-3.56921941e-01 -7.31774092e-01 1.31003332e+00 -4.95156378e-01
7.68691719e-01 -1.30502617e+00 -4.45802152e-01 6.48760021e-01
-1.58265308e-01 1.68930188e-01 -8.46154168e-02 7.98646271e-01
-1.16661215e+00 -1.21482417e-01 -6.29937137e-03 -9.65696454e-01
-1.25242949e+00 -7.67229646e-02 5.08937597e-01 5.34087837e-01
-7.74206460e-01 5.62157750e-01 1.38771623e-01 -2.49754995e-01
-3.28774184e-01 -5.31087279e-01 3.75423759e-01 -1.42630845e-01
2.88711667e-01 7.62835383e-01 4.26766992e-01 -6.01627290e-01
-1.34955704e-01 1.35046458e+00 5.29914021e-01 -3.55479449e-01
1.32080793e+00 -4.77570742e-01 -4.15617526e-01 7.95258999e-01
9.24866855e-01 8.43167305e-01 -1.48787594e+00 -4.07776348e-02
-7.40182459e-01 -9.74246740e-01 3.91975313e-01 -3.85222733e-01
-7.79474199e-01 8.62837076e-01 2.28526443e-01 5.64111590e-01
1.13452280e+00 -2.40454629e-01 6.48176730e-01 3.24159116e-01
6.61420465e-01 -9.95610595e-01 -2.39593923e-01 2.46179864e-01
9.38288152e-01 -7.42019534e-01 2.06074670e-01 -7.42496848e-01
3.14941145e-02 1.32536376e+00 -1.80385038e-02 -1.01334803e-01
1.00989771e+00 3.37593406e-01 -1.66566968e-02 -2.87497520e-01
-2.28109077e-01 -4.89060022e-02 5.05100012e-01 1.06224561e+00
2.35377118e-01 -7.48390853e-02 1.60750225e-01 -1.84372082e-01
3.38405490e-01 -1.45066008e-01 7.12019801e-01 8.42780292e-01
-3.78640532e-01 -1.03644347e+00 -9.30610657e-01 3.82867843e-01
-6.77631497e-02 2.54805148e-01 -3.13095637e-02 3.52347374e-01
-9.04729366e-02 1.02292621e+00 1.63156986e-01 1.02359965e-01
5.98913550e-01 -1.40811771e-01 1.13561380e+00 -7.33900070e-01
4.03341316e-02 5.31066597e-01 3.48863631e-01 -4.42788482e-01
-8.72545540e-01 -1.08395445e+00 -6.93594396e-01 -3.01389515e-01
-5.23704767e-01 -4.29667741e-01 1.02089763e+00 5.09816110e-01
1.58947945e-01 -1.43958777e-01 8.88756096e-01 -1.25215530e+00
-2.38716342e-02 -6.57690287e-01 -1.08794832e+00 1.82969898e-01
6.60257220e-01 -9.38605845e-01 -6.41850531e-01 1.91155374e-01] | [9.790326118469238, -2.975803852081299] |
8a4fd56d-2274-48e8-9c4f-2e712b6d9c36 | sabia-portuguese-large-language-models | 2304.07880 | null | https://arxiv.org/abs/2304.07880v2 | https://arxiv.org/pdf/2304.07880v2.pdf | Sabiá: Portuguese Large Language Models | As the capabilities of language models continue to advance, it is conceivable that "one-size-fits-all" model will remain as the main paradigm. For instance, given the vast number of languages worldwide, many of which are low-resource, the prevalent practice is to pretrain a single model on multiple languages. In this paper, we add to the growing body of evidence that challenges this practice, demonstrating that monolingual pretraining on the target language significantly improves models already extensively trained on diverse corpora. More specifically, we further pretrain GPT-J and LLaMA models on Portuguese texts using 3% or less of their original pretraining budget. Few-shot evaluations on Poeta, a suite of 14 Portuguese datasets, reveal that our models outperform English-centric and multilingual counterparts by a significant margin. Our best model, Sabi\'a-65B, performs on par with GPT-3.5-turbo. By evaluating on datasets originally conceived in the target language as well as translated ones, we study the contributions of language-specific pretraining in terms of 1) capturing linguistic nuances and structures inherent to the target language, and 2) enriching the model's knowledge about a domain or culture. Our results indicate that the majority of the benefits stem from the domain-specific knowledge acquired through monolingual pretraining. | ['Thales Sales Almeida', 'Rodrigo Nogueira', 'Hugo Abonizio', 'Ramon Pires'] | 2023-04-16 | null | null | null | null | ['culture'] | ['speech'] | [-9.31091979e-02 5.01689203e-02 -5.41013241e-01 -2.69160777e-01
-9.14079309e-01 -8.35452795e-01 7.58086205e-01 1.42020673e-01
-8.38419199e-01 9.23847377e-01 4.38045561e-01 -6.46180093e-01
1.15648434e-01 -4.29831684e-01 -8.78416598e-01 -7.76867718e-02
3.01012576e-01 8.59902620e-01 -9.06386822e-02 -7.28988588e-01
-2.04743534e-01 2.90715754e-01 -9.48249400e-01 3.97158206e-01
1.18223977e+00 3.52665633e-01 4.41375226e-01 2.34390154e-01
-2.23416373e-01 7.71465540e-01 -3.54415566e-01 -7.50078917e-01
1.86030149e-01 -2.08986998e-01 -8.49396646e-01 -2.43182823e-01
3.40943307e-01 -1.31392218e-02 -5.60193583e-02 8.81535292e-01
2.59934068e-01 -1.71093583e-01 4.72382367e-01 -5.90922534e-01
-9.81643677e-01 1.26498902e+00 -4.62666482e-01 2.77668685e-01
2.98138142e-01 1.96131244e-01 1.09398890e+00 -9.90390062e-01
9.05995369e-01 1.30173194e+00 8.38413358e-01 3.90575200e-01
-1.36927426e+00 -6.40665889e-01 2.85121471e-01 -3.35856266e-02
-1.50571966e+00 -6.96694613e-01 7.14773476e-01 -4.45217162e-01
1.35911989e+00 -1.01468461e-02 4.55545515e-01 1.36586905e+00
1.30115569e-01 7.55399406e-01 1.43329775e+00 -8.29918325e-01
-4.27684665e-01 5.92407763e-01 -9.55238342e-02 4.89712745e-01
3.09354782e-01 1.41035736e-01 -5.39656818e-01 7.35169426e-02
3.92668515e-01 -6.22295022e-01 -1.87476322e-01 -3.83235775e-02
-1.27051568e+00 8.77714694e-01 1.25652164e-01 8.05582523e-01
-1.59467235e-01 -3.26527119e-01 4.58760381e-01 4.12656784e-01
5.58810592e-01 6.50448740e-01 -9.41440582e-01 -2.63479233e-01
-9.77723956e-01 1.17157651e-02 8.19054484e-01 9.22390819e-01
8.30078304e-01 1.02606297e-01 2.61625320e-01 1.15608001e+00
2.10542187e-01 6.80871010e-01 7.76167333e-01 -5.73968410e-01
7.41957068e-01 5.50407350e-01 -3.93593535e-02 -6.44942462e-01
-3.01497906e-01 -6.88807786e-01 -4.22855139e-01 -3.83006513e-01
6.01936698e-01 -3.07961583e-01 -7.68508017e-01 2.06089783e+00
-1.16494045e-01 -3.37082475e-01 2.77425677e-01 4.60819036e-01
4.89450485e-01 6.32967830e-01 2.48013020e-01 -9.46751088e-02
1.20187557e+00 -9.55264330e-01 -3.56411904e-01 -8.64632845e-01
8.85800600e-01 -1.00558901e+00 1.50601065e+00 2.87258834e-01
-1.05382001e+00 -6.82905495e-01 -8.73109341e-01 -2.96768516e-01
-7.01308966e-01 3.33541423e-01 6.97076201e-01 7.40365028e-01
-1.07596505e+00 2.14094594e-01 -6.41553342e-01 -7.06008673e-01
1.10618975e-02 2.39342794e-01 -4.83718097e-01 -3.02902997e-01
-1.33732903e+00 1.33103156e+00 8.28094840e-01 -1.20283946e-01
-7.03093052e-01 -8.40356529e-01 -7.90975749e-01 -1.01013452e-01
2.83410847e-01 -4.73647743e-01 1.29361367e+00 -1.26775551e+00
-1.43046486e+00 1.20581424e+00 4.76315059e-03 -4.59885269e-01
5.28046966e-01 -3.07655752e-01 -6.68960690e-01 -1.68435439e-01
3.62372935e-01 7.76280642e-01 2.10956156e-01 -1.23948753e+00
-7.14364946e-01 -1.76990762e-01 1.08470358e-01 3.19142014e-01
-5.03196299e-01 4.16330338e-01 -4.35854793e-01 -6.24855042e-01
-4.02490199e-01 -9.25608695e-01 -1.05816014e-01 -6.89822137e-01
-1.37945592e-01 -1.20898508e-01 3.81774336e-01 -1.08054757e+00
1.26415086e+00 -2.11400485e+00 8.45385417e-02 5.00706472e-02
-2.57050455e-01 2.95375079e-01 -3.68270010e-01 6.28246427e-01
4.65580337e-02 4.01943833e-01 -1.61931843e-01 -3.88557404e-01
4.13776748e-02 5.19805968e-01 -2.24492610e-01 1.26099169e-01
4.21251386e-01 1.00876188e+00 -8.58641803e-01 -3.62028241e-01
-4.57155854e-02 4.67125446e-01 -7.11107254e-01 -2.57967114e-01
-2.14559421e-01 6.62234247e-01 -7.87259191e-02 3.70064825e-01
3.59009415e-01 -3.49098034e-02 6.36069953e-01 1.44102529e-01
-3.86991799e-01 7.32502580e-01 -7.50953257e-01 1.85373878e+00
-7.86653399e-01 5.14187098e-01 -2.25030705e-02 -7.75521278e-01
7.05685258e-01 3.42028826e-01 2.53967136e-01 -8.87407541e-01
1.20028965e-01 8.63502324e-01 5.75598836e-01 -3.29328448e-01
6.60683692e-01 -3.95510346e-01 -3.48843127e-01 2.17420533e-01
1.57236189e-01 -7.21765459e-02 4.93363529e-01 -6.88978508e-02
6.25980735e-01 3.09805542e-01 6.00684166e-01 -7.92966723e-01
6.31313145e-01 2.93669581e-01 5.88091969e-01 4.56251234e-01
-1.14647783e-02 2.41133198e-01 1.21810049e-01 -1.62444562e-01
-1.04559660e+00 -9.82800126e-01 -3.46142441e-01 1.38781273e+00
-4.35116500e-01 -2.99120724e-01 -6.38545871e-01 -6.18760407e-01
2.17784382e-02 1.04694450e+00 -5.03581703e-01 9.02574137e-02
-9.78664100e-01 -7.09519327e-01 8.18054616e-01 4.20996010e-01
3.68478566e-01 -9.47678387e-01 -2.19097108e-01 3.28051955e-01
-2.59482324e-01 -1.29653406e+00 -2.79440343e-01 2.27844343e-01
-7.00252533e-01 -6.39080167e-01 -5.92984915e-01 -9.59394217e-01
2.63067037e-01 -1.02748387e-02 1.50410450e+00 -1.48816571e-01
3.79648983e-01 3.23473960e-01 -3.11621547e-01 -4.16418135e-01
-8.75300407e-01 6.84326589e-01 4.62273024e-02 -3.77279341e-01
6.78676486e-01 -4.03156370e-01 2.01549634e-01 -1.12618998e-01
-7.14503169e-01 1.13041461e-01 6.87854052e-01 7.44385540e-01
3.86898249e-01 -5.16775213e-02 6.79302156e-01 -1.22712469e+00
5.86174369e-01 -6.46643639e-01 -2.61197448e-01 3.90166968e-01
-4.06974584e-01 -4.25092876e-02 9.40887868e-01 -4.24010754e-01
-1.26445460e+00 -3.71776134e-01 -3.10831934e-01 -3.90673317e-02
-2.10940152e-01 1.11632204e+00 -2.23870188e-01 1.45336986e-01
7.56422162e-01 9.96399149e-02 -3.87388498e-01 -7.02770174e-01
4.67442304e-01 6.28108025e-01 5.47698081e-01 -1.11408901e+00
7.04119802e-01 1.49208650e-01 -5.86214066e-01 -1.09496045e+00
-8.54682982e-01 -2.96193153e-01 -8.00895810e-01 2.09636912e-01
7.70460784e-01 -1.15581143e+00 3.76356244e-02 2.28452787e-01
-1.14686239e+00 -6.29585266e-01 -3.25882554e-01 6.41514599e-01
-1.45110250e-01 1.64282694e-01 -6.64353251e-01 -3.30992043e-01
-2.09479723e-02 -1.14511704e+00 7.23469615e-01 -5.58352517e-03
-3.28161865e-01 -1.57366300e+00 3.46487105e-01 3.66909921e-01
4.29575890e-01 -1.21754669e-01 1.37474096e+00 -9.61714923e-01
-1.32347554e-01 4.63727564e-02 -3.14503126e-02 5.71951032e-01
1.61731690e-02 -1.16230913e-01 -9.70857143e-01 -4.93978947e-01
-9.26763266e-02 -4.22473341e-01 4.16972101e-01 5.36144711e-02
2.67540753e-01 -9.30892676e-02 -5.77070080e-02 5.22848129e-01
1.56883919e+00 1.75319370e-02 1.74173981e-01 6.04516566e-01
5.64990819e-01 6.57182574e-01 1.79740846e-01 -4.14928347e-02
8.05210769e-01 4.33760047e-01 -1.66639820e-01 -7.73703456e-02
-3.17841858e-01 -5.72875559e-01 6.89852834e-01 1.65934777e+00
-1.46331871e-02 -1.14622280e-01 -1.44421923e+00 8.23561370e-01
-1.41768491e+00 -6.61232591e-01 4.21013236e-02 2.07535696e+00
1.21126437e+00 2.48799041e-01 1.00129642e-01 -2.84930527e-01
3.07010055e-01 1.47904204e-02 -2.90344030e-01 -5.55802941e-01
-6.48063958e-01 4.48394924e-01 3.47466171e-01 7.10987866e-01
-7.51435757e-01 1.45960402e+00 6.65576935e+00 7.31802881e-01
-1.40153861e+00 3.20232302e-01 4.81740952e-01 2.16898277e-01
-5.02796173e-01 1.67987630e-01 -1.02607727e+00 2.95733422e-01
1.27074695e+00 -4.65226412e-01 6.28032207e-01 6.54723406e-01
1.17332026e-01 9.71807986e-02 -1.23210812e+00 4.83547360e-01
2.11835399e-01 -9.82543051e-01 2.20524430e-01 1.95856392e-01
1.02296996e+00 6.73426628e-01 1.05978236e-01 8.55815411e-01
6.01004243e-01 -1.11318076e+00 1.07998073e+00 1.44716635e-01
7.90830791e-01 -5.69142818e-01 6.96136892e-01 6.05167031e-01
-9.04581904e-01 4.35194522e-02 -3.94730687e-01 -2.44396433e-01
9.27283391e-02 2.41038740e-01 -9.09281909e-01 7.60857821e-01
4.51176941e-01 7.40747631e-01 -1.06027865e+00 6.07135296e-01
-3.96479070e-01 9.58198845e-01 -4.03853267e-01 2.53141761e-01
4.19342875e-01 -2.03145981e-01 4.49944466e-01 1.65175653e+00
3.37620676e-01 -3.85284990e-01 4.70410079e-01 5.77423871e-01
-1.95547968e-01 7.29653597e-01 -7.11205602e-01 -3.72701794e-01
4.46763992e-01 8.70753169e-01 -3.35900277e-01 -3.91981691e-01
-1.02995288e+00 6.90244138e-01 6.80044711e-01 3.33114922e-01
-6.35742247e-01 -5.52890636e-02 3.90565038e-01 3.51960398e-02
2.29768887e-01 -4.29199010e-01 -3.34614843e-01 -1.35974717e+00
-4.92163487e-02 -1.46639168e+00 3.49026710e-01 -6.68033779e-01
-1.37094092e+00 7.87102103e-01 -2.60112286e-02 -7.10447907e-01
-3.13259602e-01 -9.30873394e-01 -5.44684418e-02 1.22113025e+00
-1.77617741e+00 -1.63210464e+00 3.27398866e-01 4.63930130e-01
5.74419320e-01 -3.82831872e-01 8.73466551e-01 5.77292144e-01
-4.99621928e-01 6.64288700e-01 1.81432560e-01 2.78767169e-01
8.53030980e-01 -1.00603604e+00 2.56051898e-01 1.02181005e+00
5.31504989e-01 9.30933475e-01 5.61360598e-01 -6.11150920e-01
-1.05612588e+00 -7.68135726e-01 1.41238546e+00 -6.61313355e-01
1.24838567e+00 -2.92099893e-01 -9.76581216e-01 1.18846965e+00
6.75309598e-01 -5.71186066e-01 7.28175938e-01 5.30158043e-01
-5.33849001e-01 -8.46246767e-05 -7.56399512e-01 7.69712746e-01
9.80329633e-01 -8.12780142e-01 -9.49477792e-01 1.48837447e-01
6.71549261e-01 -3.00080746e-01 -1.16767204e+00 3.36453974e-01
4.41794097e-01 -7.21995294e-01 6.52506113e-01 -7.96133995e-01
4.29794282e-01 -3.86769995e-02 -5.09400189e-01 -1.59833634e+00
-3.67342204e-01 -6.26208603e-01 5.22437096e-01 1.41015470e+00
8.35762203e-01 -7.25483716e-01 3.95512491e-01 1.53707236e-01
-3.45914304e-01 -3.69774759e-01 -8.68054271e-01 -9.03928459e-01
9.10915852e-01 -5.51892221e-01 5.61517656e-01 1.36707878e+00
-4.98092249e-02 7.55944908e-01 -2.48806961e-02 -5.30648604e-02
1.74652547e-01 -2.22334459e-01 8.19645762e-01 -1.14360952e+00
-5.31395257e-01 -6.98107898e-01 6.47520050e-02 -9.38925326e-01
5.07594228e-01 -1.38452029e+00 -1.40273646e-01 -1.31437409e+00
2.54488111e-01 -6.53601229e-01 -2.45809242e-01 6.09522283e-01
-1.94992706e-01 2.04352111e-01 3.54540408e-01 1.43554643e-01
-1.82764977e-01 2.70256847e-01 1.09448838e+00 1.54103851e-02
-2.14063853e-01 -4.41521674e-01 -9.88178551e-01 9.02814507e-01
8.31805170e-01 -2.59738952e-01 -3.47935736e-01 -1.17788768e+00
5.12924731e-01 -3.02071333e-01 -1.04019217e-01 -8.98832023e-01
1.95826893e-03 -2.34080493e-01 1.76975969e-02 -1.37879506e-01
1.92660928e-01 -7.26066768e-01 1.23808809e-01 3.02219629e-01
-1.43610403e-01 3.72580975e-01 7.71643758e-01 -4.98609468e-02
-3.08570474e-01 -3.16135317e-01 6.94235802e-01 -2.35812560e-01
-8.35580170e-01 -1.09687448e-01 -3.25320512e-01 6.27645135e-01
5.36007583e-01 -1.82209745e-01 -4.87958759e-01 -3.98605056e-02
-4.59120303e-01 4.45553893e-03 6.21191323e-01 6.40638053e-01
-2.30355173e-01 -1.09154844e+00 -1.05982375e+00 3.02468419e-01
2.82581329e-01 -3.63218069e-01 -9.78356525e-02 8.46776903e-01
-4.51422989e-01 8.35997283e-01 -2.43440419e-01 -4.46183145e-01
-7.21977711e-01 3.60095173e-01 2.85129428e-01 -5.77225626e-01
-2.17696011e-01 7.69105673e-01 1.77914411e-01 -9.77433681e-01
-2.51653463e-01 -3.07795137e-01 -5.82959317e-02 -4.87889946e-02
-3.34399417e-02 1.17375046e-01 3.12430523e-02 -1.12478399e+00
-3.15919578e-01 4.21958536e-01 -3.76079351e-01 -4.83190149e-01
1.32221532e+00 -8.51008967e-02 -1.80476755e-01 7.41697788e-01
9.55699146e-01 7.66724646e-01 -7.52959192e-01 -3.87123436e-01
4.09937382e-01 -7.26213828e-02 -1.50097638e-01 -1.26663673e+00
-6.28991425e-01 9.21342015e-01 5.39014526e-02 -1.66944206e-01
8.42220128e-01 -1.04309702e-02 5.50474405e-01 3.45871210e-01
5.36677480e-01 -1.28142357e+00 -3.52911353e-01 1.01783049e+00
7.46426284e-01 -1.15009892e+00 -1.59559861e-01 -1.54240012e-01
-6.76205695e-01 7.07101405e-01 5.23947060e-01 -1.35818019e-03
5.87371469e-01 2.40809470e-01 3.90562564e-01 1.24013163e-01
-7.59699583e-01 -1.68375507e-01 2.73062170e-01 5.64011395e-01
1.04030478e+00 3.07781309e-01 -4.77089435e-01 7.35338211e-01
-7.80744255e-01 -1.04543224e-01 2.82320470e-01 7.45016038e-01
-3.17950994e-01 -1.50901711e+00 -2.64926016e-01 1.00156054e-01
-6.80120885e-01 -5.44245362e-01 -4.09357727e-01 1.42669654e+00
4.44713891e-01 8.56363118e-01 -8.46232697e-02 -7.03788549e-02
4.00110334e-01 3.94988865e-01 5.99780440e-01 -9.18631554e-01
-9.21552062e-01 1.79650471e-01 4.58218932e-01 -2.90990267e-02
-3.67644250e-01 -9.03561771e-01 -8.70153666e-01 -2.98713475e-01
-2.98270746e-03 1.99173912e-01 5.23855209e-01 1.08701277e+00
1.85079396e-01 3.43403399e-01 7.92291984e-02 -4.54036355e-01
-5.66581011e-01 -1.15192509e+00 -2.61317760e-01 3.24462295e-01
-1.02507159e-01 -3.97178173e-01 -1.75460786e-01 1.44788206e-01] | [10.916749954223633, 9.917023658752441] |
5435dc9b-b9e3-48a1-89c1-d13c3debe7a8 | ufo-unified-feature-optimization | 2207.10341 | null | https://arxiv.org/abs/2207.10341v1 | https://arxiv.org/pdf/2207.10341v1.pdf | UFO: Unified Feature Optimization | This paper proposes a novel Unified Feature Optimization (UFO) paradigm for training and deploying deep models under real-world and large-scale scenarios, which requires a collection of multiple AI functions. UFO aims to benefit each single task with a large-scale pretraining on all tasks. Compared with the well known foundation model, UFO has two different points of emphasis, i.e., relatively smaller model size and NO adaptation cost: 1) UFO squeezes a wide range of tasks into a moderate-sized unified model in a multi-task learning manner and further trims the model size when transferred to down-stream tasks. 2) UFO does not emphasize transfer to novel tasks. Instead, it aims to make the trimmed model dedicated for one or more already-seen task. With these two characteristics, UFO provides great convenience for flexible deployment, while maintaining the benefits of large-scale pretraining. A key merit of UFO is that the trimming process not only reduces the model size and inference consumption, but also even improves the accuracy on certain tasks. Specifically, UFO considers the multi-task training and brings two-fold impact on the unified model: some closely related tasks have mutual benefits, while some tasks have conflicts against each other. UFO manages to reduce the conflicts and to preserve the mutual benefits through a novel Network Architecture Search (NAS) method. Experiments on a wide range of deep representation learning tasks (i.e., face recognition, person re-identification, vehicle re-identification and product retrieval) show that the model trimmed from UFO achieves higher accuracy than its single-task-trained counterpart and yet has smaller model size, validating the concept of UFO. Besides, UFO also supported the release of 17 billion parameters computer vision (CV) foundation model which is the largest CV model in the industry. | ['Jingdong Wang', 'Errui Ding', 'Jingtuo Liu', 'Junyu Han', 'Haocheng Feng', 'Lufei Liu', 'Jian Wang', 'Jinwen Chen', 'Zhigang Wang', 'Xinyu Zhang', 'Gang Zhang', 'Nan Peng', 'Bi Li', 'Deli Yu', 'Yifan Sun', 'Teng Xi'] | 2022-07-21 | null | null | null | null | ['vehicle-re-identification'] | ['computer-vision'] | [-1.35770589e-02 -1.07437097e-01 -1.55553982e-01 -3.47559720e-01
-4.28915381e-01 -1.50330722e-01 3.22547436e-01 -3.21677893e-01
-2.84246176e-01 4.25858468e-01 -3.06082904e-01 -1.03542162e-02
-4.68107224e-01 -7.75168002e-01 -6.85629845e-01 -6.67748153e-01
-4.41651195e-02 7.21989155e-01 -2.74183638e-02 -3.00587773e-01
-2.58890778e-01 6.81166410e-01 -1.82830000e+00 -7.56059811e-02
1.02228093e+00 1.24431884e+00 6.38174891e-01 -4.20133285e-02
1.16322882e-01 1.48576081e-01 -6.81449771e-01 -7.22266674e-01
4.39033151e-01 3.45146269e-01 -5.90400398e-01 -4.78196479e-02
4.44169998e-01 -3.53099465e-01 -3.21855754e-01 8.58880341e-01
6.41325593e-01 2.54396081e-01 5.02798557e-01 -1.66783118e+00
-7.53818095e-01 4.53108937e-01 -4.66372937e-01 1.99081562e-02
-1.95514023e-01 1.40430033e-01 8.84265125e-01 -8.91861975e-01
1.99511096e-01 1.22515273e+00 6.51254475e-01 7.27572739e-01
-1.04979336e+00 -1.06574631e+00 3.51271003e-01 -5.04868515e-02
-1.43589687e+00 -4.91595536e-01 7.07768381e-01 -3.39843452e-01
1.06244981e+00 2.86283046e-01 3.36168826e-01 1.21323252e+00
6.28127158e-02 8.72140586e-01 5.74658990e-01 -3.39844488e-02
2.28076279e-02 4.06112701e-01 2.86581099e-01 5.21279812e-01
5.00345230e-01 4.90651205e-02 -4.26545888e-01 -1.70654878e-02
6.61364496e-01 3.54964286e-01 -2.26184472e-01 -3.45665574e-01
-9.88643110e-01 7.44267702e-01 3.43752027e-01 4.55140620e-01
-4.58562702e-01 3.64297330e-02 5.19787788e-01 3.57255161e-01
4.61496443e-01 5.58056474e-01 -7.20720232e-01 6.09763674e-02
-7.47662544e-01 1.39316767e-01 5.45516372e-01 9.42923844e-01
1.07848084e+00 4.62230772e-01 -1.44320950e-01 1.03829980e+00
1.45358965e-01 6.12714350e-01 7.57188439e-01 -6.72074497e-01
5.01095116e-01 7.70512342e-01 -1.51902705e-01 -1.03141201e+00
-5.11433184e-01 -8.33121121e-01 -1.04829299e+00 5.96778989e-02
5.67631721e-02 -1.79927781e-01 -8.78935397e-01 2.18652844e+00
1.02489755e-01 5.53616602e-03 -3.12438849e-02 6.10158682e-01
7.56001830e-01 5.60695827e-01 1.43162049e-02 -1.45492172e-02
1.26317084e+00 -1.23589444e+00 -6.00150585e-01 -5.32053530e-01
5.07546902e-01 -4.79446262e-01 1.03928101e+00 3.75088632e-01
-8.46968651e-01 -9.82442737e-01 -1.11756504e+00 1.84749901e-01
-5.98153830e-01 1.78761706e-01 1.03299916e+00 8.05503726e-01
-1.01654339e+00 4.13676828e-01 -3.79356176e-01 -2.46574134e-01
4.85752672e-01 6.96078897e-01 -4.52410340e-01 -2.29318976e-01
-1.35603929e+00 9.14402664e-01 3.98080885e-01 2.17430785e-01
-1.05848920e+00 -7.69931972e-01 -8.03927779e-01 4.93726403e-01
5.14785588e-01 -8.92270207e-01 9.00234938e-01 -1.12048423e+00
-1.37368608e+00 5.10603905e-01 -1.67725921e-01 -3.52041394e-01
3.56507152e-01 -3.56371492e-01 -7.11584926e-01 -1.33708224e-01
6.10419065e-02 7.71242976e-01 1.20365322e+00 -1.20139706e+00
-7.05648482e-01 -3.57889265e-01 1.70176685e-01 1.85403764e-01
-1.07948780e+00 -1.87171593e-01 -6.90010786e-01 -6.36362791e-01
-3.86709958e-01 -8.98882866e-01 -1.47844944e-02 -2.66717881e-01
-8.74166936e-02 -5.25566339e-01 1.10963964e+00 -4.56541359e-01
1.44315541e+00 -2.33572960e+00 1.85219422e-01 1.24008641e-01
5.05498528e-01 5.30481815e-01 -4.29677874e-01 8.38996023e-02
-1.77803159e-01 1.20372176e-01 -3.33509594e-02 -6.37176692e-01
1.97859228e-01 3.32429230e-01 -7.04784021e-02 3.23983431e-02
1.42895117e-01 1.10585892e+00 -5.85710466e-01 -2.45036721e-01
1.60288811e-03 3.30348611e-01 -4.08574909e-01 6.80819228e-02
4.18011583e-02 9.39678103e-02 -5.56873083e-01 8.24559212e-01
8.62501204e-01 -3.11085254e-01 4.20508794e-02 -3.16620111e-01
1.20517991e-01 -2.95290768e-01 -1.20462167e+00 1.60985661e+00
-6.42207325e-01 3.55571002e-01 1.55542105e-01 -1.15838015e+00
1.06465423e+00 1.89580843e-01 6.89432263e-01 -8.56975794e-01
1.57857761e-01 2.47456685e-01 -3.91139351e-02 -3.81169409e-01
4.85467643e-01 1.65781930e-01 -1.27366319e-01 3.52323204e-01
2.56223857e-01 3.55226547e-01 1.34464860e-01 -6.13092482e-02
7.97509611e-01 -1.19010665e-01 6.04027472e-02 -3.73707503e-01
4.92906332e-01 -4.34863985e-01 7.32615650e-01 6.31973386e-01
-2.41098180e-01 2.87957847e-01 -1.34258000e-02 -5.94435811e-01
-6.95723236e-01 -8.60449672e-01 -2.34577835e-01 1.35844016e+00
2.42528662e-01 -2.78857172e-01 -5.75741351e-01 -8.86000276e-01
5.36299109e-01 5.16758740e-01 -7.40824223e-01 -5.15638113e-01
-5.47220945e-01 -8.63560021e-01 5.67356884e-01 5.22478580e-01
7.67452657e-01 -1.02610219e+00 -6.29756629e-01 9.00705904e-02
-1.10107385e-01 -9.30109739e-01 -6.31180227e-01 1.92952484e-01
-7.72036791e-01 -9.61756527e-01 -8.17336857e-01 -8.96308780e-01
4.49404597e-01 6.17679834e-01 9.29605901e-01 1.52433179e-02
-6.80154115e-02 2.64704347e-01 -2.05042288e-01 -5.19358158e-01
-6.30081519e-02 3.18734765e-01 4.79659259e-01 4.31993693e-01
4.50497866e-01 -4.47970003e-01 -2.50593543e-01 6.23017073e-01
-8.60204220e-01 -1.73105136e-01 8.40646744e-01 1.12124431e+00
4.02887195e-01 2.46909380e-01 1.06810427e+00 -5.77465117e-01
7.09497511e-01 -4.78165835e-01 -4.79985654e-01 5.49046934e-01
-9.80425835e-01 -1.13460675e-01 6.12534761e-01 -8.09747458e-01
-1.06848574e+00 -6.84134066e-02 1.27999917e-01 -9.07793522e-01
9.02072266e-02 3.71021867e-01 -4.91313219e-01 -1.05437607e-01
4.02679294e-01 3.00302356e-01 2.59024203e-01 -6.73093736e-01
8.03800970e-02 6.71701133e-01 2.35818833e-01 -5.62914371e-01
7.47520804e-01 1.09873630e-01 -2.17724621e-01 -5.62107146e-01
-5.80500126e-01 -2.42283389e-01 -4.94806588e-01 -9.58519056e-02
6.53909683e-01 -1.00177741e+00 -9.73941624e-01 6.68262601e-01
-1.01315904e+00 -1.84119388e-01 -2.45379910e-01 3.36566180e-01
-1.28950596e-01 2.71267474e-01 -2.48766243e-01 -6.79321527e-01
-5.73368967e-01 -1.32752752e+00 1.01064289e+00 3.53299946e-01
9.39738005e-02 -9.00284171e-01 -4.85835403e-01 3.94257575e-01
7.07609117e-01 -1.00526400e-01 9.77417469e-01 -7.75756001e-01
-3.65343213e-01 -2.11599618e-01 -2.97528595e-01 7.23500490e-01
2.61648297e-01 -4.17280734e-01 -1.22958386e+00 -6.63671017e-01
1.14375897e-01 -3.69848996e-01 8.18736136e-01 3.81892771e-01
1.20095718e+00 -5.99173382e-02 -4.98511195e-01 7.66617537e-01
1.13009882e+00 4.66402292e-01 4.77232903e-01 4.49954093e-01
6.56890869e-01 6.74102366e-01 6.53300703e-01 2.39806056e-01
1.91447690e-01 9.01356637e-01 4.95777518e-01 -3.06879759e-01
-1.24446191e-01 -1.90560650e-02 4.24198627e-01 6.11984611e-01
2.03316566e-02 -4.37841490e-02 -7.46439993e-01 3.85381818e-01
-1.94143116e+00 -7.29965329e-01 3.98442328e-01 2.19302225e+00
4.79842693e-01 9.28726271e-02 2.19704285e-02 -2.07669243e-01
7.22280324e-01 -8.89644958e-03 -8.92745137e-01 -2.12333903e-01
-7.82153383e-02 9.50046927e-02 2.67281324e-01 1.15259163e-01
-1.04872417e+00 8.68131578e-01 6.03481674e+00 1.19258142e+00
-1.18303406e+00 5.17478585e-01 7.19738245e-01 -2.37988129e-01
-3.24653015e-02 -5.48442543e-01 -1.08814442e+00 5.47396064e-01
7.72760153e-01 -4.32146549e-01 5.54625213e-01 1.26705658e+00
-2.43270233e-01 2.32106194e-01 -1.25388062e+00 1.37270904e+00
2.35106334e-01 -1.14850605e+00 3.62137824e-01 3.35677296e-01
4.69070196e-01 1.70861080e-01 2.60424137e-01 7.77186394e-01
6.00459473e-03 -1.07663631e+00 7.68820345e-01 2.78984457e-01
9.90556002e-01 -9.37840283e-01 9.65420902e-01 3.92369121e-01
-1.32441044e+00 -5.12673080e-01 -4.81115907e-01 2.06149682e-01
3.31672691e-02 6.75541222e-01 -4.29033369e-01 9.09426987e-01
7.78487861e-01 5.90314567e-01 -8.66994739e-01 7.84979463e-01
3.21897686e-01 1.99596584e-01 -3.18663865e-01 2.82976091e-01
7.15847090e-02 -1.92002982e-01 4.40790325e-01 1.06551874e+00
3.90953243e-01 -5.17405808e-01 3.33732814e-01 6.85490787e-01
-1.90009847e-01 -1.43037125e-01 -7.24042714e-01 1.78302556e-01
6.59596324e-01 1.35320830e+00 -2.07356542e-01 -2.77616024e-01
-3.51329803e-01 8.96696150e-01 3.30060720e-01 4.21039164e-01
-1.09029090e+00 -4.65324104e-01 9.96719778e-01 -1.02966428e-01
2.75817454e-01 2.60776784e-02 -1.95026532e-01 -1.08778048e+00
1.13503598e-01 -1.03786814e+00 2.85728246e-01 -3.35354656e-01
-1.28492987e+00 1.06173086e+00 8.53645056e-02 -1.15320563e+00
-5.77729680e-02 -6.88860655e-01 -3.69124204e-01 1.06739950e+00
-1.76012671e+00 -1.38616908e+00 -3.06399107e-01 8.29245329e-01
7.00250030e-01 -6.45158708e-01 7.90944636e-01 8.31317306e-01
-1.18797851e+00 1.15501964e+00 3.09190564e-02 -1.20495461e-01
7.88906276e-01 -6.83485866e-01 2.67908931e-01 7.09494650e-01
-1.48562387e-01 8.90290499e-01 1.27975956e-01 -4.36140150e-01
-1.46105671e+00 -1.28043282e+00 7.36883402e-01 -4.68352616e-01
3.47419590e-01 -4.94512349e-01 -8.34841430e-01 6.92238808e-01
-1.58638269e-01 -1.70562819e-01 6.08063340e-01 3.97384673e-01
-4.59542483e-01 -6.95247769e-01 -1.07867491e+00 2.20796168e-01
1.08930624e+00 -4.28796947e-01 -4.95628953e-01 3.08969021e-01
9.04951334e-01 -6.18556403e-02 -7.83316612e-01 5.55911899e-01
6.18564725e-01 -5.60716689e-01 1.12340987e+00 -6.47872090e-01
6.14153594e-02 -1.30539723e-02 -3.22933733e-01 -1.25094688e+00
-7.06106305e-01 -5.55492938e-01 -4.02917117e-01 1.36916685e+00
4.04237628e-01 -1.19351757e+00 5.28441131e-01 7.30044603e-01
-3.75086129e-01 -9.98584688e-01 -1.03046799e+00 -1.23555970e+00
-1.45140424e-01 -3.07904363e-01 1.07918167e+00 1.04975188e+00
-4.84749705e-01 4.17623192e-01 -6.49826229e-01 5.22145033e-02
5.31676292e-01 6.06278330e-02 7.85315990e-01 -1.64666975e+00
-3.66642356e-01 -5.42694509e-01 5.45039307e-03 -1.05635595e+00
1.10114381e-01 -9.32384074e-01 -3.07631820e-01 -1.18779540e+00
2.07503781e-01 -7.72074640e-01 -6.34598434e-01 8.97817135e-01
-9.22117382e-02 -4.16772626e-03 5.49848020e-01 4.77049887e-01
-4.38571423e-01 6.17817700e-01 1.24729550e+00 -2.69567102e-01
-2.58925349e-01 1.92104414e-01 -1.08389592e+00 5.37660718e-01
7.54603326e-01 -4.08905506e-01 -5.50490797e-01 -7.65993178e-01
-7.13666305e-02 -4.02429640e-01 1.85846001e-01 -9.62304056e-01
2.05371946e-01 -9.98794585e-02 3.70305479e-01 -2.21652508e-01
4.74897116e-01 -1.06543612e+00 3.06758493e-01 5.15063941e-01
1.67671546e-01 1.59029350e-01 3.50971967e-01 4.92066532e-01
-2.57167250e-01 -3.96774173e-01 7.22221732e-01 6.44171387e-02
-9.23469841e-01 4.90112156e-01 6.31041527e-02 -4.73514616e-01
1.21332133e+00 -4.40029472e-01 -3.58949661e-01 -3.29227820e-02
-5.74155390e-01 5.37028968e-01 1.35274246e-01 8.97349477e-01
5.49205422e-01 -1.32246387e+00 -5.68176270e-01 4.31544065e-01
1.06118120e-01 5.38849160e-02 4.76975530e-01 6.72482431e-01
1.41971037e-01 7.16829836e-01 -2.95983642e-01 -4.40116733e-01
-1.15890157e+00 7.57353008e-01 2.63891995e-01 -5.99004209e-01
-3.97974283e-01 7.59757042e-01 7.26950586e-01 -5.18082261e-01
3.64683032e-01 -3.53038199e-02 -3.08923751e-01 3.62768859e-01
5.81098855e-01 3.96114409e-01 1.99161276e-01 -5.59712529e-01
-3.62120926e-01 4.08851773e-01 -3.51392865e-01 5.97252667e-01
1.34450150e+00 -4.24981005e-02 5.50930724e-02 7.69526809e-02
1.15401912e+00 -2.46253982e-01 -1.29213881e+00 -2.26250529e-01
-2.50724196e-01 -3.84645402e-01 2.50525296e-01 -9.86851275e-01
-1.45369744e+00 7.49953389e-01 6.97741926e-01 5.15472256e-02
1.33330834e+00 -1.83186233e-01 1.00409389e+00 5.48189938e-01
6.46972716e-01 -1.21006322e+00 2.80237436e-01 5.71685672e-01
1.03620505e+00 -1.04869688e+00 -1.35691047e-01 -2.77752101e-01
-7.90024161e-01 8.96124005e-01 1.05840647e+00 3.29697192e-01
6.06425405e-01 1.73345283e-01 -3.56381685e-01 -1.93086088e-01
-5.76553166e-01 -1.41620740e-01 2.90784687e-01 7.51699507e-01
-1.73441187e-01 2.19572987e-02 7.70714954e-02 1.05206943e+00
6.61237910e-02 5.60440384e-02 -1.21800289e-01 6.09547853e-01
-3.37380558e-01 -1.06518698e+00 -1.73256502e-01 5.77771962e-01
-6.73973039e-02 -1.13233775e-01 -2.98644453e-02 9.39475775e-01
5.38584113e-01 8.03727865e-01 6.63936045e-03 -7.18351483e-01
4.92585659e-01 2.11196616e-01 1.24854214e-01 -5.01375079e-01
-8.73745680e-01 -8.32577944e-02 -1.91839915e-02 -5.01197934e-01
-1.36244014e-01 -3.38523656e-01 -8.49205017e-01 -3.86002004e-01
-5.37532091e-01 9.57085118e-02 6.54789984e-01 8.33042443e-01
8.78602505e-01 7.85437763e-01 7.90654778e-01 -9.49600101e-01
-7.90794075e-01 -1.00284755e+00 -6.82179093e-01 4.77003545e-01
8.47851634e-02 -1.12966692e+00 -1.52165174e-01 -2.43280947e-01] | [9.379958152770996, 3.245589256286621] |
0fc2e3a5-cb46-494d-9c63-6f71de88143d | robust-person-identification-a-wifi-vision | 2210.00127 | null | https://arxiv.org/abs/2210.00127v1 | https://arxiv.org/pdf/2210.00127v1.pdf | Robust Person Identification: A WiFi Vision-based Approach | Person re-identification (Re-ID) has become increasingly important as it supports a wide range of security applications. Traditional person Re-ID mainly relies on optical camera-based systems, which incur several limitations due to the changes in the appearance of people, occlusions, and human poses. In this work, we propose a WiFi vision-based system, 3D-ID, for person Re-ID in 3D space. Our system leverages the advances of WiFi and deep learning to help WiFi devices see, identify, and recognize people. In particular, we leverage multiple antennas on next-generation WiFi devices and 2D AoA estimation of the signal reflections to enable WiFi to visualize a person in the physical environment. We then leverage deep learning to digitize the visualization of the person into 3D body representation and extract both the static body shape and dynamic walking patterns for person Re-ID. Our evaluation results under various indoor environments show that the 3D-ID system achieves an overall rank-1 accuracy of 85.3%. Results also show that our system is resistant to various attacks. The proposed 3D-ID is thus very promising as it could augment or complement camera-based systems. | ['Jie Yang', 'Yili Ren'] | 2022-09-30 | null | null | null | null | ['person-identification'] | ['computer-vision'] | [-1.06757790e-01 -6.34620786e-01 1.74949169e-01 -1.01736479e-01
-3.80654693e-01 -5.60021162e-01 3.30122173e-01 -3.07798147e-01
-1.26368165e-01 2.31994122e-01 5.17877400e-01 1.05228096e-01
1.66711569e-01 -7.98384488e-01 -4.93112296e-01 -4.12234306e-01
-1.33108914e-01 1.55397311e-01 -1.23902507e-01 1.14933237e-01
-1.47961318e-01 5.04283786e-01 -1.43473399e+00 -3.50443460e-02
4.86864835e-01 9.67729390e-01 -5.02064526e-01 9.59170401e-01
4.23928559e-01 -4.53189276e-02 -6.47780359e-01 -4.45179582e-01
4.70032424e-01 -9.91050303e-02 -4.92539518e-02 -2.53877103e-01
1.04261780e+00 -5.99768102e-01 -7.51376688e-01 6.88492417e-01
1.03437388e+00 1.35730088e-01 4.29176092e-01 -1.38581109e+00
-5.40435553e-01 -1.09016802e-03 -7.92612851e-01 7.24470913e-02
1.38656259e+00 6.00016005e-02 3.19499075e-01 -7.08541155e-01
-1.28290653e-01 1.44755268e+00 1.29704356e+00 6.81253016e-01
-8.64154756e-01 -9.95216548e-01 -1.97652001e-02 2.33348683e-01
-1.61010289e+00 -5.28349519e-01 5.98297000e-01 -4.44788098e-01
5.35598576e-01 3.23323637e-01 9.83334780e-01 1.51902759e+00
-1.03783801e-01 5.27525902e-01 8.78292024e-01 -4.32114512e-01
-7.24236742e-02 -3.05908799e-01 2.25282699e-01 6.38666332e-01
7.50598013e-01 4.19101208e-01 -8.40860963e-01 -2.80479193e-01
7.73933589e-01 5.58369517e-01 -3.50278497e-01 -1.96963027e-01
-1.26466608e+00 1.93065867e-01 4.51322973e-01 -5.04381321e-02
-8.96953717e-02 4.86961931e-01 8.80075023e-02 -1.58133298e-01
-1.71182619e-03 7.96164125e-02 2.24817932e-01 -3.44143957e-01
-6.82268083e-01 3.79384756e-01 6.24661505e-01 8.40843379e-01
2.52014220e-01 1.12313963e-01 -1.74862757e-01 6.29362226e-01
6.58478022e-01 1.49746263e+00 2.02673450e-01 -7.87999809e-01
5.14744997e-01 3.27529967e-01 4.70881641e-01 -1.36618376e+00
-5.95797718e-01 -5.89227021e-01 -1.02595162e+00 -1.38366697e-02
5.66816330e-01 -5.58601260e-01 -7.06895471e-01 1.63394606e+00
4.52524394e-01 8.03517342e-01 -2.50445604e-01 1.03066814e+00
8.35655391e-01 1.78412884e-01 -6.85817227e-02 3.97297204e-01
1.47243989e+00 -7.31464505e-01 -3.74471694e-01 -3.56617957e-01
1.95630696e-02 -4.68043864e-01 6.82334781e-01 4.94592339e-01
-6.85044765e-01 -1.03193891e+00 -1.11124587e+00 4.71411824e-01
2.59162430e-02 1.92470491e-01 3.84681463e-01 1.46350443e+00
-7.89763153e-01 5.93480794e-03 -7.93187559e-01 -7.65189052e-01
3.11502784e-01 4.68179882e-01 -3.55842143e-01 -1.56133726e-01
-1.14628947e+00 3.24705213e-01 -3.39274585e-01 2.26095408e-01
-5.88104308e-01 -4.94392514e-01 -7.38535881e-01 6.00719266e-02
-2.93354988e-02 -1.13234818e+00 1.03489566e+00 -4.23401207e-01
-1.35072613e+00 6.41803503e-01 -4.34479088e-01 -4.10773665e-01
4.78899986e-01 -7.77954459e-01 -1.07662332e+00 1.17221391e-02
1.48559257e-01 7.70445764e-02 9.05766368e-01 -1.15350997e+00
-7.26843238e-01 -7.56527066e-01 8.08621421e-02 1.42656818e-01
-4.83390480e-01 -3.96434754e-01 -6.09673798e-01 -6.33922160e-01
1.85458794e-01 -1.30135214e+00 1.10658564e-01 4.23890091e-02
-5.33368289e-01 2.74078548e-01 5.19884646e-01 -6.57209575e-01
1.17002153e+00 -2.15195227e+00 -4.32869703e-01 5.38584232e-01
4.09972399e-01 2.45729700e-01 2.39815071e-01 1.31836548e-01
2.16478139e-01 -1.65293947e-01 4.58526522e-01 -4.64894116e-01
7.73667991e-02 -3.79874945e-01 -1.48638770e-01 6.74871385e-01
-7.39742935e-01 6.66292191e-01 -8.45113337e-01 -6.90838397e-02
4.99541402e-01 8.40775788e-01 -4.92030174e-01 5.05544245e-02
7.87236869e-01 4.36719418e-01 -4.39861894e-01 7.65017271e-01
9.43644702e-01 -1.05537102e-02 7.40889534e-02 -4.00535345e-01
2.14879755e-02 -1.31033823e-01 -1.39591587e+00 1.46907282e+00
-4.03052062e-01 5.14330447e-01 -6.85008094e-02 -6.27874553e-01
8.81473422e-01 2.46828511e-01 4.60416555e-01 -5.99894881e-01
-2.28543561e-02 -1.45369962e-01 -3.50504607e-01 -4.65445369e-01
1.68703824e-01 4.64155972e-01 -3.26103330e-01 5.64508379e-01
-3.79897028e-01 8.38881314e-01 -2.29533449e-01 7.93608779e-05
1.06238139e+00 7.94800818e-02 2.75854409e-01 -1.06494036e-02
5.97911477e-01 -6.18487418e-01 4.86029357e-01 1.19284451e+00
-3.77566636e-01 5.39490759e-01 -4.87232685e-01 -7.08092570e-01
-4.87800539e-01 -1.47913051e+00 2.47149035e-01 9.35662270e-01
6.45256639e-01 -5.72786689e-01 -5.75365245e-01 -5.37525654e-01
4.34191912e-01 1.00771315e-01 -4.97240812e-01 -1.23670168e-01
-6.19834721e-01 -7.20587254e-01 9.93419886e-01 7.10812986e-01
1.06862414e+00 -1.29556999e-01 -5.48337460e-01 6.77417219e-02
-6.30671263e-01 -1.15280795e+00 -6.85011029e-01 -5.93915761e-01
-3.50030541e-01 -1.07892799e+00 -1.00199378e+00 -5.35168111e-01
3.18640739e-01 1.06313789e+00 6.16930604e-01 -4.18747328e-02
-1.22370876e-01 1.12863362e+00 -7.22309276e-02 -3.79438519e-01
3.97048891e-01 -1.27322793e-01 8.60094190e-01 4.05683815e-01
6.91691458e-01 -7.22190261e-01 -1.03423262e+00 5.89241624e-01
2.72026956e-01 -3.57248813e-01 1.22980528e-01 3.33495080e-01
1.18270777e-01 7.57602751e-02 1.85475945e-01 -2.73474872e-01
3.59974921e-01 -4.14198428e-01 -2.10385412e-01 2.17208266e-01
-4.04773742e-01 -4.82670188e-01 3.02266508e-01 -4.01678681e-01
-9.89824653e-01 1.02019854e-01 -2.09780768e-01 -1.20853044e-01
-4.40961570e-01 -1.34898096e-01 -5.69715381e-01 -1.90682516e-01
7.76725054e-01 4.22017984e-02 -4.58948880e-01 -5.52438259e-01
7.44493604e-02 1.02240038e+00 9.36561704e-01 -4.17566925e-01
1.07748795e+00 8.88985395e-01 6.56757280e-02 -1.01490057e+00
-6.68750703e-01 -8.58783841e-01 -4.50420499e-01 -5.18381238e-01
8.20170522e-01 -1.45908260e+00 -1.37792051e+00 8.53013277e-01
-8.98657441e-01 1.53831393e-02 3.70294005e-01 5.63651323e-01
8.29845145e-02 6.48812294e-01 -2.88114309e-01 -9.39331949e-01
-5.93700528e-01 -6.51085317e-01 1.20195663e+00 6.43366277e-01
-4.71820772e-01 -6.88149571e-01 2.69704461e-01 5.73399961e-01
2.85371393e-01 3.20037872e-01 1.93022445e-01 -1.49098903e-01
-2.84870684e-01 -4.64532167e-01 -1.81369018e-02 -4.34066981e-01
3.88446689e-01 -5.15835464e-01 -1.33176470e+00 -4.83227879e-01
-5.67316353e-01 2.38085240e-01 5.36089540e-01 5.60097456e-01
9.46569026e-01 -2.13460281e-01 -9.50437427e-01 1.17442441e+00
9.53047216e-01 8.08543041e-02 5.35046875e-01 4.88867879e-01
1.24343002e+00 1.65210202e-01 1.78851321e-01 7.75043607e-01
6.47634566e-01 1.14311159e+00 2.95198470e-01 -1.75173238e-01
-2.73234010e-01 -4.99796897e-01 4.89134431e-01 6.07393719e-02
-5.45369327e-01 -5.46404004e-01 -8.43175113e-01 3.86373885e-02
-1.72903490e+00 -1.25457251e+00 -1.63312942e-01 2.41132641e+00
-1.21168420e-01 3.18318754e-02 7.26579070e-01 2.55593300e-01
1.06909442e+00 2.72175521e-02 -5.30624151e-01 5.72652668e-02
1.53224856e-01 8.32604710e-03 6.53942645e-01 3.93017679e-01
-1.44327712e+00 3.73940974e-01 6.03510332e+00 1.80136979e-01
-9.66228068e-01 -1.13519765e-01 -4.46486697e-02 -1.45166859e-01
1.35005981e-01 -6.64763391e-01 -1.15794790e+00 6.80596590e-01
6.62827194e-01 1.86546072e-01 4.42202300e-01 7.77376890e-01
1.03450991e-01 3.98510247e-02 -1.01828206e+00 1.93603945e+00
3.94157529e-01 -1.03655648e+00 -2.83178627e-01 3.59540969e-01
4.64952201e-01 -2.36758217e-01 1.51403487e-01 6.28486425e-02
1.99726611e-01 -8.20696592e-01 6.14964426e-01 5.84213972e-01
9.77526188e-01 -7.23078728e-01 5.17962396e-01 -2.57892814e-02
-1.79659915e+00 -2.88764179e-01 1.97616797e-02 -2.90313840e-01
3.16787869e-01 4.25092012e-01 -6.08163714e-01 3.52594554e-01
1.28920603e+00 7.88645267e-01 -5.82884371e-01 1.22432685e+00
-7.16975331e-02 6.07884705e-01 -6.09795868e-01 1.39049351e-01
-5.43402314e-01 2.21006215e-01 5.31705439e-01 1.40276289e+00
7.98869908e-01 4.36415486e-02 4.15815532e-01 2.46712238e-01
4.47905399e-02 -4.79951501e-01 -6.70664251e-01 5.19448936e-01
7.87035525e-01 9.23312843e-01 -2.92325467e-01 -4.93865758e-02
-3.46650422e-01 1.18065596e+00 -2.46713474e-01 5.06856024e-01
-8.34813118e-01 -4.53215361e-01 1.11444259e+00 4.55850422e-01
2.03990236e-01 -6.43428445e-01 -2.62095314e-02 -1.25750339e+00
-1.56519681e-01 -6.70630574e-01 4.88230258e-01 -4.68596846e-01
-1.28081477e+00 2.10406348e-01 -2.62215972e-01 -1.51084018e+00
-2.53545165e-01 -4.98563319e-01 -5.42702436e-01 6.78091884e-01
-8.75686407e-01 -1.33216655e+00 -1.03431654e+00 9.74952638e-01
6.65021166e-02 -3.82094026e-01 8.47398281e-01 7.04263747e-01
-5.67220211e-01 1.28431880e+00 2.44043872e-01 5.18607557e-01
8.31793666e-01 -1.09205866e+00 8.99348676e-01 1.11075652e+00
3.52835536e-01 9.92176771e-01 4.16958094e-01 -8.54035437e-01
-1.53388512e+00 -8.55874181e-01 5.92053354e-01 -6.26182377e-01
-6.39993772e-02 -6.23390019e-01 -9.07969922e-02 6.70881391e-01
-4.24449623e-01 5.37304729e-02 1.18585849e+00 4.48535800e-01
-5.13869762e-01 -6.12655163e-01 -1.05013251e+00 6.65115178e-01
1.58561337e+00 -5.13363957e-01 -3.42355907e-01 -9.13629904e-02
1.21029831e-01 -3.12356204e-01 -4.73730296e-01 1.15191475e-01
1.52707314e+00 -9.25180495e-01 1.78356779e+00 -5.86497150e-02
-5.51594734e-01 -5.29533446e-01 -2.15830982e-01 -9.43773150e-01
-8.65795195e-01 -6.49591744e-01 -5.40094316e-01 1.22305703e+00
-4.12248850e-01 -6.72520816e-01 8.73994410e-01 5.62680960e-01
4.76210088e-01 8.40491652e-02 -7.99074650e-01 -6.66326225e-01
-8.24736118e-01 -5.89918613e-01 9.78427589e-01 5.79906762e-01
-2.15326235e-01 1.90430015e-01 -9.57373023e-01 7.80499160e-01
1.06440997e+00 8.04802105e-02 1.37204874e+00 -1.48164177e+00
-4.04417992e-01 1.61132440e-01 -6.28423870e-01 -1.57821238e+00
-3.57827187e-01 -5.85499823e-01 -2.91988462e-01 -1.37937975e+00
-4.97728214e-02 -6.68029130e-01 -3.31499219e-01 2.68181324e-01
-2.84192823e-02 6.06353164e-01 1.40269637e-01 1.86667651e-01
-5.43859839e-01 2.53096223e-01 5.18450201e-01 -4.66123790e-01
-2.96075702e-01 6.09968305e-01 -8.35652053e-01 9.94214535e-01
6.88285887e-01 -1.90058704e-02 -7.97654092e-02 -6.97624266e-01
2.17079416e-01 -3.08771342e-01 9.47553277e-01 -1.75487947e+00
2.56951839e-01 1.78059861e-01 1.25020981e+00 -5.04516721e-01
6.09759569e-01 -9.22543168e-01 3.05201560e-01 5.71321726e-01
1.41348317e-01 4.24557664e-02 1.21279925e-01 8.08503687e-01
5.71214795e-01 3.23657930e-01 3.06501269e-01 -9.97599587e-02
-6.65799737e-01 4.05622929e-01 -3.88706326e-01 -1.91714510e-01
5.58160186e-01 -5.82908809e-01 -1.99959934e-01 -7.12275326e-01
-5.21857381e-01 9.02416743e-03 4.13014978e-01 4.06433046e-01
6.36217773e-01 -1.60581791e+00 -3.84370029e-01 4.45291996e-01
1.00125633e-01 -5.12623489e-01 4.60600257e-01 2.85161674e-01
-4.58630234e-01 4.47604686e-01 -3.70041877e-01 -8.09322119e-01
-1.56107652e+00 2.19587699e-01 4.49644119e-01 2.42673159e-01
-8.13957632e-01 8.02438378e-01 1.17221177e-01 -2.21318647e-01
4.75558579e-01 1.20365880e-01 -2.60411322e-01 -3.18977863e-01
1.00440621e+00 9.96372700e-01 -3.45570385e-01 -9.05019701e-01
-9.12941158e-01 1.18259418e+00 3.05331320e-01 -1.73332587e-01
8.46609294e-01 -6.19897068e-01 6.72974825e-01 -1.75578728e-01
6.32668257e-01 5.40583193e-01 -1.27331793e+00 -2.59475350e-01
-4.07241911e-01 -7.42552459e-01 -2.81314850e-01 -7.02458322e-01
-8.68512869e-01 9.12454486e-01 1.42316282e+00 -1.77580237e-01
8.28314900e-01 -3.30630541e-01 1.21373904e+00 6.30189896e-01
7.18389988e-01 -7.24168241e-01 3.77430879e-02 2.60785043e-01
2.48910323e-01 -1.04676986e+00 2.49583423e-01 -2.13633463e-01
-1.25003040e-01 1.01057434e+00 5.89033723e-01 1.21141545e-01
5.93666017e-01 1.26620352e-01 2.41053626e-01 4.06936631e-02
3.68792534e-01 -3.27588290e-01 3.59605074e-01 1.44468975e+00
2.23323926e-02 1.44311652e-01 5.98807454e-01 8.91755641e-01
-4.41692114e-01 -1.25216171e-01 -5.36011383e-02 6.27443671e-01
-1.95768744e-01 -8.12515199e-01 -9.82875347e-01 1.51734158e-01
-4.00517285e-01 3.47016841e-01 -2.67842859e-01 4.94238466e-01
5.95755935e-01 1.29329407e+00 -1.49519265e-01 -9.04039860e-01
3.07926714e-01 -2.23963454e-01 7.54388750e-01 -2.06483051e-01
-3.23387772e-01 -7.41830319e-02 2.88176350e-02 -6.00059986e-01
-5.57548821e-01 -5.95837295e-01 -7.42696643e-01 -6.18259549e-01
5.64516224e-02 -1.86358675e-01 4.22362864e-01 8.01986814e-01
5.78887165e-01 2.52908736e-01 4.64599311e-01 -1.13817048e+00
1.50773255e-02 -5.00531256e-01 -5.32427728e-01 2.90920019e-01
4.96228009e-01 -8.81964386e-01 2.95316149e-02 6.09436817e-02] | [6.827883243560791, 0.5062359571456909] |
0944b745-57a4-4b18-a697-bfa335df868d | what-do-you-meme-generating-explanations-for | 2212.00715 | null | https://arxiv.org/abs/2212.00715v2 | https://arxiv.org/pdf/2212.00715v2.pdf | What do you MEME? Generating Explanations for Visual Semantic Role Labelling in Memes | Memes are powerful means for effective communication on social media. Their effortless amalgamation of viral visuals and compelling messages can have far-reaching implications with proper marketing. Previous research on memes has primarily focused on characterizing their affective spectrum and detecting whether the meme's message insinuates any intended harm, such as hate, offense, racism, etc. However, memes often use abstraction, which can be elusive. Here, we introduce a novel task - EXCLAIM, generating explanations for visual semantic role labeling in memes. To this end, we curate ExHVV, a novel dataset that offers natural language explanations of connotative roles for three types of entities - heroes, villains, and victims, encompassing 4,680 entities present in 3K memes. We also benchmark ExHVV with several strong unimodal and multimodal baselines. Moreover, we posit LUMEN, a novel multimodal, multi-task learning framework that endeavors to address EXCLAIM optimally by jointly learning to predict the correct semantic roles and correspondingly to generate suitable natural language explanations. LUMEN distinctly outperforms the best baseline across 18 standard natural language generation evaluation metrics. Our systematic evaluation and analyses demonstrate that characteristic multimodal cues required for adjudicating semantic roles are also helpful for generating suitable explanations. | ['Tanmoy Chakraborty', 'Md. Shad Akhtar', 'Preslav Nakov', 'Tharun Suresh', 'Siddhant Agarwal', 'Shivam Sharma'] | 2022-12-01 | null | null | null | null | ['marketing', 'semantic-role-labeling'] | ['miscellaneous', 'natural-language-processing'] | [ 3.86762768e-01 3.32725734e-01 -3.05694431e-01 -1.78775787e-01
-5.56497812e-01 -8.00161719e-01 1.28566480e+00 5.04950821e-01
3.61720733e-02 9.25555885e-01 8.90697956e-01 -1.63330734e-01
2.21185118e-01 -7.39471674e-01 -5.42967856e-01 -4.17800277e-01
4.17339504e-01 2.86695123e-01 -2.89426506e-01 -7.12155581e-01
4.53981608e-01 -1.98323335e-02 -1.61958218e+00 1.02408934e+00
6.04618847e-01 7.68258333e-01 -4.23502713e-01 7.00146973e-01
-5.29534101e-01 1.12572217e+00 -9.17496502e-01 -1.17716336e+00
-3.49037647e-01 -6.44558489e-01 -8.88306201e-01 1.56834736e-01
6.60549879e-01 -1.04928993e-01 -3.20774317e-01 7.35847414e-01
5.11350751e-01 -1.52004194e-02 1.10989761e+00 -1.78911209e+00
-1.25600207e+00 8.81766617e-01 -5.95226705e-01 1.76780015e-01
8.03203523e-01 2.41472527e-01 1.42775548e+00 -1.00933766e+00
8.95650864e-01 1.89273202e+00 3.24390739e-01 8.22063863e-01
-1.43677104e+00 -7.52253532e-01 5.26884161e-02 1.67231292e-01
-8.44973564e-01 -2.20478848e-01 8.52339566e-01 -6.58525467e-01
3.94046843e-01 6.56618476e-01 4.32834864e-01 1.88844705e+00
-9.65088140e-04 8.87411118e-01 1.18670428e+00 3.92425060e-02
-3.76560092e-02 3.99391323e-01 1.35992184e-01 6.92047596e-01
-8.24284628e-02 -4.84715074e-01 -9.14336026e-01 -5.68568766e-01
3.36471289e-01 -6.26998767e-02 -1.39104635e-01 1.90762207e-01
-1.45534515e+00 1.17770004e+00 6.08258784e-01 7.28163347e-02
-1.97170317e-01 3.88687760e-01 5.71199358e-01 2.57084221e-01
6.63133979e-01 8.20872605e-01 2.48404980e-01 1.07578464e-01
-2.79400527e-01 3.31238478e-01 5.68699837e-01 5.75092256e-01
7.16038406e-01 -1.93580776e-01 -7.09900677e-01 1.05042624e+00
2.24870235e-01 6.22510612e-01 8.53347927e-02 -6.84788406e-01
4.79265481e-01 1.01472795e+00 2.22550735e-01 -1.63402116e+00
-3.50355804e-01 7.65669495e-02 -8.44006419e-01 -2.80373782e-01
1.34613439e-01 -2.42440358e-01 -6.09655201e-01 1.80381322e+00
3.25874120e-01 -5.63581251e-02 5.68255782e-02 1.02477241e+00
1.52042639e+00 1.07796061e+00 6.91720307e-01 5.14343195e-02
1.61261511e+00 -7.06260800e-01 -8.36912334e-01 -4.82011259e-01
2.80184776e-01 -9.23673272e-01 1.50882483e+00 -1.50064528e-02
-7.60226369e-01 -3.10923100e-01 -8.50365222e-01 -3.19802910e-01
-7.26092875e-01 4.13885154e-02 8.86628270e-01 2.55241334e-01
-8.60111833e-01 1.43893570e-01 2.66669154e-01 -4.54513311e-01
3.77675682e-01 -9.52362046e-02 -4.61463600e-01 2.78033525e-01
-1.49636900e+00 6.88708007e-01 2.11587787e-01 -1.00069016e-01
-6.62879944e-01 -6.34701014e-01 -8.51021707e-01 -3.58838767e-01
2.62949735e-01 -8.62006903e-01 8.61500382e-01 -1.15104449e+00
-8.83905470e-01 1.27131987e+00 -1.07799627e-01 -1.99883983e-01
3.66354227e-01 -1.09926306e-01 -6.93322182e-01 4.08944160e-01
5.03537178e-01 1.12827396e+00 1.19740343e+00 -1.70237446e+00
-2.81163067e-01 -1.61168590e-01 4.84469563e-01 1.42533973e-01
-7.25733399e-01 2.40265802e-01 4.38524000e-02 -9.79392409e-01
-2.99827307e-01 -8.28511238e-01 1.14108518e-01 -9.77496579e-02
-9.49698031e-01 -4.25497323e-01 8.28261495e-01 -6.27053201e-01
1.21291435e+00 -2.07889104e+00 1.71925619e-01 2.07799878e-02
8.14572215e-01 -4.04626757e-01 -2.79529124e-01 8.27122271e-01
3.19220573e-02 5.04433513e-01 -1.41588956e-01 -3.36152911e-01
3.59542966e-01 -7.46797398e-02 -8.46954584e-01 1.15774065e-01
3.45972925e-01 1.19039989e+00 -9.82502520e-01 -5.46200097e-01
-4.09014337e-02 5.42352378e-01 -3.50766122e-01 2.39919722e-01
-4.12214607e-01 1.97712079e-01 -3.25612694e-01 6.50159597e-01
4.14703369e-01 -6.59902334e-01 1.23147309e-01 -2.54915982e-01
2.02728182e-01 1.34091200e-02 -5.55989981e-01 9.51054335e-01
-3.35878700e-01 9.80748653e-01 -7.90811107e-02 -3.22484940e-01
8.65217447e-01 3.13604437e-02 1.30675018e-01 -6.80210292e-01
1.93512857e-01 -8.78212750e-02 -5.63582718e-01 -6.66792214e-01
7.95797706e-01 -4.45341945e-01 -5.74917257e-01 6.52984560e-01
-4.32810903e-01 2.75569279e-02 5.20818494e-02 8.13347638e-01
8.49648535e-01 -2.15543881e-01 2.06874788e-01 -2.88411835e-03
2.88701892e-01 9.00148302e-02 -1.95835531e-02 7.07221091e-01
-4.84137461e-02 5.56366861e-01 1.03915751e+00 -3.78647178e-01
-8.35489690e-01 -1.20706356e+00 2.69685894e-01 1.63002491e+00
7.50202894e-01 -5.49071193e-01 -4.32929754e-01 -7.98349202e-01
2.10239589e-01 8.96753311e-01 -9.83617187e-01 -1.17474683e-01
-2.38270149e-01 -9.38233376e-01 5.69397628e-01 1.69979006e-01
2.79822201e-01 -1.16219318e+00 -2.03572765e-01 9.00898781e-03
-7.41028428e-01 -1.09900045e+00 -4.08770829e-01 -4.81730342e-01
-2.06907526e-01 -1.23058474e+00 -4.35416341e-01 -5.73113859e-01
6.87293828e-01 4.44691896e-01 1.35358942e+00 2.98239708e-01
-9.02687460e-02 5.15080512e-01 -4.13152099e-01 -2.82824755e-01
-6.48038149e-01 -1.86701283e-01 -2.52814740e-01 5.44294536e-01
2.91010559e-01 -3.06311220e-01 -6.03827178e-01 2.14112133e-01
-1.00827920e+00 4.77294058e-01 2.44662419e-01 7.62809753e-01
1.01047374e-01 -6.31202161e-01 8.00465405e-01 -1.29403305e+00
1.12394345e+00 -1.21471298e+00 3.42486650e-01 3.23684365e-01
-2.14344725e-01 -1.46736234e-01 5.31604767e-01 -4.72697258e-01
-1.05940568e+00 -3.67260665e-01 8.84914398e-02 -1.13059729e-01
-1.27238750e-01 2.17942968e-01 7.27143884e-02 2.80009270e-01
9.16310787e-01 6.33585826e-03 -1.14874206e-01 1.67633742e-02
6.90229177e-01 6.21765316e-01 3.99603516e-01 -5.08140922e-01
9.03621972e-01 7.15705335e-01 -1.95504144e-01 -9.24896300e-01
-9.67766404e-01 -4.79386955e-01 1.40613839e-01 -6.66448414e-01
1.01414156e+00 -8.48374367e-01 -9.95688796e-01 2.93578357e-01
-1.55586016e+00 -2.09401157e-02 3.00220072e-01 -3.31483811e-01
-1.82385057e-01 2.59011209e-01 -7.34160125e-01 -6.96740091e-01
-3.31253976e-01 -6.43147409e-01 1.22691095e+00 2.89884899e-02
-6.43339276e-01 -1.01710081e+00 -1.35032028e-01 8.89471471e-01
2.71387070e-01 7.91402459e-01 1.14332461e+00 -6.99999809e-01
-3.06274801e-01 1.40018553e-01 -6.53590977e-01 -2.67027617e-01
-1.10748604e-01 1.50751427e-01 -9.70422745e-01 1.57638296e-01
-5.65692246e-01 -8.57122064e-01 9.76487100e-01 -1.72990888e-01
1.07713294e+00 -9.64890540e-01 -2.84648269e-01 -3.27015333e-02
1.12241912e+00 -1.56636819e-01 6.54427826e-01 2.42408723e-01
8.65081787e-01 1.17381287e+00 6.08117044e-01 6.26463473e-01
6.36103272e-01 4.59965944e-01 6.60991251e-01 -2.98228979e-01
-1.14189185e-01 -7.25513220e-01 4.79433477e-01 3.69197249e-01
9.51174945e-02 -6.05145991e-01 -6.89946055e-01 4.77601230e-01
-2.00878835e+00 -1.29537284e+00 -6.33910000e-01 1.38715899e+00
6.97922528e-01 -1.97835296e-01 2.37387240e-01 -2.46749595e-01
9.91690099e-01 6.55030191e-01 -2.87371337e-01 -4.24125552e-01
-6.22775376e-01 -3.03457618e-01 -3.29275578e-02 5.72496891e-01
-1.16239059e+00 9.45917547e-01 5.90631056e+00 8.12604189e-01
-9.24377799e-01 3.56837898e-01 9.33285058e-01 -4.19287756e-02
-1.17950046e+00 -3.08566511e-01 -5.74086308e-01 5.21867692e-01
4.76216376e-01 1.90994143e-02 3.24249595e-01 6.30225539e-01
2.19759032e-01 1.60137400e-01 -9.01233733e-01 1.27550685e+00
5.89387238e-01 -1.70115578e+00 5.09251714e-01 -2.89147228e-01
7.67543077e-01 -6.16528034e-01 3.96120787e-01 2.11057439e-01
-3.60750430e-03 -1.17532969e+00 1.12281263e+00 2.31796518e-01
7.82814801e-01 -5.75279236e-01 4.45136607e-01 -1.80150717e-01
-7.90072560e-01 -1.56391516e-01 -1.44615933e-01 -2.33085811e-01
2.35061571e-01 3.66456002e-01 -1.03220129e+00 -9.91247818e-02
1.35425881e-01 8.57170880e-01 -8.91274035e-01 3.27399284e-01
-6.29296482e-01 4.42148864e-01 2.29286313e-01 -6.59450710e-01
4.07811970e-01 1.40313819e-01 6.93642676e-01 1.59773517e+00
1.80275757e-02 1.87465265e-01 -6.61005080e-03 1.07066572e+00
-5.13843238e-01 3.69098514e-01 -7.18300343e-01 -6.10408127e-01
4.79534656e-01 1.42382848e+00 -9.15350616e-01 -3.24337542e-01
-6.05054758e-02 1.28717494e+00 2.67260492e-01 4.35276866e-01
-1.16938150e+00 -5.99671118e-02 6.15807474e-01 3.46881151e-01
-5.73479712e-01 2.96081603e-01 -4.31227088e-01 -1.10368943e+00
-3.42952490e-01 -8.46844077e-01 5.43955207e-01 -1.10512149e+00
-1.81896627e+00 5.77310324e-01 -6.21018633e-02 -8.49719644e-01
-2.79596187e-02 -6.04400694e-01 -6.23841703e-01 4.19519186e-01
-1.22308612e+00 -1.73127878e+00 -4.33454931e-01 4.80524540e-01
7.05712736e-01 1.53027829e-02 4.93434042e-01 1.65956348e-01
-3.95003408e-01 5.08023143e-01 -5.91696441e-01 -8.03121030e-02
8.70819211e-01 -1.24804819e+00 2.10856542e-01 5.63178122e-01
1.39467344e-01 4.55595613e-01 1.05491185e+00 -7.49430835e-01
-1.20300901e+00 -1.06208253e+00 1.08594322e+00 -8.23100507e-01
1.11644363e+00 -6.46308303e-01 -6.65451527e-01 5.05205154e-01
5.32487094e-01 -7.07755506e-01 9.09261882e-01 1.94032729e-01
-7.10706294e-01 3.88652623e-01 -8.95474136e-01 1.01888585e+00
1.19132733e+00 -7.72003531e-01 -5.90433657e-01 7.90164232e-01
9.26555336e-01 9.06130672e-02 -3.24015498e-01 -1.68148190e-01
3.67897481e-01 -9.79019403e-01 1.06853175e+00 -1.04032290e+00
1.40717638e+00 -7.90839568e-02 -7.48853199e-03 -1.24594116e+00
-2.27402255e-01 -7.38686860e-01 2.76201870e-02 1.50125492e+00
7.59587884e-01 -4.48018312e-01 4.64869529e-01 7.33092248e-01
-3.94420847e-02 -4.03965712e-01 -5.23693085e-01 3.50866094e-02
-4.44817513e-01 -3.63799185e-01 5.36960304e-01 1.39600432e+00
1.20436110e-01 1.01050425e+00 -8.65533710e-01 -6.18485548e-02
5.50276697e-01 2.13320211e-01 9.15894151e-01 -1.01267409e+00
-7.02016354e-02 -6.20499253e-01 -1.71864167e-01 -5.11004746e-01
5.25334954e-01 -9.49183106e-01 -2.62670189e-01 -1.71761858e+00
7.30937421e-01 -8.88687819e-02 1.94595709e-01 6.85590982e-01
-4.18347478e-01 7.92903066e-01 3.89007896e-01 7.75580034e-02
-7.97606766e-01 5.01064956e-01 1.40601683e+00 -5.19749403e-01
1.72955561e-02 -5.36522150e-01 -1.31739044e+00 7.22175241e-01
5.79145849e-01 -2.19823048e-01 -5.27268946e-01 -2.68237382e-01
1.01099825e+00 -3.24899219e-02 9.99585390e-01 -1.80243626e-01
-2.54875481e-01 -4.76447195e-01 1.59997538e-01 -1.06703833e-01
5.70255458e-01 -2.82421619e-01 1.96249381e-01 1.50704131e-01
-8.02334607e-01 6.58896640e-02 6.37629777e-02 7.82838762e-01
-2.39896178e-01 9.54796746e-02 5.78157246e-01 5.23926653e-02
-9.59242880e-01 3.06935553e-02 -4.40992326e-01 3.32353354e-01
1.04811525e+00 -2.71746874e-01 -1.07635462e+00 -8.41879070e-01
-6.29241824e-01 2.00248882e-01 3.33339363e-01 9.53649938e-01
9.00565743e-01 -1.63804305e+00 -1.08629847e+00 -4.30965304e-01
5.39314389e-01 -9.47176695e-01 5.28344870e-01 4.83560920e-01
-2.69813985e-01 -1.53526559e-01 -1.15562044e-01 -1.63396403e-01
-1.27352619e+00 5.11557877e-01 -1.23818181e-01 2.36099079e-01
-3.42696458e-01 9.85172093e-01 5.95091641e-01 -3.10216486e-01
-2.17987314e-01 3.35452735e-01 -6.45746946e-01 7.88934946e-01
7.36478984e-01 3.43807966e-01 -5.01666605e-01 -1.00635397e+00
-2.86890477e-01 1.29581928e-01 3.09804261e-01 -5.82406111e-02
1.05459833e+00 -3.09020668e-01 -5.01165628e-01 5.01299739e-01
1.14383256e+00 2.71529853e-01 -8.06256533e-01 3.36395085e-01
-6.17746115e-02 -6.31315410e-01 -4.64047223e-01 -1.11834311e+00
-6.51015818e-01 8.33559990e-01 4.41586711e-02 6.78378105e-01
7.28466451e-01 5.40759623e-01 9.17627096e-01 -2.35010963e-02
-6.30977601e-02 -9.57003236e-01 7.89390445e-01 2.03292623e-01
1.54584432e+00 -1.48944294e+00 -3.71319681e-01 -6.54161930e-01
-1.21443129e+00 9.25547898e-01 8.78007114e-01 3.76796633e-01
2.46186480e-02 -2.70087808e-01 7.89742321e-02 -7.07147479e-01
-8.49802673e-01 -5.91165349e-02 5.38641095e-01 4.49379146e-01
5.05577445e-01 3.49295020e-01 -4.13669020e-01 7.34622002e-01
-2.56518811e-01 -8.56878221e-01 6.99642658e-01 2.47595340e-01
-3.27449590e-01 -4.86087114e-01 -6.27104640e-01 2.85260350e-01
-3.46581519e-01 -1.71502665e-01 -1.32471383e+00 6.57206774e-01
2.13345096e-01 1.09935892e+00 -9.33136232e-03 -4.91807908e-01
2.79289298e-02 -4.49496321e-02 4.29111719e-02 -4.99521285e-01
-7.20963299e-01 -2.68523842e-01 6.03081346e-01 -3.74617934e-01
-5.57485342e-01 -2.40603268e-01 -1.30722916e+00 -6.42782688e-01
2.96395421e-01 -2.14155223e-02 4.81878310e-01 7.75408387e-01
3.94045651e-01 2.79131293e-01 5.59317768e-01 -5.93420804e-01
-9.09710974e-02 -7.27353096e-01 -2.07733735e-01 1.16692698e+00
2.41558135e-01 -7.49607980e-01 -4.79079068e-01 -2.97183748e-02] | [8.517391204833984, 10.659186363220215] |
71a8501f-6167-4e4b-a6b2-98544a2c7a86 | semi-supervised-object-detection-based-on | 2210.10998 | null | https://arxiv.org/abs/2210.10998v1 | https://arxiv.org/pdf/2210.10998v1.pdf | Semi-supervised object detection based on single-stage detector for thighbone fracture localization | The thighbone is the largest bone supporting the lower body. If the thighbone fracture is not treated in time, it will lead to lifelong inability to walk. Correct diagnosis of thighbone disease is very important in orthopedic medicine. Deep learning is promoting the development of fracture detection technology. However, the existing computer aided diagnosis (CAD) methods baesd on deep learning rely on a large number of manually labeled data, and labeling these data costs a lot of time and energy. Therefore, we develop a object detection method with limited labeled image quantity and apply it to the thighbone fracture localization. In this work, we build a semi-supervised object detection(SSOD) framework based on single-stage detector, which including three modules: adaptive difficult sample oriented (ADSO) module, Fusion Box and deformable expand encoder (Dex encoder). ADSO module takes the classification score as the label reliability evaluation criterion by weighting, Fusion Box is designed to merge similar pseudo boxes into a reliable box for box regression and Dex encoder is proposed to enhance the adaptability of image augmentation. The experiment is conducted on the thighbone fracture dataset, which includes 3484 training thigh fracture images and 358 testing thigh fracture images. The experimental results show that the proposed method achieves the state-of-the-art AP in thighbone fracture detection at different labeled data rates, i.e. 1%, 5% and 10%. Besides, we use full data to achieve knowledge distillation, our method achieves 86.2% AP50 and 52.6% AP75. | ['Shaoquan Wang', 'Yueming Zhang', 'Bin Guan', 'Guoshan Zhanga', 'Jinkun Yao', 'Jinman Wei'] | 2022-10-20 | null | null | null | null | ['semi-supervised-object-detection', 'image-augmentation'] | ['computer-vision', 'computer-vision'] | [ 4.74219210e-03 1.75138086e-01 -4.65891242e-01 -2.06517592e-01
-9.07618761e-01 3.83118987e-01 -2.31770352e-01 -1.62307113e-01
-6.26136005e-01 6.07967556e-01 -1.28131593e-02 3.84170637e-02
5.83157875e-02 -8.94197762e-01 -5.32065570e-01 -7.79850483e-01
-5.46246171e-02 7.09020853e-01 7.45864749e-01 -6.94389939e-02
-2.37212449e-01 2.55949527e-01 -1.14781857e+00 4.60518390e-01
7.40277708e-01 1.31841063e+00 2.14171484e-01 3.38116854e-01
1.84239119e-01 8.54638577e-01 -4.44961935e-01 -1.65956154e-01
2.24267140e-01 -2.18135774e-01 -6.10548496e-01 2.79743522e-01
-2.63807923e-01 -7.21905589e-01 -5.60079396e-01 8.39640021e-01
9.28021729e-01 -1.86036810e-01 7.26870239e-01 -1.04747140e+00
-3.58326226e-01 5.51544189e-01 -9.22622323e-01 5.06489217e-01
-1.97116897e-01 2.73821533e-01 4.28280383e-01 -1.06395721e+00
4.08944130e-01 1.13653004e+00 8.46787393e-01 4.98388320e-01
-8.75170052e-01 -1.01756978e+00 -4.29197073e-01 4.10034657e-01
-1.64035106e+00 -1.02504283e-01 7.50063062e-01 -3.84309173e-01
4.50555712e-01 -9.53873806e-03 7.55390108e-01 8.43874872e-01
6.30476236e-01 8.24452400e-01 1.00004613e+00 -3.62994760e-01
1.72639713e-01 -2.79358685e-01 1.04163177e-01 1.08122873e+00
3.39414507e-01 2.94411898e-01 -3.07593346e-01 2.19009817e-01
1.06539750e+00 3.56482625e-01 -1.15681358e-01 -3.29831511e-01
-8.10264349e-01 9.85265315e-01 1.00581431e+00 5.16900942e-02
-3.51613224e-01 8.19808021e-02 5.86075842e-01 -1.42144249e-03
2.30526447e-01 -2.55051106e-01 -2.24824890e-01 2.12713748e-01
-7.85213888e-01 1.54611766e-01 2.26870313e-01 8.24915051e-01
4.17392105e-01 -1.39345154e-01 -2.65135765e-01 9.41578865e-01
5.53130329e-01 5.86988688e-01 8.81771147e-01 -5.19275308e-01
3.13163012e-01 9.79797840e-01 -2.85991848e-01 -7.56821752e-01
-7.13448107e-01 -5.35121441e-01 -9.73130405e-01 4.12325799e-01
6.90977424e-02 -3.35660465e-02 -1.51217055e+00 1.23765790e+00
4.46774393e-01 -2.05486745e-01 -3.42983663e-01 1.24059927e+00
1.08566248e+00 3.26843172e-01 1.66830763e-01 -2.54740447e-01
1.83467770e+00 -1.08178222e+00 -7.08036840e-01 -1.25977173e-01
8.53213847e-01 -5.60980678e-01 1.08248544e+00 3.47690701e-01
-8.41426194e-01 -6.97545707e-01 -1.38241041e+00 -1.88172847e-01
-2.62146443e-03 4.82462376e-01 4.79695112e-01 5.95941126e-01
-2.37200856e-01 2.83010900e-01 -1.22495949e+00 -2.68225782e-02
9.22990203e-01 6.52266324e-01 -4.22234088e-01 -3.21213871e-01
-1.16972077e+00 8.70861053e-01 5.19760787e-01 1.99095115e-01
-9.91744637e-01 -4.09266442e-01 -7.23532140e-01 -1.47164628e-01
4.49401617e-01 -7.58821130e-01 1.25176597e+00 -5.29620469e-01
-1.18128788e+00 8.85378957e-01 4.11463916e-01 -3.55651915e-01
6.33383453e-01 -3.47582668e-01 -4.73366112e-01 3.25328767e-01
3.02744895e-01 6.82647824e-01 5.36611676e-01 -8.91315758e-01
-6.97021604e-01 -8.66458356e-01 -2.45045781e-01 2.47244716e-01
-2.88256437e-01 -8.96115974e-02 -4.36324149e-01 -6.57724738e-01
4.25585032e-01 -8.72823656e-01 -2.73790151e-01 4.89472032e-01
-3.11312288e-01 -3.54598612e-01 8.60736370e-01 -8.01497996e-01
1.32146728e+00 -2.17481375e+00 -2.03531682e-02 2.36059666e-01
3.31608206e-01 6.59123287e-02 4.71666306e-01 -1.11229815e-01
-5.33870384e-02 -1.33276761e-01 -4.52577710e-01 -1.59858733e-01
-3.38560015e-01 4.75068003e-01 4.78119880e-01 5.07605255e-01
2.93698519e-01 6.80725873e-01 -5.95970690e-01 -1.23020661e+00
1.44402087e-01 3.51019919e-01 -7.43309736e-01 1.34690791e-01
2.17365101e-01 1.94221660e-01 -5.61243653e-01 1.02088726e+00
7.13546336e-01 -3.90703052e-01 -2.58621752e-01 -3.63230616e-01
3.66857529e-01 -1.34011000e-01 -1.45849407e+00 2.00421214e+00
-1.23152345e-01 -5.68328500e-02 -1.71400473e-01 -8.23407650e-01
9.24106836e-01 2.47301817e-01 7.36377895e-01 -5.49147069e-01
6.41573787e-01 5.19158959e-01 2.11440220e-01 -9.31245863e-01
-2.54606754e-01 -5.70948720e-01 6.55819327e-02 1.29999265e-01
1.20336331e-01 1.68942109e-01 1.65740624e-01 -4.47409511e-05
1.17771208e+00 1.04963914e-01 3.63176584e-01 -1.45780995e-01
5.05452156e-01 1.09237656e-01 7.15833664e-01 1.71318665e-01
-4.17046487e-01 7.53736675e-01 9.33796465e-02 -5.67015290e-01
-9.53326881e-01 -1.34626877e+00 -4.37371641e-01 5.76133847e-01
3.10391635e-01 -1.21625751e-01 -6.47567332e-01 -8.63202989e-01
8.98602456e-02 -1.03099287e-01 -7.29314804e-01 -7.66000032e-01
-7.74397016e-01 -9.41591322e-01 4.54391003e-01 1.10828316e+00
9.06739056e-01 -1.11685312e+00 -7.35217750e-01 1.43724158e-01
-1.51917189e-01 -6.67803645e-01 -4.32738006e-01 2.01452985e-01
-1.05154848e+00 -1.19496882e+00 -9.89240944e-01 -1.12290525e+00
6.26540720e-01 -5.74967414e-02 6.01215541e-01 2.67041296e-01
-8.18869770e-01 -2.68142015e-01 -3.42591852e-01 -3.78821820e-01
-1.21836744e-01 1.32704377e-01 -4.48479364e-03 -2.29871571e-01
4.75434214e-01 -3.24336737e-01 -1.03760076e+00 3.84529799e-01
-7.61799037e-01 1.65007472e-01 1.26930761e+00 1.20041287e+00
7.54770041e-01 1.56665578e-01 3.30168843e-01 -7.40920901e-01
2.39659846e-01 -2.92268932e-01 1.13580458e-01 -2.82003009e-03
-8.63522887e-01 -1.32092312e-01 1.56665176e-01 -5.21472275e-01
-9.48975682e-01 2.24373668e-01 -4.43967968e-01 -3.48088026e-01
8.88596624e-02 5.52209020e-01 -1.22659363e-01 -4.17998657e-02
8.71169329e-01 5.44628762e-02 3.18704009e-01 -5.03073931e-01
1.07711134e-02 9.61918354e-01 6.18497789e-01 -4.45079982e-01
4.17638689e-01 4.42515552e-01 3.08490731e-02 -3.08086544e-01
-6.74342453e-01 -5.43883860e-01 -4.53349233e-01 -4.11757976e-01
9.01109099e-01 -1.03173387e+00 -5.31451106e-01 2.26972893e-01
-6.78022861e-01 4.48404327e-02 -4.48676258e-01 8.09732020e-01
-5.46328902e-01 3.97556514e-01 -9.45716381e-01 -5.18557131e-01
-7.27319896e-01 -1.43809235e+00 1.26898491e+00 3.12011719e-01
7.52462000e-02 -1.51177689e-01 -1.08389936e-01 5.86438954e-01
-1.49117365e-01 2.77518898e-01 8.30046356e-01 -2.66712070e-01
-2.48198539e-01 -3.83505136e-01 -4.45047528e-01 3.57682168e-01
8.35211948e-02 -4.99961644e-01 -6.35636866e-01 -2.19392329e-01
1.66329369e-01 -4.98188317e-01 9.69615936e-01 4.46436226e-01
1.13992226e+00 8.53548348e-02 -5.38222373e-01 4.89028990e-01
1.42490637e+00 3.94398004e-01 7.65852988e-01 4.96552050e-01
5.93830824e-01 1.98779643e-01 8.35007131e-01 4.01940465e-01
2.66025156e-01 5.39842129e-01 5.29357255e-01 -3.58199686e-01
-5.08732498e-01 -7.12457672e-02 6.01292551e-02 6.81460619e-01
-3.44821274e-01 1.14742652e-01 -1.06199908e+00 3.33425283e-01
-1.72284126e+00 -5.38560390e-01 -2.04150349e-01 1.77305639e+00
1.11860466e+00 6.34224355e-01 2.67404407e-01 7.05324769e-01
6.21827006e-01 -5.02040923e-01 -5.87004900e-01 3.72203171e-01
3.71553540e-01 4.89230067e-01 4.98474747e-01 -1.37054864e-02
-1.23462474e+00 4.89260286e-01 5.42756367e+00 1.10721660e+00
-9.42629755e-01 4.33456630e-01 6.56093240e-01 -1.23661004e-01
3.91261011e-01 -3.97367835e-01 -5.88673353e-01 6.30636930e-01
4.25771475e-01 2.82611907e-01 -2.05006391e-01 1.11949503e+00
1.08290292e-01 -1.49915561e-01 -7.85267651e-01 1.06289816e+00
-3.80842835e-02 -9.64180589e-01 -7.78960856e-03 -4.18177340e-03
1.49681598e-01 -2.61440545e-01 -9.93954912e-02 4.85982031e-01
3.47471833e-02 -9.36216533e-01 3.75300437e-01 2.99816668e-01
8.88173223e-01 -8.20301175e-01 1.28405261e+00 2.13086501e-01
-1.17285550e+00 -2.86112875e-01 -3.96484703e-01 9.52886790e-02
2.04634875e-01 4.34055775e-01 -7.03350842e-01 5.66044807e-01
9.70169306e-01 4.43085700e-01 -4.59452659e-01 1.15921903e+00
-1.14265710e-01 5.31672955e-01 -4.62329090e-01 1.25226781e-01
-5.33338934e-02 1.75123319e-01 8.79109576e-02 7.97430336e-01
1.63337409e-01 4.10111934e-01 5.85670173e-01 3.80384624e-01
-1.39230758e-01 3.79031330e-01 -1.71575509e-02 3.65388274e-01
2.40209207e-01 1.08115268e+00 -8.67629290e-01 -2.64740795e-01
-3.84384096e-01 7.56817639e-01 1.39318943e-01 -2.54096270e-01
-1.08289289e+00 -2.32142717e-01 -1.26577884e-01 4.71517354e-01
2.20000800e-02 9.65694115e-02 -4.84806180e-01 -7.61633813e-01
2.13321112e-02 -6.97471619e-01 8.11549664e-01 -7.57827044e-01
-1.06646979e+00 2.98230767e-01 -1.96200132e-01 -1.27661419e+00
3.25073540e-01 -7.67853558e-01 -1.80895954e-01 3.69499683e-01
-1.14891207e+00 -1.37900531e+00 -6.36456132e-01 5.35863996e-01
6.40899241e-01 -3.09114903e-02 6.14914656e-01 7.68644035e-01
-8.00845921e-01 5.67906737e-01 -2.58233547e-01 5.71496665e-01
6.60321891e-01 -1.00948489e+00 -3.47845018e-01 5.13062775e-01
-2.31742173e-01 2.23591253e-01 4.18675691e-01 -7.43787110e-01
-9.99455452e-01 -9.55894053e-01 3.56881022e-01 -5.98126277e-03
2.32356474e-01 7.03839287e-02 -6.50024533e-01 5.00049770e-01
-4.05341834e-01 4.83495891e-01 6.40096366e-01 -3.28879684e-01
2.24428996e-02 -2.98051834e-01 -1.39726722e+00 3.50665659e-01
1.17101038e+00 6.27681166e-02 -6.22391760e-01 4.00912166e-01
7.03330457e-01 -7.64850259e-01 -1.21339250e+00 1.05929089e+00
7.22339809e-01 -6.51906252e-01 1.07761931e+00 -5.05011559e-01
6.82584882e-01 -3.84569138e-01 -1.74864724e-01 -5.19523084e-01
-4.54407662e-01 3.08917344e-01 -2.38662332e-01 7.77706265e-01
1.01704068e-01 -1.34277046e-01 1.28797138e+00 2.95005411e-01
-6.26813531e-01 -1.41615963e+00 -1.17655063e+00 -7.83796310e-01
-2.00749990e-02 -1.33561730e-01 2.74885684e-01 5.41137576e-01
-4.04705964e-02 3.23045135e-01 -1.82571426e-01 2.73390673e-02
4.95371103e-01 -1.38802230e-01 5.11534750e-01 -1.21120620e+00
-2.73435116e-01 -1.68986529e-01 -8.67050111e-01 -8.46221268e-01
-6.81824744e-01 -7.26948142e-01 1.38983831e-01 -1.56543183e+00
4.99634713e-01 -4.29416329e-01 -6.19679391e-01 5.38657844e-01
-1.25160143e-01 6.78431153e-01 -2.63164163e-01 2.97351420e-01
-3.14896792e-01 6.76908314e-01 1.57152259e+00 -4.04362082e-01
-3.92415449e-02 -2.16776412e-02 -2.32707962e-01 7.66447008e-01
5.96213639e-01 -7.00097859e-01 -3.24471414e-01 -2.55674154e-01
-2.98755229e-01 1.72526520e-02 3.92800838e-01 -1.36851835e+00
1.42218083e-01 3.87518853e-02 7.89197385e-01 -7.90605009e-01
4.25913095e-01 -9.33920145e-01 -4.70343046e-02 1.33529902e+00
8.96452144e-02 -1.64549336e-01 -4.50513251e-02 5.97653627e-01
-7.44260028e-02 -1.91100225e-01 1.08537328e+00 -3.16879600e-01
-7.35077918e-01 5.82624853e-01 -1.88928470e-01 7.95890465e-02
1.27044237e+00 -3.76035333e-01 3.79976556e-02 3.57386231e-01
-1.10672510e+00 2.98396468e-01 -4.39602584e-02 2.12368920e-01
8.72013032e-01 -1.64821374e+00 -6.98582351e-01 1.19788080e-01
2.97957510e-01 4.67661917e-01 4.62729037e-01 9.87443507e-01
-9.36257184e-01 3.62941809e-02 -3.44016135e-01 -9.86704290e-01
-1.25104535e+00 5.26323855e-01 3.71021897e-01 -4.80594397e-01
-9.92498577e-01 9.19937372e-01 -3.67679931e-02 -1.68708805e-02
2.16678336e-01 -4.10890043e-01 -2.03685462e-01 -2.62304723e-01
4.45806593e-01 3.70553792e-01 1.82667911e-01 -4.33964670e-01
-4.67769653e-01 5.13464093e-01 -3.21101755e-01 1.58031017e-01
1.12815082e+00 2.53698617e-01 3.40528637e-01 1.92649484e-01
1.03199160e+00 -5.39078295e-01 -9.57045078e-01 -2.46870354e-01
-2.64456481e-01 -3.04175586e-01 2.53394514e-01 -7.59330750e-01
-1.20712960e+00 8.66208792e-01 1.44408858e+00 -3.81084859e-01
1.06600881e+00 2.08268866e-01 1.24073374e+00 2.13503793e-01
5.21484494e-01 -1.02453792e+00 4.49927479e-01 -2.95130372e-01
8.48180592e-01 -1.42775786e+00 3.82916301e-01 -6.88772023e-01
-4.22900110e-01 1.05728221e+00 1.03589058e+00 -4.12214786e-01
9.25472379e-01 3.91986370e-01 -1.84416752e-02 -3.94026995e-01
-1.95841998e-01 -1.74185321e-01 1.12012178e-01 5.25445640e-01
2.02155054e-01 1.52032256e-01 -7.73324907e-01 1.09929681e+00
-9.38482881e-02 5.26424050e-01 3.71828340e-02 1.14765465e+00
-7.58275032e-01 -8.02704036e-01 -4.08762813e-01 7.71916628e-01
-4.63183880e-01 2.43103653e-01 1.79516643e-01 1.10735273e+00
6.79724514e-01 6.47841096e-01 -1.23381838e-01 -6.30914509e-01
3.88652027e-01 -8.01193789e-02 4.93439555e-01 -7.03687370e-01
-2.31494531e-01 2.41163060e-01 -1.29917905e-01 -2.34755844e-01
-3.35107952e-01 -4.60467666e-01 -1.78811038e+00 -2.73082312e-02
-7.68729746e-01 -9.56368074e-02 2.78057337e-01 9.96944964e-01
-1.31705895e-01 9.45814371e-01 3.88049185e-01 -3.70502383e-01
-6.99029267e-01 -1.29834342e+00 -7.46024907e-01 4.13789630e-01
-3.24900821e-02 -1.33390200e+00 1.46633431e-01 1.62029788e-01] | [14.926115036010742, -2.241628885269165] |
1fb86b69-b7dc-4ce9-8f20-7de76e212ee6 | cross-lingual-character-level-neural | 1708.09157 | null | https://arxiv.org/abs/1708.09157v3 | https://arxiv.org/pdf/1708.09157v3.pdf | Cross-lingual, Character-Level Neural Morphological Tagging | Even for common NLP tasks, sufficient supervision is not available in many languages -- morphological tagging is no exception. In the work presented here, we explore a transfer learning scheme, whereby we train character-level recurrent neural taggers to predict morphological taggings for high-resource languages and low-resource languages together. Learning joint character representations among multiple related languages successfully enables knowledge transfer from the high-resource languages to the low-resource ones, improving accuracy by up to 30% | ['Ryan Cotterell', 'Georg Heigold'] | 2017-08-30 | null | null | null | null | ['morphological-tagging'] | ['natural-language-processing'] | [ 1.08586468e-01 8.81944671e-02 -4.84680772e-01 -8.82496014e-02
-1.02898037e+00 -7.24759817e-01 1.41107559e-01 3.63292962e-01
-8.97927105e-01 1.00451839e+00 3.66230339e-01 -5.49872220e-01
2.26840615e-01 -7.05965996e-01 -5.06923735e-01 -3.12475294e-01
4.56866175e-02 6.22596204e-01 4.43279535e-01 -1.73201039e-02
-1.97139114e-01 3.46665204e-01 -7.56791413e-01 3.70418519e-01
9.13951814e-01 2.53299445e-01 5.43054879e-01 3.29990357e-01
-5.33600211e-01 5.44732094e-01 -2.97435403e-01 -8.40931118e-01
7.37291947e-02 -1.92737803e-01 -1.02256262e+00 -1.89400524e-01
6.24670647e-02 4.50876057e-01 -9.77844968e-02 1.11856210e+00
2.27521941e-01 -1.90403014e-02 5.70936680e-01 -1.60638571e-01
-1.03169191e+00 1.03357470e+00 -3.88284624e-01 3.33607197e-01
3.12641591e-01 -2.35933870e-01 1.32195103e+00 -9.32512879e-01
9.18815613e-01 9.94290709e-01 8.17686737e-01 5.76995432e-01
-1.28807104e+00 -4.16664451e-01 1.37688145e-01 -1.20198928e-01
-1.55076170e+00 -3.71882260e-01 6.06607556e-01 -2.18346030e-01
1.71531284e+00 -3.60498399e-01 4.11598921e-01 9.49372768e-01
-1.07348800e-01 6.89562976e-01 1.11156631e+00 -7.43730664e-01
-2.93339908e-01 2.50834703e-01 4.31224704e-02 9.06040490e-01
4.62232113e-01 -2.09780127e-01 -6.29178584e-01 -3.58427279e-02
8.30899596e-01 -3.59615862e-01 1.39325619e-01 7.64978826e-02
-1.12127662e+00 8.35797966e-01 1.61886245e-01 1.07582951e+00
-4.77410555e-01 -2.86621630e-01 6.28012896e-01 3.94975096e-01
8.81410182e-01 6.56623363e-01 -8.79722834e-01 -7.00248852e-02
-5.77997565e-01 -3.72375160e-01 7.35432744e-01 1.01476359e+00
9.75259125e-01 4.29150462e-01 1.54656023e-01 1.21192062e+00
2.21674576e-01 4.44822401e-01 8.66611898e-01 -5.25179841e-02
7.45202899e-01 5.47624946e-01 -1.97146833e-01 -4.07186300e-01
-3.91422629e-01 -2.52761722e-01 -5.81716001e-01 -5.21775365e-01
2.66347229e-01 -5.10058761e-01 -8.15497816e-01 1.74939179e+00
2.20041554e-02 1.23351835e-01 4.96094942e-01 4.06720817e-01
6.86483502e-01 8.23835373e-01 7.99066544e-01 -2.70792544e-01
1.48193002e+00 -7.47401416e-01 -5.22676826e-01 -5.29493749e-01
9.79417920e-01 -9.56121624e-01 1.00198150e+00 5.58212399e-02
-1.18063974e+00 -4.51025069e-01 -8.03690791e-01 -1.49739802e-01
-6.43493950e-01 4.02028233e-01 8.77906859e-01 5.91167867e-01
-1.26624584e+00 6.60540402e-01 -8.37042868e-01 -7.49901056e-01
6.91324770e-02 2.59124845e-01 -7.37418413e-01 -2.76017319e-02
-1.26194024e+00 1.22943842e+00 9.51900899e-01 -2.10445285e-01
-5.42738378e-01 -6.51032567e-01 -1.03987718e+00 -1.99386090e-01
1.37364417e-02 -3.42407107e-01 1.08082247e+00 -9.07504857e-01
-1.64932048e+00 1.52984464e+00 -1.70763023e-02 -5.56986928e-01
3.95756885e-02 -3.73051524e-01 -6.30091906e-01 -1.97239429e-01
-1.29935727e-01 6.24191761e-01 2.17248067e-01 -7.87671089e-01
-6.72905207e-01 -6.45112023e-02 -3.90530348e-01 2.45191514e-01
-5.82663715e-01 7.92626441e-01 -4.50836509e-01 -7.96059430e-01
-1.53455004e-01 -7.00017095e-01 -3.89709949e-01 -7.61237562e-01
-1.20575070e-01 -5.68915248e-01 3.51027489e-01 -1.05886042e+00
9.32291269e-01 -1.88210523e+00 2.48895422e-01 -1.27866790e-01
-5.02801299e-01 7.39475369e-01 -4.17577416e-01 4.53330249e-01
2.76997369e-02 1.94086522e-01 -2.89700836e-01 -4.43114698e-01
-2.25294620e-01 7.15136647e-01 -2.48345703e-01 3.49289060e-01
8.43576789e-01 9.67126608e-01 -9.64782894e-01 -4.88762170e-01
-1.46259961e-04 6.39641464e-01 -2.20877051e-01 3.58588934e-01
-9.67755094e-02 3.80741477e-01 -2.30218634e-01 6.14542365e-01
1.47291318e-01 2.06378132e-01 7.11203754e-01 3.65150839e-01
-3.19313377e-01 1.02034354e+00 -5.00801861e-01 1.90943766e+00
-1.03244746e+00 2.21387357e-01 -1.72598213e-01 -8.39320719e-01
1.14011860e+00 5.91033280e-01 2.58661836e-01 -5.05213916e-01
-6.04053028e-02 5.48649251e-01 4.75820117e-02 -2.63145953e-01
6.92485631e-01 -7.07402349e-01 -4.60650653e-01 2.42645428e-01
6.11683488e-01 1.22992210e-01 3.56359892e-02 -3.36268932e-01
1.06323159e+00 4.03861284e-01 8.19106340e-01 -2.18650892e-01
5.52657068e-01 -5.32811843e-02 8.77782226e-01 2.37934381e-01
-9.52871330e-03 2.29791328e-01 1.39080212e-01 -3.66152823e-01
-1.07749856e+00 -1.04805541e+00 -1.99496895e-01 1.68828261e+00
-5.56472480e-01 -3.60017061e-01 -3.83692175e-01 -8.86802137e-01
-1.77867129e-01 5.76578200e-01 -3.15340012e-01 1.16336092e-01
-9.60051000e-01 -7.55318105e-01 8.98120046e-01 7.61306763e-01
-2.51267087e-02 -1.69059265e+00 5.57979457e-02 4.00836587e-01
9.60923061e-02 -1.47423029e+00 -1.55751839e-01 7.23632157e-01
-9.92531598e-01 -5.66265464e-01 -8.38240027e-01 -1.42250049e+00
5.50847471e-01 -2.86136150e-01 1.22301400e+00 -3.79401594e-02
-4.53160070e-02 -2.65710503e-01 -4.62028116e-01 -3.12169254e-01
-7.83990145e-01 5.83083928e-01 4.51554656e-02 -1.92879081e-01
5.64732254e-01 -5.34352422e-01 3.52931827e-01 -1.47777691e-01
-5.40740192e-01 -2.62741745e-01 1.02376080e+00 4.42514002e-01
7.38363206e-01 -5.04925549e-01 5.81031024e-01 -1.34000015e+00
2.15792179e-01 -4.55760300e-01 -6.14630282e-01 3.61333132e-01
-1.64672524e-01 1.85269028e-01 8.54067564e-01 -5.07292032e-01
-1.15642786e+00 2.74897069e-01 -5.42616248e-01 -1.65148117e-02
-5.00478029e-01 7.34023273e-01 -4.07053441e-01 1.31025136e-01
4.12412971e-01 2.36731544e-01 -8.70458066e-01 -1.06366944e+00
5.07787108e-01 5.96378505e-01 6.48392677e-01 -6.80610597e-01
6.68353856e-01 -1.82198986e-01 -4.65251774e-01 -1.05460918e+00
-1.00776219e+00 -6.67787194e-01 -1.22254956e+00 4.77685124e-01
1.23108852e+00 -9.85901535e-01 1.30288256e-02 2.17073128e-01
-1.32822287e+00 -4.25077528e-01 -4.14825678e-01 5.97505331e-01
-3.31911892e-01 2.88480699e-01 -1.18235528e+00 -5.73943436e-01
-4.41210717e-01 -5.67972779e-01 8.25934827e-01 7.12157041e-03
-8.05791020e-02 -1.47603524e+00 5.73064089e-01 -1.59676507e-01
2.16349453e-01 -3.50553840e-01 1.21913850e+00 -1.23228788e+00
-1.25795364e-01 -2.00039789e-01 -1.49703205e-01 2.36586109e-01
2.27433220e-01 -2.89757133e-01 -7.63840377e-01 -2.19792902e-01
-5.91026068e-01 -5.31621158e-01 7.27026641e-01 -3.53256911e-02
6.16402328e-01 -1.26520842e-01 -3.02557290e-01 4.63056654e-01
1.35743761e+00 -2.15145156e-01 5.05119979e-01 2.42257744e-01
7.40983963e-01 5.51838040e-01 4.31158453e-01 4.73381430e-02
4.10285413e-01 4.89586979e-01 -3.42215776e-01 -8.23901743e-02
-3.17182153e-01 -5.73698997e-01 7.47594595e-01 1.79297185e+00
-2.33231694e-01 -7.09590763e-02 -1.10655499e+00 9.33422625e-01
-1.53640044e+00 -4.55680609e-01 8.16662535e-02 2.25411272e+00
1.31493914e+00 5.08559644e-02 -6.77248870e-04 -5.17118990e-01
8.64616334e-01 1.30920023e-01 -4.95600922e-04 -6.28717601e-01
-2.68381894e-01 5.81746042e-01 4.61467206e-01 5.46291709e-01
-1.34851456e+00 1.83373964e+00 7.04332161e+00 6.76627398e-01
-9.08531606e-01 5.28756082e-01 1.84781905e-02 4.81296659e-01
-2.54179031e-01 1.49582148e-01 -1.06833160e+00 8.68943632e-02
1.60983562e+00 -1.01750001e-01 6.28422871e-02 7.17331886e-01
-3.37549865e-01 3.78429502e-01 -7.54504502e-01 6.15989149e-01
1.04762085e-01 -8.35989892e-01 2.95031905e-01 7.53162652e-02
6.35623157e-01 6.79154456e-01 -2.60855556e-01 5.81094384e-01
1.12173676e+00 -8.70947838e-01 3.64427090e-01 2.40418777e-01
9.89249527e-01 -9.96896803e-01 8.15396190e-01 2.28719965e-01
-1.30358183e+00 4.82058257e-01 -8.48205984e-01 1.78422928e-01
2.76375651e-01 4.22602832e-01 -1.14070904e+00 8.00763607e-01
2.33090937e-01 7.32519329e-01 -6.20655656e-01 8.80546331e-01
-8.40074062e-01 1.07794082e+00 -2.02174202e-01 -1.13503583e-01
3.48915249e-01 -2.30958074e-01 3.47746938e-01 2.00103164e+00
2.59520233e-01 -1.62322611e-01 5.50813496e-01 2.60383129e-01
-4.90909070e-01 8.90853345e-01 -7.97929645e-01 -3.34354371e-01
1.48149297e-01 1.44154537e+00 -8.02041411e-01 -2.34251320e-01
-6.44654810e-01 1.19687808e+00 1.19467056e+00 1.17623089e-02
-3.50843877e-01 -4.35380757e-01 5.82314432e-01 -2.04020232e-01
5.20763338e-01 -5.40240943e-01 -5.48833907e-02 -1.31337512e+00
-3.52906376e-01 -5.34100890e-01 6.78351402e-01 -3.75331730e-01
-1.73982632e+00 8.48587573e-01 -4.20091599e-01 -7.83141553e-01
-5.07762671e-01 -8.85304153e-01 -2.98370242e-01 1.17679834e+00
-1.83482754e+00 -1.65503526e+00 7.81626284e-01 4.38395619e-01
3.43463778e-01 -5.86460412e-01 1.42881680e+00 3.39468837e-01
-4.51796949e-01 6.28805876e-01 8.42426121e-02 6.47847116e-01
7.57257819e-01 -1.35330594e+00 7.45165169e-01 9.56670403e-01
8.97965133e-01 6.42161369e-01 3.57889980e-01 -7.45334983e-01
-1.02745783e+00 -1.19289219e+00 1.73474038e+00 -2.88891971e-01
1.17693019e+00 -4.86878574e-01 -1.36995912e+00 9.79545295e-01
4.25478190e-01 2.14666307e-01 1.18296874e+00 7.85710752e-01
-8.53171110e-01 4.21235472e-01 -7.19945371e-01 5.56457400e-01
1.09535515e+00 -1.17334890e+00 -1.09826553e+00 3.40385348e-01
7.41759539e-01 -8.12962949e-02 -1.02463830e+00 2.22176343e-01
-2.78591625e-02 -1.54637486e-01 7.01202571e-01 -1.13062561e+00
1.68584716e-02 -9.42610130e-02 -1.40483364e-01 -1.43008828e+00
-4.72472131e-01 -5.04778862e-01 2.27537811e-01 1.66672325e+00
7.95444608e-01 -5.51560760e-01 5.95167160e-01 -9.52096432e-02
-2.71440029e-01 -6.88902810e-02 -1.14751351e+00 -9.45840955e-01
5.90263665e-01 -5.05238116e-01 4.33969796e-02 1.24755478e+00
2.98091441e-01 6.94673359e-01 -3.38296413e-01 2.96509683e-01
1.80569384e-02 4.01689336e-02 2.52379663e-02 -1.27955246e+00
-6.24174237e-01 -3.61051828e-01 -5.28681755e-01 -6.99057639e-01
1.08720934e+00 -1.59203601e+00 2.84902126e-01 -1.41233182e+00
3.83294791e-01 -4.48393524e-01 -6.41225457e-01 1.17489028e+00
-3.23365837e-01 4.94332701e-01 1.23268440e-01 2.77388487e-02
-7.20766902e-01 3.18270326e-01 4.66123521e-01 1.75653502e-01
-1.35299876e-01 -1.80787593e-01 -4.79830235e-01 9.23671603e-01
6.76015794e-01 -7.56911337e-01 2.66091049e-01 -6.89593613e-01
3.10401469e-01 -1.16315968e-01 -3.39132786e-01 -9.58697319e-01
-4.42895293e-02 4.30157222e-02 3.25104713e-01 -1.71325609e-01
1.23092808e-01 -4.99002725e-01 -2.54552931e-01 3.40514421e-01
-2.73608834e-01 2.23653525e-01 4.71399724e-01 2.89698422e-01
-2.04878077e-01 -6.97078288e-01 7.34371364e-01 -3.58427972e-01
-8.96187484e-01 2.65085071e-01 -6.75353050e-01 3.39656711e-01
5.39123952e-01 2.68690079e-01 -1.50796641e-02 1.11575253e-01
-7.42710054e-01 -3.92930716e-01 4.95933145e-01 5.30753911e-01
1.60792038e-01 -1.15298104e+00 -9.18981314e-01 7.58920088e-02
2.68212467e-01 -4.40168411e-01 -1.25729814e-01 3.45534652e-01
-2.77622879e-01 5.73325038e-01 -2.70820200e-01 1.69919282e-02
-1.13378227e+00 6.09760582e-01 8.37993845e-02 -7.66505837e-01
-6.45351589e-01 1.00974405e+00 -1.78713575e-01 -8.76483321e-01
-1.28514975e-01 5.41281365e-02 -3.79876435e-01 1.36248479e-02
3.51007670e-01 -2.90089399e-01 2.74493039e-01 -1.13699293e+00
-5.20533979e-01 4.64369416e-01 -2.55133420e-01 -1.61429316e-01
1.65808582e+00 1.10528536e-01 -1.63430184e-01 8.02006125e-01
1.07142282e+00 5.37729859e-01 -8.16959023e-01 -5.67644000e-01
8.72126758e-01 1.20474324e-01 -2.15440691e-01 -7.13601410e-01
-6.49800122e-01 1.05167067e+00 5.44349179e-02 -2.65877455e-01
7.83998072e-01 3.12576234e-01 7.96523035e-01 5.83435297e-01
6.23556256e-01 -1.08404362e+00 -3.52699637e-01 1.31443548e+00
4.32393849e-01 -9.50995684e-01 -2.06867456e-01 -3.45400870e-01
-8.28614235e-01 8.22703004e-01 3.32629263e-01 -2.54849255e-01
3.99639785e-01 4.41857219e-01 2.80612826e-01 5.68132363e-02
-7.27766693e-01 -7.52839684e-01 4.51440424e-01 8.89620662e-01
1.14160049e+00 2.20818043e-01 -3.24256182e-01 6.96246326e-01
-1.66912302e-01 -3.84312570e-01 1.76015690e-01 7.53086507e-01
-4.55061406e-01 -1.81800640e+00 4.09477316e-02 5.71449921e-02
-9.19809580e-01 -6.36566937e-01 -5.99257112e-01 7.24175274e-01
1.57545567e-01 6.80426836e-01 3.12990397e-01 -8.39767829e-02
2.72181243e-01 6.22615278e-01 3.90192717e-01 -1.28113234e+00
-9.83217359e-01 1.76337972e-01 4.44717169e-01 -1.66027129e-01
-5.40972412e-01 -8.11466515e-01 -1.38973057e+00 1.86712131e-01
-2.00255349e-01 3.07149172e-01 4.54956681e-01 9.40277398e-01
7.65960217e-02 2.61970043e-01 2.42644116e-01 -7.21492231e-01
-2.40987524e-01 -1.09246230e+00 -7.75618792e-01 2.11067572e-01
-2.52089083e-01 -2.80319721e-01 8.46274346e-02 5.99680818e-04] | [10.476880073547363, 9.930620193481445] |
637c564e-cc7c-4e25-87e4-8183e87dc6f9 | high-resolution-face-age-editing | 2005.04410 | null | https://arxiv.org/abs/2005.04410v1 | https://arxiv.org/pdf/2005.04410v1.pdf | High Resolution Face Age Editing | Face age editing has become a crucial task in film post-production, and is also becoming popular for general purpose photography. Recently, adversarial training has produced some of the most visually impressive results for image manipulation, including the face aging/de-aging task. In spite of considerable progress, current methods often present visual artifacts and can only deal with low-resolution images. In order to achieve aging/de-aging with the high quality and robustness necessary for wider use, these problems need to be addressed. This is the goal of the present work. We present an encoder-decoder architecture for face age editing. The core idea of our network is to create both a latent space containing the face identity, and a feature modulation layer corresponding to the age of the individual. We then combine these two elements to produce an output image of the person with a desired target age. Our architecture is greatly simplified with respect to other approaches, and allows for continuous age editing on high resolution images in a single unified model. | ['Gilles Puy', 'Yann Gousseau', 'Alasdair Newson', 'Xu Yao', 'Pierre Hellier'] | 2020-05-09 | null | null | null | null | ['face-age-editing'] | ['computer-vision'] | [ 6.59476876e-01 5.29876389e-02 2.23331392e-01 -3.99442762e-01
-2.60267466e-01 -2.50120252e-01 6.31312609e-01 -5.15925646e-01
-3.43140692e-01 6.90041959e-01 1.29411340e-01 2.28226796e-01
3.42712104e-01 -6.37977660e-01 -6.53814852e-01 -7.05773652e-01
1.33712992e-01 8.63025561e-02 -2.25133710e-02 -3.11765121e-03
1.43864125e-01 6.96323395e-01 -1.83019149e+00 2.30230510e-01
8.02810967e-01 1.04201996e+00 2.63273884e-02 8.86819184e-01
2.08529085e-01 7.07362592e-01 -7.92908549e-01 -8.93608332e-01
3.79974216e-01 -6.64998472e-01 -5.37451625e-01 3.15122008e-01
1.08773267e+00 -7.23435104e-01 -5.12151182e-01 1.11105061e+00
7.35896230e-01 2.96654589e-02 6.54469728e-01 -1.27704895e+00
-9.47999716e-01 9.96913984e-02 -7.05015600e-01 -1.21066973e-01
4.89967167e-01 -2.94403210e-02 5.52221298e-01 -6.95312023e-01
6.63136125e-01 1.39374828e+00 4.86251384e-01 1.10566676e+00
-1.23133755e+00 -7.42885053e-01 9.45639685e-02 2.02662438e-01
-1.33364582e+00 -9.41653848e-01 7.23551571e-01 -2.96536595e-01
5.34821808e-01 1.97796941e-01 8.01805377e-01 1.31882906e+00
2.28677660e-01 4.33526039e-01 1.18382370e+00 -2.93377608e-01
-5.16294427e-02 1.64979193e-02 -5.77701688e-01 7.43928969e-01
-1.89068355e-02 5.90785630e-02 -6.93483531e-01 7.11572990e-02
1.14652872e+00 -2.00287580e-01 -2.61722684e-01 -4.53204066e-01
-7.88163662e-01 4.92749065e-01 2.86026269e-01 -1.54724317e-02
-2.91620404e-01 3.46312016e-01 1.03765249e-01 4.86586690e-01
5.85235059e-01 4.16737258e-01 -2.03116331e-02 -9.81552452e-02
-1.13007331e+00 5.18198729e-01 5.72369277e-01 7.85996079e-01
4.02996957e-01 3.16670686e-01 -8.61840546e-02 9.79660869e-01
1.02073573e-01 5.98059058e-01 1.74584523e-01 -1.32441688e+00
9.95679572e-02 3.45017821e-01 4.50053019e-03 -9.60881889e-01
-1.10144705e-01 -1.68794245e-01 -1.02377617e+00 9.59376395e-01
4.18399602e-01 -8.36843103e-02 -1.03869772e+00 2.11071944e+00
2.72451282e-01 1.91157639e-01 -8.73467922e-02 7.43979752e-01
5.15157104e-01 6.38608158e-01 1.42043561e-01 -4.09256995e-01
1.31983280e+00 -6.89737022e-01 -1.02162814e+00 -3.59334350e-01
-3.92174236e-02 -9.10579145e-01 9.04473543e-01 4.23128992e-01
-1.49985957e+00 -5.39020538e-01 -1.19202924e+00 -3.90177011e-01
-2.06272215e-01 2.44294629e-01 5.08667946e-01 7.93703735e-01
-1.35701323e+00 7.50818968e-01 -5.15397310e-01 -3.07731390e-01
6.18519902e-01 5.87047040e-01 -6.87120438e-01 4.17860821e-02
-1.00687408e+00 1.12631953e+00 -7.75152370e-02 1.75909519e-01
-7.12906241e-01 -7.41865098e-01 -9.81152713e-01 -2.18509972e-01
2.40578175e-01 -9.76648092e-01 1.28393841e+00 -1.37988520e+00
-1.73663366e+00 1.14887059e+00 -1.30131751e-01 -3.58922094e-01
7.15837777e-01 -4.26125139e-01 -5.02118945e-01 1.37309819e-01
-1.35381222e-01 9.52483237e-01 1.55892897e+00 -1.28444850e+00
-4.95262444e-01 -4.70643342e-01 -1.99268349e-02 3.37335378e-01
-4.19924051e-01 3.95656228e-01 -6.28712833e-01 -9.22536254e-01
-1.61702201e-01 -8.17819953e-01 8.84825289e-02 7.07345724e-01
-6.26768321e-02 2.32504234e-01 1.00165081e+00 -1.08944142e+00
1.22928059e+00 -2.14781165e+00 5.57884455e-01 -1.42738447e-01
4.41954106e-01 3.93262446e-01 -6.50318414e-02 1.76589444e-01
-2.80442923e-01 -9.14350823e-02 -3.05337101e-01 -6.99601412e-01
-2.54457980e-01 -7.11651966e-02 -1.21999301e-01 5.84815085e-01
4.37216669e-01 7.14610159e-01 -6.55629575e-01 -6.00348592e-01
1.91359133e-01 7.18257189e-01 -6.06802881e-01 3.59045804e-01
-1.14618123e-01 4.41227406e-01 4.41433489e-02 5.08386791e-01
6.33181989e-01 2.03605086e-01 -5.52601041e-03 -4.14600909e-01
-5.90203777e-02 -2.17831314e-01 -1.12821484e+00 1.63147724e+00
-3.66644830e-01 6.40001595e-01 4.17852044e-01 -3.19235563e-01
8.00499201e-01 4.05297279e-01 4.84917462e-01 -6.87376857e-01
9.17342827e-02 1.17030159e-01 2.75738649e-02 -4.41339135e-01
6.59374595e-01 -2.62782931e-01 9.32689533e-02 3.66232574e-01
1.09041214e-01 -3.36255014e-01 1.73010036e-01 1.50294468e-01
6.63776577e-01 2.15974927e-01 1.70965716e-01 1.57709315e-01
5.76309979e-01 -6.97259665e-01 4.71938491e-01 1.08822584e-01
-2.11487263e-01 9.66246724e-01 5.72605014e-01 -3.37410718e-01
-1.48795354e+00 -1.11674428e+00 1.21171981e-01 6.76416636e-01
-5.46710864e-02 -2.12239251e-01 -9.91615713e-01 -4.78521347e-01
-1.94740713e-01 5.33352733e-01 -7.72395372e-01 -4.36410308e-01
-7.58515477e-01 -3.55924040e-01 4.90218133e-01 3.27783883e-01
5.53509295e-01 -1.13415194e+00 -3.62771958e-01 4.19205474e-03
4.83133085e-02 -1.09893680e+00 -7.05226898e-01 -4.49600309e-01
-5.33340394e-01 -8.03606033e-01 -1.11384380e+00 -7.86749899e-01
9.60539341e-01 -6.32066876e-02 9.41891551e-01 1.60916224e-01
-4.94157165e-01 2.39971593e-01 -8.18894655e-02 -4.51491773e-01
-6.81361616e-01 -8.05822387e-02 3.20665598e-01 4.41920161e-01
-2.35895529e-01 -8.33594739e-01 -6.90211177e-01 1.45174846e-01
-9.03539062e-01 2.43331820e-01 5.92429459e-01 7.24854946e-01
2.30797559e-01 1.54465809e-02 5.78311086e-01 -7.64926016e-01
6.30648851e-01 2.29299635e-01 -5.59525788e-01 6.56616315e-02
-4.06000733e-01 -3.64362821e-02 5.99135518e-01 -5.21323085e-01
-1.26620996e+00 1.61463559e-01 -3.66414815e-01 -5.02259731e-01
1.67817697e-01 -1.02143131e-01 -4.69471604e-01 -3.31313550e-01
5.88675559e-01 4.07484099e-02 3.50945979e-01 -3.80927444e-01
4.93253022e-01 6.35785460e-01 7.72305965e-01 -2.89484024e-01
1.03929806e+00 4.59075838e-01 8.60105678e-02 -8.52753460e-01
-6.07011557e-01 2.81708598e-01 -6.74523652e-01 -4.78425503e-01
8.45802069e-01 -8.37506235e-01 -6.53589725e-01 9.67514813e-01
-1.08490944e+00 -2.64365762e-01 -3.81899476e-01 1.11670807e-01
-6.35734200e-01 3.62237215e-01 -4.83077377e-01 -8.23561490e-01
-2.83546150e-01 -1.11272728e+00 1.10740614e+00 3.45448792e-01
-2.52786368e-01 -6.98352277e-01 -1.21391334e-01 2.13938326e-01
4.15344954e-01 3.34534675e-01 7.50917017e-01 3.03692073e-01
-4.02853966e-01 -2.04511225e-01 -2.75564402e-01 6.40094757e-01
1.61892742e-01 3.00092280e-01 -1.10611379e+00 -4.24936801e-01
-3.66293713e-02 -3.60857964e-01 7.40638852e-01 2.95732826e-01
1.21709263e+00 -2.26786628e-01 -2.10935101e-02 8.16552460e-01
1.12467504e+00 2.18384996e-01 1.01351976e+00 -4.86595072e-02
7.81997740e-01 6.12969398e-01 4.41691369e-01 2.69179344e-01
5.76906875e-02 9.13060963e-01 3.92533779e-01 -2.53813416e-01
-4.95573103e-01 -2.54156142e-01 5.58573544e-01 5.19501090e-01
-3.53725702e-01 -4.74488661e-02 -4.97541070e-01 3.48543733e-01
-1.44056094e+00 -1.17421305e+00 3.05531025e-01 2.21969581e+00
9.27944481e-01 -1.02348782e-01 8.22988153e-02 1.96167484e-01
6.61193013e-01 3.41796637e-01 -5.23118496e-01 -6.41988397e-01
-1.16905510e-01 3.13020706e-01 3.19479138e-01 3.73714566e-01
-1.12008035e+00 9.33518291e-01 6.87880278e+00 5.52879989e-01
-1.20170522e+00 -1.49466008e-01 5.54982364e-01 -2.68273383e-01
5.14080711e-02 -2.92060077e-01 -6.34060264e-01 3.69512856e-01
6.77744627e-01 -2.67948806e-01 6.85142636e-01 5.86092353e-01
9.25263539e-02 -7.71184191e-02 -1.13383520e+00 1.29119086e+00
4.59126592e-01 -1.07204032e+00 1.23956487e-01 1.55320421e-01
6.84117794e-01 -8.34775448e-01 5.35706699e-01 -2.98599042e-02
-1.32922351e-01 -1.27935672e+00 8.67615998e-01 5.33951998e-01
1.52526450e+00 -7.33643651e-01 2.82049179e-01 -2.43302323e-02
-9.94324923e-01 -1.34757429e-01 -2.84115434e-01 4.99386936e-02
3.78822923e-01 3.09075594e-01 -2.65268385e-01 3.64235520e-01
5.56847394e-01 6.46442115e-01 -5.71117938e-01 8.16635072e-01
-3.52432847e-01 2.85684895e-02 -3.79631892e-02 4.85827208e-01
-4.46771413e-01 -1.64910749e-01 4.77025002e-01 6.88113332e-01
4.82087135e-01 -7.84242675e-02 -2.98266709e-01 6.11587465e-01
-5.15005469e-01 -2.97499038e-02 -6.85513377e-01 -9.30837616e-02
3.36280853e-01 1.18207610e+00 -3.43728721e-01 -3.81576359e-01
-3.11943769e-01 1.61246717e+00 5.39646149e-01 1.06924877e-01
-8.82693291e-01 -2.93350279e-01 1.04607701e+00 2.87901223e-01
8.92130658e-02 -2.76716113e-01 -6.99930117e-02 -1.09073317e+00
1.60775661e-01 -1.24530697e+00 -4.79137525e-02 -8.73723090e-01
-1.06044221e+00 5.89161456e-01 -1.11400202e-01 -1.01952517e+00
-3.37514102e-01 -5.76681018e-01 -4.64771926e-01 9.87456501e-01
-1.20925534e+00 -1.37937272e+00 -3.74991268e-01 5.32573700e-01
5.40352046e-01 -1.82583004e-01 9.92353320e-01 5.85322440e-01
-6.85920000e-01 8.81982863e-01 -3.00303191e-01 -4.77310345e-02
1.09807277e+00 -1.18745124e+00 5.84350765e-01 1.06818712e+00
-3.07335779e-02 4.90673304e-01 8.18913102e-01 -6.79856777e-01
-1.33571315e+00 -1.05880535e+00 8.82167101e-01 -3.48930925e-01
1.99887246e-01 -5.99331141e-01 -7.41659820e-01 6.07569158e-01
2.77648389e-01 -3.10849518e-01 2.55627275e-01 -1.78656191e-01
-4.01147187e-01 -3.08548003e-01 -1.20612466e+00 1.10215569e+00
1.21096063e+00 -6.18434250e-01 -3.41113538e-01 -3.84498388e-02
4.83683467e-01 -5.08422792e-01 -8.58155549e-01 2.06416383e-01
8.16910446e-01 -1.08213770e+00 1.11734796e+00 -1.25703797e-01
5.79123914e-01 -3.14781189e-01 3.10796589e-01 -1.34560418e+00
-3.33145976e-01 -9.23990190e-01 -3.25930744e-01 1.43961251e+00
1.11581489e-01 -3.52975845e-01 8.64852726e-01 6.08844459e-01
-1.04058599e-02 -5.58952510e-01 -8.14011216e-01 -5.17958760e-01
-1.19704053e-01 -5.44028729e-02 6.17884517e-01 6.19464040e-01
-5.24851620e-01 3.50822538e-01 -1.02019739e+00 9.79413688e-02
7.82015145e-01 -2.22141266e-01 9.16421294e-01 -1.06584585e+00
-9.33555141e-02 -3.96965206e-01 -6.09088480e-01 -8.25440824e-01
1.51270300e-01 -4.39570844e-01 -1.08295076e-01 -1.35250473e+00
6.33807406e-02 -3.00649609e-02 2.29190901e-01 2.42244199e-01
-2.70451784e-01 6.39652550e-01 2.88946331e-01 -2.53005661e-02
-4.49137017e-02 6.14232004e-01 1.54695964e+00 9.12440047e-02
-3.20298895e-02 -9.49835218e-03 -7.78712511e-01 6.58500135e-01
9.01075542e-01 -1.70116141e-01 -4.00394976e-01 -4.41178381e-01
2.10147589e-01 -8.30165520e-02 3.44426155e-01 -1.22821093e+00
7.22402483e-02 7.82309920e-02 8.24344873e-01 -2.33042508e-01
9.02778149e-01 -7.92422116e-01 4.03535426e-01 4.28607166e-01
-1.84194729e-01 1.85112864e-01 -4.18773033e-02 3.28036010e-01
-1.58206657e-01 -2.81589311e-02 1.21454883e+00 7.45869130e-02
-6.15875304e-01 5.75683832e-01 -1.64384708e-01 -2.68656224e-01
1.17962551e+00 -3.64389360e-01 -1.26081124e-01 -7.04360068e-01
-7.43503213e-01 -4.94687408e-02 9.45814312e-01 5.73987663e-01
6.84025407e-01 -1.32386816e+00 -8.58321011e-01 5.23469388e-01
-1.11479320e-01 -2.60662943e-01 4.49116796e-01 4.74608779e-01
-6.13494813e-01 -2.10672349e-01 -6.80330276e-01 -1.90829068e-01
-1.83423674e+00 5.98255813e-01 4.37048703e-01 1.46871563e-02
-5.64339340e-01 8.25098932e-01 3.32497507e-01 7.44314864e-02
2.52556711e-01 2.03525335e-01 -2.63204277e-01 3.27743328e-04
8.63891840e-01 2.53831208e-01 -2.24580586e-01 -8.68024170e-01
2.79224222e-03 7.49014735e-01 -2.24501207e-01 -3.05170625e-01
1.16319418e+00 -2.03499869e-01 -2.21239865e-01 1.81365475e-01
1.04914272e+00 9.98729989e-02 -1.58278096e+00 3.63578349e-02
-4.94721740e-01 -7.35575855e-01 -1.15712754e-01 -5.42769492e-01
-1.23802280e+00 8.88231874e-01 6.87629461e-01 -1.33281080e-02
1.42138457e+00 -3.17165583e-01 6.42112792e-01 -1.64802700e-01
2.29958653e-01 -1.10606968e+00 1.27286732e-01 1.98839635e-01
1.20587409e+00 -9.66519952e-01 3.42893541e-01 -5.52754760e-01
-5.75877249e-01 9.59121764e-01 7.27656603e-01 -3.30055282e-02
3.07837874e-01 2.49248639e-01 1.50411472e-01 -2.47724727e-02
-6.26737714e-01 -1.74665693e-02 3.18466425e-01 8.67308021e-01
4.38316345e-01 -2.09365249e-01 -2.30255648e-01 1.07222944e-01
-3.84817034e-01 3.97454761e-02 3.74231130e-01 8.14267635e-01
-2.23648816e-01 -1.48419297e+00 -4.30339903e-01 5.19011676e-01
-7.07769930e-01 1.51280880e-01 -5.01031995e-01 5.66371858e-01
2.63033509e-01 6.15426064e-01 6.36187270e-02 -3.13224822e-01
4.06634510e-01 1.75374150e-01 8.82948637e-01 -4.82433915e-01
-5.64953208e-01 -1.78017512e-01 1.40852585e-01 -6.31740868e-01
-2.84424722e-01 -7.03931332e-01 -6.77217364e-01 -5.88212967e-01
-8.84039253e-02 -3.14422816e-01 7.35544920e-01 4.03016686e-01
2.24001035e-01 5.60541034e-01 6.58511281e-01 -1.11512208e+00
-2.81741053e-01 -8.36437047e-01 -6.34499490e-01 5.62019706e-01
2.72812754e-01 -6.02725446e-01 -2.26208329e-01 5.48244178e-01] | [12.61470890045166, -0.26702800393104553] |
6d0c3c01-b7d5-422d-8e3b-87d87cfe328b | experimental-design-for-any-p-norm | 2305.01942 | null | https://arxiv.org/abs/2305.01942v1 | https://arxiv.org/pdf/2305.01942v1.pdf | Experimental Design for Any $p$-Norm | We consider a general $p$-norm objective for experimental design problems that captures some well-studied objectives (D/A/E-design) as special cases. We prove that a randomized local search approach provides a unified algorithm to solve this problem for all $p$. This provides the first approximation algorithm for the general $p$-norm objective, and a nice interpolation of the best known bounds of the special cases. | ['Hong Zhou', 'Robert Wang', 'Lap Chi Lau'] | 2023-05-03 | null | null | null | null | ['experimental-design'] | ['methodology'] | [ 6.70137554e-02 -9.00431871e-02 -7.66713142e-01 -1.81134284e-01
-8.14227104e-01 -4.28630322e-01 -2.35709608e-01 -2.22623169e-01
-4.25563723e-01 9.99221623e-01 -2.81474441e-01 -3.84840965e-01
-7.72814989e-01 -3.52663040e-01 -9.32916641e-01 -9.21352208e-01
-5.13379514e-01 3.95660520e-01 -2.08221093e-01 -2.20076069e-01
7.50324070e-01 7.55672395e-01 -1.51758051e+00 -4.02419120e-01
5.98399639e-01 1.38638914e+00 2.11295739e-01 7.32429147e-01
3.44448149e-01 -2.20082849e-01 -5.31654596e-01 -9.28324014e-02
8.88967156e-01 -5.01934767e-01 -9.37450230e-01 1.56316802e-01
5.37202716e-01 -8.46812353e-02 -4.67327237e-03 1.11678302e+00
1.04331005e+00 2.31351525e-01 4.94851619e-01 -1.26842773e+00
-3.16031426e-01 4.30917054e-01 -1.04413557e+00 1.27962986e-02
3.14306557e-01 3.07173818e-01 1.42028964e+00 -6.58830822e-01
6.40246570e-01 1.30735457e+00 7.05512702e-01 3.64119262e-01
-1.55511296e+00 -4.64401245e-01 -1.09326025e-03 -1.04323149e-01
-1.61045182e+00 -3.73636931e-01 7.78544426e-01 -2.66252428e-01
8.60984087e-01 3.68148714e-01 3.03308964e-01 5.32299876e-01
2.10244432e-01 8.81901324e-01 1.21238971e+00 -2.84683973e-01
4.95871276e-01 -1.18012778e-01 2.16958001e-01 8.11958551e-01
5.21564066e-01 3.36997062e-01 -2.16912016e-01 -2.62288600e-01
9.01363134e-01 -3.92450839e-01 -3.02605808e-01 -6.14732265e-01
-9.64422762e-01 9.43790972e-01 1.12013817e-01 3.78563434e-01
-2.70722598e-01 6.19006574e-01 1.73726693e-01 5.25996208e-01
2.00174004e-01 1.08408999e+00 -8.66345406e-01 -1.37955368e-01
-8.88130367e-01 7.62548685e-01 7.53280222e-01 1.13821590e+00
4.90060985e-01 1.73268199e-01 -1.09829336e-01 6.01997018e-01
8.21138397e-02 7.00181067e-01 -1.78266913e-01 -1.27684450e+00
4.04860318e-01 1.04547553e-01 5.97512960e-01 -1.01421916e+00
-5.22985220e-01 -5.04810512e-01 -5.95055819e-01 2.14977026e-01
4.12928462e-01 -2.66594887e-01 -2.20030084e-01 1.84569752e+00
3.06742668e-01 -2.81370804e-02 -5.45276642e-01 6.07924223e-01
1.21920407e-02 7.43573427e-01 -3.76974881e-01 -8.83815587e-01
1.04686248e+00 -6.50500417e-01 -8.06092024e-01 -4.31919582e-02
4.92655665e-01 -7.64777124e-01 1.01759827e+00 4.91093427e-01
-1.38798344e+00 -3.12281340e-01 -1.18967438e+00 3.02686572e-01
-1.48347527e-01 8.71120617e-02 6.17307663e-01 7.82013118e-01
-1.29218149e+00 5.32499254e-01 -1.55408299e-02 -1.05605602e-01
2.37874597e-01 9.16672945e-01 -6.51373789e-02 -1.88032150e-01
-6.62571311e-01 5.95378637e-01 1.71208054e-01 2.67851055e-01
-1.06774342e+00 -1.14531505e+00 -4.65360373e-01 -8.69304985e-02
6.31123185e-01 -4.83587444e-01 9.95197296e-01 -2.03999996e-01
-1.72573960e+00 8.83528531e-01 2.81447824e-02 -1.15738362e-01
3.67311448e-01 1.15757875e-01 1.73900470e-01 -1.33002818e-01
-2.13115916e-01 3.94758254e-01 6.67165220e-01 -8.28949928e-01
-4.27025348e-01 -5.05278528e-01 1.39442995e-01 -4.88015525e-02
1.44275606e-01 4.07711864e-01 1.46121442e-01 -6.04292929e-01
-3.22881311e-01 -8.53514314e-01 -8.21876705e-01 1.08905502e-01
-3.98459405e-01 -3.12786967e-01 4.74630564e-01 -2.33421624e-01
1.68193579e+00 -1.75862658e+00 4.80464369e-01 4.91117984e-01
9.33731273e-02 9.66742560e-02 -4.57543403e-01 4.34036553e-01
-2.78842211e-01 2.23366737e-01 -1.26906306e-01 -9.29277688e-02
4.52296525e-01 -2.37514805e-02 4.33204882e-02 8.66760254e-01
-2.82682538e-01 7.84951925e-01 -5.47698736e-01 -2.01138794e-01
-1.18286155e-01 -2.91031808e-01 -9.58289564e-01 9.81910452e-02
-3.35070878e-01 -3.99789102e-02 -4.77722853e-01 2.95260608e-01
8.12281668e-01 2.17193849e-02 1.61193952e-01 3.64140235e-02
-3.73099834e-01 -2.54465163e-01 -1.67319798e+00 1.63886988e+00
-5.42640746e-01 4.63789910e-01 9.03073728e-01 -1.79153049e+00
8.58458936e-01 2.71884441e-01 1.00756073e+00 -3.73225302e-01
5.30491471e-01 5.32527804e-01 -3.08924288e-01 -5.01910150e-01
1.63322389e-01 -3.69608670e-01 -3.44516069e-01 4.98935133e-01
4.51136678e-02 -3.83240879e-01 2.32151464e-01 -4.39193457e-01
1.18566561e+00 -4.80342597e-01 4.25919443e-01 -1.23670959e+00
4.63085681e-01 -2.84506500e-01 5.97167909e-01 8.52994561e-01
-3.86043906e-01 2.28251278e-01 7.46774435e-01 -1.77609146e-01
-1.08205521e+00 -7.15243399e-01 -2.05683008e-01 8.74453604e-01
-6.25575008e-03 -3.09258550e-01 -6.03143930e-01 -5.18017232e-01
2.74330616e-01 4.24254060e-01 -5.83971083e-01 1.84092820e-02
-6.78553402e-01 -7.74044096e-01 1.39492124e-01 3.76007676e-01
2.85503030e-01 -4.36314106e-01 -3.43913853e-01 2.71172434e-01
2.44931877e-01 -6.96169138e-01 -9.35086310e-01 4.12941664e-01
-9.35149789e-01 -1.27148926e+00 -9.67524529e-01 -1.15176725e+00
6.31470263e-01 1.27793342e-01 9.48218882e-01 -2.78446585e-01
-6.71482027e-01 4.86243755e-01 3.11178684e-01 -1.69080436e-01
1.66098759e-01 -3.94181341e-01 1.48869172e-01 -2.61339098e-01
-1.04149707e-01 -2.45130479e-01 -8.34535539e-01 8.41571748e-01
-5.83034515e-01 -8.49811614e-01 2.63335049e-01 9.34920073e-01
8.73046279e-01 4.77870256e-01 5.60834348e-01 -4.77730483e-01
8.66509795e-01 -1.96045384e-01 -1.41132355e+00 2.32324645e-01
-7.66460598e-01 4.09493715e-01 8.43936741e-01 -4.45570260e-01
-5.25302827e-01 -5.58213219e-02 -1.83221832e-01 -2.83817053e-01
2.69538015e-01 4.91098404e-01 -4.12025183e-01 -5.71489930e-01
6.13345683e-01 -1.64502054e-01 7.92391971e-03 -6.87006235e-01
2.25212321e-01 4.25024033e-01 -4.32955734e-02 -1.11928439e+00
6.22046530e-01 1.18549012e-01 5.03961682e-01 -8.86530101e-01
-4.25539017e-01 -5.42288780e-01 1.33788690e-01 2.02110380e-01
5.55365384e-01 -2.53991991e-01 -1.38076973e+00 -1.19248793e-01
-8.17198753e-01 -4.56076801e-01 -7.73045897e-01 4.87068295e-01
-1.27500117e+00 1.21337153e-01 -2.85813421e-01 -7.84852803e-01
1.61681045e-02 -1.53695595e+00 1.00718153e+00 -1.12231262e-01
1.47829756e-01 -9.10442293e-01 4.20931518e-01 -3.33051234e-02
4.40456420e-01 3.59103709e-01 9.42142129e-01 -1.83605969e-01
-4.58687782e-01 -1.91526935e-01 6.19246960e-02 3.87690514e-01
-4.26925153e-01 -4.63659056e-02 -2.13048682e-01 -4.53014761e-01
3.81495357e-01 -1.32773936e-01 3.73178363e-01 1.27413547e+00
1.38493502e+00 -6.50087297e-01 -3.16224575e-01 4.47238058e-01
1.77680445e+00 6.13053143e-01 4.46732551e-01 -9.31621119e-02
-9.55109671e-02 2.35413507e-01 6.14296496e-01 7.27806032e-01
-2.74980575e-01 8.39953184e-01 2.29353428e-01 -4.75460514e-02
3.85284245e-01 1.38429016e-01 9.23356339e-02 4.46581483e-01
-4.78457361e-02 -2.58362651e-01 -5.04369915e-01 6.80805206e-01
-1.76769030e+00 -8.52781117e-01 1.15759782e-01 2.40407085e+00
8.87457788e-01 -5.91464527e-02 4.07824785e-01 2.49482021e-01
9.41692114e-01 2.39171982e-01 -4.04242516e-01 -8.72701108e-01
-1.15075558e-01 1.09060907e+00 1.08281600e+00 4.36655968e-01
-8.67257357e-01 3.24883968e-01 9.42623520e+00 1.46573675e+00
-6.03341639e-01 -3.66498381e-02 7.10074842e-01 -2.42320001e-01
-2.78260529e-01 -2.22439036e-01 -7.86530077e-01 3.30840409e-01
8.16909373e-01 -4.76208419e-01 6.29702330e-01 1.09274817e+00
5.54422796e-01 2.61643641e-02 -1.27562475e+00 1.29743230e+00
-3.08306128e-01 -1.35391676e+00 -4.84807700e-01 4.01368409e-01
1.05592549e+00 -6.41713202e-01 4.30337340e-01 -1.60461236e-02
2.69870549e-01 -1.04017627e+00 4.12450522e-01 -1.57059893e-01
9.43381250e-01 -9.60282922e-01 4.44948554e-01 1.25290483e-01
-1.28475654e+00 -3.15061390e-01 -5.62865973e-01 5.03033735e-02
1.18073776e-01 6.11116707e-01 -1.63099393e-01 3.06811213e-01
3.82595748e-01 2.77070880e-01 2.50940435e-02 1.32910717e+00
1.77800611e-01 2.27310911e-01 -7.38719225e-01 -2.90983737e-01
3.30122083e-01 -3.66300315e-01 6.92425013e-01 8.12814951e-01
6.61297619e-01 3.29361945e-01 1.20558493e-01 7.81425655e-01
-2.30298743e-01 3.39069158e-01 -7.32707441e-01 -2.04781756e-01
2.04383552e-01 5.74154615e-01 -4.21624869e-01 3.23297620e-01
-1.11710787e-01 4.17341799e-01 -1.28427610e-01 3.10433567e-01
-8.96843493e-01 -9.78234470e-01 1.16740072e+00 -2.34432295e-02
5.83697557e-01 -3.76001537e-01 -2.78453648e-01 -8.49162042e-01
-5.19865751e-02 -9.97793376e-01 4.85268950e-01 -2.74306178e-01
-1.07685459e+00 -1.37888685e-01 1.32113889e-01 -9.94844258e-01
8.41939598e-02 -1.12722063e+00 -3.05356055e-01 6.32454336e-01
-1.01918340e+00 -2.08943978e-01 6.02698565e-01 6.03841841e-01
4.14445490e-01 1.35094998e-02 4.68858540e-01 3.30412149e-01
-7.32620001e-01 9.24495518e-01 4.90252286e-01 -5.52474260e-01
3.40013117e-01 -1.37303090e+00 -3.96697342e-01 8.69878054e-01
-4.41493899e-01 6.86183810e-01 1.19092751e+00 -4.96990047e-02
-2.11221552e+00 -5.50331056e-01 5.82961679e-01 1.85436711e-01
7.66256690e-01 -3.49600278e-02 6.28023893e-02 5.07138014e-01
1.63589612e-01 -5.52845225e-02 6.09565675e-01 2.30470911e-01
7.33785778e-02 -5.28906822e-01 -1.42472613e+00 7.24370956e-01
1.45077169e+00 -3.54831249e-01 -1.55780658e-01 4.86788839e-01
5.63895047e-01 -3.86098683e-01 -1.15004003e+00 3.81672412e-01
3.52915794e-01 -3.20328802e-01 1.27323222e+00 -5.95050812e-01
1.42061993e-01 -2.12972507e-01 -3.27488422e-01 -1.23403680e+00
-1.70246258e-01 -1.37578237e+00 1.17947318e-01 4.93680209e-01
6.01853371e-01 -5.36429822e-01 6.11993790e-01 5.22464991e-01
-5.01859635e-02 -9.34742510e-01 -1.33270466e+00 -1.27986407e+00
2.29786351e-01 -4.12985504e-01 4.66750503e-01 6.97772264e-01
3.95918071e-01 1.55161574e-01 -6.01565063e-01 -3.27996135e-01
8.35170031e-01 2.04037189e-01 4.92009729e-01 -8.35716248e-01
-5.90363860e-01 -1.08064044e+00 -5.03199637e-01 -1.61534846e+00
1.78673103e-01 -6.71063781e-01 3.43077689e-01 -1.07983577e+00
-8.26107189e-02 -5.77591002e-01 -2.05652758e-01 -1.63408816e-01
-1.28396656e-02 1.71017293e-02 -2.26203263e-01 -5.28355479e-01
-5.94886601e-01 4.69202787e-01 1.28173614e+00 -3.04995090e-01
-2.75729746e-01 3.08848053e-01 -8.50296438e-01 2.36122951e-01
5.49233317e-01 -3.18001032e-01 -4.40982640e-01 -2.92600840e-01
4.83176827e-01 3.01454931e-01 -1.98867097e-01 -7.29301274e-01
-1.11883692e-02 -8.19188118e-01 -2.65266806e-01 -3.86540294e-01
-1.02386519e-01 -1.05048943e+00 1.61902651e-01 7.06656873e-01
-6.37792647e-01 2.85084724e-01 5.02533428e-02 3.89786541e-01
-1.02434732e-01 -4.20092762e-01 1.08219278e+00 1.02163367e-02
-2.43254289e-01 7.56510317e-01 -1.70110449e-01 3.80150825e-01
1.18495703e+00 -1.47573531e-01 -4.66622263e-01 -2.05039248e-01
-8.09222221e-01 4.08182651e-01 2.44690895e-01 -2.09997788e-01
3.83616447e-01 -1.39116085e+00 -5.04106343e-01 -9.45631191e-02
-1.97544292e-01 -3.24486643e-01 1.56625628e-01 1.02864349e+00
-7.24669397e-01 6.51117384e-01 -1.21389084e-01 -3.69715720e-01
-8.79653752e-01 9.96205270e-01 4.30292517e-01 -4.61441338e-01
7.27519244e-02 1.12256086e+00 1.31649837e-01 -3.12902741e-02
2.51122266e-01 -3.92518878e-01 3.37937474e-01 -1.62094295e-01
4.96018052e-01 7.82539725e-01 -1.31169200e-01 -7.00142086e-02
-3.95574927e-01 8.04827273e-01 4.84612823e-01 -1.90456018e-01
1.55830717e+00 -7.03886822e-02 -2.10527033e-01 9.75568518e-02
1.86623573e+00 1.52311385e-01 -8.94936562e-01 -1.67600378e-01
-3.28691415e-02 -5.73980629e-01 1.04612254e-01 -2.24781021e-01
-1.30439639e+00 5.51555097e-01 5.30830801e-01 7.05439374e-02
1.38773382e+00 -1.41156316e-01 4.74649608e-01 5.18373370e-01
7.48188794e-01 -1.43760812e+00 -2.13889498e-02 2.81829506e-01
1.07687902e+00 -5.69172084e-01 2.92491138e-01 -3.05201411e-01
-5.28813042e-02 9.53083038e-01 2.80692041e-01 -4.57672477e-01
1.08725810e+00 4.64920402e-01 -7.49778926e-01 -1.07396930e-01
-5.24846852e-01 -2.58009642e-01 7.21984804e-02 3.51704508e-01
1.75087303e-01 1.75885353e-02 -1.29602063e+00 5.19956529e-01
4.32601199e-02 4.53299843e-03 8.15009624e-02 1.05354249e+00
-6.56788588e-01 -1.20502877e+00 -3.20603192e-01 2.91198045e-01
-5.82632422e-01 1.48810387e-01 -9.07291099e-02 7.32336402e-01
-1.40286699e-01 9.06139910e-01 -2.76078522e-01 -2.10050151e-01
5.58849335e-01 -8.68824050e-02 8.90337229e-01 -8.82410631e-02
-5.21796644e-01 2.46080145e-01 -1.83868352e-02 -1.21111476e+00
-5.23278177e-01 -4.13364530e-01 -5.88321686e-01 -8.31085920e-01
-5.46818674e-01 4.76726264e-01 7.31802523e-01 6.26098275e-01
2.94975847e-01 2.12986574e-01 1.24648845e+00 -6.70299411e-01
-8.91853869e-01 -4.12955374e-01 -1.22829759e+00 2.60972441e-03
2.30835736e-01 -7.69403696e-01 -6.81061685e-01 -2.84923583e-01] | [6.54437255859375, 4.317537784576416] |
4205d148-d015-470c-9c29-ce4df8c1a2dd | separable-hovernet-and-instance-yolo-for | 2203.00262 | null | https://arxiv.org/abs/2203.00262v1 | https://arxiv.org/pdf/2203.00262v1.pdf | Separable-HoverNet and Instance-YOLO for Colon Nuclei Identification and Counting | Nuclear segmentation, classification and quantification within Haematoxylin & Eosin stained histology images enables the extraction of interpretable cell-based features that can be used in downstream explainable models in computational pathology (CPath). However, automatic recognition of different nuclei is faced with a major challenge in that there are several different types of nuclei, some of them exhibiting large intraclass variability. In this work, we propose an approach that combine Separable-HoverNet and Instance-YOLOv5 to indentify colon nuclei small and unbalanced. Our approach can achieve mPQ+ 0.389 on the Segmentation and Classification-Preliminary Test Dataset and r2 0.599 on the Cellular Composition-Preliminary Test Dataset on ISBI 2022 CoNIC Challenge. | ['Dong Hu', 'Min Wu', 'Lijian Mao', 'Liukun Zhang', 'Chunhui Lin'] | 2022-03-01 | null | null | null | null | ['explainable-models', 'nuclear-segmentation'] | ['computer-vision', 'medical'] | [ 1.42244086e-01 -4.91873268e-03 -2.28928402e-01 -1.32329121e-01
-9.09853935e-01 -7.94429243e-01 2.86912650e-01 6.99616134e-01
-5.02655029e-01 1.04564512e+00 -2.06068326e-02 -2.64776319e-01
-3.77219588e-01 -5.91173172e-01 1.14702638e-02 -1.26899052e+00
4.47253361e-02 8.71243060e-01 1.66781038e-01 8.75796005e-02
3.67326945e-01 8.21766794e-01 -9.17852402e-01 4.80501443e-01
5.65510690e-01 5.72112858e-01 1.33619122e-02 1.28069699e+00
-1.08387517e-02 7.63159156e-01 -3.44009399e-01 -1.84761629e-01
-3.24511349e-01 -3.59702617e-01 -1.04335487e+00 -8.02489966e-02
3.08703780e-02 2.90387183e-01 1.80325955e-01 9.18148279e-01
6.34331465e-01 -3.48218471e-01 1.35458148e+00 -1.09620976e+00
1.03588171e-01 5.74916720e-01 -7.06276774e-01 7.14276195e-01
-9.21807736e-02 2.10147828e-01 8.77129197e-01 -4.96793777e-01
1.04076934e+00 8.14245284e-01 7.55008340e-01 4.83835250e-01
-1.59915769e+00 -3.52841079e-01 -7.91090906e-01 2.20854029e-01
-1.56209111e+00 3.57923023e-02 -4.91717421e-02 -6.42837405e-01
8.93015444e-01 7.97402322e-01 7.39776731e-01 2.15855256e-01
4.09588963e-01 5.71874797e-01 1.38985598e+00 -2.54550219e-01
8.64254907e-02 1.71019718e-01 3.14112365e-01 5.69910407e-01
3.99685293e-01 -1.90792501e-01 -1.98628336e-01 -1.27292961e-01
5.41258752e-01 -1.53705180e-01 -1.41899362e-01 1.14925995e-01
-1.19686747e+00 8.38035405e-01 2.75350839e-01 6.99607968e-01
-1.39779136e-01 3.14562589e-01 6.66125298e-01 -1.89609583e-02
3.64094079e-01 4.33218777e-01 -4.84363288e-01 1.63761467e-01
-1.01795530e+00 -1.41334468e-02 5.77372611e-01 3.51766258e-01
3.94044548e-01 -3.92290592e-01 -1.03442529e-02 5.45773923e-01
3.56995404e-01 2.34619603e-01 8.01653743e-01 -6.67337060e-01
-4.21231121e-01 8.13136041e-01 -4.07582670e-01 -8.48341167e-01
-1.00731671e+00 -4.19813931e-01 -1.16551697e+00 9.03959051e-02
8.42885733e-01 1.83365017e-01 -9.23722744e-01 1.09889793e+00
5.25687575e-01 -8.86143222e-02 5.28198443e-02 8.13423157e-01
8.42369616e-01 1.22108512e-01 5.19261897e-01 -5.30432016e-02
1.56812298e+00 -6.02343798e-01 -4.97375399e-01 2.68190145e-01
1.23608279e+00 -8.30385327e-01 4.39771056e-01 3.23340833e-01
-6.11127496e-01 -1.06195331e-01 -7.98227847e-01 -1.19512707e-01
-5.70660114e-01 2.38494918e-01 8.52024496e-01 6.45034254e-01
-1.07758248e+00 3.75975430e-01 -8.31919372e-01 -6.34210110e-01
6.07484162e-01 7.55084872e-01 -6.65064156e-01 1.67666942e-01
-5.58105230e-01 8.63778293e-01 7.51493573e-01 1.57143220e-01
-4.25199151e-01 -8.91078889e-01 -3.65208268e-01 -2.06264287e-01
-9.87795740e-02 -6.50334656e-01 7.61342943e-01 -4.17773753e-01
-9.11780179e-01 1.37014508e+00 -1.85765862e-01 -5.06861269e-01
5.21943271e-01 7.77427077e-01 3.35347354e-02 5.05684435e-01
2.10886542e-02 9.09856379e-01 8.47131293e-03 -1.08530676e+00
-8.76915097e-01 -4.49805140e-01 -4.10922438e-01 1.50200039e-01
2.25907385e-01 -1.32973611e-01 -6.75088987e-02 -4.43917751e-01
3.17192942e-01 -1.03622460e+00 -4.84162092e-01 1.00076675e-01
-6.01760983e-01 -1.35831624e-01 5.08815587e-01 -7.77318835e-01
8.60838652e-01 -1.91197872e+00 1.37056306e-01 6.17398798e-01
7.17409134e-01 9.19320583e-02 2.19035879e-01 -5.88632151e-02
-1.17014870e-01 7.61245608e-01 -5.03548123e-02 1.30944410e-02
-1.29859462e-01 9.73492116e-02 3.90791655e-01 8.32678437e-01
2.89368093e-01 9.08556700e-01 -7.97336698e-01 -1.05783880e+00
2.83153445e-01 5.21952569e-01 -4.33933198e-01 -2.49390051e-01
1.47923768e-01 5.57174563e-01 -2.58706778e-01 9.01971638e-01
5.87964892e-01 -2.56400287e-01 4.16028559e-01 -1.29959762e-01
2.21512526e-01 -3.53846550e-01 -8.65174949e-01 9.77734387e-01
-4.21757139e-02 7.46899903e-01 -3.34987640e-02 -7.32977271e-01
5.70446849e-01 3.54473799e-01 6.66859984e-01 -2.13960499e-01
5.80068588e-01 5.08207500e-01 4.94693488e-01 -2.54220426e-01
1.44897744e-01 -7.49236941e-01 1.99240237e-01 6.37083203e-02
7.83295929e-02 -4.37210798e-01 5.86629629e-01 -2.63217594e-02
1.22316372e+00 -4.45924014e-01 6.41324341e-01 -6.55138552e-01
9.68749404e-01 2.41237924e-01 4.32179570e-01 2.31466621e-01
-6.76670074e-01 8.38352025e-01 9.47782099e-01 -2.09760517e-01
-1.00835085e+00 -7.22740173e-01 -5.17011225e-01 6.83685124e-01
-1.66432828e-01 -9.46758687e-02 -5.99843264e-01 -6.63442314e-01
-4.41191159e-02 2.43899360e-01 -8.69980454e-01 1.73419118e-01
-3.29526871e-01 -1.32719588e+00 9.12335515e-01 4.11364079e-01
-1.27650142e-01 -9.37222064e-01 -3.66815090e-01 8.43798295e-02
-3.06677550e-01 -9.45615947e-01 -3.88651639e-02 5.50827801e-01
-9.37448919e-01 -1.59484243e+00 -5.00210643e-01 -8.16323519e-01
1.05266881e+00 -1.70710728e-01 9.21301663e-01 6.17928565e-01
-9.48108017e-01 -9.82458591e-02 -2.21097812e-01 -3.73277068e-01
-6.91922724e-01 2.38419667e-01 -2.79145688e-01 -4.50768024e-01
4.17611152e-01 -2.09500968e-01 -6.56195760e-01 2.36179098e-01
-9.27178323e-01 -4.17233668e-02 6.05672300e-01 1.10082150e+00
1.25108266e+00 3.06043237e-01 3.23330790e-01 -1.12272656e+00
-2.78840512e-02 -6.01411462e-01 -2.36926734e-01 1.21475272e-01
-4.78386700e-01 -2.66410172e-01 6.48612261e-01 -2.23721161e-01
-5.41838229e-01 1.27935484e-01 -8.60612169e-02 1.44784629e-01
-2.83616960e-01 5.44747710e-01 2.59697229e-01 -2.99063385e-01
6.21960640e-01 1.89142600e-02 1.23662822e-01 5.86533081e-03
-1.32933304e-01 4.92936969e-01 3.40822905e-01 -6.55776784e-02
6.31538093e-01 7.21569598e-01 8.19107711e-01 -8.08833480e-01
-5.85930884e-01 -9.69870090e-01 -8.30312788e-01 -1.34934839e-02
1.02423954e+00 -6.10439062e-01 -8.31796288e-01 3.44913214e-01
-6.49195910e-01 -3.48671019e-01 -1.34838402e-01 5.48080862e-01
-5.84383547e-01 1.65030092e-01 -9.25140798e-01 -5.76537788e-01
-6.53940201e-01 -1.20525002e+00 9.02395844e-01 3.91853660e-01
-6.78834200e-01 -1.29587770e+00 3.42420280e-01 5.66918433e-01
1.22858606e-01 6.35934412e-01 1.20711315e+00 -9.64807749e-01
-6.99955598e-02 -4.06174660e-01 -4.05549258e-01 -1.26332194e-01
-1.24840416e-01 5.79402447e-01 -8.02007556e-01 -1.07945278e-01
-4.64273334e-01 -2.39270404e-01 8.14039409e-01 6.46265566e-01
1.09144604e+00 7.88508635e-03 -6.38055801e-01 7.10436881e-01
1.64201033e+00 -5.64806210e-03 6.55016780e-01 4.18326229e-01
4.79022920e-01 7.11589873e-01 6.70093000e-01 1.17440708e-02
9.82306302e-02 1.09567277e-01 1.94706976e-01 -4.13444430e-01
-2.72557884e-01 1.62730649e-01 -3.34289581e-01 5.09651661e-01
-2.07998946e-01 -2.89016455e-01 -1.27100325e+00 7.40953386e-01
-1.46601450e+00 -8.90598655e-01 -7.92942286e-01 1.46125960e+00
9.65410173e-01 2.69617662e-02 2.44735200e-02 4.96413052e-01
7.46787012e-01 -5.85732996e-01 -2.06199437e-01 -3.58962238e-01
-3.62075835e-01 3.72952431e-01 5.73168635e-01 4.65946645e-01
-8.81397009e-01 6.99089170e-01 6.50276184e+00 1.06135798e+00
-1.01477766e+00 -9.28497687e-02 1.30761480e+00 3.28045756e-01
8.85441974e-02 -1.75517812e-01 -7.98643887e-01 2.63782203e-01
9.94195759e-01 -2.34679490e-01 -4.05380316e-02 2.72161365e-01
3.16270560e-01 -5.90409935e-01 -8.30576897e-01 6.21410012e-01
-1.84783041e-01 -1.26237488e+00 -2.11725652e-01 3.41490179e-01
4.98576462e-01 1.26421861e-02 -7.25919008e-02 4.68578339e-02
-5.08838817e-02 -1.39748120e+00 1.66758224e-01 4.94787335e-01
1.00186455e+00 -7.70838797e-01 1.53866959e+00 2.34791562e-01
-8.87816191e-01 3.17063332e-01 -3.90859336e-01 3.43315065e-01
-3.58675867e-01 3.70961547e-01 -1.64080036e+00 3.56622219e-01
3.39225382e-01 3.54308128e-01 -6.69179261e-01 1.30620468e+00
4.68860306e-02 8.05726826e-01 -3.86817932e-01 -2.69339949e-01
-4.66945628e-03 2.09039208e-02 1.79295212e-01 1.48627853e+00
1.97828323e-01 2.91043788e-01 -1.73271939e-01 5.17036140e-01
3.30735058e-01 4.75906968e-01 7.27323145e-02 -2.35486880e-01
1.63009822e-01 1.95689821e+00 -1.76360714e+00 -1.37221605e-01
2.47641698e-01 4.46815491e-01 6.85886219e-02 -1.54972568e-01
-7.28965044e-01 -1.54146299e-01 2.49898344e-01 3.48237664e-01
-7.54255652e-02 2.17941105e-01 -4.17802066e-01 -7.06515014e-01
-8.86893094e-01 -7.17401922e-01 6.64515316e-01 -4.38455701e-01
-1.41162825e+00 2.74126798e-01 -3.53626519e-01 -1.02806866e+00
3.72143053e-02 -8.47707629e-01 -4.25445437e-01 6.92724586e-01
-1.45983660e+00 -1.37458909e+00 -1.85597077e-01 6.96385687e-04
1.89085156e-01 2.39454955e-01 1.01807857e+00 5.44725209e-02
-4.11514938e-01 5.76208293e-01 2.28158668e-01 1.53623655e-01
6.34185314e-01 -1.81415892e+00 -4.91539419e-01 1.45297527e-01
-1.93401232e-01 2.91397303e-01 7.40546584e-01 -4.36801493e-01
-9.08951938e-01 -1.15594637e+00 8.88377249e-01 -4.47093517e-01
7.73166895e-01 2.08616987e-01 -7.38594353e-01 4.00898635e-01
-1.30481035e-01 1.66111916e-01 1.53840494e+00 -2.54898995e-01
1.55774742e-01 2.63428271e-01 -1.45794308e+00 4.05400097e-01
1.60053521e-01 -2.25129440e-01 -1.11197472e-01 4.71914828e-01
-2.30419189e-02 -5.62506557e-01 -1.28200686e+00 1.81731239e-01
6.24038339e-01 -6.95639491e-01 8.00822437e-01 -4.89449561e-01
2.42618993e-01 -6.34019077e-01 7.32926652e-02 -9.47900772e-01
-3.27698767e-01 -8.59700143e-02 5.17134070e-01 9.99275923e-01
6.08436227e-01 -3.82618904e-01 1.11371481e+00 2.78014451e-01
-4.85222414e-03 -9.52714860e-01 -1.10016418e+00 -2.98234522e-01
2.85861701e-01 -1.06561296e-01 1.93416551e-01 9.64490652e-01
2.82757670e-01 1.81191474e-01 3.40811968e-01 -8.24281722e-02
5.80566168e-01 -2.63523251e-01 5.57320654e-01 -1.29388428e+00
-1.24488980e-01 -6.74075902e-01 -1.00604773e+00 1.78020015e-01
1.29492790e-01 -1.36654770e+00 -1.85203686e-01 -1.35377192e+00
8.07640553e-01 -5.45968294e-01 -3.81723315e-01 5.57020187e-01
-1.80754423e-01 7.77163446e-01 -4.45705876e-02 1.48503959e-01
-4.80062217e-01 -1.70495823e-01 1.28793371e+00 -1.83337972e-01
8.00202340e-02 -2.13248029e-01 -5.99598467e-01 7.79154539e-01
9.20285225e-01 -8.77641678e-01 -4.21890542e-02 4.99295861e-01
2.48231664e-01 8.91472213e-03 4.97304380e-01 -9.52422678e-01
-2.49411091e-02 -2.69561172e-01 8.25948358e-01 -9.31217015e-01
-3.93710881e-02 -6.22416019e-01 4.67948616e-01 1.05267024e+00
-2.80922472e-01 -2.22226620e-01 2.02612579e-01 5.10981143e-01
-1.04143053e-01 -4.47715044e-01 8.64750564e-01 -1.84665665e-01
-1.58638999e-01 9.53831896e-02 -6.66479409e-01 -1.29390955e-01
1.06117260e+00 -4.73262340e-01 -5.98228157e-01 4.39608209e-02
-9.45708394e-01 3.82880211e-01 5.49859822e-01 -5.05566537e-01
1.85979202e-01 -1.02847850e+00 -8.53221536e-01 -3.22247833e-01
3.33768362e-03 5.17155156e-02 5.51138699e-01 1.49055684e+00
-1.26314092e+00 5.96004486e-01 -2.97210038e-01 -8.96418154e-01
-1.69278097e+00 5.86769469e-02 6.57420337e-01 -8.85252297e-01
-1.24884937e-02 1.11029208e+00 5.80659881e-02 -3.81743819e-01
-4.52596277e-01 -2.38727897e-01 -6.01998448e-01 3.65517616e-01
1.90275192e-01 5.31783164e-01 6.87997639e-02 -9.82042134e-01
-4.22813207e-01 4.67026979e-01 -3.06221899e-02 2.08743408e-01
1.29773819e+00 -2.40037162e-02 -6.24698281e-01 4.04570132e-01
1.25834858e+00 -6.02533408e-02 -3.94966990e-01 3.30449611e-01
2.35888168e-01 -1.77208900e-01 1.66655019e-01 -9.42731738e-01
-9.18031335e-01 7.81175435e-01 7.57284701e-01 1.72978878e-01
1.08174098e+00 -1.07009940e-01 4.86383349e-01 5.39726876e-02
2.05963030e-02 -9.51989353e-01 -4.25046831e-01 1.23105645e-01
3.82404894e-01 -1.22170150e+00 3.83092821e-01 -7.57213056e-01
-4.48294371e-01 1.45925081e+00 5.03578424e-01 1.24896124e-01
4.68862206e-01 3.66161257e-01 1.72481030e-01 -4.37581062e-01
-8.90149236e-01 -1.98396116e-01 3.52596879e-01 5.98993719e-01
7.96416640e-01 3.67337644e-01 -8.48684847e-01 7.41165996e-01
-4.27950233e-01 -4.94739898e-02 7.23174870e-01 7.13199854e-01
-3.06221783e-01 -9.51809704e-01 -3.10751647e-01 7.01353073e-01
-9.11121607e-01 1.31925449e-01 -4.86646414e-01 1.22076261e+00
2.40508825e-01 6.00578904e-01 5.75782657e-02 -1.09903745e-01
-1.58045471e-01 -2.03006506e-01 3.82687539e-01 -5.88561475e-01
-9.19709146e-01 7.07139194e-01 1.48730660e-02 -1.27824384e-03
-6.00009143e-01 -7.70581365e-01 -1.93261087e+00 -3.95524770e-01
-7.66498923e-01 1.74415872e-01 4.15486187e-01 8.50097418e-01
-2.69251943e-01 6.42731190e-01 4.10410702e-01 -3.62602383e-01
-1.09793179e-01 -1.03053129e+00 -1.11346889e+00 2.47512951e-01
2.61718452e-01 -3.75778615e-01 -7.02771187e-01 4.46306676e-01] | [15.098689079284668, -3.170558214187622] |
cd0d09f0-fda4-4c7e-8afd-6ea02ad0ac05 | neural-progressive-hedging-enforcing | 2202.13436 | null | https://arxiv.org/abs/2202.13436v1 | https://arxiv.org/pdf/2202.13436v1.pdf | Neural-Progressive Hedging: Enforcing Constraints in Reinforcement Learning with Stochastic Programming | We propose a framework, called neural-progressive hedging (NP), that leverages stochastic programming during the online phase of executing a reinforcement learning (RL) policy. The goal is to ensure feasibility with respect to constraints and risk-based objectives such as conditional value-at-risk (CVaR) during the execution of the policy, using probabilistic models of the state transitions to guide policy adjustments. The framework is particularly amenable to the class of sequential resource allocation problems since feasibility with respect to typical resource constraints cannot be enforced in a scalable manner. The NP framework provides an alternative that adds modest overhead during the online phase. Experimental results demonstrate the efficacy of the NP framework on two continuous real-world tasks: (i) the portfolio optimization problem with liquidity constraints for financial planning, characterized by non-stationary state distributions; and (ii) the dynamic repositioning problem in bike sharing systems, that embodies the class of supply-demand matching problems. We show that the NP framework produces policies that are better than deep RL and other baseline approaches, adapting to non-stationarity, whilst satisfying structural constraints and accommodating risk measures in the resulting policies. Additional benefits of the NP framework are ease of implementation and better explainability of the policies. | ['Duc Thien Nguyen', 'Shiau Hong Lim', 'Laura Wynter', 'Supriyo Ghosh'] | 2022-02-27 | null | null | null | null | ['portfolio-optimization'] | ['time-series'] | [-2.45654643e-01 2.89328575e-01 -7.06981659e-01 -8.95273536e-02
-7.97813535e-01 -5.80537975e-01 5.39287567e-01 -6.34094998e-02
-4.72920150e-01 9.94293034e-01 2.78835267e-01 -8.95070374e-01
-6.57841086e-01 -7.60685265e-01 -5.55734336e-01 -6.41388953e-01
-4.07139927e-01 8.07995558e-01 -1.02839500e-01 -1.56530097e-01
2.55533159e-01 5.19503236e-01 -1.09532893e+00 -1.74805447e-01
5.24824798e-01 1.44909787e+00 5.29207885e-02 3.63777310e-01
2.09217399e-01 9.96217132e-01 -1.66326761e-01 -3.96060884e-01
6.66662574e-01 1.92283332e-01 -5.04433870e-01 -8.23606029e-02
-4.14843053e-01 -6.98546529e-01 -1.53412193e-01 8.66722047e-01
3.65735114e-01 5.79078019e-01 7.50897467e-01 -1.43998373e+00
-1.84075862e-01 6.10135436e-01 -4.36274618e-01 3.82784843e-01
-1.59600172e-02 3.95454735e-01 1.48934615e+00 -4.27872747e-01
-1.39344661e-02 1.27252758e+00 5.01180887e-01 4.61901814e-01
-1.26428223e+00 -2.82571256e-01 6.84728861e-01 -1.98147267e-01
-7.36238003e-01 -3.51518363e-01 2.10589811e-01 -4.70100820e-01
1.27539718e+00 1.20003924e-01 4.77538019e-01 8.79378855e-01
4.10950929e-01 9.25484776e-01 1.05373359e+00 -1.67458162e-01
6.22604847e-01 1.24865025e-01 -3.46975923e-01 1.05967999e-01
3.67443375e-02 7.83135772e-01 -2.61986285e-01 -4.08447802e-01
8.05827975e-01 9.77944583e-04 1.16661876e-01 -4.88392979e-01
-6.04447365e-01 1.09745073e+00 -1.59826681e-01 -2.51342475e-01
-8.15918446e-01 2.48982042e-01 4.12271678e-01 4.40677464e-01
4.15751964e-01 6.23312891e-01 -6.56343460e-01 -2.36286610e-01
-9.72823441e-01 7.47564852e-01 1.13306201e+00 9.34291005e-01
2.59598047e-02 4.94450122e-01 -5.06678879e-01 3.01725090e-01
3.62655640e-01 5.35324633e-01 3.81431907e-01 -1.39509726e+00
9.22779083e-01 -2.75693923e-01 9.30577517e-01 -6.94243252e-01
-4.84994292e-01 -4.48195308e-01 -2.27363557e-01 4.31767136e-01
5.58386207e-01 -5.07626355e-01 -4.80722994e-01 1.98746574e+00
1.51150286e-01 1.43480785e-02 1.80502504e-01 5.26269078e-01
-4.45639074e-01 7.57226586e-01 -1.38546273e-01 -7.29483783e-01
1.03601539e+00 -7.00842679e-01 -6.04541719e-01 -2.22407398e-03
2.39991590e-01 -2.88493305e-01 9.35130537e-01 6.15589082e-01
-1.54370093e+00 1.52471989e-01 -6.03836179e-01 7.84060895e-01
1.09171584e-01 -4.86967742e-01 5.97013354e-01 7.32785583e-01
-1.00854301e+00 8.74073505e-01 -9.77911234e-01 2.70469576e-01
2.92168528e-01 4.31400061e-01 4.39475864e-01 5.26455283e-01
-1.26302230e+00 1.07204533e+00 3.31123441e-01 -6.75835535e-02
-1.13650572e+00 -9.90585804e-01 -5.53808093e-01 5.19070685e-01
1.17487764e+00 -3.37934196e-01 2.00572133e+00 -9.39798534e-01
-2.03513432e+00 1.92809999e-02 7.19192386e-01 -8.61252487e-01
9.63204265e-01 -3.24156851e-01 -1.66942403e-01 -2.16278825e-02
-1.78559542e-01 2.00919565e-02 7.95557380e-01 -7.96108305e-01
-8.04736853e-01 8.97672772e-02 1.98509961e-01 5.06072223e-01
-7.23857209e-02 1.83858976e-01 1.92461431e-01 -8.54655743e-01
-6.13603055e-01 -1.01159763e+00 -6.70573354e-01 -5.81507564e-01
-3.46612304e-01 -3.29770111e-02 2.43601918e-01 -8.34888577e-01
1.36238515e+00 -1.76634431e+00 -6.45136684e-02 6.66941404e-01
-4.97789472e-01 -2.21298039e-02 1.79305464e-01 6.00839496e-01
-1.26391649e-01 1.06508709e-01 -1.86968476e-01 -3.55097234e-01
6.43490016e-01 2.74575084e-01 -8.75585914e-01 3.74031574e-01
-5.44607453e-02 7.37850606e-01 -7.39510357e-01 6.72987476e-02
2.47709248e-02 -4.89421099e-01 -6.69757605e-01 3.66227359e-01
-7.69052565e-01 1.94102257e-01 -5.24325490e-01 5.80936253e-01
5.43756709e-02 2.85643861e-02 2.79434025e-01 6.37993336e-01
-1.55951649e-01 2.62778819e-01 -1.38788033e+00 8.82852018e-01
-6.77463531e-01 -1.59997314e-01 2.78410882e-01 -1.23487592e+00
4.15101409e-01 3.10921848e-01 7.86005378e-01 -6.54210806e-01
-9.91992578e-02 7.30630308e-02 -3.30334336e-01 -4.02740508e-01
5.32959402e-01 -5.93361795e-01 -3.30532908e-01 1.04116642e+00
-3.31148684e-01 -6.06172495e-02 -1.40700072e-01 2.57323473e-03
8.70818555e-01 3.22170794e-01 5.11159480e-01 -4.70966846e-01
-1.83066159e-01 -3.07453662e-01 9.15760934e-01 8.78536999e-01
-1.61186740e-01 -4.82912622e-02 1.03795767e+00 -1.23550236e-01
-1.07287276e+00 -1.04400170e+00 1.78757295e-01 1.07486653e+00
-4.33717698e-01 4.22873706e-01 -4.11605835e-01 -6.48110092e-01
7.12344885e-01 1.08217108e+00 -4.47378576e-01 7.81537369e-02
-3.92589986e-01 -9.16035712e-01 4.68531772e-02 7.26925910e-01
1.82967901e-01 -1.12588036e+00 -9.31261361e-01 6.08759880e-01
2.37567842e-01 -7.71170735e-01 -6.89300060e-01 3.69193494e-01
-6.26837075e-01 -8.79248083e-01 -6.97682977e-01 -3.82104218e-02
3.71589750e-01 -3.28417480e-01 1.02584171e+00 -4.81658816e-01
4.72041965e-02 7.28867114e-01 2.61692911e-01 -4.57644224e-01
-2.87891477e-01 -2.75199383e-01 4.08504277e-01 -1.22659216e-02
-2.25728735e-01 -3.73601228e-01 -5.83334506e-01 3.29038411e-01
-8.35443974e-01 -2.41334885e-01 2.29394123e-01 9.79021549e-01
5.71354449e-01 3.95455897e-01 1.05522382e+00 -8.12471688e-01
7.09859550e-01 -7.48114824e-01 -1.20817435e+00 5.29101729e-01
-9.90349889e-01 2.04509020e-01 6.34904027e-01 -5.62009752e-01
-1.42503142e+00 -1.29994214e-01 3.23816657e-01 -3.55697930e-01
3.72940063e-01 6.49814665e-01 -1.97427943e-01 3.38509202e-01
3.79153006e-02 1.30369952e-02 8.38914737e-02 -3.76862913e-01
2.70783931e-01 3.24581742e-01 5.05170405e-01 -1.06613064e+00
6.06532872e-01 1.46409750e-01 8.96079838e-02 -8.45756382e-02
-7.55151153e-01 -7.67388567e-02 -2.67182086e-02 -2.39994042e-02
3.74119431e-01 -9.47897613e-01 -1.06981015e+00 2.68526375e-01
-4.11757946e-01 -9.77242410e-01 -6.95410192e-01 4.88381326e-01
-1.47698069e+00 1.31045088e-01 -7.68309772e-01 -1.56804800e+00
-1.00682788e-01 -8.68736327e-01 4.17394757e-01 1.57986522e-01
-4.61159274e-03 -1.25047374e+00 1.95249170e-01 -6.53643757e-02
5.11679351e-01 2.97717452e-01 1.20513344e+00 -7.46478081e-01
-4.39894199e-01 2.61217412e-02 3.23091477e-01 3.26749712e-01
-4.79378551e-02 -5.06042801e-02 -5.59660256e-01 -6.10627949e-01
2.12255687e-01 -4.30061042e-01 5.30109882e-01 6.81778550e-01
1.22514284e+00 -1.08358312e+00 6.52423427e-02 3.40977281e-01
1.38843727e+00 5.31806886e-01 2.41536558e-01 6.49713457e-01
2.45620031e-02 9.25051033e-01 1.00412667e+00 1.11583436e+00
4.39928889e-01 7.87802339e-01 5.98392367e-01 2.68511921e-01
9.37073946e-01 -4.63950902e-01 6.72043264e-01 6.85595274e-02
1.60914034e-01 -1.11704841e-01 -8.05699706e-01 4.28635508e-01
-2.23364782e+00 -1.26086259e+00 7.62393057e-01 2.52061105e+00
9.00986075e-01 4.48474407e-01 8.60491991e-01 -2.74493128e-01
3.49550217e-01 -4.39369678e-02 -1.10883069e+00 -8.50914478e-01
1.44129455e-01 2.30527930e-02 9.87007916e-01 6.47411823e-01
-9.22363758e-01 6.37028039e-01 6.95913076e+00 7.04624712e-01
-6.93438292e-01 -1.54969648e-01 1.15521264e+00 -5.47458529e-01
-4.29498315e-01 -3.01609170e-02 -8.66468489e-01 6.85631871e-01
1.37105000e+00 -4.63720977e-01 8.24021161e-01 8.26902092e-01
6.06975317e-01 -6.63797259e-02 -1.11776769e+00 3.62885743e-01
-5.59841216e-01 -1.17064440e+00 -5.61193824e-01 2.70878643e-01
8.38152587e-01 -1.60343394e-01 6.48820996e-01 6.08205497e-01
1.05005574e+00 -1.05971670e+00 1.04935777e+00 7.86176145e-01
6.93317294e-01 -1.40797210e+00 4.41732228e-01 4.40772772e-01
-7.75940418e-01 -9.48047400e-01 -9.33333039e-02 2.49947589e-02
4.58923787e-01 4.68213111e-01 -5.06083548e-01 2.98932821e-01
4.60824758e-01 1.34732395e-01 3.54696274e-01 8.98057759e-01
-1.25900745e-01 5.30685246e-01 -3.93499196e-01 2.29676232e-01
5.08889258e-01 -3.33963335e-01 4.50370491e-01 8.97301555e-01
2.87970603e-01 1.74568310e-01 6.06829762e-01 8.66147041e-01
3.22831124e-01 -6.94276690e-02 -5.48025489e-01 -1.07475191e-01
4.24318641e-01 8.37091923e-01 -5.79151869e-01 -8.50494877e-02
-3.70397717e-01 1.97332129e-01 2.79876292e-02 5.89387357e-01
-8.20092916e-01 -1.41889408e-01 6.84761465e-01 -8.26766938e-02
5.91210842e-01 -2.30726197e-01 -1.79999650e-01 -8.66862178e-01
-6.54738918e-02 -1.14071107e+00 7.31477857e-01 -1.15309708e-01
-1.32031727e+00 -4.61881459e-02 2.87531406e-01 -8.66008997e-01
-9.80682015e-01 -5.34679770e-01 -6.54751003e-01 9.13972139e-01
-1.79725814e+00 -5.59866071e-01 8.77147436e-01 4.92459714e-01
4.13589358e-01 -2.99000859e-01 5.33484936e-01 -1.63698554e-01
-8.82328033e-01 4.11862224e-01 6.13183379e-01 -5.95328748e-01
2.98453987e-01 -1.71875954e+00 3.27277452e-01 7.50745654e-01
-2.91309178e-01 2.05812708e-01 7.65950322e-01 -7.27353036e-01
-1.20243514e+00 -9.39324558e-01 3.36856782e-01 -2.57811695e-01
1.07037771e+00 -1.34522453e-01 -5.16607642e-01 6.60222173e-01
-1.68108881e-01 -2.67368555e-01 5.50266564e-01 -9.43552181e-02
2.78511588e-02 3.13780755e-02 -1.26383603e+00 4.32899922e-01
5.75798094e-01 -3.67878675e-01 -4.36076313e-01 5.26835859e-01
7.60312259e-01 -4.26322669e-01 -8.29295695e-01 1.55810565e-01
5.40485084e-01 -5.80779850e-01 8.89015257e-01 -1.19339335e+00
9.13408771e-02 3.27863365e-01 -3.22607756e-01 -1.38563132e+00
-4.17078674e-01 -1.48146975e+00 -5.59883177e-01 1.06164396e+00
5.20371258e-01 -7.81524956e-01 7.14276433e-01 1.22299445e+00
1.29691616e-01 -8.90088797e-01 -1.23017323e+00 -1.29315293e+00
5.04762888e-01 -3.46680850e-01 8.95063937e-01 6.19848251e-01
7.02909306e-02 -4.48385090e-01 -6.76939726e-01 2.17698231e-01
7.32051373e-01 2.58162469e-01 3.10502648e-01 -7.16452122e-01
-1.16217363e+00 -6.72941267e-01 4.97334838e-01 -7.56760359e-01
5.70098221e-01 -4.32603687e-01 1.99592322e-01 -9.28396046e-01
-2.11164039e-02 -8.09012830e-01 -7.97195971e-01 6.12269640e-01
2.50927992e-02 -7.20093131e-01 4.37775820e-01 4.71813232e-02
-5.36385357e-01 7.91078389e-01 6.03919864e-01 1.27597764e-01
-5.29634535e-01 8.21782231e-01 -8.36170912e-01 6.04507267e-01
9.03229833e-01 -3.55283171e-01 -6.32720709e-01 9.07220244e-02
4.24352139e-01 7.93264985e-01 1.54557571e-01 -2.44686842e-01
-3.41892280e-02 -1.12402225e+00 -1.20013230e-01 -4.75857079e-01
1.21736908e-02 -7.27109015e-01 2.76353329e-01 6.86765254e-01
-6.21906996e-01 6.33170664e-01 1.89481944e-01 8.22912216e-01
2.24187553e-01 -1.92046210e-01 7.07510650e-01 -8.36243108e-02
-3.73147607e-01 5.37668407e-01 -5.88087618e-01 3.94643217e-01
1.26518798e+00 1.87606990e-01 -2.92783137e-03 -8.45937908e-01
-7.50955403e-01 8.34877372e-01 1.67216942e-01 1.34793788e-01
3.67760658e-01 -1.21822488e+00 -4.87202466e-01 -2.24978775e-01
-4.14285481e-01 -1.63938999e-01 5.16830608e-02 7.69517362e-01
9.77717862e-02 6.10575020e-01 -8.00571516e-02 6.83329552e-02
-5.37761867e-01 7.91987777e-01 7.26732731e-01 -9.66773033e-01
-5.57879388e-01 6.31683350e-01 1.83077574e-01 -8.44138414e-02
5.41178942e-01 -3.50301147e-01 -6.11529090e-02 1.72810256e-01
5.65331578e-01 5.58142662e-01 -9.04557258e-02 -3.69129665e-02
-3.95464152e-02 -2.00208440e-01 -1.66190401e-01 -7.57586718e-01
1.56707883e+00 -2.03779057e-01 4.68263596e-01 4.02827412e-01
3.41693342e-01 -4.06771660e-01 -2.18804073e+00 -5.71566999e-01
6.50346816e-01 -4.51684743e-01 1.79282874e-01 -1.02820039e+00
-1.14074731e+00 4.65484142e-01 2.13189915e-01 2.10703045e-01
8.82357419e-01 -7.00083613e-01 7.33819664e-01 4.01691765e-01
5.57628155e-01 -1.74180984e+00 3.93305086e-02 4.22705352e-01
7.18032181e-01 -7.60372400e-01 1.21181551e-02 3.51161212e-01
-1.22481334e+00 7.89225996e-01 2.26079389e-01 -1.73831165e-01
7.85927355e-01 4.59065229e-01 -3.63661915e-01 2.88274884e-01
-1.19650114e+00 1.65780395e-01 4.84845005e-02 4.06513721e-01
-4.03844714e-01 4.69142497e-01 9.82886553e-02 8.60019922e-01
-1.51568456e-02 -2.97303647e-01 5.33246875e-01 1.08162129e+00
-3.92556667e-01 -9.77842033e-01 -3.02677423e-01 5.56654751e-01
-9.22468960e-01 1.47849666e-02 9.79297683e-02 7.28568912e-01
-6.48348033e-01 7.76207447e-01 1.12700596e-01 3.14814836e-01
3.80545676e-01 5.93386367e-02 2.42708117e-01 -6.23031080e-01
-6.20162189e-01 2.85341412e-01 2.53715754e-01 -9.22047257e-01
3.03303227e-02 -1.12218785e+00 -1.03286219e+00 -2.66303807e-01
-2.47035876e-01 6.32335469e-02 4.19903100e-01 1.10560846e+00
5.92608079e-02 2.86467016e-01 1.29902947e+00 -7.54208088e-01
-2.03693819e+00 -2.25254536e-01 -1.04381502e+00 -3.13026160e-02
4.89421278e-01 -8.56248856e-01 -3.58643293e-01 -5.58526039e-01] | [4.262150764465332, 2.5767571926116943] |
a3ed8adb-8029-47cf-98b8-5bfcd20372d3 | how-object-information-improves-skeleton | 2306.05844 | null | https://arxiv.org/abs/2306.05844v1 | https://arxiv.org/pdf/2306.05844v1.pdf | How Object Information Improves Skeleton-based Human Action Recognition in Assembly Tasks | As the use of collaborative robots (cobots) in industrial manufacturing continues to grow, human action recognition for effective human-robot collaboration becomes increasingly important. This ability is crucial for cobots to act autonomously and assist in assembly tasks. Recently, skeleton-based approaches are often used as they tend to generalize better to different people and environments. However, when processing skeletons alone, information about the objects a human interacts with is lost. Therefore, we present a novel approach of integrating object information into skeleton-based action recognition. We enhance two state-of-the-art methods by treating object centers as further skeleton joints. Our experiments on the assembly dataset IKEA ASM show that our approach improves the performance of these state-of-the-art methods to a large extent when combining skeleton joints with objects predicted by a state-of-the-art instance segmentation model. Our research sheds light on the benefits of combining skeleton joints with object information for human action recognition in assembly tasks. We analyze the effect of the object detector on the combination for action classification and discuss the important factors that must be taken into account. | ['Horst-Michael Gross', 'Markus Eisenbach', 'Sebastian Baake', 'Mona Köhler', 'Dustin Aganian'] | 2023-06-09 | null | null | null | null | ['skeleton-based-action-recognition', 'action-classification', 'action-recognition-in-videos', 'action-recognition'] | ['computer-vision', 'computer-vision', 'computer-vision', 'computer-vision'] | [ 1.32973179e-01 1.28254980e-01 -5.08991443e-02 -1.68298125e-01
-2.99167693e-01 -1.93607569e-01 4.31749254e-01 -3.53575833e-02
-4.10092056e-01 4.10238445e-01 -1.71852335e-01 2.69974023e-01
-1.49222121e-01 -5.57724237e-01 -5.51402390e-01 -6.61511660e-01
1.17149577e-01 8.40071023e-01 7.70979226e-01 -2.88905442e-01
2.85030156e-01 8.85198474e-01 -1.70535409e+00 3.05157006e-01
8.82880270e-01 8.13218474e-01 5.61945438e-01 7.98593402e-01
-3.10206097e-02 8.62243116e-01 -7.70397663e-01 -1.09198406e-01
4.17527258e-01 -5.91409504e-01 -1.00467098e+00 5.42159855e-01
1.31914854e-01 -1.78007409e-01 2.66832467e-02 8.63259435e-01
1.20900288e-01 3.20579886e-01 8.16683710e-01 -1.32586932e+00
-1.95257157e-01 3.94266367e-01 -4.55193311e-01 -2.82135397e-01
4.40852463e-01 3.24569821e-01 7.65015721e-01 -5.95483363e-01
8.85295749e-01 1.46450782e+00 5.02162337e-01 3.03457439e-01
-9.63734388e-01 -3.97328526e-01 2.63314068e-01 4.26044405e-01
-9.12625074e-01 -1.95247754e-01 7.95126915e-01 -5.70759296e-01
9.39908326e-01 -5.03479056e-02 8.03129673e-01 7.42025018e-01
4.17357624e-01 1.27004850e+00 8.74415338e-01 -5.78105986e-01
1.57591239e-01 -3.15495998e-01 -5.17439879e-02 9.84953701e-01
2.42567629e-01 -3.39125246e-01 -4.29450750e-01 2.51870722e-01
1.20203841e+00 1.09036282e-01 2.12007508e-01 -9.20527697e-01
-1.45371532e+00 5.81987202e-01 4.52187926e-01 5.86436868e-01
-7.31624484e-01 3.13211471e-01 4.18175638e-01 1.35781884e-01
3.09979409e-01 7.48424530e-01 -6.21757090e-01 -3.40187937e-01
-4.13840652e-01 3.39489102e-01 8.46402109e-01 8.62866104e-01
5.52040815e-01 -4.01326537e-01 5.51867438e-03 1.01732290e+00
2.98086673e-01 2.03100994e-01 1.90669343e-01 -1.45537770e+00
3.53546262e-01 9.43054318e-01 2.24995360e-01 -6.33964479e-01
-5.73389769e-01 1.10677533e-01 -3.10537070e-01 6.83239996e-01
9.00213659e-01 8.85653943e-02 -1.09693182e+00 1.04016364e+00
4.60979104e-01 -3.65323782e-01 -9.70289484e-02 8.94379020e-01
2.01519743e-01 1.50678307e-01 1.35249421e-01 1.24081001e-01
1.40701163e+00 -1.53444529e+00 -8.01009953e-01 -1.87144414e-01
6.72572672e-01 -9.23941791e-01 7.56928682e-01 7.40927577e-01
-8.82910073e-01 -9.17578578e-01 -9.13261533e-01 3.31933312e-02
-3.51113230e-01 4.40878510e-01 7.98559904e-01 4.70475167e-01
-4.09077972e-01 9.93435621e-01 -1.44469547e+00 -8.95540655e-01
4.34623778e-01 5.55890858e-01 -5.21874487e-01 -3.08242366e-02
-5.82928181e-01 1.39724004e+00 3.50929022e-01 7.33034627e-04
-6.72321498e-01 -6.60401210e-02 -7.66731381e-01 -2.90914655e-01
7.77144313e-01 -6.54100597e-01 1.60575628e+00 -6.75915480e-01
-1.82685745e+00 6.05452597e-01 1.60196841e-01 -3.85374993e-01
8.18436325e-01 -6.05420589e-01 3.92241068e-02 4.87678468e-01
1.42107859e-01 8.00931692e-01 8.05568695e-01 -1.42241168e+00
-8.45989406e-01 -5.76050341e-01 9.82595086e-02 2.62414753e-01
1.39465496e-01 8.76266137e-02 -4.80026841e-01 -4.19271797e-01
3.74321312e-01 -1.34953845e+00 -4.83165264e-01 3.85682195e-01
-2.68058270e-01 -4.84714538e-01 9.60361004e-01 -5.98898888e-01
5.73592901e-01 -1.97097659e+00 5.46147466e-01 -8.32650736e-02
-4.53206711e-02 2.53996283e-01 5.76001368e-02 5.27420044e-01
2.98052579e-01 -4.38791245e-01 -2.02797338e-01 -5.32530606e-01
2.26337719e-03 6.06976449e-01 4.49352801e-01 4.47771311e-01
1.63878217e-01 6.93120360e-01 -9.97103035e-01 -7.45724976e-01
6.06482863e-01 2.57250201e-02 -2.76579648e-01 3.02942514e-01
-2.25780755e-01 9.12484944e-01 -6.89194143e-01 8.68359089e-01
1.36500984e-01 1.00610264e-01 1.57469481e-01 7.53615145e-03
-3.92604619e-02 -5.52841239e-02 -1.17844832e+00 1.90356052e+00
-3.67133766e-01 1.71768919e-01 3.99433434e-01 -1.05930912e+00
9.61230040e-01 3.64000022e-01 9.53000128e-01 -1.09312475e-01
1.30157501e-01 4.09508318e-01 3.13289106e-01 -6.58716440e-01
4.30638552e-01 -7.77215958e-02 4.56296001e-03 2.44537607e-01
2.33883470e-01 -6.63744748e-01 4.60532010e-01 -2.30064660e-01
1.20278823e+00 6.45421863e-01 5.04477262e-01 9.30830464e-03
5.51286936e-01 3.22382063e-01 4.05219018e-01 5.81355333e-01
-5.29706776e-01 5.29443145e-01 3.42876673e-01 -3.59027296e-01
-8.14259827e-01 -9.04820621e-01 1.98333085e-01 1.16298211e+00
2.96552837e-01 -1.21775679e-02 -8.51628244e-01 -1.09469616e+00
1.18005611e-01 5.08221507e-01 -4.59870130e-01 -1.38121799e-01
-7.79130876e-01 -3.91733527e-01 2.80044049e-01 9.08123910e-01
4.94817734e-01 -1.45465648e+00 -1.10374606e+00 4.25177068e-01
-8.09314996e-02 -1.26800585e+00 2.45924238e-02 3.09273601e-01
-1.15656269e+00 -1.43196619e+00 -8.23288500e-01 -7.02320814e-01
6.68703735e-01 2.54410028e-01 6.51119173e-01 7.77135743e-03
-5.08201480e-01 7.34561861e-01 -7.01688826e-01 -6.46827459e-01
-7.60990083e-01 1.91562429e-01 2.05237627e-01 -1.67059511e-01
4.64248769e-02 -4.23966795e-01 -3.14889193e-01 7.52506256e-01
-5.96932948e-01 4.46470492e-02 1.02987933e+00 5.12651742e-01
2.40825206e-01 9.76254568e-02 2.63476342e-01 -8.05370450e-01
3.31493080e-01 8.34046677e-02 -2.22135812e-01 3.06037098e-01
-3.02791804e-01 1.14287212e-01 3.91705930e-01 -4.70308065e-01
-1.30923963e+00 6.10086381e-01 5.87085858e-02 -5.26260436e-01
-5.65199852e-01 1.12142675e-01 -3.05372208e-01 -1.09281612e-03
3.42778772e-01 -2.63943613e-01 3.12540889e-01 -7.37876654e-01
4.15594548e-01 5.95701516e-01 5.58642805e-01 -6.39752150e-01
3.42963368e-01 5.17892778e-01 2.07446650e-01 -8.82752717e-01
-5.36816776e-01 -9.77400661e-01 -1.51404238e+00 -5.94394207e-01
1.25394046e+00 -4.20408338e-01 -6.85893953e-01 8.71953607e-01
-1.45229912e+00 -3.97460550e-01 -4.24862504e-01 6.57865822e-01
-1.06076825e+00 4.42548573e-01 -8.40056896e-01 -9.70373869e-01
1.66205093e-01 -1.36782193e+00 1.31979179e+00 -1.11462191e-01
-4.18022960e-01 -5.63784957e-01 -7.77159929e-02 8.80122662e-01
-1.63389787e-01 2.59183317e-01 5.78635454e-01 -7.21472383e-01
-4.43637192e-01 -3.27417910e-01 1.10228837e-01 4.95043039e-01
3.79698634e-01 -1.78977493e-02 -5.25764644e-01 1.99098065e-01
-3.62060778e-03 -2.78924435e-01 6.85897291e-01 1.43925413e-01
8.07769775e-01 2.10786432e-01 -5.65352857e-01 -2.02459306e-01
7.46124208e-01 3.17933172e-01 5.92545807e-01 3.16879511e-01
8.55038404e-01 1.06054187e+00 1.29642379e+00 2.97884315e-01
-3.37173715e-02 9.36604559e-01 6.93022907e-01 1.19558200e-01
-1.77548036e-01 6.88461289e-02 5.18991351e-01 7.68672764e-01
-7.55531132e-01 3.29307280e-02 -1.01324105e+00 4.12116349e-01
-2.31022072e+00 -7.19815254e-01 -3.93594474e-01 1.74702418e+00
4.88203883e-01 2.22798049e-01 2.94010043e-01 2.99839288e-01
7.19718039e-01 -1.82676554e-01 -4.25568938e-01 -9.19925943e-02
4.71801281e-01 2.07543885e-03 5.21328926e-01 7.87218213e-02
-1.29265344e+00 1.10258925e+00 6.00697136e+00 6.62091434e-01
-5.64005554e-01 7.82009140e-02 -2.42390595e-02 4.47421223e-01
8.51504564e-01 -2.35044345e-01 -5.79971313e-01 4.14320454e-02
4.23279285e-01 5.85705936e-01 6.85306787e-02 1.25591147e+00
8.38166773e-02 -7.00808465e-01 -1.30084658e+00 7.25174367e-01
6.50559887e-02 -6.84313595e-01 -3.69274348e-01 9.47988331e-02
6.66463733e-01 -2.73562491e-01 -4.54547912e-01 2.72762179e-01
5.05415618e-01 -6.23514295e-01 9.02617335e-01 5.59881151e-01
-6.52918294e-02 -4.00217146e-01 9.74583268e-01 4.59101766e-01
-1.28870916e+00 -2.78883696e-01 -2.02854335e-01 -2.37680078e-01
5.61051846e-01 2.58986980e-01 -1.11104500e+00 7.82272696e-01
4.65383172e-01 6.83508337e-01 -5.68422139e-01 1.03143632e+00
-6.13202691e-01 1.31655931e-01 -2.36574978e-01 -4.61259373e-02
1.25979602e-01 -3.55383426e-01 5.22058964e-01 8.45856786e-01
1.31600961e-01 -3.63726467e-02 7.25785255e-01 6.71279490e-01
4.33616966e-01 -6.16288781e-02 -4.75441039e-01 -2.30960205e-01
6.82438239e-02 1.13979435e+00 -1.48253417e+00 -2.45200068e-01
-2.55697876e-01 1.33883035e+00 1.31529659e-01 -4.74318489e-02
-7.04543829e-01 -2.61578709e-01 5.19902170e-01 1.39069051e-01
4.97855127e-01 -8.26555610e-01 -3.21043193e-01 -5.70268393e-01
-4.74589039e-03 -8.07024240e-01 1.10568553e-01 -7.58864820e-01
-1.08277988e+00 2.56005470e-02 2.52502501e-01 -1.31327391e+00
-3.39924723e-01 -1.09882963e+00 -2.48422265e-01 1.95164725e-01
-7.65881717e-01 -1.53556931e+00 -1.95644200e-01 1.88931361e-01
1.00798726e+00 5.10170907e-02 7.13397443e-01 -1.07684001e-01
-3.01082820e-01 -1.91030934e-01 -2.84013927e-01 3.66658866e-01
6.49107575e-01 -1.17146969e+00 1.89641118e-01 5.94418347e-01
3.14601302e-01 4.35984761e-01 7.48021722e-01 -1.01218259e+00
-1.28564739e+00 -6.58683360e-01 4.27460253e-01 -6.49949729e-01
6.44940555e-01 -1.47091746e-01 -6.24465942e-01 7.19322026e-01
-6.92732558e-02 -4.05026153e-02 1.96073756e-01 2.08449095e-01
2.10010991e-01 2.05083176e-01 -1.07303047e+00 5.16432047e-01
1.25292301e+00 -2.11595356e-01 -8.83144736e-01 3.73758823e-01
4.63800818e-01 -3.82775486e-01 -9.37642694e-01 5.24907231e-01
6.26896799e-01 -9.77943897e-01 9.44504201e-01 -4.92684811e-01
4.13672090e-01 -3.87176424e-01 9.13790166e-02 -1.16569388e+00
-8.77531096e-02 -2.68153578e-01 -1.15701139e-01 7.33261228e-01
4.37165685e-02 -3.80620927e-01 9.45788205e-01 4.44052100e-01
-4.93800581e-01 -7.33659387e-01 -7.69146562e-01 -1.18050253e+00
-3.14986378e-01 -3.28633875e-01 8.84988382e-02 7.12925553e-01
1.25446811e-01 2.96049953e-01 -6.45012856e-02 -1.35807604e-01
1.55082643e-01 7.84020964e-03 1.17827976e+00 -1.59755850e+00
-2.64147639e-01 -3.24387610e-01 -8.03894460e-01 -1.05461717e+00
2.53657103e-01 -4.20230001e-01 5.29898942e-01 -1.97085333e+00
-9.10021663e-02 -2.81096905e-01 -1.64803881e-02 7.15190709e-01
9.94678214e-03 1.36357084e-01 5.83996296e-01 3.63865077e-01
-9.06195939e-01 4.98793781e-01 1.71564341e+00 4.88186302e-03
-2.89594710e-01 3.41746360e-01 1.95201993e-01 1.15913689e+00
7.24090576e-01 -4.10534292e-01 -4.31709997e-02 -9.44992900e-02
-3.33714336e-01 -5.34294210e-02 3.21467519e-01 -1.42965746e+00
3.60995054e-01 -1.97350487e-01 2.65996456e-01 -5.16014218e-01
6.33327246e-01 -1.00687778e+00 1.58076342e-02 8.87344897e-01
-4.36907485e-02 -3.90670523e-02 -1.93545625e-01 6.16780758e-01
-1.46273509e-01 -4.78579611e-01 6.85513675e-01 -5.60962856e-01
-8.14005375e-01 -2.48132765e-01 -7.22489893e-01 -5.15043676e-01
1.28359628e+00 -4.89099115e-01 2.53257126e-01 -2.05852479e-01
-9.50936496e-01 4.94981147e-02 6.27921104e-01 6.72686100e-01
2.76035130e-01 -8.83365750e-01 -3.02838236e-01 2.52652750e-03
1.22135349e-01 1.77310169e-01 -1.07475609e-01 1.22024190e+00
-8.08422744e-01 4.05306965e-01 -5.29725909e-01 -7.68212616e-01
-1.40776098e+00 3.61351848e-01 5.17447218e-02 -5.50930977e-01
-6.27433360e-01 7.30549574e-01 -1.03913108e-02 -6.80147409e-01
2.16562524e-01 -6.43452168e-01 -2.19358385e-01 7.47514367e-02
9.61967558e-03 1.01623178e+00 8.40305164e-02 -6.17590725e-01
-2.89052993e-01 5.53506672e-01 3.52796726e-02 -2.55683273e-01
1.27371168e+00 1.27484217e-01 -2.21789405e-01 6.87120020e-01
5.98530710e-01 -1.75621673e-01 -1.39992988e+00 2.19298810e-01
2.93997198e-01 -3.92002851e-01 -4.68678862e-01 -7.22734153e-01
-8.85031760e-01 7.83114672e-01 3.84201616e-01 8.76248777e-02
6.34025455e-01 3.08278531e-01 6.89085782e-01 7.12592423e-01
9.51912045e-01 -1.56373620e+00 6.19226575e-01 6.37717664e-01
9.93722856e-01 -1.30477214e+00 2.64229923e-01 -9.85543013e-01
-8.86947513e-01 1.27175713e+00 7.04195082e-01 -1.74010113e-01
3.07114512e-01 6.19667470e-02 1.56698063e-01 -2.66182393e-01
-2.16325074e-02 -4.23717111e-01 1.76589668e-01 5.97145200e-01
2.12762609e-01 1.24887101e-01 -2.93618917e-01 3.68278444e-01
3.82603444e-02 -3.63735072e-02 3.06559771e-01 1.55056334e+00
-7.15100884e-01 -1.45834756e+00 -7.35516846e-01 2.92501628e-01
-3.48348051e-01 7.40989685e-01 -7.23527610e-01 1.11851680e+00
3.33374500e-01 9.51450109e-01 -1.16514951e-01 -1.65508121e-01
5.94405174e-01 2.75997400e-01 9.00776982e-01 -1.00013554e+00
-5.95433533e-01 1.01095811e-02 2.41932467e-01 -7.88731754e-01
-8.09233546e-01 -1.12844694e+00 -1.72584426e+00 4.21193272e-01
-4.86187518e-01 -1.24604844e-01 8.23102176e-01 1.18342555e+00
8.06940570e-02 6.31837547e-01 1.00416420e-02 -1.53056765e+00
-6.32547021e-01 -1.16573584e+00 -7.38222599e-01 4.85182583e-01
-2.71320939e-01 -1.17375040e+00 4.39014845e-02 2.98833489e-01] | [7.4779486656188965, 0.13686013221740723] |
4c538e64-eb6c-4534-a708-19e1bf87088a | summary-level-training-of-sentence-rewriting | 1909.08752 | null | https://arxiv.org/abs/1909.08752v3 | https://arxiv.org/pdf/1909.08752v3.pdf | Summary Level Training of Sentence Rewriting for Abstractive Summarization | As an attempt to combine extractive and abstractive summarization, Sentence Rewriting models adopt the strategy of extracting salient sentences from a document first and then paraphrasing the selected ones to generate a summary. However, the existing models in this framework mostly rely on sentence-level rewards or suboptimal labels, causing a mismatch between a training objective and evaluation metric. In this paper, we present a novel training signal that directly maximizes summary-level ROUGE scores through reinforcement learning. In addition, we incorporate BERT into our model, making good use of its ability on natural language understanding. In extensive experiments, we show that a combination of our proposed model and training procedure obtains new state-of-the-art performance on both CNN/Daily Mail and New York Times datasets. We also demonstrate that it generalizes better on DUC-2002 test set. | ['Sang-goo Lee', 'Sanghwan Bae', 'Taeuk Kim', 'Jihoon Kim'] | 2019-09-19 | summary-level-training-of-sentence-rewriting-1 | https://aclanthology.org/D19-5402 | https://aclanthology.org/D19-5402.pdf | ws-2019-11 | ['extractive-document-summarization'] | ['natural-language-processing'] | [ 3.56248915e-01 1.93205476e-01 -4.60580319e-01 -5.29989779e-01
-1.17073286e+00 -5.85056722e-01 7.73641348e-01 2.18927145e-01
-5.92274249e-01 1.03326690e+00 7.26709366e-01 -2.03317806e-01
2.43874416e-01 -6.24200284e-01 -6.34499788e-01 -2.41695330e-01
3.79407942e-01 1.20626271e-01 4.99643981e-02 -4.79962587e-01
5.30592442e-01 2.31267944e-01 -1.02623260e+00 4.21131164e-01
1.29084861e+00 5.54469883e-01 3.72527868e-01 7.13981330e-01
-1.59511358e-01 9.73563671e-01 -8.83480370e-01 -6.25751853e-01
1.28577262e-01 -7.63312101e-01 -9.85158861e-01 1.81205496e-01
4.99269426e-01 -3.39858770e-01 -4.73663688e-01 9.66087043e-01
5.01879036e-01 2.97374129e-01 6.32049024e-01 -7.35414505e-01
-9.35719430e-01 8.74177814e-01 -4.57323611e-01 4.57467079e-01
4.43715304e-01 2.04803735e-01 1.52105093e+00 -8.12106550e-01
6.71134353e-01 9.86710072e-01 3.11996996e-01 6.55855775e-01
-1.01074278e+00 -3.35502803e-01 4.43521410e-01 2.04399422e-01
-8.30544293e-01 -5.47779799e-01 9.35323298e-01 4.38502654e-02
1.28154600e+00 2.06587836e-01 4.59086955e-01 1.14867735e+00
4.10447806e-01 1.17081118e+00 9.33461905e-01 -5.63847899e-01
1.17040105e-01 -9.46006551e-02 3.48991454e-01 7.34891713e-01
4.29990292e-02 -2.10683003e-01 -5.68000793e-01 8.02429691e-02
2.70691633e-01 -1.25968486e-01 -4.10233259e-01 7.05880970e-02
-1.15325582e+00 8.61579955e-01 5.54965258e-01 3.05935204e-01
-4.77110147e-01 1.13576360e-01 5.06855965e-01 2.79863000e-01
5.78629911e-01 8.04837227e-01 -8.81807953e-02 -1.64915681e-01
-1.20192194e+00 2.74881899e-01 7.46005714e-01 8.60908568e-01
6.19502962e-01 9.62513462e-02 -7.30135202e-01 9.82486606e-01
-8.71693268e-02 2.39078775e-01 6.21275961e-01 -8.70725572e-01
7.15942383e-01 4.80459988e-01 5.78779057e-02 -7.61002481e-01
-3.18653911e-01 -7.97850490e-01 -8.22149515e-01 -1.34831905e-01
-5.64040951e-02 -2.56972939e-01 -9.07425106e-01 1.83082426e+00
-2.56790429e-01 1.68265104e-02 3.62114757e-01 7.56855428e-01
9.08278167e-01 7.87072539e-01 -7.70591199e-02 -4.68444407e-01
1.08909726e+00 -1.52376747e+00 -7.30665863e-01 -4.70268518e-01
5.16358137e-01 -5.23396909e-01 1.28006804e+00 2.43824795e-01
-1.39195859e+00 -5.81321239e-01 -1.38076568e+00 -3.79429191e-01
-6.86502159e-02 3.92484576e-01 3.62099767e-01 2.40105122e-01
-1.04073048e+00 8.40546012e-01 -6.04804218e-01 -3.84364754e-01
3.24370354e-01 3.74365039e-02 -4.51420946e-03 7.44969249e-02
-1.18411446e+00 1.02790296e+00 2.59949625e-01 -3.07830162e-02
-6.74823582e-01 -3.70635182e-01 -7.92640448e-01 3.78907889e-01
5.23985624e-01 -1.00849402e+00 1.58335817e+00 -8.30593109e-01
-1.82834280e+00 7.12598562e-01 -3.23650688e-01 -8.42821300e-01
4.27003026e-01 -4.93827283e-01 -1.88811257e-01 2.56505579e-01
1.85370803e-01 7.17291594e-01 6.24984384e-01 -1.06069732e+00
-6.34588122e-01 -1.55999184e-01 4.22224224e-01 3.75554025e-01
-4.56265181e-01 -1.25395536e-01 -2.86849171e-01 -7.15720713e-01
-2.77033120e-01 -6.80508852e-01 -2.39937738e-01 -6.02248967e-01
-5.49398005e-01 -4.57963169e-01 3.09947848e-01 -7.50713885e-01
1.48259485e+00 -1.81061351e+00 4.68833834e-01 -3.79762590e-01
1.56785280e-01 3.13729972e-01 -3.48199069e-01 6.92657948e-01
2.27791369e-01 1.64533406e-01 -4.65733737e-01 -6.31051481e-01
-3.48116159e-02 6.94985210e-04 -6.32175088e-01 6.19244501e-02
5.42312920e-01 1.08809364e+00 -1.11671269e+00 -5.15768766e-01
-1.82029292e-01 -5.70472404e-02 -5.14056504e-01 1.26388162e-01
-4.16106373e-01 1.06466055e-01 -4.43694711e-01 3.33924979e-01
3.44217390e-01 -3.36709112e-01 7.42208734e-02 -1.83585379e-02
-8.13540146e-02 7.74656415e-01 -4.88677114e-01 1.95701611e+00
-5.94034493e-01 6.54968441e-01 -2.86556572e-01 -9.81487215e-01
9.33135808e-01 1.33489534e-01 -8.54214095e-03 -7.38668025e-01
-5.35884313e-02 1.71374500e-01 -1.01451380e-02 -3.61327529e-01
1.08556068e+00 -8.13816935e-02 -8.35712329e-02 5.95569432e-01
1.85954332e-01 -3.46366197e-01 5.67094326e-01 4.89564925e-01
1.15619898e+00 2.24362791e-01 5.46796322e-01 -1.05520360e-01
6.10734701e-01 -6.76117763e-02 5.09263515e-01 9.55604672e-01
-4.95182350e-02 9.43582594e-01 8.10764134e-01 -2.32647192e-02
-1.04791677e+00 -8.45476329e-01 2.92800456e-01 9.12135482e-01
-3.39484178e-02 -5.62444031e-01 -7.27212965e-01 -1.12131965e+00
-3.79733384e-01 1.28356540e+00 -4.51962948e-01 -2.69569725e-01
-8.17282498e-01 -4.52966124e-01 5.20045817e-01 5.29715538e-01
5.28044641e-01 -1.31978071e+00 -6.09929979e-01 2.73856819e-01
-3.60836715e-01 -1.09069109e+00 -8.14090014e-01 2.49589272e-02
-8.37225199e-01 -6.59171462e-01 -9.35308695e-01 -7.53372729e-01
4.36527938e-01 5.27246237e-01 1.38370192e+00 1.62259325e-01
2.24979609e-01 1.06068365e-01 -4.54323351e-01 -4.42497253e-01
-6.24628544e-01 7.49592781e-01 -2.21865416e-01 -1.71546832e-01
3.79859544e-02 -4.61442471e-01 -5.11069834e-01 -3.16105038e-01
-8.72435272e-01 2.46860638e-01 8.39148819e-01 1.09099913e+00
1.92656010e-01 -5.88603139e-01 1.18119717e+00 -9.15418923e-01
1.25807798e+00 -2.64591187e-01 -2.11357698e-01 5.39153755e-01
-7.21589148e-01 2.16011584e-01 9.82063055e-01 -2.65808944e-02
-1.21438444e+00 -2.75135845e-01 -1.80821985e-01 -2.46115059e-01
9.55173224e-02 7.98580825e-01 6.61735162e-02 6.02462947e-01
7.08393812e-01 5.14786243e-01 -8.77913386e-02 -2.42062792e-01
4.92330790e-01 7.40904629e-01 6.92515254e-01 -4.16231066e-01
6.03978336e-01 2.06945434e-01 -3.43539596e-01 -6.39476418e-01
-1.24850500e+00 -2.57682443e-01 -5.58021784e-01 -4.71794754e-02
6.66319013e-01 -7.38915503e-01 -2.37557635e-01 1.95180863e-01
-1.45697379e+00 -8.81767869e-02 -4.02486056e-01 3.27472746e-01
-6.48036003e-01 6.92030251e-01 -6.44114971e-01 -6.15172625e-01
-7.64052272e-01 -1.02333403e+00 1.08070397e+00 4.63497818e-01
-2.91168004e-01 -7.26613581e-01 1.47027299e-01 3.47723663e-01
2.97206312e-01 -1.12208631e-02 9.69138622e-01 -7.81334519e-01
-4.70541060e-01 -1.42102823e-01 -2.63746321e-01 5.46926975e-01
1.98890686e-01 -1.21473419e-02 -7.45057583e-01 -2.36119419e-01
-2.42889505e-02 -5.12630761e-01 1.42408001e+00 2.95262933e-01
9.48113799e-01 -3.17999125e-01 1.84131972e-02 2.57993072e-01
1.20409727e+00 -1.16283796e-03 7.13865042e-01 3.53397250e-01
3.47042024e-01 5.59764922e-01 7.00171113e-01 3.12257826e-01
3.78886700e-01 5.00579834e-01 1.81247741e-01 -4.80975136e-02
-2.26412266e-01 -3.12000901e-01 4.93775189e-01 8.90160203e-01
1.14867419e-01 -5.80831230e-01 -5.92695832e-01 5.88436663e-01
-2.10281181e+00 -1.15367532e+00 3.10908943e-01 2.02973318e+00
1.12628937e+00 5.33406615e-01 1.36257693e-01 -2.75714725e-01
5.87054849e-01 7.35055387e-01 -5.43454707e-01 -5.80309510e-01
-3.38019520e-01 1.93024263e-01 2.11686492e-01 6.47216618e-01
-8.37846935e-01 1.18536246e+00 6.30134535e+00 8.23460698e-01
-1.11343586e+00 -7.29819387e-02 7.19199121e-01 -2.82562256e-01
-5.54115593e-01 -2.41934080e-02 -7.26449966e-01 2.57153481e-01
8.68770421e-01 -5.69812894e-01 3.02492321e-01 5.96542358e-01
3.42653602e-01 1.94780214e-03 -1.17929327e+00 5.24298668e-01
4.64512855e-01 -1.61872244e+00 1.67468056e-01 -2.29715228e-01
8.91662598e-01 -5.00182547e-02 -1.61288306e-01 6.14733577e-01
2.29827493e-01 -9.48363781e-01 7.12844431e-01 5.37363827e-01
4.49272007e-01 -7.77799904e-01 6.92068040e-01 5.16045094e-01
-5.82897186e-01 5.09647168e-02 -3.86811107e-01 -2.31540412e-01
3.48733902e-01 4.51561213e-01 -9.48581755e-01 8.54976416e-01
6.89520165e-02 9.18406904e-01 -7.35808909e-01 8.87923896e-01
-5.53703368e-01 6.70869648e-01 1.17740780e-01 -5.42690814e-01
4.69534963e-01 -2.07379803e-01 6.88943863e-01 1.44184124e+00
1.90471992e-01 -1.67975612e-02 1.62334234e-01 9.58703041e-01
-5.24519980e-01 1.39283314e-01 -5.36973119e-01 -8.88914168e-02
2.87706673e-01 1.23803771e+00 -5.74502349e-01 -6.32348180e-01
-4.80398059e-01 1.10728776e+00 8.07410419e-01 4.52873915e-01
-8.37988496e-01 -5.61392844e-01 2.52256781e-01 -3.05981487e-01
3.58826786e-01 -2.80383468e-01 -5.30737221e-01 -1.50924301e+00
2.53249437e-01 -9.22521651e-01 4.82878499e-02 -7.51025319e-01
-1.12773037e+00 7.88833499e-01 -2.59815641e-02 -1.25045669e+00
-5.08652210e-01 -1.93829417e-01 -8.68156314e-01 7.21191883e-01
-1.81612170e+00 -8.82579029e-01 -8.18091854e-02 -1.27762079e-01
1.21819937e+00 -2.30517596e-01 6.04917526e-01 -1.17845267e-01
-6.80433035e-01 5.47129452e-01 7.62423202e-02 -3.77421565e-02
7.60836065e-01 -1.46075845e+00 7.90665686e-01 1.10916638e+00
2.48364538e-01 5.49315393e-01 8.52378190e-01 -4.70166177e-01
-9.71143365e-01 -9.06749308e-01 1.16880965e+00 -1.40476391e-01
4.91400242e-01 -1.53717607e-01 -8.51307511e-01 6.07746184e-01
9.22297478e-01 -6.64303362e-01 4.62338686e-01 1.46931306e-01
-2.20300958e-01 -7.37723038e-02 -7.87679970e-01 1.00526941e+00
1.03879404e+00 -2.94225872e-01 -1.00122428e+00 5.00432134e-01
9.68337178e-01 -4.70678240e-01 -4.64609802e-01 3.43823075e-01
3.29434514e-01 -1.06012583e+00 6.22496665e-01 -8.84833872e-01
1.23139930e+00 7.64361918e-02 6.67379722e-02 -1.73517382e+00
-4.48842257e-01 -6.80248380e-01 -7.78409317e-02 1.35206878e+00
6.62955582e-01 -3.99206221e-01 6.43513501e-01 2.38901392e-01
-5.53503573e-01 -1.09102082e+00 -7.26279557e-01 -6.91093326e-01
2.94660807e-01 3.33186202e-02 4.23409909e-01 5.63574970e-01
2.36702681e-01 1.03326416e+00 -3.89478505e-01 -2.51515269e-01
2.95905083e-01 3.94911379e-01 7.14345396e-01 -7.72587240e-01
-3.58633220e-01 -8.32440913e-01 2.40261123e-01 -1.56197834e+00
5.67732573e-01 -1.02577913e+00 2.40119115e-01 -1.88231468e+00
4.15901780e-01 2.43854046e-01 -3.21937740e-01 2.66962856e-01
-5.24356782e-01 -8.04628432e-02 4.99124527e-01 8.37431028e-02
-8.33796561e-01 9.49523449e-01 1.28233325e+00 -3.84832293e-01
-2.66681463e-01 2.36929748e-02 -1.06092775e+00 4.80766207e-01
8.13728631e-01 -2.92533338e-01 -3.09319168e-01 -6.45734608e-01
3.95054892e-02 2.41771027e-01 1.35882363e-01 -7.81851411e-01
1.76641166e-01 -2.20349401e-01 1.23859674e-01 -7.83153713e-01
1.99143037e-01 -1.48552641e-01 -6.86176360e-01 3.47999692e-01
-9.32476759e-01 2.05358520e-01 3.31160203e-02 6.53530061e-01
-3.02467734e-01 -6.70978010e-01 6.19565010e-01 -2.69135624e-01
-3.62434149e-01 -1.56192437e-01 -2.49531701e-01 1.83133021e-01
6.26662135e-01 2.93455236e-02 -6.44998610e-01 -6.98319137e-01
-3.64505351e-01 4.49311256e-01 3.62160087e-01 5.64503670e-01
6.68393731e-01 -1.17037964e+00 -1.16259992e+00 -2.74664819e-01
2.49474458e-02 -1.34163439e-01 1.10594526e-01 6.67328715e-01
-3.88526112e-01 6.63570523e-01 -5.50695062e-02 -3.31113338e-01
-1.07127321e+00 3.56600940e-01 1.46855816e-01 -7.21074045e-01
-6.09098375e-01 5.40961742e-01 1.38183802e-01 -2.79865474e-01
2.08187059e-01 -4.67873573e-01 -2.33074144e-01 2.64333207e-02
4.08535033e-01 2.72383571e-01 1.55004874e-01 -2.84308076e-01
-9.93804261e-02 1.48926869e-01 -5.14986455e-01 -3.16052526e-01
1.23677897e+00 -1.73055857e-01 7.75770023e-02 3.93196046e-01
1.10184216e+00 1.40499026e-01 -1.02976966e+00 -4.01878178e-01
6.76025301e-02 -2.29828507e-01 -1.73733041e-01 -8.42245340e-01
-6.50005698e-01 8.72415245e-01 -1.28348887e-01 3.47662956e-01
1.13831306e+00 -9.33659598e-02 1.04926002e+00 7.58147120e-01
-1.98907475e-03 -1.19727468e+00 3.68926018e-01 7.03472793e-01
1.23224938e+00 -1.24438071e+00 1.66081607e-01 -2.12241814e-01
-9.17816341e-01 1.20782828e+00 6.74737811e-01 -3.99784982e-01
-8.61189663e-02 -1.65451348e-01 -1.15027435e-01 4.82094884e-02
-1.12536824e+00 -7.40752369e-02 2.50422239e-01 1.90653875e-01
6.66196942e-01 -1.56789958e-01 -8.98521304e-01 6.78530812e-01
-4.56783354e-01 1.22960314e-01 9.82041240e-01 8.36014986e-01
-6.91119492e-01 -1.02433825e+00 1.15732320e-01 5.48156202e-01
-5.18494904e-01 -3.28981370e-01 -6.00526094e-01 4.69474882e-01
-6.48177207e-01 9.83733237e-01 -4.06812169e-02 -2.65111625e-01
5.35354435e-01 1.35510281e-01 4.50101227e-01 -8.86407733e-01
-7.88555264e-01 1.28897294e-01 3.51374507e-01 -9.77770537e-02
-2.75034547e-01 -5.73581398e-01 -1.28346264e+00 -4.91845794e-02
-3.47703427e-01 2.92071313e-01 4.24934626e-01 1.03259373e+00
5.32402158e-01 8.76182795e-01 8.46302986e-01 -7.53409564e-01
-1.16204977e+00 -1.30283248e+00 -3.57588679e-01 3.47380996e-01
4.15521026e-01 -1.21299393e-01 -1.84800014e-01 -9.87436175e-02] | [12.220596313476562, 9.303250312805176] |
6b6d10fd-b3c5-4702-af72-151810df2718 | mitigating-molecular-aggregation-in-drug | 2306.02206 | null | https://arxiv.org/abs/2306.02206v1 | https://arxiv.org/pdf/2306.02206v1.pdf | Mitigating Molecular Aggregation in Drug Discovery with Predictive Insights from Explainable AI | As the importance of high-throughput screening (HTS) continues to grow due to its value in early stage drug discovery and data generation for training machine learning models, there is a growing need for robust methods for pre-screening compounds to identify and prevent false-positive hits. Small, colloidally aggregating molecules are one of the primary sources of false-positive hits in high-throughput screens, making them an ideal candidate to target for removal from libraries using predictive pre-screening tools. However, a lack of understanding of the causes of molecular aggregation introduces difficulty in the development of predictive tools for detecting aggregating molecules. Herein, we present an examination of the molecular features differentiating datasets of aggregating and non-aggregating molecules, as well as a machine learning approach to predicting molecular aggregation. Our method uses explainable graph neural networks and counterfactuals to reliably predict and explain aggregation, giving additional insights and design rules for future screening. The integration of this method in HTS approaches will help combat false positives, providing better lead molecules more rapidly and thus accelerating drug discovery cycles. | ['Rebecca L. Davis', 'Pascal Friederich', 'Kaitlin A. Isfeld', 'Jonas Teufel', 'Hunter Sturm'] | 2023-06-03 | null | null | null | null | ['drug-discovery'] | ['medical'] | [ 5.22018969e-01 -1.71903893e-01 -1.83249250e-01 -4.89424542e-02
-5.85411012e-01 -8.15123439e-01 3.92154664e-01 9.81508791e-01
-1.89069360e-01 1.41857839e+00 7.37360725e-03 -6.59658015e-01
-6.56398907e-02 -8.03626359e-01 -5.43013215e-01 -7.23538041e-01
-3.52387249e-01 6.50532663e-01 2.26051167e-01 -8.51010613e-04
3.91348213e-01 9.05648291e-01 -8.62836897e-01 4.76579726e-01
1.19899261e+00 6.62293911e-01 2.19667733e-01 5.65953791e-01
8.95332992e-02 5.35026073e-01 -7.34215617e-01 -4.49506849e-01
-7.15282261e-02 -7.15301991e-01 -5.69785714e-01 -2.64672160e-01
2.71492712e-02 1.56455621e-01 3.63883972e-01 6.55704975e-01
7.07761228e-01 -7.00793564e-02 7.34975636e-01 -4.18830454e-01
-8.03258300e-01 9.53088775e-02 -2.46551216e-01 1.49625450e-01
5.18573701e-01 2.93971151e-01 1.01358712e+00 -6.89946711e-01
5.40141046e-01 9.57929373e-01 3.57912153e-01 5.43404460e-01
-1.56264973e+00 -4.70480025e-01 -2.80716270e-01 1.66139416e-02
-8.99409950e-01 -3.76914918e-01 3.98260951e-01 -7.97943115e-01
9.08671081e-01 6.17600083e-01 8.06150079e-01 7.80092895e-01
5.27139008e-01 4.36306745e-01 1.15536880e+00 -1.78242713e-01
6.97671056e-01 5.49054407e-02 -2.29482129e-01 5.45555472e-01
6.51251197e-01 1.25599787e-01 -4.95543361e-01 -6.05980933e-01
5.26260972e-01 7.57765695e-02 -2.02828005e-01 -3.01747501e-01
-7.98651099e-01 9.26316381e-01 3.75471622e-01 2.86887050e-01
-4.16848928e-01 -3.09067696e-01 9.99676958e-02 -2.77982473e-01
4.44207668e-01 1.32615542e+00 -6.29104972e-01 8.41913968e-02
-6.93605065e-01 2.89544940e-01 6.14102662e-01 -1.15058683e-01
3.98654252e-01 4.31451797e-02 2.41023768e-02 4.26546127e-01
1.52402057e-03 2.13871643e-01 4.21619713e-01 -4.46236998e-01
5.54448776e-02 9.76430297e-01 3.38618875e-01 -6.35116935e-01
-5.69129109e-01 -5.87569475e-01 -3.60157192e-01 2.61696845e-01
4.84778017e-01 -1.12955108e-01 -7.27403522e-01 1.48265791e+00
2.84388095e-01 -4.02832553e-02 9.56997350e-02 6.40115201e-01
3.12937945e-01 4.72058922e-01 4.45307046e-01 -6.83747053e-01
9.38627601e-01 -2.02572793e-01 -1.23185866e-01 2.63323992e-01
8.75161409e-01 -7.85113156e-01 8.09252739e-01 4.62620884e-01
-8.53878975e-01 -6.78293174e-03 -8.71328175e-01 4.39189374e-01
-5.71890712e-01 -7.03565031e-02 9.77240443e-01 6.98416293e-01
-3.67094517e-01 1.21412921e+00 -7.09390521e-01 -1.81736812e-01
8.48244667e-01 9.02969718e-01 -3.28086376e-01 1.09912746e-01
-8.45489025e-01 9.40775514e-01 2.72427589e-01 1.84541997e-02
-5.61569989e-01 -7.75960982e-01 -4.60829556e-01 -1.32444143e-01
9.76767391e-02 -5.44600666e-01 5.92458546e-01 -5.70288658e-01
-1.37155211e+00 2.92665511e-01 -2.87833065e-01 -6.26652420e-01
1.51214540e-01 6.72724321e-02 -2.43568450e-01 -1.33421436e-01
1.94062635e-01 3.11740279e-01 2.13532403e-01 -9.05976176e-01
-4.32830423e-01 -6.12151384e-01 -4.11611527e-01 2.05697063e-02
6.67718127e-02 -8.52540061e-02 7.25109220e-01 -7.92502388e-02
-2.12987736e-01 -8.15544784e-01 -5.66374421e-01 -5.47523797e-01
-6.65202618e-01 -1.30454868e-01 9.28983450e-01 -3.78710449e-01
8.27955306e-01 -1.55059528e+00 1.48650095e-01 3.79221827e-01
5.47098339e-01 6.48916781e-01 -1.04646005e-01 6.52194321e-01
-1.42816514e-01 4.79879111e-01 -1.30879125e-02 5.76912820e-01
-7.21172154e-01 -3.01931351e-01 -2.55842537e-01 3.84041667e-01
5.59713542e-01 9.68036532e-01 -1.10289454e+00 4.31915000e-02
3.10417891e-01 4.24705088e-01 -3.21979851e-01 1.19417086e-01
-4.60507333e-01 8.22608590e-01 -6.85993731e-01 7.64075458e-01
1.93756312e-01 -5.49035847e-01 2.58470565e-01 1.22593418e-01
-3.44786316e-01 3.86592984e-01 -4.15554583e-01 7.18856692e-01
6.29653642e-03 3.41861516e-01 -8.50460827e-01 -6.53059483e-01
8.42706978e-01 8.02095234e-02 6.46702945e-01 -8.46423566e-01
1.35042474e-01 4.46950406e-01 5.02439916e-01 -1.45472318e-01
-9.09922123e-02 -3.71090889e-01 4.21004266e-01 8.93158764e-02
-4.21657234e-01 2.53785461e-01 5.17993987e-01 3.08981929e-02
1.15292597e+00 -1.49182990e-01 3.56007308e-01 -1.62719399e-01
5.06150067e-01 1.92067176e-01 4.71761674e-01 2.09580928e-01
9.87620205e-02 2.93849796e-01 5.34645736e-01 -7.63913333e-01
-1.01320457e+00 -7.07023621e-01 -1.56368434e-01 5.99357247e-01
-9.94383022e-02 -3.96696776e-01 -4.71751958e-01 -3.93012315e-01
1.74932376e-01 4.28031951e-01 -5.69785058e-01 -2.23386139e-01
-1.90621570e-01 -1.30946136e+00 1.69213057e-01 8.54789317e-02
-1.26525564e-02 -6.41538262e-01 -2.89729208e-01 6.41257405e-01
2.72201121e-01 -6.90127492e-01 1.40612990e-01 3.38443965e-01
-8.57105672e-01 -1.67503142e+00 -9.50264275e-01 -2.69596428e-01
6.87823355e-01 9.70390439e-02 6.86140299e-01 1.03173546e-01
-5.71177840e-01 -3.65453988e-01 -1.84779018e-01 -9.05936956e-01
-8.02617490e-01 -1.47061571e-01 1.38057888e-01 -2.43912742e-01
3.66822004e-01 -4.24871713e-01 -1.04762971e+00 3.03036660e-01
-4.87561494e-01 -1.19372860e-01 8.91860843e-01 7.64925003e-01
7.33773649e-01 -1.11869648e-01 8.67201567e-01 -9.43010151e-01
9.97757673e-01 -2.91151464e-01 -7.86968648e-01 3.58127028e-01
-7.16071784e-01 2.44333744e-01 9.13976073e-01 -4.78510022e-01
-6.55234218e-01 1.04013927e-01 -5.16042896e-02 4.63207275e-01
-3.30180228e-02 6.26970112e-01 -1.28889292e-01 -4.86311227e-01
8.59488368e-01 6.76463991e-02 1.54726043e-01 -3.27305228e-01
-6.55296370e-02 2.65405387e-01 -2.01126844e-01 -4.88662004e-01
4.17661548e-01 3.26184034e-01 5.76183796e-01 -1.12126088e+00
-5.50510347e-01 -3.78177345e-01 -2.51600474e-01 1.42933100e-01
7.70737410e-01 -5.80766082e-01 -1.05317986e+00 1.19915968e-02
-9.78955269e-01 -3.08269888e-01 -1.19234361e-01 5.75950861e-01
-1.50068998e-01 4.10815418e-01 -3.83229375e-01 -7.58191705e-01
-4.27417636e-01 -1.46598208e+00 7.45838642e-01 3.35287839e-01
-5.53847432e-01 -8.27411711e-01 4.57362652e-01 5.43310583e-01
2.81491786e-01 7.11698890e-01 1.24361789e+00 -9.02382314e-01
-8.41575980e-01 -5.43758273e-01 -1.25101684e-02 -6.79419711e-02
4.06338722e-01 2.48872459e-01 -6.63414001e-01 -9.18609351e-02
-4.71111149e-01 -2.90016562e-01 6.38749897e-01 7.07282245e-01
7.66302288e-01 -4.24717784e-01 -5.57282209e-01 1.33872107e-01
1.28415871e+00 7.08246112e-01 7.82989204e-01 2.22495064e-01
4.88513440e-01 4.34808820e-01 1.93899214e-01 3.05360943e-01
-1.51809782e-01 6.30086005e-01 2.41302237e-01 -3.16397458e-01
2.10315123e-01 -1.40027106e-01 1.14840344e-02 7.14670792e-02
-6.24090612e-01 -1.72204897e-01 -7.01421142e-01 -2.82758418e-02
-1.57018304e+00 -1.21024966e+00 -5.84333479e-01 2.74696088e+00
9.57987130e-01 -6.31044656e-02 6.16809368e-01 -8.89090747e-02
5.01761794e-01 -5.04082680e-01 -7.21063256e-01 -4.91403162e-01
-4.54454869e-01 5.37044585e-01 5.45114875e-01 4.32460189e-01
-8.09079587e-01 7.70037830e-01 7.01196957e+00 5.02787888e-01
-1.39635086e+00 -6.06647730e-01 1.20703924e+00 3.22378501e-02
-1.05779827e-01 2.42599770e-01 -9.29197073e-01 4.71094161e-01
1.02542639e+00 -3.97226721e-01 1.94421783e-01 5.40508151e-01
7.17973351e-01 -3.61799031e-01 -1.05021572e+00 3.74062687e-01
-4.11652297e-01 -2.02551341e+00 2.81844676e-01 4.74871635e-01
9.13396597e-01 -2.42144629e-01 9.25628282e-03 -4.06258225e-01
2.83209980e-01 -1.08843696e+00 8.02812651e-02 2.04235986e-01
6.07736468e-01 -7.49158084e-01 7.48881817e-01 1.65578753e-01
-7.49499440e-01 -5.57973981e-02 -4.96522099e-01 -3.61456931e-01
-4.68281321e-02 8.48122001e-01 -1.13473570e+00 1.37847021e-01
-9.62031260e-02 2.76517570e-01 -6.76077783e-01 1.26914394e+00
1.49260640e-01 3.10026288e-01 -1.83278710e-01 -5.53253293e-01
-8.29048157e-02 -4.16252017e-01 2.43329644e-01 6.76853061e-01
1.15291953e-01 -1.74522139e-02 5.66731580e-02 9.01951849e-01
2.74997577e-02 3.05864275e-01 -4.76953447e-01 -6.46137774e-01
3.81281346e-01 8.69839311e-01 -1.09797359e+00 -4.46262658e-02
-3.73752475e-01 5.68962097e-01 2.31277213e-01 2.87224412e-01
-3.50232124e-01 -1.97057486e-01 3.94487858e-01 6.03553772e-01
-1.01364538e-01 -1.76023826e-01 -4.23077941e-01 -9.06071663e-01
-4.33425844e-01 -8.10319960e-01 2.49344304e-01 -2.96792626e-01
-1.09073520e+00 1.99087560e-01 -3.16049814e-01 -6.93142354e-01
1.69306397e-02 -7.37684250e-01 -7.45641828e-01 9.14687037e-01
-1.11560011e+00 -8.50702882e-01 2.01125845e-01 -1.17663778e-02
2.89383978e-01 -2.32542023e-01 1.07436681e+00 -1.74996257e-01
-5.81765592e-01 -2.47210301e-02 4.12697136e-01 -4.12797689e-01
4.84537363e-01 -1.11590779e+00 2.16801628e-01 5.39210141e-01
9.64903310e-02 7.06554651e-01 9.08583641e-01 -1.07082474e+00
-1.21965837e+00 -8.34297240e-01 5.50578773e-01 -6.26902640e-01
7.25043535e-01 -2.12893009e-01 -8.23615372e-01 2.96533555e-02
-3.15604448e-01 -5.24107039e-01 1.38072288e+00 3.31774145e-01
-5.91424480e-02 1.76883817e-01 -9.07751620e-01 7.04455733e-01
3.60786855e-01 -2.26866473e-02 6.02502301e-02 7.75849700e-01
2.33926892e-01 -2.39076734e-01 -7.85360575e-01 2.88021833e-01
5.66685259e-01 -1.01094317e+00 9.09232616e-01 -8.39065969e-01
3.76243740e-01 -3.28843206e-01 4.07199085e-01 -1.15551996e+00
-4.19765234e-01 -6.87158048e-01 4.17493396e-02 4.57222849e-01
1.01295197e+00 -6.43439293e-01 1.34105742e+00 1.00666392e+00
2.83638924e-01 -1.19118381e+00 -7.32765853e-01 -7.14406252e-01
2.56138235e-01 3.67187530e-01 2.88543373e-01 6.11973822e-01
2.90337533e-01 6.66878223e-01 -1.39571846e-01 -2.39124298e-01
5.87478340e-01 7.23567158e-02 5.91005266e-01 -1.65966916e+00
-3.97289962e-01 -5.28362453e-01 -6.53070211e-01 -3.49717289e-01
-2.55226970e-01 -5.15520394e-01 -5.77573836e-01 -1.46687841e+00
3.81814897e-01 -1.88719705e-01 7.79165924e-02 2.61050344e-01
-1.93712428e-01 2.46558085e-01 -2.55750626e-01 2.95455039e-01
-3.97229910e-01 2.60318756e-01 1.19221115e+00 -1.51458308e-01
-7.40697443e-01 2.74535596e-01 -1.07262993e+00 2.96160161e-01
9.57357526e-01 -3.63555223e-01 -3.38520080e-01 4.26385045e-01
5.77372789e-01 -2.38848642e-01 1.69104069e-01 -9.36496317e-01
-2.47674704e-01 -5.64910293e-01 8.86393011e-01 -1.78728119e-01
1.14009306e-01 -4.00075674e-01 3.98915291e-01 7.32767820e-01
-2.90975124e-01 -3.82498205e-01 2.12635145e-01 6.96120203e-01
6.88404441e-02 1.50762349e-02 7.90828824e-01 -2.90603638e-01
-3.90536487e-02 3.00329149e-01 -5.19933224e-01 -2.94651389e-01
1.09407830e+00 -6.25316739e-01 -5.22842288e-01 -1.24738194e-01
-7.71991968e-01 -7.39106163e-02 6.85754240e-01 -1.09106101e-01
5.64279199e-01 -7.40660310e-01 -4.57396030e-01 -1.12388376e-02
8.41618553e-02 -1.97852701e-01 7.99239576e-02 8.37470353e-01
-7.66368985e-01 6.61228299e-01 -7.25661740e-02 -3.03538710e-01
-1.51831555e+00 5.14417350e-01 3.68050843e-01 -3.21126282e-01
2.84636747e-02 7.11031973e-01 3.27108651e-01 4.26005214e-01
-2.87575364e-01 1.48755237e-01 -1.37243390e-01 -3.37029904e-01
6.96906269e-01 4.23952907e-01 2.39274055e-01 -3.78082931e-01
-3.59207809e-01 2.25926802e-01 -4.36439037e-01 4.90030795e-01
1.45406222e+00 4.50197458e-01 -1.42316118e-01 1.13988616e-01
8.78222406e-01 6.03870004e-02 -9.66756284e-01 5.21560073e-01
2.48191833e-01 -3.24348509e-01 -2.35849470e-01 -9.73848403e-01
-1.78832904e-01 6.94826305e-01 4.55831558e-01 3.31342101e-01
4.81998116e-01 1.54348433e-01 5.50652146e-01 5.27475953e-01
2.26592392e-01 -8.67263079e-01 2.27966234e-01 9.11699980e-02
7.30022848e-01 -1.11946237e+00 4.66632336e-01 -5.05533814e-01
-3.40969056e-01 1.16671753e+00 1.96488068e-01 1.03923924e-01
1.25974551e-01 -1.35999113e-01 -2.80328780e-01 -4.98720258e-01
-6.41268849e-01 -1.57427117e-02 2.22925305e-01 7.36836076e-01
9.06303763e-01 1.97447181e-01 -6.84806883e-01 2.25281045e-01
1.26256347e-01 -1.35677373e-02 6.26847267e-01 9.07484114e-01
-7.14574873e-01 -1.64634264e+00 -1.59332275e-01 8.18762958e-01
-7.19456494e-01 -2.02210128e-01 -1.33545780e+00 5.01451254e-01
-2.53999233e-01 8.25714529e-01 -4.48305786e-01 -1.75716579e-01
2.96997994e-01 -2.01173052e-02 6.15244150e-01 -6.07781827e-01
-2.27812082e-01 3.17246437e-01 1.82130367e-01 -1.45526499e-01
-2.71864265e-01 -4.10971671e-01 -1.13647413e+00 -2.57517725e-01
-9.88066792e-01 6.34735048e-01 5.83689034e-01 9.81590331e-01
8.41349781e-01 1.54965058e-01 6.29106402e-01 -7.81806827e-01
-9.84152779e-02 -5.26546896e-01 -5.73112905e-01 3.16052794e-01
-1.00767940e-01 -4.91551757e-01 -1.49570361e-01 1.36437297e-01] | [5.049355506896973, 5.656099796295166] |
f5ab05ac-c48e-4ac0-bdb8-1c6d50fd3f4e | sharpy-shape-reconstruction-and-hand-pose | 2303.10042 | null | https://arxiv.org/abs/2303.10042v1 | https://arxiv.org/pdf/2303.10042v1.pdf | ShaRPy: Shape Reconstruction and Hand Pose Estimation from RGB-D with Uncertainty | Despite their potential, markerless hand tracking technologies are not yet applied in practice to the diagnosis or monitoring of the activity in inflammatory musculoskeletal diseases. One reason is that the focus of most methods lies in the reconstruction of coarse, plausible poses for gesture recognition or AR/VR applications, whereas in the clinical context, accurate, interpretable, and reliable results are required. Therefore, we propose ShaRPy, the first RGB-D Shape Reconstruction and hand Pose tracking system, which provides uncertainty estimates of the computed pose to guide clinical decision-making. Our method requires only a light-weight setup with a single consumer-level RGB-D camera yet it is able to distinguish similar poses with only small joint angle deviations. This is achieved by combining a data-driven dense correspondence predictor with traditional energy minimization, optimizing for both, pose and hand shape parameters. We evaluate ShaRPy on a keypoint detection benchmark and show qualitative results on recordings of a patient. | ['Marc Stamminger', 'Bernhard Egger', 'Martin Vossiek', 'Georg Schett', 'Arnd Kleyer', 'Simon Heinrich', 'Johanna Bräunig', 'Birte Coppers', 'Anna-Maria Liphardt', 'Vanessa Wirth'] | 2023-03-17 | null | null | null | null | ['pose-tracking', 'keypoint-detection', 'gesture-recognition'] | ['computer-vision', 'computer-vision', 'computer-vision'] | [ 2.61805773e-01 -3.36230942e-03 -1.24326564e-01 4.03761938e-02
-1.04250944e+00 -3.80348325e-01 1.95784837e-01 2.60391623e-01
-5.54612994e-01 4.23257291e-01 2.11320609e-01 -1.41685247e-01
-3.75603676e-01 -1.10141166e-01 -3.42682034e-01 -6.33319080e-01
6.62397891e-02 9.84277844e-01 2.66496003e-01 -6.92947581e-02
6.82585388e-02 7.62355924e-01 -1.26096129e+00 -9.45927482e-03
3.43511313e-01 9.91992414e-01 2.02337742e-01 8.35552812e-01
3.38813394e-01 3.31613094e-01 -2.74788499e-01 -5.33549897e-02
2.65363365e-01 -2.82714814e-01 -6.65578425e-01 1.58833697e-01
3.76442552e-01 -5.24370074e-01 -8.76738131e-02 5.07060826e-01
1.07782543e+00 -6.55471385e-02 4.17797476e-01 -7.45995045e-01
4.01174158e-01 -6.79086894e-02 -4.48304743e-01 -1.23481974e-01
9.03394699e-01 9.73994732e-02 5.96330047e-01 -8.75307977e-01
1.14555204e+00 8.85751843e-01 8.48576248e-01 5.81812739e-01
-1.28837907e+00 -2.36569196e-01 -1.70791179e-01 -1.57168396e-02
-1.47249734e+00 -3.17295045e-01 6.26311064e-01 -4.95224714e-01
8.00592661e-01 7.40085006e-01 1.01312029e+00 1.18847704e+00
4.63911563e-01 6.15506947e-01 1.26286149e+00 -5.46441317e-01
3.73597026e-01 -1.57216609e-01 -2.22016603e-01 5.22479951e-01
1.58344612e-01 7.38767982e-02 -7.04593718e-01 -1.55304894e-01
1.04387951e+00 3.02950740e-01 -6.46388412e-01 -6.91477954e-01
-1.37264311e+00 3.61342937e-01 3.58170837e-01 3.34147722e-01
-6.89511120e-01 3.22567403e-01 4.95583639e-02 -1.90419868e-01
2.12288499e-01 1.95467740e-01 -3.58720034e-01 -7.46703625e-01
-8.37582231e-01 2.21268684e-01 6.34044468e-01 8.31765592e-01
-1.46560803e-01 -5.54138422e-01 -1.88004017e-01 1.43632218e-01
5.22782266e-01 5.73907375e-01 2.45463207e-01 -7.36064613e-01
2.99497694e-01 6.05938613e-01 1.80497602e-01 -8.67718399e-01
-8.98310840e-01 -3.38636577e-01 -5.50087929e-01 4.64805603e-01
6.44443333e-01 8.57171193e-02 -1.03364158e+00 1.10507667e+00
7.57616282e-01 3.09801903e-02 -5.25177658e-01 1.43435657e+00
3.54992598e-01 -1.76815882e-01 -1.61789924e-01 -2.41846710e-01
1.49324763e+00 -3.41860205e-01 -5.62298477e-01 -2.65079230e-01
5.68108678e-01 -9.32391286e-01 1.02115822e+00 9.33266103e-01
-1.06833851e+00 -4.59053293e-02 -6.81036711e-01 -5.24541959e-02
1.28308445e-01 1.95360452e-01 4.20476645e-01 7.62800455e-01
-5.56486368e-01 7.94454455e-01 -1.58299732e+00 -3.57214183e-01
1.38815567e-01 6.77480280e-01 -5.51750600e-01 2.11732015e-02
-5.86393476e-01 1.16060579e+00 -6.81358352e-02 5.68133414e-01
-3.08748335e-01 -7.30045378e-01 -5.53127170e-01 -4.93004322e-01
5.90710580e-01 -7.00150967e-01 1.09077549e+00 -1.76340044e-01
-1.79492331e+00 8.81548166e-01 -9.07098576e-02 2.97344066e-02
1.22248137e+00 -7.55394220e-01 1.02522708e-01 2.86706656e-01
-4.81461525e-01 2.29410395e-01 8.88556778e-01 -1.13263679e+00
-2.20895469e-01 -7.79671431e-01 -3.37404042e-01 2.64454603e-01
1.36804104e-01 -1.37132883e-01 -7.96693921e-01 -5.30479610e-01
6.22131288e-01 -1.26244140e+00 -4.38117176e-01 4.52503204e-01
-5.75788140e-01 2.79134303e-01 4.88789529e-01 -1.16283250e+00
1.01025772e+00 -1.81710219e+00 5.59854925e-01 5.85944831e-01
3.06054533e-01 -1.98829416e-02 4.96999562e-01 1.59361511e-01
1.65480480e-01 -4.22139615e-01 -3.81374322e-02 -4.21830356e-01
-1.33028045e-01 1.43721536e-01 4.97854277e-02 8.72243285e-01
-1.44918427e-01 7.53929377e-01 -7.38717198e-01 -6.12881303e-01
6.55478776e-01 1.00851703e+00 -5.89815378e-01 1.85031146e-01
-4.65426594e-02 1.13176990e+00 -6.18212163e-01 8.95328879e-01
3.19670260e-01 -1.69337437e-01 3.78381342e-01 -4.89356637e-01
1.48359938e-02 2.65295863e-01 -1.69288123e+00 2.39879346e+00
-6.82364106e-01 2.58994758e-01 1.97943270e-01 -2.53257841e-01
3.97443026e-01 4.61958945e-01 8.04430127e-01 -5.32842755e-01
6.23789668e-01 6.06654584e-01 -8.43902752e-02 -4.11657482e-01
1.76166758e-01 -1.10446148e-01 1.04979396e-01 4.40595031e-01
-5.07702708e-01 -2.68595248e-01 -6.40026689e-01 -1.81295916e-01
1.19901967e+00 5.41445851e-01 3.00956845e-01 1.10629544e-01
1.43345147e-01 5.81599846e-02 1.82108223e-01 2.73520470e-01
1.22414552e-01 1.15714431e+00 6.96298778e-02 -1.04288168e-01
-8.53082418e-01 -1.04696751e+00 -2.26314008e-01 4.07095939e-01
1.19729877e-01 -2.29508653e-01 -6.63450599e-01 -4.24502164e-01
1.41054198e-01 1.92742035e-01 -4.65676069e-01 1.12449273e-01
-8.18806589e-01 -2.42645189e-01 1.41354412e-01 6.93121552e-01
-2.90157527e-01 -7.61505425e-01 -1.43443668e+00 3.35050285e-01
1.57642692e-01 -1.02360833e+00 -3.34893972e-01 2.56441712e-01
-1.07859719e+00 -1.30500042e+00 -1.05233133e+00 -2.71652907e-01
6.57633841e-01 -3.43352079e-01 8.73523593e-01 2.66717412e-02
-8.97705078e-01 8.43700707e-01 -3.29372913e-01 -1.14708863e-01
-1.54443443e-01 -6.89333528e-02 1.41691834e-01 -3.32825631e-01
-2.64638126e-01 -5.26596189e-01 -9.88183677e-01 4.25104797e-01
-4.59966719e-01 -3.18173356e-02 7.41565585e-01 6.69962406e-01
9.41862881e-01 -6.01112127e-01 -1.83809742e-01 -5.22671402e-01
4.21267450e-01 1.89784080e-01 -4.63907987e-01 2.50623316e-01
-6.20068908e-01 4.37456444e-02 -6.74659386e-02 -6.33534074e-01
-7.54381835e-01 6.92762136e-01 -3.87525469e-01 -6.02063119e-01
-1.60404712e-01 3.65197390e-01 1.41470835e-01 -2.49470040e-01
6.17343128e-01 9.64506902e-03 1.62699774e-01 -7.65836835e-01
2.82084107e-01 6.49071455e-01 7.64227450e-01 -4.96036083e-01
5.79601228e-01 5.20336926e-01 4.22159284e-01 -8.70542169e-01
-2.92977810e-01 -7.22105086e-01 -1.00137329e+00 -3.45596552e-01
7.47193098e-01 -5.44410408e-01 -1.16893268e+00 2.83626527e-01
-1.17866755e+00 -2.17794925e-01 -3.47950101e-01 8.80422950e-01
-8.94187391e-01 3.63760054e-01 -3.43727797e-01 -1.03913820e+00
-5.49684882e-01 -1.31055093e+00 1.74899995e+00 -1.62378877e-01
-6.20986700e-01 -5.86593628e-01 1.58286661e-01 3.13953429e-01
1.75610438e-01 7.27645636e-01 4.04861659e-01 -2.06946447e-01
-4.69249934e-01 -6.52275145e-01 2.27849782e-01 -1.91590279e-01
1.67116642e-01 -2.51059473e-01 -8.22037399e-01 -3.57605457e-01
-8.96468535e-02 -2.50679493e-01 7.83261433e-02 4.48800176e-01
1.00178742e+00 6.67613298e-02 -4.72181737e-01 5.02243459e-01
1.33934212e+00 -1.84023425e-01 5.71508408e-01 1.93288103e-01
8.75380754e-01 5.42115033e-01 8.14682007e-01 5.59630096e-01
1.04648657e-01 1.30364466e+00 3.67992669e-01 -3.10645602e-03
-1.99615985e-01 8.45933426e-03 -7.29518831e-02 6.50505960e-01
-7.04506457e-01 2.08807513e-01 -1.16307294e+00 2.41958171e-01
-1.70172763e+00 -2.43231058e-01 -2.98786193e-01 2.59767938e+00
9.48810637e-01 1.02279566e-01 1.75393149e-01 5.04395843e-01
1.85382962e-01 -4.88552988e-01 -5.25976479e-01 2.19712645e-01
5.66084862e-01 7.18343794e-01 6.04773760e-01 4.12979156e-01
-5.66298783e-01 3.99394393e-01 5.83817720e+00 5.07293642e-01
-1.33943498e+00 1.88392118e-01 -1.33115843e-01 -4.45794344e-01
3.23919132e-02 -2.29586959e-01 -4.57668573e-01 1.48369297e-01
6.14314675e-01 4.77719992e-01 9.80429798e-02 7.85930336e-01
3.97622406e-01 -5.08896232e-01 -1.30485761e+00 1.27066505e+00
-1.76598847e-01 -1.16520953e+00 -6.21812224e-01 2.63433248e-01
9.98372957e-02 -1.21343650e-01 -1.20124318e-01 -4.51357424e-01
-3.11070919e-01 -1.07961774e+00 7.53015041e-01 9.15277898e-01
8.54213834e-01 -5.13709247e-01 4.68114525e-01 4.28169578e-01
-1.02079511e+00 3.72941017e-01 1.83433756e-01 2.73961961e-01
3.31383467e-01 4.11996305e-01 -9.91474032e-01 5.30057907e-01
5.87749422e-01 1.85201004e-01 -3.65061671e-01 1.08444071e+00
-2.42932796e-01 1.94174156e-01 -7.00992942e-01 1.36915833e-01
-3.73329490e-01 1.30776703e-01 7.41503656e-01 8.76439631e-01
3.78632724e-01 1.71569705e-01 2.49613956e-01 5.19134939e-01
5.08193970e-01 3.06210071e-02 -1.30072340e-01 1.67230591e-01
1.93812817e-01 1.21157861e+00 -9.38597262e-01 4.19382393e-01
1.33167610e-01 1.31200385e+00 -1.80707410e-01 -1.01877503e-01
-5.78511834e-01 -5.39493263e-02 5.43301642e-01 5.79643369e-01
1.44433826e-02 -5.61725736e-01 -4.82944936e-01 -8.85071754e-01
5.09376645e-01 -9.11877453e-01 1.72325090e-01 -6.82102740e-01
-7.25904942e-01 4.68901336e-01 9.65593383e-03 -1.41901731e+00
-8.02449584e-01 -8.86966050e-01 -1.47944972e-01 1.05064130e+00
-1.10125089e+00 -1.24762988e+00 -5.59652805e-01 6.89441562e-01
2.31718764e-01 4.61602777e-01 1.10384309e+00 1.79822907e-01
-3.18168968e-01 4.21878815e-01 -1.43179581e-01 -1.30439758e-01
6.47041857e-01 -1.27960360e+00 1.05573401e-01 3.42158884e-01
1.04100198e-01 6.36757433e-01 1.01890755e+00 -7.31191039e-01
-2.18864226e+00 -3.44813466e-01 5.72026014e-01 -8.54255915e-01
3.67701322e-01 -3.32861483e-01 -5.57652354e-01 4.24375206e-01
-5.13777018e-01 5.01276910e-01 3.86358321e-01 -2.40513850e-02
7.09822625e-02 3.31680655e-01 -1.34322143e+00 2.61089921e-01
1.15449977e+00 -5.57836056e-01 -4.33159202e-01 3.71573508e-01
1.34023771e-01 -1.31327927e+00 -1.25213623e+00 2.51466215e-01
1.13551295e+00 -5.92390537e-01 1.15419900e+00 -4.13029253e-01
5.26169315e-02 -1.58587828e-01 -4.88628633e-02 -9.48014736e-01
2.51394421e-01 -6.61704242e-01 -3.55418384e-01 6.95813715e-01
-2.87030973e-02 -2.81933278e-01 1.28473186e+00 8.23231936e-01
1.29436255e-01 -1.01293111e+00 -1.39289582e+00 -7.48352766e-01
-3.77061397e-01 -8.01251113e-01 7.00723976e-02 4.32780415e-01
3.56514975e-02 -2.27869540e-01 -2.72569150e-01 3.38805676e-01
7.92219698e-01 2.42831141e-01 8.11848223e-01 -1.37156212e+00
-6.10112548e-01 -2.33058915e-01 -6.58165395e-01 -8.53874922e-01
-4.90815341e-01 -3.55744928e-01 2.74384618e-01 -1.46932006e+00
-1.37838140e-01 -6.69212461e-01 1.02635957e-01 2.93119341e-01
1.23434827e-01 4.34286773e-01 -8.75945576e-03 3.01031888e-01
-1.20529726e-01 1.40533984e-01 1.22304332e+00 2.03648731e-01
-3.64806980e-01 2.70608693e-01 -1.80313340e-03 7.69611478e-01
2.49724403e-01 -5.05807519e-01 -1.25810415e-01 -1.76261142e-01
3.76021445e-01 2.40778551e-01 7.14926958e-01 -9.26366091e-01
2.17361644e-01 -1.16284199e-01 4.96368021e-01 -7.11230814e-01
5.40901959e-01 -1.30115068e+00 5.11310160e-01 8.82308066e-01
-3.55878659e-02 -5.51781505e-02 -3.79390619e-03 4.60605174e-01
1.68207526e-01 5.77134602e-02 5.69030106e-01 4.17618128e-03
-4.78548706e-01 2.11958274e-01 1.49521336e-01 -1.56599030e-01
9.21068311e-01 -5.66336036e-01 3.14927667e-01 -1.82656497e-01
-1.19324505e+00 -2.60008335e-01 5.62357783e-01 3.42769176e-01
7.80908823e-01 -1.25061429e+00 -4.79228497e-01 2.74892850e-03
1.29350647e-01 5.75933099e-01 1.18722513e-01 1.40414238e+00
-7.35479593e-01 4.21274781e-01 -6.96851015e-02 -1.03711545e+00
-1.52869713e+00 1.25537246e-01 4.32367116e-01 -3.11500669e-01
-1.01047945e+00 6.84130371e-01 -4.07424241e-01 -1.97537974e-01
3.96418720e-01 -5.88134289e-01 1.66345716e-01 -1.29650533e-01
4.10116643e-01 5.39787054e-01 7.34291017e-01 -5.93600869e-01
-8.74751985e-01 9.30233121e-01 2.78760165e-01 -4.05660242e-01
1.32330358e+00 -9.20249447e-02 4.43396360e-01 3.55092704e-01
9.44190443e-01 1.84497416e-01 -1.36976385e+00 -1.30108586e-02
1.42528251e-01 -7.08480537e-01 3.60029817e-01 -9.55512345e-01
-9.13127065e-01 9.34682190e-01 1.14564228e+00 -4.05406505e-01
1.10122550e+00 1.39626727e-01 7.78556824e-01 1.91605195e-01
8.20249796e-01 -9.67923880e-01 -4.48832065e-02 -8.69427398e-02
1.19819975e+00 -9.72800672e-01 3.69546264e-01 -6.45845294e-01
-3.53740811e-01 1.10606015e+00 4.28844243e-02 6.44090325e-02
5.78992486e-01 5.78176379e-01 2.06743419e-01 -3.12122345e-01
-6.90816715e-02 -1.81980431e-01 7.51090944e-01 7.06240058e-01
5.91357648e-01 6.07885458e-02 -2.84810454e-01 3.02016348e-01
-8.88663530e-02 2.59927124e-01 3.06170434e-02 1.53691924e+00
-4.50062193e-02 -1.22003639e+00 -6.92100585e-01 1.37548164e-01
-5.23169935e-01 3.14692140e-01 -2.32248411e-01 9.27006960e-01
-4.89341728e-02 3.88288438e-01 -1.12787381e-01 -4.02274013e-01
9.00532603e-01 -8.05502161e-02 1.16236961e+00 -6.85317993e-01
-7.38668501e-01 4.18887407e-01 -1.53181180e-01 -1.16996586e+00
-3.13377708e-01 -8.98801804e-01 -1.38075447e+00 1.96337812e-02
-4.82902259e-01 -4.53831077e-01 1.18684590e+00 1.07731879e+00
2.10714728e-01 4.61597651e-01 1.68779194e-01 -1.25044584e+00
-7.24605680e-01 -7.21249878e-01 -6.72507703e-01 3.93287569e-01
4.01939780e-01 -1.01718140e+00 7.73649439e-02 -2.36238003e-01] | [6.736443042755127, -0.806222677230835] |
b0f4777d-f50d-4b37-86c6-7fd70fe66919 | clue-me-in-semi-supervised-fgvc-with-out-of | 2112.02825 | null | https://arxiv.org/abs/2112.02825v1 | https://arxiv.org/pdf/2112.02825v1.pdf | Clue Me In: Semi-Supervised FGVC with Out-of-Distribution Data | Despite great strides made on fine-grained visual classification (FGVC), current methods are still heavily reliant on fully-supervised paradigms where ample expert labels are called for. Semi-supervised learning (SSL) techniques, acquiring knowledge from unlabeled data, provide a considerable means forward and have shown great promise for coarse-grained problems. However, exiting SSL paradigms mostly assume in-distribution (i.e., category-aligned) unlabeled data, which hinders their effectiveness when re-proposed on FGVC. In this paper, we put forward a novel design specifically aimed at making out-of-distribution data work for semi-supervised FGVC, i.e., to "clue them in". We work off an important assumption that all fine-grained categories naturally follow a hierarchical structure (e.g., the phylogenetic tree of "Aves" that covers all bird species). It follows that, instead of operating on individual samples, we can instead predict sample relations within this tree structure as the optimization goal of SSL. Beyond this, we further introduced two strategies uniquely brought by these tree structures to achieve inter-sample consistency regularization and reliable pseudo-relation. Our experimental results reveal that (i) the proposed method yields good robustness against out-of-distribution data, and (ii) it can be equipped with prior arts, boosting their performance thus yielding state-of-the-art results. Code is available at https://github.com/PRIS-CV/RelMatch. | ['Jun Guo', 'Yi-Zhe Song', 'Zhanyu Ma', 'Dongliang Chang', 'Ruoyi Du'] | 2021-12-06 | null | null | null | null | ['fine-grained-image-classification'] | ['computer-vision'] | [-4.86770831e-02 -1.13738864e-03 -4.43765283e-01 -5.99704027e-01
-3.59455884e-01 -7.74640024e-01 7.90668786e-01 -3.00172642e-02
-2.86143541e-01 7.80886889e-01 -6.63666278e-02 -3.16228807e-01
-8.42199922e-02 -5.51008999e-01 -7.30141819e-01 -8.04944456e-01
1.32240551e-02 4.62415963e-01 1.66520089e-01 1.50569350e-01
1.54651418e-01 4.71161187e-01 -1.86214399e+00 1.99813917e-01
8.81141365e-01 9.48008120e-01 1.64529055e-01 2.55166352e-01
-2.99453408e-01 8.44621480e-01 -3.07999134e-01 -4.61348921e-01
2.11557418e-01 -3.63227367e-01 -8.83747160e-01 3.23938698e-01
6.77176178e-01 -9.92126465e-02 3.67047964e-03 1.25139284e+00
9.02581066e-02 -7.77159408e-02 1.07077992e+00 -1.38604510e+00
-6.68338120e-01 5.31354785e-01 -7.00664759e-01 2.78611723e-02
-1.77689850e-01 1.22843392e-01 1.24019802e+00 -9.18545008e-01
6.37226999e-01 1.08018899e+00 8.14041317e-01 5.09241819e-01
-1.54779959e+00 -4.97268289e-01 4.30344164e-01 2.81901091e-01
-1.62834096e+00 -2.40611643e-01 7.55850494e-01 -8.02574217e-01
5.21425247e-01 3.39637071e-01 5.11445224e-01 1.05455506e+00
-1.30525351e-01 8.13244760e-01 1.59749830e+00 -4.20073122e-01
4.42732304e-01 3.47646594e-01 4.27044928e-01 6.49026811e-01
2.06703305e-01 2.34001637e-01 -3.53167742e-01 5.03874980e-02
6.54604673e-01 -4.88948524e-02 -1.44972146e-01 -8.14605355e-01
-9.20956552e-01 8.74453187e-01 6.82826102e-01 3.58974993e-01
-4.57044132e-02 -2.63557047e-01 3.62687737e-01 8.21728557e-02
6.77585244e-01 2.40039736e-01 -4.41059411e-01 3.17911029e-01
-1.19607580e+00 -7.22269863e-02 6.82689011e-01 9.52784836e-01
1.18497884e+00 -9.41826105e-02 -6.79668263e-02 9.94799256e-01
3.94876897e-01 3.38874370e-01 4.06521469e-01 -5.92774808e-01
8.41193646e-02 7.45055437e-01 -1.96409971e-03 -8.01491141e-01
-3.80346805e-01 -6.48242414e-01 -1.07073593e+00 3.08665782e-01
6.26036704e-01 2.07026035e-01 -1.04630136e+00 1.71774697e+00
4.55905080e-01 1.08002722e-01 -3.27313721e-01 9.83754694e-01
6.89545035e-01 2.60212630e-01 1.68020800e-01 -2.68592805e-01
1.27533448e+00 -9.49256599e-01 -3.31927478e-01 -2.38878131e-01
4.11601692e-01 -3.59531313e-01 1.22347331e+00 4.74339247e-01
-3.74243766e-01 -7.99368083e-01 -9.87878442e-01 9.65968743e-02
-4.92575288e-01 2.94946909e-01 6.71974957e-01 7.64157593e-01
-9.81271088e-01 5.74576139e-01 -8.72303545e-01 -2.57917225e-01
5.56756318e-01 1.28293261e-01 -5.24947524e-01 -1.63869206e-02
-7.30206251e-01 7.02584445e-01 4.94323730e-01 2.91174918e-01
-9.40308213e-01 -6.59031928e-01 -6.26676917e-01 -4.34054099e-02
5.81595480e-01 -4.83551353e-01 9.65457201e-01 -1.15766931e+00
-1.05889297e+00 1.21949410e+00 -8.47575665e-02 -4.12831038e-01
5.35735488e-01 4.17220360e-03 -2.02825904e-01 1.64783597e-01
2.63751656e-01 7.39698410e-01 1.05112863e+00 -1.62707448e+00
-6.99664176e-01 -4.95221764e-01 -3.47356834e-02 -1.00928500e-01
-3.25379014e-01 -7.70857707e-02 -2.93591768e-01 -6.72405899e-01
9.11382064e-02 -9.67197657e-01 -6.72570542e-02 3.27734083e-01
-5.49787521e-01 -3.60101402e-01 5.80044806e-01 -3.43843848e-01
1.29128003e+00 -2.18822455e+00 1.29311278e-01 2.21007869e-01
4.37716216e-01 4.10783350e-01 1.06247842e-01 4.45764661e-01
-1.04322858e-01 2.29090244e-01 -4.45271075e-01 -3.92610133e-01
1.29482120e-01 3.93934458e-01 -2.99296498e-01 6.63182616e-01
2.66816974e-01 7.49141037e-01 -7.67631710e-01 -6.90780461e-01
4.10632074e-01 2.72078604e-01 -4.52664346e-01 3.57880384e-01
-3.23842585e-01 5.96416295e-01 -2.25413665e-01 7.45921671e-01
7.24726498e-01 -5.51371634e-01 3.87221515e-01 -3.07225138e-01
-2.83710301e-01 -1.75420374e-01 -1.16195297e+00 1.40756154e+00
-2.47920066e-01 3.69438708e-01 1.70667887e-01 -1.16799963e+00
1.02601719e+00 -1.46927863e-01 1.88571915e-01 -2.91718602e-01
1.54414833e-01 2.15700760e-01 -1.44643523e-02 -2.09603980e-01
1.10142164e-01 -2.29761288e-01 1.36387765e-01 1.10416904e-01
2.83639818e-01 -6.84764981e-02 2.90953636e-01 6.32330403e-02
5.45456171e-01 4.07274455e-01 6.78911507e-01 -5.66591084e-01
5.13061166e-01 1.23660363e-01 5.66346884e-01 5.86719155e-01
-3.60409319e-01 5.03746688e-01 4.11141306e-01 -3.77178669e-01
-9.56058919e-01 -1.10843325e+00 -4.51884508e-01 1.10088420e+00
1.74523860e-01 -3.38420093e-01 -7.26815403e-01 -8.69629085e-01
1.31543070e-01 5.74993432e-01 -9.18280423e-01 1.16940327e-01
-7.57076591e-02 -6.32407904e-01 4.31579530e-01 5.08148551e-01
4.21512216e-01 -7.67438829e-01 -3.98424327e-01 -4.75955345e-02
-9.41644758e-02 -9.27564800e-01 -2.13645846e-01 5.43098092e-01
-7.44792283e-01 -1.24934566e+00 -6.30788863e-01 -6.55847967e-01
7.00942695e-01 3.32890958e-01 1.13022339e+00 5.74190654e-02
-2.60238498e-01 1.53404206e-01 -5.69353223e-01 -7.53244087e-02
-2.05520675e-01 -8.78427401e-02 8.40174779e-02 3.54334146e-01
4.52871501e-01 -6.07967436e-01 -4.92756128e-01 6.51020706e-01
-7.07009017e-01 3.98535803e-02 6.22383237e-01 1.12953269e+00
9.25860345e-01 8.88803334e-04 5.01742244e-01 -1.08097088e+00
-3.01705254e-03 -4.47179765e-01 -6.41559720e-01 4.79723662e-01
-7.28034556e-01 2.04316437e-01 9.29873884e-01 -4.03282970e-01
-9.89276528e-01 1.19157210e-01 -1.56941429e-01 -7.03816473e-01
-6.32790744e-01 3.45723897e-01 -2.90637821e-01 -1.16389319e-01
7.78523445e-01 1.86366290e-01 2.93696969e-04 -6.67511165e-01
4.67095107e-01 8.79813254e-01 5.54325998e-01 -7.26621330e-01
6.92430675e-01 5.53361654e-01 -1.62623078e-01 -9.25503314e-01
-1.03268850e+00 -7.12500393e-01 -9.95359778e-01 -2.82776654e-01
7.04920650e-01 -9.56840515e-01 -5.40035248e-01 3.17881703e-01
-4.52857435e-01 -4.24589366e-01 -2.42795512e-01 3.77390265e-01
-5.26688457e-01 6.24117434e-01 -3.95383149e-01 -8.84844422e-01
-6.22240640e-02 -8.47294688e-01 9.95498180e-01 2.17372805e-01
-1.26549974e-01 -1.02417541e+00 -2.68886965e-02 4.17486876e-01
1.43328562e-01 1.37185350e-01 7.24225700e-01 -6.87321365e-01
-4.11696702e-01 1.60421222e-01 -4.39567089e-01 5.75545847e-01
2.06506029e-01 1.63870201e-01 -1.25794625e+00 -4.24474686e-01
-1.74289286e-01 -5.99725306e-01 7.79510498e-01 2.71107972e-01
1.28309047e+00 -1.88461319e-01 -2.91920871e-01 7.69196212e-01
1.45368743e+00 -2.74426609e-01 2.66654909e-01 1.50243372e-01
8.06844711e-01 9.03764069e-01 8.19547355e-01 5.07667840e-01
3.41864198e-01 6.98250055e-01 5.18565059e-01 -6.36337176e-02
-3.83861095e-01 -4.89067942e-01 -1.27236601e-02 6.75005972e-01
-1.03067540e-01 -8.04478489e-03 -8.74385595e-01 5.47500253e-01
-1.88530254e+00 -8.70805264e-01 -3.46690178e-01 2.20249796e+00
9.02752459e-01 9.36655626e-02 2.18484610e-01 7.17918649e-02
7.30500281e-01 3.38596702e-02 -6.71911240e-01 -4.37913425e-02
-1.98488355e-01 1.00046910e-01 3.80632281e-01 3.55224937e-01
-1.42697001e+00 9.82560337e-01 5.15674353e+00 1.28435576e+00
-1.12357199e+00 -5.92660904e-02 5.91962934e-01 3.65967184e-01
-1.21254109e-01 -1.94633864e-02 -1.04092836e+00 5.18975437e-01
6.16713464e-01 1.71017364e-01 3.97377014e-01 1.01589227e+00
-1.78812183e-02 -1.19017929e-01 -1.08192956e+00 1.07362759e+00
-8.92239660e-02 -9.60250556e-01 -1.50780454e-01 3.79421115e-02
6.92445755e-01 -1.31528318e-01 -2.30603695e-01 3.53386939e-01
3.10263813e-01 -9.04066622e-01 9.31258023e-01 2.06576914e-01
1.14071453e+00 -4.69416261e-01 6.26460910e-01 6.54328823e-01
-1.33348656e+00 8.96025300e-02 -5.17488420e-01 -6.53619766e-02
-2.42634594e-01 6.41072989e-01 -6.91267252e-01 8.30349088e-01
8.24820399e-01 8.57073247e-01 -9.34240222e-01 9.32343662e-01
-4.67238456e-01 8.08656335e-01 -2.65641659e-01 4.65598591e-02
2.12578878e-01 -2.42157653e-01 4.84696403e-02 1.17122483e+00
-1.39749944e-01 -1.65054843e-01 4.23372954e-01 7.72189319e-01
2.16931552e-01 1.17858075e-01 -4.25305039e-01 1.03120534e-02
3.85883272e-01 1.38349235e+00 -1.05425107e+00 -3.18481922e-01
-4.62258846e-01 6.96637690e-01 7.60611117e-01 1.90444857e-01
-7.52214849e-01 8.31769407e-02 4.95472342e-01 2.81799119e-02
5.68736792e-01 8.49866681e-03 -3.89672667e-01 -1.41966832e+00
-1.36408255e-01 -8.84058297e-01 4.92322713e-01 -4.98218775e-01
-1.99246371e+00 6.39682531e-01 1.63238689e-01 -1.53529596e+00
-1.89266756e-01 -8.13742936e-01 -6.11490607e-02 6.87117159e-01
-1.44625521e+00 -1.58963549e+00 -2.58956194e-01 7.30811954e-01
4.22825724e-01 1.46151647e-01 7.87911594e-01 1.86571822e-01
-5.27768254e-01 6.00375175e-01 2.54118919e-01 3.19684744e-02
7.29862332e-01 -1.44016898e+00 1.63261350e-02 9.40013051e-01
4.56388354e-01 6.80226266e-01 5.89165866e-01 -6.01676345e-01
-1.13002241e+00 -1.08608186e+00 5.90337396e-01 -2.27350324e-01
9.15897250e-01 -5.80869853e-01 -1.03630531e+00 5.70073962e-01
-1.72854707e-01 3.23692709e-01 7.89087772e-01 3.50086123e-01
-7.92188644e-01 -1.63862184e-01 -1.11690998e+00 3.58629048e-01
1.12772989e+00 -7.24519968e-01 -6.17463946e-01 1.89901114e-01
1.68553680e-01 -4.35770005e-02 -7.60033548e-01 4.37770128e-01
4.91961420e-01 -1.13570225e+00 9.16496933e-01 -4.61247981e-01
3.43976974e-01 -6.92114353e-01 -2.87684560e-01 -1.31094182e+00
-3.53010148e-01 -3.38080861e-02 -2.72078570e-02 1.33868361e+00
1.63470939e-01 -5.55349350e-01 8.31550479e-01 4.28867042e-01
-3.06599975e-01 -5.85919797e-01 -7.59043336e-01 -1.01390231e+00
4.45457473e-02 -2.56507188e-01 1.92480296e-01 1.06854880e+00
-2.06750050e-01 4.10883218e-01 -5.62940478e-01 2.59612650e-01
9.51388776e-01 5.16568124e-01 8.41444552e-01 -1.37048090e+00
-4.75993276e-01 -4.64785308e-01 -3.61882657e-01 -1.04474461e+00
2.40568772e-01 -9.20904160e-01 2.51085430e-01 -1.21056235e+00
2.65553445e-01 -6.25919998e-01 -2.87709236e-02 7.35899866e-01
-3.14793587e-01 5.18636405e-01 1.57939434e-01 4.94774252e-01
-5.74579000e-01 5.29839158e-01 1.01884902e+00 -3.36968824e-02
5.72191067e-02 8.01343750e-03 -6.16879880e-01 7.58870006e-01
7.78462648e-01 -2.77273357e-01 -4.29523945e-01 2.16660518e-02
-2.61507869e-01 -1.76555291e-01 5.32203794e-01 -8.73837173e-01
6.86237216e-02 -1.76945820e-01 3.06440562e-01 -5.75181305e-01
1.46643862e-01 -9.87560153e-01 3.02762389e-01 3.67463231e-01
-1.67966291e-01 -5.87575853e-01 4.53398079e-02 5.94738424e-01
-4.33113396e-01 -2.44337469e-01 9.81573641e-01 -6.42973110e-02
-1.09261429e+00 2.40045518e-01 -8.86238664e-02 2.16649383e-01
9.00645196e-01 -2.02462107e-01 -3.44204456e-01 1.03051655e-01
-7.23089278e-01 1.61124945e-01 6.64304614e-01 1.34798050e-01
1.88411355e-01 -9.52538431e-01 -6.20754719e-01 2.59608626e-01
4.50861543e-01 -9.99727994e-02 3.45149755e-01 7.22325385e-01
-3.74217898e-01 3.69570732e-01 -3.69821757e-01 -9.28961039e-01
-1.32867062e+00 9.71923649e-01 1.07633732e-01 -2.27131367e-01
-4.43882942e-01 8.96372855e-01 5.30200243e-01 -5.78861356e-01
2.75939882e-01 -1.13999449e-01 -2.81193227e-01 2.52286077e-01
4.23166156e-01 4.94133718e-02 3.54633331e-02 -7.30614841e-01
-4.95756209e-01 6.74551725e-01 -2.93783303e-02 3.58085364e-01
1.18072391e+00 -1.84811220e-01 -1.38730571e-01 5.74383140e-01
1.02633035e+00 7.20761493e-02 -1.55142140e+00 -2.41970837e-01
1.90176249e-01 -6.12059176e-01 -1.02172829e-01 -8.11214209e-01
-7.90767014e-01 1.04633248e+00 5.02795756e-01 3.88772786e-01
1.09606135e+00 1.67631581e-01 -6.60122484e-02 7.98038393e-02
7.51656055e-01 -7.14808047e-01 -2.48139724e-01 1.53895929e-01
6.53255343e-01 -1.49306571e+00 7.09076226e-02 -7.05963254e-01
-6.62483752e-01 9.53246713e-01 6.06413126e-01 -1.69270083e-01
6.45256162e-01 6.75130114e-02 2.27755252e-02 -1.04608752e-01
-5.55551291e-01 -6.26534402e-01 5.41741490e-01 7.69782066e-01
4.07454580e-01 3.74674916e-01 -2.27741748e-01 5.45064032e-01
-1.37477592e-01 3.05649713e-02 1.36019811e-01 7.13221252e-01
-4.17561263e-01 -1.06179559e+00 -4.23660457e-01 5.16743839e-01
-1.40306443e-01 -1.18468873e-01 -5.10479331e-01 9.96074259e-01
2.91937947e-01 9.88385141e-01 -9.43268687e-02 -4.15699542e-01
1.34997234e-01 -5.91920596e-03 4.53259915e-01 -7.01976895e-01
-5.79587460e-01 1.49217471e-01 1.16873525e-01 -3.95481437e-01
-6.88674450e-01 -5.53231657e-01 -7.89480209e-01 -3.00829858e-01
-4.35326070e-01 2.04279214e-01 4.02041942e-01 8.61124039e-01
5.80325909e-02 1.42167613e-01 6.40285552e-01 -7.95949101e-01
-6.85942471e-01 -1.00236762e+00 -8.21004152e-01 4.50944334e-01
1.81251749e-01 -9.58746135e-01 -5.88516533e-01 1.88200027e-01] | [9.59261417388916, 2.6681599617004395] |
1d1bcdbf-f8a2-4fdf-aed7-7bc363abaaa7 | the-case-for-fully-bayesian-optimisation-in | 2208.13960 | null | https://arxiv.org/abs/2208.13960v1 | https://arxiv.org/pdf/2208.13960v1.pdf | The case for fully Bayesian optimisation in small-sample trials | While sample efficiency is the main motive for use of Bayesian optimisation when black-box functions are expensive to evaluate, the standard approach based on type II maximum likelihood (ML-II) may fail and result in disappointing performance in small-sample trials. The paper provides three compelling reasons to adopt fully Bayesian optimisation (FBO) as an alternative. First, failures of ML-II are more commonplace than implied by the existing studies using the contrived settings. Second, FBO is more robust than ML-II, and the price of robustness is almost trivial. Third, FBO has become simple to implement and fast enough to be practical. The paper supports the argument using relevant experiments, which reflect the current practice regarding models, algorithms, and software platforms. Since the benefits seem to outweigh the costs, researchers should consider adopting FBO for their applications so that they can guard against potential failures that end up wasting precious research resources. | ['Yuji Saikai'] | 2022-08-30 | null | null | null | null | ['bayesian-optimisation'] | ['methodology'] | [ 1.23985112e-01 2.13573217e-01 -2.36239716e-01 -5.01260817e-01
-7.23530412e-01 -4.05029297e-01 6.26900077e-01 2.10196212e-01
-9.14121866e-01 8.74050677e-01 1.21383905e-01 -8.88582230e-01
-5.37748814e-01 -3.88962030e-01 -6.02804661e-01 -6.62782967e-01
1.45771444e-01 2.15147346e-01 -2.98952349e-02 2.64047146e-01
7.75252342e-01 4.48096871e-01 -1.46683049e+00 -5.40336609e-01
6.40769362e-01 5.10639787e-01 1.02074213e-01 4.62011814e-01
2.36541986e-01 4.26540047e-01 -6.95850253e-01 -5.52415729e-01
2.94342220e-01 -3.55763197e-01 -4.74766672e-01 -1.17189966e-01
-5.75299710e-02 -6.00752592e-01 2.53842384e-01 7.52180994e-01
8.61712873e-01 9.15018842e-02 5.74388146e-01 -1.26487446e+00
-1.76004589e-01 3.28651935e-01 -4.65790510e-01 7.59207681e-02
4.29465473e-01 4.88986999e-01 7.19789147e-01 -3.76623958e-01
2.77457416e-01 1.38725865e+00 6.66855931e-01 1.65330872e-01
-1.32833993e+00 -7.41445005e-01 -8.44248384e-03 -3.58322799e-01
-1.40787137e+00 -8.14560235e-01 1.49912998e-01 -5.12958705e-01
9.74254191e-01 5.61101496e-01 7.27672279e-01 8.04445744e-01
4.82182205e-01 2.73245931e-01 1.51786196e+00 -4.09645468e-01
4.34610695e-01 3.08968067e-01 -9.86402631e-02 2.13540614e-01
1.17235506e+00 2.78831363e-01 -3.96501809e-01 -5.55338740e-01
6.91100776e-01 -3.29570115e-01 -1.80330530e-01 -2.39775881e-01
-9.03805912e-01 1.00108755e+00 -1.99111819e-01 2.22758263e-01
-2.55697459e-01 1.60901412e-01 3.73723686e-01 1.56274661e-01
2.82248139e-01 6.35719001e-01 -3.79075706e-01 -7.23446548e-01
-1.02506042e+00 3.92742932e-01 8.62814188e-01 4.46420938e-01
5.49646839e-02 -3.80645916e-02 2.59602666e-01 6.58456564e-01
9.74802792e-01 2.70427436e-01 3.94848943e-01 -1.12186337e+00
5.76915927e-02 2.43164450e-01 4.25869256e-01 -9.54335034e-01
-3.10548067e-01 -4.11679924e-01 -2.61045784e-01 4.51025337e-01
7.20381975e-01 -1.69462532e-01 -4.27547723e-01 1.36921346e+00
1.99500784e-01 -3.48793566e-01 -1.99302837e-01 7.64912426e-01
2.73487121e-01 4.68859196e-01 1.98480606e-01 -5.59312463e-01
1.07938063e+00 -1.56323627e-01 -7.82661378e-01 -5.36233723e-01
7.38012731e-01 -9.77001786e-01 1.02533066e+00 5.55077493e-01
-1.09812975e+00 -1.27157301e-01 -1.17892170e+00 4.43059236e-01
-5.28179780e-02 -3.90026063e-01 8.88894439e-01 1.50041509e+00
-7.16986299e-01 6.47707522e-01 -9.62141514e-01 -5.00861406e-01
1.24809973e-01 4.57007259e-01 9.57670622e-03 4.22561504e-02
-6.69137836e-01 1.05051160e+00 4.30363476e-01 4.65210855e-01
-2.74256706e-01 -3.68598402e-01 -6.50350213e-01 -5.45642599e-02
3.55407387e-01 -3.80317420e-01 1.13695860e+00 -7.00769067e-01
-1.39793944e+00 3.42743129e-01 4.25470993e-02 -1.38032988e-01
6.61127925e-01 -1.36007547e-01 -1.81466579e-01 -1.37669206e-01
2.06897855e-01 4.96579200e-01 5.81472337e-01 -8.83957982e-01
-5.00108123e-01 -2.33545691e-01 -1.87964201e-01 2.19154805e-01
9.33358073e-02 4.90974456e-01 5.62120713e-02 -3.66316497e-01
2.66966194e-01 -7.15286076e-01 -2.42187545e-01 -3.70976865e-01
5.11221066e-02 -2.19699815e-01 2.96620637e-01 -5.45162141e-01
1.36208570e+00 -2.06124306e+00 -6.92914963e-01 1.69640943e-01
-2.95861423e-01 6.89741671e-02 2.71051943e-01 4.56880540e-01
-4.11209278e-02 4.95504707e-01 4.81410474e-02 3.41642588e-01
3.10287923e-01 6.22392967e-02 6.72290921e-02 8.23913336e-01
1.22768376e-02 5.53434849e-01 -8.10583115e-01 -4.55705583e-01
2.66018897e-01 2.61232972e-01 -3.76471758e-01 -1.51797682e-01
3.11136007e-01 9.38631445e-02 -2.82770067e-01 6.30160213e-01
6.23914540e-01 -1.59764156e-01 3.47050458e-01 2.83463359e-01
-4.92304772e-01 6.22420788e-01 -1.56477118e+00 7.95731366e-01
-1.27550755e-02 6.48663759e-01 4.53484282e-02 -9.26574945e-01
6.31075919e-01 4.81902122e-01 9.06704366e-02 -5.96273661e-01
1.41285703e-01 4.58786279e-01 5.67847788e-01 -4.61564690e-01
2.40782157e-01 -2.66908020e-01 7.87866712e-02 5.67739964e-01
-3.06711048e-01 -5.01441181e-01 9.44435745e-02 -1.69863731e-01
9.30611849e-01 5.96775651e-01 5.86857021e-01 -5.77485025e-01
-1.85269460e-01 -2.45389059e-01 5.91209412e-01 1.03072190e+00
-2.86356241e-01 3.30687165e-01 4.70332265e-01 1.81225836e-02
-7.73686349e-01 -6.45173132e-01 -5.48774481e-01 7.03420401e-01
-2.06990167e-01 -1.13457926e-01 -4.47208732e-01 -3.04600239e-01
-5.22320159e-02 9.48265135e-01 -1.92498833e-01 6.93852156e-02
-7.01288581e-02 -1.15848100e+00 3.44081432e-01 2.49994218e-01
2.25999177e-01 -5.51007390e-01 -1.21388865e+00 4.82832521e-01
1.16593339e-01 -5.67149937e-01 -1.73178930e-02 7.73402154e-02
-1.21350312e+00 -7.69780636e-01 -8.52841854e-01 -9.31005739e-03
4.84668761e-01 3.55561256e-01 6.91776633e-01 1.97585672e-01
-1.06739596e-01 4.52330768e-01 -1.27791569e-01 -7.32395589e-01
-2.92006463e-01 -4.55008835e-01 -1.47005320e-01 -4.58403111e-01
4.93142068e-01 -3.44070584e-01 -4.32548195e-01 4.09531534e-01
-8.91186416e-01 -4.79273766e-01 7.68330753e-01 6.91331029e-01
1.08870654e-03 2.01223478e-01 9.13553774e-01 -6.66964889e-01
7.95867264e-01 -3.81764352e-01 -8.14269006e-01 1.14104114e-01
-1.29638064e+00 -1.01861067e-01 -1.94622472e-01 -4.05503422e-01
-1.13894844e+00 -2.51493484e-01 1.16970111e-02 4.64315146e-01
-2.56592780e-01 8.13031852e-01 -2.58724820e-02 -1.68864697e-01
7.23210931e-01 -4.59289014e-01 1.93537205e-01 -3.16879809e-01
-2.77889848e-01 7.36977100e-01 -1.31271243e-01 -3.89927238e-01
3.83526474e-01 1.77703843e-01 -1.67862192e-01 -1.05874217e+00
-5.18888593e-01 -3.47298145e-01 5.41654415e-02 -2.30755076e-01
4.81923372e-01 -8.49575400e-01 -8.72517765e-01 1.06343642e-01
-6.81386888e-01 -3.89097810e-01 8.92127827e-02 1.08217096e+00
-5.48326790e-01 6.05642796e-01 -7.34359920e-02 -1.35609174e+00
1.56828105e-01 -1.41718721e+00 5.87272465e-01 3.63337249e-01
-7.51787543e-01 -8.25868726e-01 -1.65517539e-01 5.82945824e-01
3.32681984e-01 2.04526842e-01 6.07277572e-01 -5.24039268e-01
-3.89834762e-01 -5.60399890e-01 -1.12433024e-01 4.06998172e-02
1.24124378e-01 4.23492104e-01 -8.84978592e-01 -4.92506683e-01
3.14021051e-01 -2.94904888e-01 9.51974019e-02 6.89831316e-01
6.52767837e-01 -4.40770954e-01 -2.03917399e-01 -4.96174097e-02
1.37054694e+00 4.79474127e-01 3.90523851e-01 7.61718929e-01
-8.72749910e-02 7.70507991e-01 9.12229300e-01 4.92987216e-01
3.52478743e-01 7.13770449e-01 9.57297608e-02 1.01940461e-01
4.09155369e-01 -2.19811797e-02 5.22999346e-01 6.08014107e-01
1.06342442e-01 5.36079705e-03 -9.01605725e-01 4.83216465e-01
-1.68321490e+00 -7.77477443e-01 -3.51867527e-01 2.80549479e+00
7.76389301e-01 7.16441035e-01 3.71605307e-01 7.20906407e-02
5.30720174e-01 -7.29778856e-02 -9.68989506e-02 -8.34275723e-01
1.54178947e-01 -8.04010499e-03 7.42817044e-01 3.92405450e-01
-6.68463647e-01 1.10598184e-01 7.52007484e+00 6.31842315e-01
-7.05437899e-01 1.29781425e-01 7.05658495e-01 -1.61138222e-01
-1.06567815e-01 6.01380467e-01 -6.27776027e-01 5.80511510e-01
1.35459721e+00 -1.30808249e-01 1.60832793e-01 5.95647275e-01
6.97802663e-01 -1.07740521e+00 -8.39477301e-01 4.22129720e-01
-3.51857603e-01 -7.98775673e-01 -7.31726587e-01 3.36763471e-01
3.41425240e-01 -2.39237621e-01 -1.36708155e-01 2.35855550e-01
2.37349868e-01 -1.10764885e+00 8.47635031e-01 2.28099316e-01
3.14997286e-01 -7.40873456e-01 9.99322772e-01 5.05522370e-01
-3.93249631e-01 8.06483068e-03 -5.11521399e-01 -5.61778843e-01
2.24035040e-01 5.24164498e-01 -1.05392981e+00 3.81846547e-01
7.96676159e-01 -1.22360229e-01 -6.15796685e-01 1.51196253e+00
-1.28493786e-01 1.16864455e+00 -7.11963773e-01 -2.72448599e-01
3.13630342e-01 -2.92435974e-01 5.18851757e-01 1.08156085e+00
2.26958796e-01 -2.13046178e-01 -4.13547158e-01 4.32159096e-01
8.07152867e-01 1.89708441e-01 -7.80203581e-01 -5.48342705e-01
6.59061491e-01 9.06704128e-01 -1.15858698e+00 -8.44610203e-03
-6.82790697e-01 1.54379740e-01 -3.11063796e-01 2.23149016e-01
-4.57418144e-01 -3.79063725e-01 1.60339996e-02 6.88681975e-02
2.59817258e-04 -4.54554670e-02 -7.04600453e-01 -5.15578032e-01
-2.46394593e-02 -1.12389886e+00 2.60440856e-01 -6.83004498e-01
-6.72829628e-01 -1.31797850e-01 5.20951867e-01 -6.89335763e-01
-4.19218540e-01 -4.74376172e-01 -2.33246163e-01 9.41198230e-01
-7.86669075e-01 -5.21276951e-01 2.97382802e-01 -2.79571265e-01
3.20983857e-01 3.68109047e-01 6.24170065e-01 5.40048182e-02
-5.38367510e-01 5.54136515e-01 2.23803058e-01 -4.40463662e-01
7.47078836e-01 -9.62147057e-01 9.76035818e-02 7.42722034e-01
-2.04991266e-01 1.10085332e+00 1.00242484e+00 -9.29824352e-01
-1.18435121e+00 -8.75744671e-02 8.89622033e-01 -4.12599623e-01
6.14879310e-01 -1.84642613e-01 -6.02206051e-01 5.33991218e-01
7.01077655e-02 -9.02148545e-01 6.90118611e-01 3.19302201e-01
2.02616468e-01 2.18966633e-01 -1.11744773e+00 8.06686223e-01
3.32549036e-01 -1.22098766e-01 -6.68199062e-01 2.20799401e-01
2.07306385e-01 3.76519412e-02 -1.09432721e+00 5.59917748e-01
8.66182148e-01 -8.98879528e-01 7.40945160e-01 -1.00315958e-01
-2.88314879e-01 -1.74968287e-01 -1.53686196e-01 -8.15165281e-01
1.77554023e-02 -9.13616240e-01 4.83875901e-01 1.32111943e+00
5.79230487e-01 -1.03855371e+00 5.45907319e-01 1.19376802e+00
1.69623658e-01 -5.69149792e-01 -1.06070817e+00 -8.90681922e-01
-4.78070276e-03 -6.49939239e-01 4.56353962e-01 9.37041223e-01
-1.41292699e-02 1.29993603e-01 -2.31912255e-01 -2.40791626e-02
6.44407630e-01 -3.28904659e-01 8.04484725e-01 -1.23781741e+00
-6.02851808e-01 -5.04868507e-01 -3.01032364e-01 -5.18909633e-01
-4.65466052e-01 -1.49548367e-01 1.15199633e-01 -1.41601562e+00
2.26983771e-01 -4.02576238e-01 3.67116332e-02 3.18772912e-01
-3.15404147e-01 -4.81223166e-02 1.13227867e-01 5.14646992e-02
-4.05295081e-02 9.28680599e-02 1.03077912e+00 3.41109753e-01
-2.20283583e-01 3.26564640e-01 -1.07866478e+00 8.47045422e-01
7.87892282e-01 -7.32912362e-01 -3.91065866e-01 8.79691839e-02
5.89233935e-01 1.59251943e-01 3.30954671e-01 -5.84117234e-01
2.29459442e-03 -4.89304304e-01 3.48181456e-01 -4.04906005e-01
3.09838533e-01 -7.96722353e-01 7.16331065e-01 6.04019642e-01
-1.11000881e-01 3.26735713e-02 2.80959010e-01 1.76620796e-01
2.30060354e-01 -1.12589610e+00 5.39860129e-01 2.77159386e-03
-7.66467080e-02 -5.17500937e-01 -8.54758978e-01 -1.99043438e-01
8.56097519e-01 -6.78412855e-01 -1.58742577e-01 -6.06673896e-01
-4.03389990e-01 -1.58475582e-02 6.88585877e-01 4.34194878e-02
2.36058921e-01 -8.13098550e-01 -4.95575339e-01 -8.66654292e-02
-1.04548052e-01 -3.77092838e-01 4.08649556e-02 1.29414284e+00
-6.29393697e-01 6.02741539e-01 9.62382481e-02 -5.08826435e-01
-9.90640640e-01 2.86882728e-01 -2.14548055e-02 1.60281137e-01
-6.57234371e-01 5.93487740e-01 -9.07765776e-02 -1.90440476e-01
4.23029095e-01 7.99338371e-02 2.52004594e-01 1.60631344e-01
2.91348219e-01 6.54242039e-01 1.50934532e-02 -2.16660663e-01
-3.64478111e-01 -1.91524327e-02 -4.34418507e-02 -7.65475273e-01
1.32774878e+00 -1.99947193e-01 -6.56080544e-02 5.71148515e-01
6.02895975e-01 7.24748075e-02 -1.22273076e+00 4.48157519e-01
4.29740250e-01 -7.78390169e-01 3.25329930e-01 -8.52002740e-01
-2.25215480e-01 6.79944217e-01 5.86327910e-01 2.66767919e-01
9.33428407e-01 -4.42765296e-01 -2.45912418e-01 3.37530285e-01
3.44447434e-01 -1.35323215e+00 -4.38513368e-01 -2.43894383e-01
6.79737687e-01 -9.66607332e-01 7.28918254e-01 -6.22947551e-02
-3.59582514e-01 7.77010798e-01 1.05326846e-01 3.14775944e-01
5.02010643e-01 5.94691262e-02 -1.11862577e-01 -8.32755715e-02
-6.77809119e-01 3.66921909e-02 -2.56073866e-02 4.56473768e-01
6.31462395e-01 -9.20519046e-03 -1.19806731e+00 4.71145868e-01
-1.15949988e-01 1.44785896e-01 7.93057919e-01 1.28083348e+00
-4.44886506e-01 -1.18934500e+00 -8.97790492e-01 6.77408934e-01
-8.77709687e-01 1.13253228e-01 -3.31289440e-01 1.35110462e+00
-3.27533722e-01 1.46379662e+00 -2.24233687e-01 1.02858357e-01
-8.75427872e-02 1.30574271e-01 4.59226906e-01 -3.66104275e-01
-4.42921340e-01 7.52512038e-01 4.57552791e-01 -4.01525527e-01
-6.20523155e-01 -1.02050972e+00 -5.53931475e-01 -2.31005952e-01
-7.80127406e-01 2.65391529e-01 9.24934626e-01 1.14807093e+00
1.33455873e-01 6.49577528e-02 3.63431752e-01 -5.64185321e-01
-1.08000457e+00 -1.03764105e+00 -5.91469824e-01 -3.52455795e-01
-3.86010818e-02 -8.61533344e-01 -5.06174326e-01 -3.84874731e-01] | [7.795362949371338, 5.096977710723877] |
a1fa9427-70b8-4cbe-bed0-bc770c1aba75 | targeting-underrepresented-populations-in | 2108.12112 | null | https://arxiv.org/abs/2108.12112v1 | https://arxiv.org/pdf/2108.12112v1.pdf | Targeting Underrepresented Populations in Precision Medicine: A Federated Transfer Learning Approach | The limited representation of minorities and disadvantaged populations in large-scale clinical and genomics research has become a barrier to translating precision medicine research into practice. Due to heterogeneity across populations, risk prediction models are often found to be underperformed in these underrepresented populations, and therefore may further exacerbate known health disparities. In this paper, we propose a two-way data integration strategy that integrates heterogeneous data from diverse populations and from multiple healthcare institutions via a federated transfer learning approach. The proposed method can handle the challenging setting where sample sizes from different populations are highly unbalanced. With only a small number of communications across participating sites, the proposed method can achieve performance comparable to the pooled analysis where individual-level data are directly pooled together. We show that the proposed method improves the estimation and prediction accuracy in underrepresented populations, and reduces the gap of model performance across populations. Our theoretical analysis reveals how estimation accuracy is influenced by communication budgets, privacy restrictions, and heterogeneity across populations. We demonstrate the feasibility and validity of our methods through numerical experiments and a real application to a multi-center study, in which we construct polygenic risk prediction models for Type II diabetes in AA population. | ['Rui Duan', 'Tianxi Cai', 'Sai Li'] | 2021-08-27 | null | null | null | null | ['data-integration'] | ['knowledge-base'] | [ 2.20490992e-01 3.17234129e-01 -7.52552569e-01 -4.90750968e-01
-1.11036694e+00 -6.27686828e-02 -6.57497644e-02 4.88498777e-01
-4.18559819e-01 1.01542997e+00 4.53958899e-01 -4.91301894e-01
-2.57641971e-01 -9.11053360e-01 -6.53217971e-01 -5.27616143e-01
-2.70539522e-01 4.79527026e-01 -6.15749300e-01 2.70670623e-01
-1.80813774e-01 -3.02965604e-02 -8.64967585e-01 1.81253850e-01
1.45244324e+00 6.46886826e-01 -5.76179922e-01 2.67345995e-01
1.28450006e-01 4.85228390e-01 -5.22661388e-01 -6.60968840e-01
4.66298342e-01 -3.83142054e-01 -4.08368677e-01 -3.47443014e-01
3.74971688e-01 -4.11189407e-01 9.85136479e-02 1.00506699e+00
9.20000970e-01 -3.81173760e-01 4.20418084e-01 -1.18965638e+00
-6.34848297e-01 8.12583685e-01 -7.67208815e-01 -4.47327673e-01
1.32438585e-01 -3.31492841e-01 6.93789482e-01 -4.07013327e-01
5.30340195e-01 1.05587435e+00 1.14329004e+00 4.36799794e-01
-1.23421478e+00 -9.69708383e-01 1.17595553e-01 -1.52090013e-01
-1.42549789e+00 -3.13523561e-01 3.78596842e-01 -7.20400274e-01
9.89532247e-02 3.18188220e-01 6.78749621e-01 8.40546787e-01
8.56660753e-02 3.57661754e-01 8.98115098e-01 -3.31468821e-01
3.47141534e-01 2.70085961e-01 2.45419413e-01 5.37951529e-01
7.83155501e-01 -5.49404956e-02 -3.94061178e-01 -1.26254678e+00
4.81374264e-01 3.55654508e-01 -2.73373157e-01 -3.49155366e-01
-1.35249567e+00 9.80922878e-01 2.62635984e-02 -1.67232364e-01
-3.97335082e-01 -4.40584838e-01 3.09711337e-01 3.20542604e-01
7.89909601e-01 -1.12854265e-01 -4.85661536e-01 9.05923396e-02
-7.54257858e-01 1.86152250e-01 1.05398345e+00 1.13456821e+00
4.39717084e-01 -4.78287667e-01 -2.42864162e-01 7.49679983e-01
5.03457189e-02 6.36178374e-01 3.63135606e-01 -9.39758062e-01
6.45152271e-01 7.58558810e-01 2.85167515e-01 -1.13851869e+00
-5.03000617e-01 -3.76288146e-01 -1.26818860e+00 -3.46250474e-01
7.29578614e-01 -8.91486526e-01 -4.28167343e-01 1.99523616e+00
5.97921669e-01 1.01578303e-01 1.12149291e-01 7.91014016e-01
4.55979615e-01 3.35917443e-01 3.52211297e-01 -4.50006634e-01
1.40739822e+00 -5.31709731e-01 -7.75734067e-01 7.66895488e-02
9.51462567e-01 -3.50791365e-01 5.29315650e-01 8.29936713e-02
-1.05583858e+00 4.31781784e-02 -6.17464900e-01 7.92779624e-02
-1.70145288e-01 2.31010780e-01 5.80275893e-01 1.17783511e+00
-7.85419106e-01 4.36129749e-01 -6.41535997e-01 -6.29776299e-01
1.08995080e+00 4.33801800e-01 -2.80751586e-01 -2.83137321e-01
-1.26176476e+00 3.95236552e-01 -1.98700458e-01 1.37592942e-01
-1.28152966e-01 -1.55021060e+00 -6.00689828e-01 1.67970464e-01
2.07547143e-01 -1.20536387e+00 6.81160510e-01 -7.28890419e-01
-1.07663012e+00 6.19539857e-01 -8.02227557e-02 -2.41121829e-01
8.73460770e-01 -7.59755075e-02 -3.52802753e-01 -1.68932974e-01
2.06502348e-01 4.73855203e-03 1.09404445e-01 -6.70234084e-01
-6.48893535e-01 -8.43895495e-01 -7.34480917e-01 1.58552453e-01
-5.41265011e-01 1.33878186e-01 1.37113452e-01 -5.38923562e-01
-1.95520654e-01 -7.85528839e-01 -7.20291913e-01 1.83414474e-01
-3.70346338e-01 2.55581111e-01 2.59512246e-01 -7.51353264e-01
1.23522067e+00 -1.99252796e+00 -3.20427194e-02 4.37838495e-01
4.99274164e-01 1.51245281e-01 3.50062326e-02 3.48124534e-01
2.18113154e-01 4.74188417e-01 -1.07189164e-01 1.74738348e-01
-2.01960012e-01 -3.57905030e-01 1.43972144e-01 6.26127183e-01
-1.26583837e-02 6.06550097e-01 -7.23273873e-01 -6.14728808e-01
-1.66768342e-01 4.24291968e-01 -9.20439899e-01 1.34271249e-01
3.08269680e-01 4.54154998e-01 -9.44481313e-01 8.49595249e-01
9.77720261e-01 -5.52611887e-01 8.44904125e-01 -1.81712527e-02
3.32507752e-02 -1.25409707e-01 -1.09885180e+00 1.34134412e+00
-7.64455125e-02 1.55253327e-02 2.54855335e-01 -1.17523158e+00
7.56583333e-01 3.25921565e-01 7.28152454e-01 -2.46035114e-01
1.00275055e-01 2.87308812e-01 2.25221872e-01 -5.54475605e-01
-1.72908887e-01 -2.55249590e-01 -1.94138065e-01 4.97914553e-01
-5.09177744e-01 9.20352578e-01 -4.35405642e-01 -5.48126772e-02
8.88615906e-01 -5.25382221e-01 6.26043439e-01 -3.93633187e-01
2.69425571e-01 -6.45825267e-02 1.27473187e+00 9.75151837e-01
-5.91999590e-01 4.26437587e-01 6.84204638e-01 -5.65025806e-01
-8.32116365e-01 -8.17387223e-01 -5.08293986e-01 8.70383203e-01
-2.22709954e-01 -2.08320394e-01 -5.71268976e-01 -6.41463518e-01
5.23192048e-01 1.07953906e-01 -6.56448543e-01 -1.37431741e-01
-3.88383478e-01 -1.54431987e+00 6.78531289e-01 1.54700458e-01
4.86628741e-01 -1.09261930e-01 -2.71762520e-01 1.80971637e-01
-2.98642606e-01 -7.90233910e-01 -4.29558873e-01 -4.55883592e-01
-9.23767507e-01 -1.20308864e+00 -1.34144604e+00 -6.86527252e-01
6.48679793e-01 -1.08389914e-01 8.06409478e-01 -2.20288709e-01
-2.47043893e-01 9.93806049e-02 -9.66056138e-02 -8.30793202e-01
-4.02868658e-01 2.22447351e-01 6.89184815e-02 2.52231777e-01
7.75085986e-01 -3.05177152e-01 -9.44489121e-01 3.58208835e-01
-3.63865703e-01 -7.42103681e-02 5.18366277e-01 9.44461405e-01
2.96737999e-01 -4.76766527e-01 1.16141021e+00 -1.61341548e+00
4.18944836e-01 -9.10934508e-01 -5.44235349e-01 5.38978100e-01
-1.00300622e+00 -4.46818441e-01 3.96724969e-01 -3.96042734e-01
-1.15146160e+00 -3.30994576e-02 3.86097848e-01 2.21765324e-01
3.10867801e-02 7.08032072e-01 -2.24717617e-01 -1.18061222e-01
4.41225380e-01 -1.32735103e-01 5.69935620e-01 -5.34059525e-01
1.88942365e-02 1.18167281e+00 -2.61153221e-01 -5.32720387e-01
1.91284671e-01 3.89688551e-01 -1.17116803e-02 -6.50343180e-01
-4.40575033e-01 -5.84019609e-02 -1.58911198e-01 1.44986674e-01
5.39559782e-01 -1.49180412e+00 -8.61480713e-01 4.47171152e-01
-6.19844019e-01 3.21527123e-02 -5.45611680e-02 1.09726775e+00
-1.88123405e-01 2.98491806e-01 -6.76717699e-01 -7.47203410e-01
-6.07910693e-01 -9.76251602e-01 8.30738187e-01 1.44395847e-02
-1.32573962e-01 -9.82710898e-01 2.28800774e-01 6.50785804e-01
5.05898416e-01 4.05831635e-01 1.41211331e+00 -7.21348405e-01
-3.29647452e-01 -3.98973018e-01 -3.57233763e-01 -2.00649828e-01
4.68827337e-01 -3.38565379e-01 -7.31310785e-01 -4.05898005e-01
-3.32268953e-01 -1.18901893e-01 3.17133218e-01 8.52603137e-01
1.17254972e+00 -4.56632584e-01 -7.34570563e-01 6.66510940e-01
1.16829073e+00 2.08520308e-01 4.78460073e-01 -3.93812582e-02
6.48002446e-01 8.96038473e-01 3.31082314e-01 9.44513440e-01
5.97135007e-01 3.40629786e-01 -2.59418935e-01 -4.64131385e-01
4.70618069e-01 -1.48890063e-01 -3.89971472e-02 7.36800551e-01
-1.89877480e-01 -3.95968445e-02 -9.87531960e-01 4.29033130e-01
-2.03610301e+00 -7.59910524e-01 -3.17599103e-02 2.69400549e+00
1.06199551e+00 -7.01671004e-01 5.54235101e-01 -5.47211230e-01
1.13897121e+00 -3.53038758e-01 -7.05315888e-01 -2.10751325e-01
-5.53803220e-02 -2.40961969e-01 7.59438634e-01 4.29968014e-02
-8.85552227e-01 1.82368383e-01 7.41788673e+00 4.50475931e-01
-8.44837189e-01 2.23832607e-01 1.27370930e+00 -2.57979840e-01
-3.06185901e-01 -2.83416957e-01 -4.78502393e-01 5.40132344e-01
1.15154147e+00 -6.93526387e-01 -9.66985822e-02 6.56894267e-01
3.57905686e-01 2.76575118e-01 -1.17839277e+00 7.02450573e-01
-6.81477487e-02 -1.47385228e+00 -4.02144752e-02 4.99782324e-01
7.48196483e-01 -1.28475189e-01 2.01706514e-01 1.63031980e-01
2.27832779e-01 -1.03382301e+00 1.22040780e-02 4.24359649e-01
8.80787015e-01 -6.28638268e-01 8.87666464e-01 2.85195261e-01
-6.52361810e-01 -3.16785306e-01 -4.98846978e-01 -5.97392917e-02
7.36840442e-02 1.11529005e+00 -7.48049259e-01 6.66397274e-01
5.99072158e-01 5.76234102e-01 -1.82937041e-01 1.06332350e+00
8.66392791e-01 7.08482862e-01 -2.13984847e-01 2.21839268e-02
-4.50168937e-01 -5.33574820e-01 9.94106382e-02 8.83826137e-01
8.03385496e-01 1.96012914e-01 9.14657488e-02 7.35947847e-01
-2.28845283e-01 7.48787761e-01 -6.84411645e-01 -1.08764619e-01
5.15093446e-01 8.89476240e-01 1.81611627e-01 -2.65680492e-01
-8.64848793e-01 2.30880544e-01 3.84025604e-01 4.39536482e-01
-6.01386368e-01 -2.41038159e-01 7.09729254e-01 2.44692162e-01
-5.84751330e-02 4.64174449e-01 -3.77035975e-01 -1.30678761e+00
-9.95306484e-03 -1.01770043e+00 1.05601776e+00 8.14085305e-02
-1.71135545e+00 1.00431271e-01 -1.40863076e-01 -1.32242894e+00
7.75291994e-02 -2.20679164e-01 -2.35625908e-01 1.10281932e+00
-1.47555244e+00 -1.01736546e+00 1.16555979e-02 3.08555543e-01
-3.93110693e-01 -4.41683173e-01 1.04304123e+00 8.20085883e-01
-1.17881083e+00 1.08785224e+00 4.14533198e-01 1.80153251e-01
1.01868451e+00 -8.05951297e-01 -2.94618785e-01 2.62845576e-01
-8.49530935e-01 7.82181382e-01 -1.55402320e-02 -8.40359509e-01
-1.49743247e+00 -1.41434610e+00 9.78492081e-01 -7.86438286e-02
2.52941400e-01 -4.73465532e-01 -7.44335830e-01 8.20366859e-01
-9.35403556e-02 6.09243810e-02 1.71312332e+00 7.21612930e-01
-4.01984930e-01 -3.99400145e-01 -1.64488494e+00 6.56836212e-01
8.21596861e-01 -2.29413509e-01 -4.87002358e-02 4.94889617e-01
5.43883801e-01 -1.31989613e-01 -1.59922361e+00 3.21796566e-01
9.46925104e-01 -7.16053784e-01 8.76455843e-01 -1.00961292e+00
3.19577843e-01 2.04377025e-01 -1.57903075e-01 -1.08181047e+00
-3.74400675e-01 -9.03659880e-01 1.73055053e-01 1.19104600e+00
7.88649797e-01 -1.08048773e+00 1.08883631e+00 1.27432442e+00
5.73229313e-01 -5.58585703e-01 -1.05632734e+00 -4.10095841e-01
7.28051424e-01 2.38652661e-01 9.51277137e-01 1.37768424e+00
4.75307643e-01 -1.51683956e-01 -5.43700874e-01 3.55041713e-01
1.03363490e+00 4.19860959e-01 6.00020766e-01 -1.38699210e+00
-1.30988777e-01 -1.27155231e-02 -3.15896600e-01 -2.44209960e-01
-1.72712635e-02 -8.02267075e-01 -4.52360511e-01 -1.05330372e+00
8.06833625e-01 -9.59860206e-01 -4.70443994e-01 3.90446335e-01
-4.54019308e-01 -2.12636530e-01 -1.88339025e-01 1.84442811e-02
-1.22301765e-01 3.46595079e-01 1.11041570e+00 -1.40971735e-01
-3.09317589e-01 2.47246936e-01 -1.19233429e+00 4.52377975e-01
8.58703911e-01 -4.19860065e-01 -3.23705345e-01 -4.10726547e-01
1.50252774e-01 4.58707243e-01 1.24539427e-01 -5.70113003e-01
1.04592346e-01 -3.53991687e-01 6.11862779e-01 -3.73855606e-02
-1.53460085e-01 -8.19537520e-01 5.00404119e-01 6.91468298e-01
-6.39625311e-01 -2.44087294e-01 -8.38965699e-02 7.87055373e-01
9.22255591e-02 3.34745586e-01 4.02909309e-01 -8.24132115e-02
2.92458773e-01 4.19296056e-01 -3.17030638e-01 2.35011801e-01
1.14555478e+00 1.94719210e-01 -7.78924286e-01 -3.44688386e-01
-5.72523296e-01 4.99052078e-01 2.82868981e-01 -9.35896039e-02
2.77034581e-01 -1.36839652e+00 -1.16113985e+00 4.72344071e-01
2.58809775e-01 -2.40774348e-01 6.22915566e-01 1.21869421e+00
-3.14193338e-01 2.03812033e-01 -6.28842488e-02 -4.04700488e-01
-1.19450068e+00 5.64636469e-01 3.54036510e-01 5.04248776e-02
-5.19354641e-01 3.13603312e-01 1.79173499e-01 -7.85887361e-01
2.53329545e-01 -3.24054569e-01 3.53982486e-02 1.00449964e-01
7.12000012e-01 7.72009790e-01 -1.04407929e-01 -3.38148326e-01
-3.35791618e-01 4.79106158e-01 -2.94127047e-01 3.37803543e-01
1.18850291e+00 -2.66109765e-01 -2.06647590e-02 1.77482083e-01
1.38032341e+00 1.30577192e-01 -5.52570879e-01 -2.86336184e-01
-2.89711297e-01 -5.92340589e-01 -5.67079112e-02 -7.19228089e-01
-1.19434094e+00 6.12508178e-01 7.23810434e-01 2.20645722e-02
1.06166780e+00 -2.99463987e-01 7.33296633e-01 1.62512630e-01
3.91025037e-01 -8.39244962e-01 -9.21026170e-01 -3.79070401e-01
2.80062318e-01 -1.44614613e+00 2.21319705e-01 -7.78585434e-01
-6.56190157e-01 5.59529305e-01 3.50945145e-01 4.40037102e-01
1.05310357e+00 2.00644180e-01 3.09547573e-01 1.75271958e-01
-7.09566891e-01 4.37879592e-01 -4.67758141e-02 9.02972817e-01
4.95068103e-01 4.40999925e-01 -6.26113415e-01 1.12227917e+00
1.91970930e-01 5.76263726e-01 6.18189454e-01 9.61250186e-01
1.35699093e-01 -1.38090277e+00 -4.25815612e-01 1.02945614e+00
-7.91520417e-01 -7.09923133e-02 -3.23456079e-01 5.32274008e-01
1.77597791e-01 8.28446805e-01 1.17533922e-01 -1.27011715e-02
3.29875082e-01 1.32370457e-01 1.33315235e-01 -3.47642034e-01
-6.35963857e-01 2.36773595e-01 2.57948488e-01 -5.16614914e-01
-5.06455600e-01 -8.96680057e-01 -7.35216737e-01 -4.94587600e-01
2.56214887e-02 3.25669169e-01 2.36796036e-01 4.35140342e-01
1.23289692e+00 1.51348636e-01 8.05303156e-01 -5.59190139e-02
-1.01847470e+00 -5.64234912e-01 -8.06534827e-01 3.03323239e-01
4.29798752e-01 -2.46026412e-01 -1.65153012e-01 -1.88712403e-01] | [6.297177314758301, 6.404817581176758] |
348e3610-7b58-4644-8bca-c95acb971bb8 | guiding-computational-stance-detection-with | 2305.19845 | null | https://arxiv.org/abs/2305.19845v1 | https://arxiv.org/pdf/2305.19845v1.pdf | Guiding Computational Stance Detection with Expanded Stance Triangle Framework | Stance detection determines whether the author of a piece of text is in favor of, against, or neutral towards a specified target, and can be used to gain valuable insights into social media. The ubiquitous indirect referral of targets makes this task challenging, as it requires computational solutions to model semantic features and infer the corresponding implications from a literal statement. Moreover, the limited amount of available training data leads to subpar performance in out-of-domain and cross-target scenarios, as data-driven approaches are prone to rely on superficial and domain-specific features. In this work, we decompose the stance detection task from a linguistic perspective, and investigate key components and inference paths in this task. The stance triangle is a generic linguistic framework previously proposed to describe the fundamental ways people express their stance. We further expand it by characterizing the relationship between explicit and implicit objects. We then use the framework to extend one single training corpus with additional annotation. Experimental results show that strategically-enriched data can significantly improve the performance on out-of-domain and cross-target evaluation. | ['Nancy F. Chen', 'Hai Leong Chieu', 'Yong Keong Yap', 'Zhengyuan Liu'] | 2023-05-31 | null | null | null | null | ['stance-detection'] | ['natural-language-processing'] | [ 2.46841714e-01 2.67112434e-01 -8.18534017e-01 -6.33910060e-01
-5.34227967e-01 -7.78586090e-01 1.14417434e+00 5.19355834e-01
-4.76123542e-01 7.80526042e-01 5.74881554e-01 -2.18975723e-01
5.48296049e-02 -8.04098189e-01 -2.40846112e-01 -3.90858382e-01
4.53917027e-01 5.92966914e-01 4.54246581e-01 -5.00366032e-01
3.14273000e-01 -9.59189981e-02 -1.43824470e+00 6.65022731e-01
8.68777931e-01 9.71451581e-01 -1.42735958e-01 -3.10969111e-02
-4.06925529e-01 7.90707827e-01 -7.72011459e-01 -9.74576294e-01
-1.90881789e-01 -4.38046515e-01 -9.32397246e-01 7.31535372e-04
3.29730839e-01 9.51004177e-02 2.69332558e-01 1.11596584e+00
3.55271101e-01 -3.26259047e-01 7.14255393e-01 -1.13741553e+00
-5.63000500e-01 8.81417572e-01 -2.96830952e-01 2.47189954e-01
5.32732546e-01 -2.00376883e-01 1.48809040e+00 -7.40381658e-01
8.45594227e-01 1.39331973e+00 6.66855812e-01 4.82810378e-01
-1.19662952e+00 -4.29209262e-01 6.60447240e-01 1.93617210e-01
-7.55937755e-01 -4.27481711e-01 1.03137004e+00 -7.03463852e-01
4.21086043e-01 4.19475496e-01 5.77000797e-01 1.64725459e+00
-4.72102851e-01 8.22643876e-01 1.41887307e+00 -4.41679239e-01
1.14469431e-01 5.35882652e-01 5.39622009e-01 3.58159572e-01
4.52279896e-01 -3.55002105e-01 -7.71603346e-01 -2.44024649e-01
1.14213221e-01 -4.55438972e-01 -3.08837235e-01 -1.90895870e-01
-1.09599853e+00 1.19825840e+00 4.27661300e-01 6.94225132e-01
-8.54910463e-02 -3.98434073e-01 5.34339309e-01 2.53399551e-01
7.95995057e-01 5.64234316e-01 -4.04876173e-01 -1.82313904e-01
-9.22643363e-01 6.75329506e-01 1.06407273e+00 4.83487755e-01
4.99362439e-01 -4.93680865e-01 -2.63799787e-01 9.06941235e-01
2.89283365e-01 3.21839809e-01 4.74988401e-01 -8.08650494e-01
6.66475534e-01 1.01639378e+00 3.23514700e-01 -1.31461000e+00
-2.97060966e-01 -5.68901241e-01 -1.73810542e-01 -6.89787045e-02
7.67386615e-01 -3.95453461e-02 -5.71822077e-02 1.96771467e+00
6.46573782e-01 -5.58474004e-01 -1.08428717e-01 9.96981919e-01
8.00164938e-01 2.17417315e-01 1.72311991e-01 -4.10127819e-01
1.69618678e+00 -6.33232951e-01 -7.19794393e-01 -6.77585125e-01
8.41925681e-01 -7.85253584e-01 1.27367127e+00 2.71813154e-01
-8.12134504e-01 -1.64458409e-01 -8.76221061e-01 -2.26902381e-01
-5.86376071e-01 2.30285921e-03 4.72611845e-01 4.62168008e-01
-4.93410319e-01 4.75277036e-01 -2.79170543e-01 -3.40691537e-01
4.70859796e-01 -8.48494694e-02 6.27361462e-02 1.30166322e-01
-1.68734610e+00 1.08503413e+00 4.61818933e-01 -2.02401783e-02
-2.11203709e-01 -6.41225874e-01 -7.30866194e-01 -3.46565068e-01
7.66744852e-01 -5.92639029e-01 1.20388019e+00 -1.26348782e+00
-1.20832908e+00 1.44221795e+00 -3.10657233e-01 -3.40568215e-01
9.04741585e-01 -4.14093703e-01 -3.92789036e-01 -1.44324392e-01
6.11061215e-01 2.52347201e-01 9.01114047e-01 -9.62937415e-01
-5.94600379e-01 -6.07769132e-01 5.27953267e-01 2.07935885e-01
-5.12816966e-01 4.62299556e-01 -6.41023666e-02 -7.17601478e-01
-7.52831846e-02 -8.39124799e-01 3.09548020e-01 -4.86824363e-02
-5.51584721e-01 -5.89088142e-01 8.21887136e-01 -4.65351552e-01
1.45244300e+00 -1.84039032e+00 2.00298473e-01 3.92994983e-03
2.88376838e-01 3.40791523e-01 4.16544378e-01 4.66170847e-01
1.07375614e-01 1.25637040e-01 -5.89745305e-02 -2.11852491e-01
2.73613870e-01 1.94153488e-01 -3.91040206e-01 3.00578296e-01
7.04368502e-02 8.24294090e-01 -9.60303664e-01 -8.51522267e-01
-2.60698825e-01 2.30734289e-01 -5.08819818e-01 -1.41230617e-02
-5.26054442e-01 3.20958912e-01 -6.79882407e-01 5.70845962e-01
1.90753162e-01 -3.67783159e-01 5.32280743e-01 -5.37398756e-01
-1.64607048e-01 1.00449479e+00 -7.34485984e-01 1.22413540e+00
-4.34348375e-01 6.67784691e-01 4.64419685e-02 -1.19131339e+00
7.64783382e-01 1.19299022e-02 9.18314010e-02 -7.60938764e-01
3.36533636e-01 4.21882808e-01 6.53155297e-02 -6.41241372e-01
2.56626278e-01 -2.52041519e-01 -4.54199433e-01 5.00808299e-01
-4.66772527e-01 2.87281245e-01 5.79137385e-01 2.11193606e-01
5.89040279e-01 9.56403911e-02 5.64011216e-01 -5.03890514e-01
6.76450789e-01 1.59290638e-02 5.96147060e-01 3.13485503e-01
-1.23560756e-01 -1.15159312e-02 9.77600515e-01 -2.27827623e-01
-5.33664286e-01 -7.65890062e-01 -1.36554986e-01 1.46863973e+00
9.49501246e-02 -5.05620480e-01 -6.91122115e-01 -1.02269113e+00
1.00689672e-01 9.22531068e-01 -8.94783914e-01 1.66905954e-01
-6.22451544e-01 -7.09649801e-01 3.92006636e-01 4.01281506e-01
4.73921508e-01 -7.81878710e-01 -7.50599325e-01 1.63822100e-01
-8.69241059e-01 -1.31673634e+00 -4.66274805e-02 -9.74451005e-02
-5.28627813e-01 -1.36576390e+00 -2.98540682e-01 -4.31970298e-01
3.97222668e-01 -1.72875628e-01 1.22268343e+00 -1.05754510e-02
2.98988223e-01 -1.70284465e-01 -4.14644510e-01 -4.43056792e-01
-4.46929365e-01 9.80676636e-02 -1.02184206e-01 1.30197674e-01
6.43353105e-01 -4.60513204e-01 -3.93188179e-01 3.65184546e-01
-5.40731013e-01 8.68988559e-02 1.33868754e-01 6.98125422e-01
7.81460106e-02 -2.99143583e-01 5.41398048e-01 -1.42325115e+00
8.78072441e-01 -4.98505056e-01 -3.32756370e-01 1.25396311e-01
-5.91545701e-01 -1.35646984e-02 2.95774817e-01 -4.54548210e-01
-1.17659616e+00 -6.72127724e-01 1.86751096e-03 1.77367359e-01
1.39801145e-01 7.53553033e-01 -3.32713872e-01 4.55179214e-01
7.93366790e-01 -1.35375872e-01 -4.16648574e-02 -4.51484531e-01
2.74762720e-01 8.06976616e-01 1.39779657e-01 -8.19124222e-01
6.22016132e-01 5.89625657e-01 -2.53375858e-01 -6.70999587e-01
-1.91647553e+00 -4.60771829e-01 -6.17983162e-01 -2.27219403e-01
5.91426790e-01 -7.67089128e-01 -6.86351061e-01 1.69035390e-01
-1.28786385e+00 -3.40722531e-01 4.56018839e-03 7.17765167e-02
-4.08020556e-01 2.69538999e-01 -3.57915878e-01 -5.34210265e-01
-1.17911920e-01 -8.23004901e-01 9.08267319e-01 -3.59683394e-01
-1.11750937e+00 -1.33781707e+00 9.92227867e-02 8.20488095e-01
1.09594747e-01 2.66707778e-01 1.09186935e+00 -1.08625174e+00
8.51695761e-02 -1.02295764e-01 -2.03219280e-01 2.37288818e-01
2.26284757e-01 -1.83655709e-01 -9.00094211e-01 8.68633315e-02
3.02107036e-01 -5.54087520e-01 6.56938493e-01 -3.24283652e-02
9.79850233e-01 -6.86107159e-01 -4.57166076e-01 6.18663523e-03
1.06330156e+00 -5.42452812e-01 1.50019914e-01 6.20826125e-01
3.92257094e-01 1.21275425e+00 9.41913426e-01 3.31463724e-01
5.00171363e-01 1.01397133e+00 2.07055002e-01 3.40619445e-01
-9.26748812e-02 -2.52284229e-01 4.47839737e-01 3.40884298e-01
-1.88390594e-02 -1.00642174e-01 -8.14294577e-01 5.52396655e-01
-1.96428323e+00 -1.09830773e+00 -2.92463124e-01 1.82648075e+00
1.30107987e+00 5.66950142e-01 4.62570786e-01 4.17060256e-01
7.07231104e-01 3.74850750e-01 -1.64884716e-01 -1.77973434e-01
-1.40682101e-01 -1.84189096e-01 -6.94714040e-02 5.82033277e-01
-1.09074783e+00 9.91348505e-01 5.66044235e+00 8.56266856e-01
-1.30432820e+00 3.89986277e-01 4.04439211e-01 -2.47172322e-02
-4.33412105e-01 -2.19019368e-01 -1.13930225e+00 5.91489434e-01
7.26178527e-01 -2.48171180e-01 -2.09509149e-01 7.78903842e-01
2.51255244e-01 -6.56062961e-02 -1.56941700e+00 7.06749499e-01
2.05988079e-01 -1.16959679e+00 -2.19394788e-02 8.94845799e-02
2.84230202e-01 -2.03346074e-01 5.28264306e-02 1.91902593e-01
4.05339599e-02 -6.84121549e-01 1.16106415e+00 -6.76763877e-02
5.66886485e-01 -1.66754887e-01 5.78834414e-01 5.32026350e-01
-7.28970170e-01 -4.97881398e-02 2.51780208e-02 -4.10433918e-01
7.88429230e-02 7.93481946e-01 -8.00728619e-01 1.59298375e-01
4.26895320e-01 8.19148839e-01 -4.66887325e-01 1.77836567e-01
-7.99920619e-01 6.63897872e-01 -2.45731533e-01 -4.04486895e-01
2.68978059e-01 1.23921474e-02 6.92931533e-01 1.18272591e+00
-5.31758517e-02 -8.55777413e-03 2.83496946e-01 1.10644913e+00
-1.77877042e-02 1.96614906e-01 -4.05427665e-01 -2.25420594e-01
4.46114987e-01 1.21459532e+00 -6.29406154e-01 -4.28138494e-01
-3.51481944e-01 6.66132092e-01 6.25369191e-01 2.26456404e-01
-9.05634105e-01 1.76820949e-01 6.28115714e-01 5.87564707e-01
1.68012097e-01 1.21686459e-01 -4.22149092e-01 -1.22689521e+00
2.03776255e-01 -8.45646977e-01 4.80495900e-01 -4.87553567e-01
-1.46183801e+00 2.46065557e-01 1.24490820e-01 -9.98705983e-01
-4.04910535e-01 -7.72365153e-01 -5.06107032e-01 6.36669457e-01
-1.61730361e+00 -1.34177077e+00 -1.70655519e-01 3.94178241e-01
5.31184733e-01 2.84695655e-01 6.52373612e-01 2.31223062e-01
-4.67742980e-01 4.74927664e-01 -4.49908972e-01 3.35941821e-01
7.83210814e-01 -9.74166930e-01 -5.38919047e-02 6.05231464e-01
2.31137387e-02 7.15962470e-01 1.19286811e+00 -6.12562537e-01
-7.27177143e-01 -9.14128721e-01 1.52499521e+00 -7.36104190e-01
1.14264059e+00 -4.30691928e-01 -8.66120934e-01 7.06828952e-01
-7.05327243e-02 -2.80241221e-01 1.03651381e+00 7.69596159e-01
-8.92791450e-01 1.21153675e-01 -1.00193977e+00 8.08351457e-01
1.38406301e+00 -5.40629447e-01 -1.19837189e+00 3.49489093e-01
5.55334866e-01 -2.26020738e-01 -5.83740890e-01 1.03881121e-01
4.68019485e-01 -1.12940657e+00 8.60513330e-01 -7.69282460e-01
9.08270776e-01 -3.35018486e-02 -3.10610563e-01 -1.17451787e+00
-1.71304047e-01 -2.88108289e-01 -4.01988588e-02 1.51666033e+00
5.91032922e-01 -6.40646338e-01 5.88042021e-01 5.07687986e-01
1.52586831e-03 -7.25537658e-01 -7.33247995e-01 -5.12062490e-01
1.68297976e-01 -5.18402755e-01 3.21628898e-01 1.20028293e+00
4.36485261e-01 6.90176785e-01 -4.97279987e-02 -1.93386272e-01
5.09464085e-01 6.65294647e-01 5.70097804e-01 -1.59733474e+00
-2.17873693e-01 -6.50522888e-01 -1.13728665e-01 -9.07281160e-01
5.61025023e-01 -9.39703643e-01 -2.37224743e-01 -1.41753554e+00
1.45590067e-01 -5.11240959e-01 4.65385281e-02 4.33513075e-01
-2.59286374e-01 2.24355116e-01 9.17959586e-02 3.49786520e-01
-6.47745192e-01 3.41446340e-01 9.07068610e-01 -1.16187945e-01
4.65399958e-02 1.38412088e-01 -1.09503436e+00 1.20120764e+00
7.05801308e-01 -4.30120707e-01 -4.08269882e-01 -3.70392501e-01
7.43888378e-01 -4.13503498e-01 5.30196786e-01 -3.43754500e-01
-1.62104353e-01 -4.11655545e-01 3.65007035e-02 -2.79464662e-01
4.44397211e-01 -6.41402900e-01 -3.21654767e-01 4.83161569e-01
-6.88383460e-01 -3.28565359e-01 -2.56325483e-01 4.91071492e-01
-3.55201215e-01 -2.79801607e-01 6.24807000e-01 -2.11716160e-01
-5.04876196e-01 -2.08101079e-01 -2.80974433e-02 6.77408338e-01
9.92182195e-01 -2.73323894e-01 -5.33342361e-01 -1.57698452e-01
-7.53802240e-01 6.97805658e-02 4.68398631e-01 4.68517154e-01
2.43097901e-01 -1.21574700e+00 -7.77497888e-01 -2.16443017e-01
3.89307946e-01 -3.36673081e-01 -1.56424448e-01 1.16059566e+00
-3.85687239e-02 4.10221934e-01 1.16269320e-01 -5.78749478e-01
-1.49142599e+00 3.32361758e-01 2.30977356e-01 -4.12440836e-01
-2.93342441e-01 7.73471296e-01 3.12783301e-01 -3.53830338e-01
-3.28874439e-02 -1.13856249e-01 -4.85587895e-01 8.06410730e-01
5.69024503e-01 2.44223028e-01 1.49270985e-02 -8.71091187e-01
-4.16819066e-01 4.08042431e-01 -7.96652585e-02 -1.44742310e-01
1.12956500e+00 -2.28074044e-01 -3.61451417e-01 8.06202054e-01
1.06929851e+00 3.71189952e-01 -6.72350168e-01 -5.61520815e-01
3.12222034e-01 -5.05283177e-01 -1.46167040e-01 -9.20173705e-01
-5.55195570e-01 9.54951882e-01 -2.98495233e-01 5.24149418e-01
4.80979234e-01 3.90198320e-01 5.84945023e-01 1.27376214e-01
1.44088075e-01 -1.29417741e+00 1.97784066e-01 4.38569933e-01
1.00506961e+00 -1.20473552e+00 -4.26258482e-02 -9.54313755e-01
-6.25894010e-01 9.54895437e-01 6.23522997e-01 2.73300022e-01
4.71967369e-01 6.80392236e-02 5.88399507e-02 -4.77609783e-01
-8.11297894e-01 -2.28975892e-01 3.03264707e-01 3.68659139e-01
8.48160625e-01 9.11414176e-02 -7.05076218e-01 7.09758461e-01
-4.77583945e-01 -2.03227609e-01 2.61687666e-01 7.21374393e-01
-4.76508409e-01 -1.18510246e+00 -2.25353181e-01 2.77129740e-01
-6.63995206e-01 -3.75692882e-02 -9.64590788e-01 8.05363953e-01
2.91182250e-01 8.78203273e-01 -3.84274237e-02 4.66312729e-02
3.30864638e-01 2.05542475e-01 5.30059993e-01 -6.39416754e-01
-6.29635572e-01 -5.26073724e-02 8.60200822e-01 -4.42837864e-01
-9.20844138e-01 -8.81704688e-01 -1.04881132e+00 -3.44658405e-01
-6.23153001e-02 1.67760611e-01 4.41615790e-01 1.37260044e+00
1.71936721e-01 3.18835288e-01 2.58830249e-01 -2.69695967e-01
-7.85667479e-01 -8.81298423e-01 -8.93373638e-02 7.72018015e-01
8.14610869e-02 -9.55657244e-01 -4.33990538e-01 -4.81938384e-02] | [8.928295135498047, 10.038768768310547] |
240d4e80-9104-4861-8bba-bc9832edd8bd | distilling-facial-knowledge-with-teacher | 2209.01115 | null | https://arxiv.org/abs/2209.01115v1 | https://arxiv.org/pdf/2209.01115v1.pdf | Distilling Facial Knowledge With Teacher-Tasks: Semantic-Segmentation-Features For Pose-Invariant Face-Recognition | This paper demonstrates a novel approach to improve face-recognition pose-invariance using semantic-segmentation features. The proposed Seg-Distilled-ID network jointly learns identification and semantic-segmentation tasks, where the segmentation task is then "distilled" (MobileNet encoder). Performance is benchmarked against three state-of-the-art encoders on a publicly available data-set emphasizing head-pose variations. Experimental evaluations show the Seg-Distilled-ID network shows notable robustness benefits, achieving 99.9% test-accuracy in comparison to 81.6% on ResNet-101, 96.1% on VGG-19 and 96.3% on InceptionV3. This is achieved using approximately one-tenth of the top encoder's inference parameters. These results demonstrate distilling semantic-segmentation features can efficiently address face-recognition pose-invariance. | ['Hafiz Malik', 'Rafi Ud Duala Refat', 'Zaid El Shair', 'Ali Hassani'] | 2022-09-02 | null | null | null | null | ['robust-face-recognition'] | ['computer-vision'] | [ 2.32135206e-01 4.27021384e-01 -5.04626811e-01 -9.70449984e-01
-1.03870916e+00 -1.95411339e-01 3.80046844e-01 -6.52475893e-01
-6.57916307e-01 4.66715306e-01 3.75534706e-02 6.85972795e-02
1.01198480e-01 -4.08037454e-01 -1.05178750e+00 -5.61648786e-01
-2.26931617e-01 5.50505102e-01 -7.88070187e-02 -6.77822670e-03
1.47979287e-02 6.27149165e-01 -1.54615629e+00 1.65956885e-01
4.70109344e-01 1.28959131e+00 -1.75302252e-01 5.16767263e-01
1.66974932e-01 2.39136264e-01 -6.59866512e-01 -7.21902251e-01
2.16396913e-01 -5.28309681e-02 -1.01990676e+00 5.33797257e-02
1.33077145e+00 -6.25758111e-01 -5.44361413e-01 1.06366456e+00
7.74991393e-01 -3.26085947e-02 4.68453497e-01 -1.25159395e+00
-7.46690750e-01 4.14259613e-01 -6.47972524e-01 2.45782942e-01
3.54594827e-01 -1.45461261e-01 7.18943357e-01 -1.07218885e+00
6.39538527e-01 1.72153127e+00 8.68910372e-01 1.13107741e+00
-9.12557602e-01 -9.53237891e-01 6.01484329e-02 1.29328072e-01
-1.74664652e+00 -8.76418054e-01 3.35258365e-01 -8.77401605e-02
9.96983469e-01 5.95383272e-02 1.35910928e-01 1.21173334e+00
7.78897926e-02 7.77234495e-01 7.54964411e-01 -1.20152257e-01
-2.70814518e-03 -2.48015344e-01 1.97161108e-01 1.07080853e+00
3.22871268e-01 1.14741758e-01 -9.13784087e-01 2.22236976e-01
5.77818751e-01 -4.82043445e-01 -3.45754996e-02 -8.18665847e-02
-4.66596872e-01 6.43640935e-01 6.37187004e-01 5.20513989e-02
8.34497958e-02 4.30600941e-01 5.49712241e-01 5.66400066e-02
6.74413562e-01 -1.58431709e-01 -5.81316888e-01 1.22556992e-01
-1.26856542e+00 1.84792176e-01 7.08160579e-01 1.18530643e+00
6.50939047e-01 5.30268908e-01 -2.13945001e-01 9.48328376e-01
7.84793317e-01 8.79744470e-01 3.79346102e-01 -5.91154635e-01
4.06211466e-01 2.81505734e-01 -4.73118186e-01 -6.78140461e-01
-5.47358692e-01 -4.88269031e-01 -6.92177534e-01 -9.97420549e-02
1.53112918e-01 -2.39674956e-01 -1.55568206e+00 1.96397328e+00
2.58606046e-01 4.93622482e-01 -1.46273956e-01 5.94587326e-01
1.34930909e+00 1.53574675e-01 4.46724713e-01 2.36082375e-01
1.72057867e+00 -7.18051195e-01 -5.77494025e-01 -3.42923462e-01
2.79756516e-01 -4.48582619e-01 5.58565378e-01 6.17046803e-02
-1.06751704e+00 -9.69126105e-01 -1.05587018e+00 -6.55307844e-02
-3.59938353e-01 4.20615494e-01 4.42772746e-01 1.28501070e+00
-1.56712413e+00 4.23925757e-01 -8.31506014e-01 -4.65948492e-01
1.10325062e+00 1.00940084e+00 -5.67500174e-01 -1.22550897e-01
-1.02768648e+00 6.30824268e-01 1.24424025e-01 6.78291395e-02
-1.19321454e+00 -8.07774961e-01 -1.08502614e+00 -3.69926952e-02
1.43615395e-01 -4.74871367e-01 1.29318881e+00 -8.03318739e-01
-1.54662204e+00 1.18619525e+00 -4.85526383e-01 -3.45611453e-01
3.69988054e-01 -2.35631227e-01 -6.79386437e-01 4.70980287e-01
2.85270303e-01 1.32804275e+00 1.00430441e+00 -9.46439326e-01
-2.87724584e-01 -9.19484735e-01 -2.41001502e-01 5.21040931e-02
-4.57418531e-01 2.66053170e-01 -9.84078646e-01 -5.72132289e-01
1.43270582e-01 -1.00595927e+00 1.92407832e-01 5.47595993e-02
-6.41328037e-01 -4.26535964e-01 1.11961985e+00 -9.18889105e-01
7.38034427e-01 -2.07057953e+00 -1.96428865e-01 1.55800849e-01
-1.24656349e-01 6.44524455e-01 -4.43112016e-01 -2.29790926e-01
-2.64227033e-01 1.67621076e-01 -1.04106255e-01 -7.34474003e-01
1.66443735e-01 1.14398813e-02 9.78944749e-02 7.64514148e-01
1.79381520e-01 1.11432350e+00 -1.57937497e-01 -3.47263396e-01
5.14908656e-02 9.29038405e-01 -9.51137483e-01 1.47162765e-01
2.77345888e-02 3.04238617e-01 -2.47533992e-01 9.98051047e-01
1.18073153e+00 1.04348920e-02 1.33098662e-01 -5.06266594e-01
3.49558800e-01 5.86134195e-03 -9.93025005e-01 1.83124971e+00
-2.04757541e-01 5.83853543e-01 1.88061967e-01 -8.95996988e-01
8.21475744e-01 1.37128532e-01 4.09290493e-01 -1.04711080e+00
4.60329711e-01 5.26935346e-02 -2.49216899e-01 -2.60548025e-01
3.62856507e-01 1.99901581e-01 -1.22732259e-01 -1.00921132e-01
8.74389291e-01 3.39350015e-01 -1.48794595e-02 -6.27198024e-04
6.58868492e-01 1.25959203e-01 -2.48764470e-01 -7.90694654e-01
7.09757924e-01 -7.77154744e-01 6.96257174e-01 6.60148442e-01
-4.08601165e-01 6.50104523e-01 3.61234486e-01 -2.33587116e-01
-4.13631171e-01 -1.13502300e+00 -6.15868509e-01 1.52994847e+00
-7.46574923e-02 -4.28310692e-01 -1.59384871e+00 -8.68543446e-01
2.25600675e-01 4.79631513e-01 -9.93701041e-01 -1.66258633e-01
-4.77407008e-01 -5.82406282e-01 1.25866544e+00 9.09447253e-01
1.08262718e+00 -7.30944395e-01 -1.78605497e-01 -2.25006968e-01
1.12945624e-01 -1.30505610e+00 -5.36987305e-01 -2.18710959e-01
-4.94003087e-01 -1.12507391e+00 -5.49446404e-01 -9.78330195e-01
7.08978117e-01 3.75190983e-03 9.66985345e-01 3.44015174e-02
-6.40848577e-01 6.13019526e-01 9.49005634e-02 -4.24510449e-01
2.50314206e-01 2.88520545e-01 2.89004296e-01 2.05896601e-01
6.03986323e-01 -6.63128197e-02 -6.08276963e-01 2.94540107e-01
-5.61442316e-01 -5.89163959e-01 2.41074100e-01 6.87467337e-01
4.73608017e-01 -4.49319065e-01 7.91061044e-01 -8.44855726e-01
-7.50606358e-02 -1.14072271e-01 -5.84170520e-01 2.83093899e-01
-3.96465778e-01 -1.00787789e-01 2.82195777e-01 8.16100091e-02
-1.14696193e+00 2.97882646e-01 -6.60878181e-01 -3.81575763e-01
-2.85808653e-01 -1.48384711e-02 -7.27553785e-01 -5.71795046e-01
3.03620905e-01 1.52608931e-01 8.87218192e-02 -4.13078815e-01
2.51086503e-01 7.02538073e-01 7.44269967e-01 -5.24562120e-01
6.10301495e-01 3.40742469e-01 -5.28124310e-02 -1.31353295e+00
-7.32151210e-01 -3.47497970e-01 -8.31183791e-01 -3.88455912e-02
1.16808784e+00 -1.24469388e+00 -1.10873616e+00 8.60856473e-01
-9.35996115e-01 -1.23208933e-01 1.98052868e-01 1.24015681e-01
-3.77255142e-01 9.95718539e-02 -5.38921773e-01 -2.72085488e-01
-4.84474540e-01 -1.49948382e+00 1.56604910e+00 4.50623959e-01
7.09356144e-02 -9.20953214e-01 -4.25279260e-01 6.88127518e-01
4.52913374e-01 -1.41001314e-01 3.62504691e-01 -8.52400243e-01
-4.78241950e-01 -5.18613011e-02 -4.30423617e-01 3.96592736e-01
-1.75522596e-01 -1.95062682e-01 -1.59550166e+00 -7.53506601e-01
-4.59109038e-01 -5.07575750e-01 9.33322906e-01 4.87159222e-01
1.74595642e+00 -3.61840218e-01 -4.21924204e-01 1.10918677e+00
1.18172646e+00 1.61073342e-01 7.22385943e-01 -1.57018483e-01
8.97589922e-01 2.88164198e-01 2.15940773e-01 2.85807401e-01
5.32230675e-01 7.96316504e-01 3.69732082e-01 -9.26749930e-02
-5.47905982e-01 -2.80508965e-01 2.03916118e-01 2.36471698e-01
4.97942716e-02 -1.50556460e-01 -9.27683890e-01 2.11247772e-01
-1.17942882e+00 -8.85632098e-01 2.38441557e-01 1.81683695e+00
6.08177543e-01 -1.31553695e-01 1.13546133e-01 -2.03938305e-01
6.68588161e-01 3.66457820e-01 -6.84832871e-01 -3.07106405e-01
1.09877057e-01 7.46688306e-01 8.14876556e-01 6.62248671e-01
-1.46492147e+00 1.32793164e+00 6.92695284e+00 6.97056353e-01
-1.18709254e+00 3.36382776e-01 8.08714926e-01 1.65982004e-02
4.66857068e-02 -7.15968192e-01 -1.65517890e+00 3.93347919e-01
1.15663409e+00 2.26777077e-01 1.91849768e-01 1.02261460e+00
-3.37768644e-01 2.10810646e-01 -1.16688907e+00 1.20927417e+00
8.12609494e-01 -1.11741912e+00 1.89666048e-01 4.95185629e-02
5.37144601e-01 1.48146376e-01 5.91709375e-01 4.68448192e-01
-7.48749524e-02 -1.49885178e+00 6.03232801e-01 1.00724958e-01
1.50595546e+00 -1.08062029e+00 5.27568519e-01 -2.99134463e-01
-1.46749675e+00 2.98703481e-02 -4.72017288e-01 7.56537497e-01
-2.35033035e-01 8.15879628e-02 -1.03048432e+00 3.98243517e-01
8.27397704e-01 6.29054964e-01 -9.82926309e-01 6.88396573e-01
-2.09283352e-01 6.56093538e-01 -2.14465484e-01 3.76624972e-01
8.82640779e-02 4.59603876e-01 9.26050022e-02 1.48156893e+00
2.24561766e-02 2.45144144e-02 1.05378456e-01 4.27332699e-01
-7.18311310e-01 -2.48329982e-01 -3.91948134e-01 2.82670528e-01
5.66566825e-01 9.85862911e-01 -5.93391359e-01 -2.10266158e-01
-3.81034106e-01 1.13425696e+00 1.40054077e-01 3.70683312e-01
-1.10102379e+00 -3.47530842e-01 1.04600501e+00 -6.77568614e-02
5.19416630e-01 1.20416589e-01 -1.31671295e-01 -7.93937862e-01
-4.16972935e-01 -8.89511287e-01 3.27825546e-01 -4.19285506e-01
-7.87407100e-01 7.89872408e-01 1.16283268e-01 -4.60945249e-01
-1.45982638e-01 -1.05846143e+00 -1.64734706e-01 6.53036058e-01
-1.39520800e+00 -1.66786337e+00 -3.43944132e-01 7.90293694e-01
7.31643856e-01 -5.82320511e-01 9.24008727e-01 5.18889964e-01
-8.97917688e-01 1.73042917e+00 -8.96202326e-02 4.95199502e-01
5.91423154e-01 -9.55962121e-01 7.68477678e-01 7.57802427e-01
9.73220095e-02 6.74954474e-01 8.02572072e-03 -5.22640347e-01
-1.63072407e+00 -1.42910278e+00 7.27479935e-01 -2.75486946e-01
-5.68076372e-02 -6.67157173e-01 -7.03211725e-01 1.01179230e+00
2.33320966e-01 3.92503709e-01 7.89808273e-01 2.73060501e-01
-7.69981325e-01 -4.73765820e-01 -1.60508871e+00 3.18843156e-01
1.42300165e+00 -7.82689631e-01 -4.96896386e-01 2.90172011e-01
3.90107810e-01 -6.91891909e-01 -7.31720209e-01 6.03112876e-01
6.80629253e-01 -6.23306811e-01 1.25953305e+00 -5.95805705e-01
-4.42860276e-02 1.34042352e-01 -3.13218027e-01 -1.08079255e+00
-4.38435197e-01 -3.60431015e-01 1.56746760e-01 1.20659733e+00
1.36406049e-01 -5.02799213e-01 1.16402423e+00 4.52445894e-01
-1.13655455e-01 -5.76293051e-01 -1.33707619e+00 -7.67777205e-01
2.73783188e-02 -3.47148150e-01 7.18352795e-01 7.07248867e-01
-5.64033985e-01 2.20637262e-01 3.12136393e-02 4.26442236e-01
9.30630624e-01 -5.26587248e-01 5.84590197e-01 -1.12951422e+00
2.36982644e-01 -2.21297264e-01 -7.31334865e-01 -9.85840559e-01
8.50909233e-01 -1.03880262e+00 -3.69793028e-02 -1.18263173e+00
2.02615246e-01 -3.26191522e-02 -3.88865590e-01 9.39439058e-01
1.04867391e-01 8.48931730e-01 3.99469584e-02 -4.15682346e-01
-7.42946804e-01 5.60672939e-01 1.04565132e+00 -3.02652389e-01
2.63397574e-01 -2.19366804e-01 -6.86777353e-01 7.07199335e-01
8.38647246e-01 -1.80144444e-01 -5.66113353e-01 -5.37614465e-01
-6.53920352e-01 -2.25125223e-01 3.26230317e-01 -1.28570080e+00
3.17890763e-01 2.75315940e-01 8.93412530e-01 -6.69274747e-01
5.26660085e-01 -5.04239976e-01 -1.82444274e-01 4.72629189e-01
-2.91002303e-01 -1.71316028e-01 7.99217939e-01 3.96488875e-01
-1.03857480e-02 1.20800234e-01 1.15424931e+00 2.51049638e-01
-1.13956976e+00 7.40140557e-01 -1.49166569e-01 3.38734657e-01
9.52503800e-01 -2.57059127e-01 -2.59174228e-01 -4.46676090e-02
-7.12615371e-01 -5.42684048e-02 -5.67929968e-02 6.87669158e-01
7.72633195e-01 -1.32275200e+00 -8.32677662e-01 8.14328790e-01
4.86015342e-02 -3.54684234e-01 3.92603874e-01 3.98700505e-01
-4.32159245e-01 8.54795992e-01 -4.59197074e-01 -8.20554912e-01
-1.61791229e+00 -9.27183926e-02 6.15010977e-01 3.72429460e-01
-2.61189491e-01 1.67228246e+00 2.16171920e-01 -6.05641067e-01
5.30876160e-01 4.29504551e-02 -6.20535500e-02 2.76094466e-01
7.16071606e-01 4.18220341e-01 4.42146003e-01 -1.21126866e+00
-1.07382834e+00 7.52784729e-01 -3.86319846e-01 1.15825847e-01
1.14907336e+00 1.29148230e-01 -1.22663490e-02 -5.15051603e-01
1.75218129e+00 -3.52447897e-01 -1.23944521e+00 8.58056266e-03
-1.96309790e-01 -4.79597688e-01 1.20556444e-01 -1.02405643e+00
-1.60647404e+00 9.69499052e-01 1.07448757e+00 -7.51901269e-01
1.01163495e+00 -9.40420181e-02 5.86073816e-01 4.25098389e-01
6.04417205e-01 -1.08853042e+00 1.47823676e-01 6.89123571e-01
8.10530901e-01 -1.16093087e+00 -6.31891191e-02 -6.37113810e-01
-3.06617439e-01 8.22246313e-01 1.05763245e+00 2.61145663e-02
8.66399944e-01 2.67661870e-01 -6.44750372e-02 -1.83548033e-01
-1.58337608e-01 -1.78701341e-01 5.59139013e-01 6.74196959e-01
6.32640183e-01 2.12821960e-01 1.22315980e-01 6.38365149e-01
-2.73305476e-01 -8.85863602e-02 -1.87463060e-01 6.73860967e-01
-3.62508893e-01 -9.44978714e-01 -1.69162691e-01 3.09243232e-01
-7.42569327e-01 9.45333764e-02 -3.24297041e-01 9.33753371e-01
2.60123283e-01 7.29963243e-01 3.25546443e-01 -4.77080613e-01
2.52312005e-01 2.12947145e-01 7.99393237e-01 -6.66891038e-01
-5.18511415e-01 -9.43504348e-02 1.30504698e-01 -1.02295494e+00
-1.19033642e-01 -6.48415208e-01 -1.26886785e+00 -5.23314953e-01
9.03566554e-03 -1.29748404e-01 8.26557159e-01 7.64660060e-01
6.34142935e-01 6.48579061e-01 5.02485216e-01 -8.64015758e-01
-7.38078654e-01 -8.70794296e-01 -5.03053248e-01 1.91308364e-01
2.64880836e-01 -9.05096591e-01 5.89555362e-03 2.00992301e-02] | [13.384145736694336, 0.6395379304885864] |
299b79be-4667-4621-bfd7-602b9c538c73 | schrodinger-spectrum-based-continuous-cuff | 2301.08439 | null | https://arxiv.org/abs/2301.08439v1 | https://arxiv.org/pdf/2301.08439v1.pdf | Schrödinger Spectrum based Continuous Cuff-less Blood Pressure Estimation using Clinically Relevant Features from PPG Signal and its Second Derivative | The presented study aims to estimate blood pressure (BP) using photoplethysmogram (PPG) signals while employing multiple machine learning models. The study proposes a novel algorithm for signal reconstruction, which utilizes the semi-classical signal analysis (SCSA) technique. The proposed algorithm optimises the semi-classical constant and eliminates the trade-off between complexity and accuracy in reconstruction. The reconstructed signals' spectral features are extracted and incorporated with clinically relevant PPG and its second derivative's (SDPPG) morphological features. The developed method was assessed using a publicly available virtual in-silico dataset with more than 4000 subjects, and the Multi-Parameter Intelligent Monitoring in Intensive Care Units dataset. Results showed that the method attained a mean absolute error of 5.37 and 2.96 mmHg for systolic and diastolic BP, respectively, using the CatBoost supervisory algorithm. This approach met the standards set by the Advancement of Medical Instrumentation, and achieved Grade A for all BP categories in the British Hypertension Society protocol. The proposed framework performs well even when applied to a combined database of the MIMIC-III and the Queensland dataset. This study also evaluates the proposed method's performance in a non-clinical setting with noisy and deformed PPG signals, to validate the efficacy of the SCSA method. The noise stress tests showed that the algorithm maintained its key feature detection, signal reconstruction capability, and estimation accuracy up to a 10 dB SNR ratio. It is believed that the proposed cuff-less BP estimation technique has the potential to perform well on resource-constrained settings due to its straightforward implementation approach. | ['Jayant Kalra', 'Sayan Sarkar', 'Aayushman Ghosh'] | 2023-01-20 | null | null | null | null | ['blood-pressure-estimation'] | ['medical'] | [ 4.28877026e-01 5.91274239e-02 4.25782919e-01 -2.22030923e-01
-6.38827443e-01 -3.00143152e-01 5.25254942e-02 1.52217373e-01
-5.34008861e-01 1.04292071e+00 4.84325364e-02 -3.04380625e-01
-5.30883312e-01 -4.99141783e-01 -1.90043867e-01 -8.01410198e-01
-5.22371233e-01 3.22530419e-01 -3.03062275e-02 4.79339669e-03
2.77302802e-01 4.91161764e-01 -1.33072603e+00 5.52109443e-02
9.95244920e-01 1.08257639e+00 -1.70461550e-01 1.14634204e+00
2.69604713e-01 2.76806146e-01 -5.82791388e-01 -2.34507948e-01
6.43355489e-01 -5.25745869e-01 -2.25667879e-01 -3.16415846e-01
6.36679649e-01 -3.62923861e-01 1.69651121e-01 5.43550432e-01
1.49047780e+00 -6.03286959e-02 2.69289464e-01 -6.20827913e-01
2.34082550e-01 3.02531838e-01 -3.18070620e-01 5.26275754e-01
2.38813251e-01 5.05190790e-01 4.21941787e-01 -4.27281499e-01
7.60286376e-02 7.30060279e-01 1.08540797e+00 1.17088512e-01
-1.83412170e+00 -4.27025110e-01 -7.65833259e-01 -5.78873046e-03
-1.34576356e+00 -5.02267301e-01 6.90980613e-01 -5.77174842e-01
7.00771153e-01 8.08475018e-01 9.72091079e-01 4.57887471e-01
5.47419488e-01 -2.34835774e-01 2.04537678e+00 -6.47570789e-01
3.07113945e-01 6.80847883e-01 1.52958199e-01 4.54983860e-01
3.21421385e-01 3.24352741e-01 -1.48482800e-01 -5.41931212e-01
7.12596714e-01 -4.24401641e-01 -3.40700060e-01 -6.81536924e-03
-7.98177898e-01 6.11408651e-01 -3.30709934e-01 4.45689321e-01
-6.24032438e-01 -2.22553313e-01 5.91486275e-01 3.93242091e-01
2.94824630e-01 2.90535897e-01 -4.14048105e-01 -5.59797704e-01
-1.11916196e+00 1.62709340e-01 1.03425527e+00 3.61179173e-01
1.52637303e-01 1.52586862e-01 -2.98350602e-01 7.83531904e-01
7.18386114e-01 5.50672352e-01 4.49429601e-01 -1.01254594e+00
1.84912726e-01 4.15412068e-01 1.85228854e-01 -1.23756754e+00
-7.45760739e-01 -5.97550273e-01 -8.44845355e-01 3.40811789e-01
6.43064618e-01 -1.85479298e-01 -5.90330005e-01 1.35340595e+00
3.76570493e-01 2.31199980e-01 -5.56713343e-02 9.56160903e-01
8.74666393e-01 1.60846412e-01 3.27719241e-01 -6.25856876e-01
1.43711245e+00 -1.10588662e-01 -6.10155523e-01 2.28088662e-01
9.86242890e-02 -6.38360739e-01 7.76403844e-01 6.87656999e-01
-1.19229293e+00 -9.65350211e-01 -9.94048417e-01 5.07388711e-01
8.16169381e-02 6.89700171e-02 1.75176874e-01 1.49343264e+00
-9.57086742e-01 9.41905916e-01 -5.65939188e-01 -3.11761945e-01
2.16021910e-01 4.23050642e-01 -1.07063003e-01 3.68106335e-01
-1.40050411e+00 1.07738841e+00 -8.54028855e-03 5.91773748e-01
-1.26289099e-01 -1.03031135e+00 -5.19551396e-01 -4.61719595e-02
-4.63131785e-01 -6.10305369e-01 5.16287684e-01 -4.10642654e-01
-1.87689424e+00 7.74858236e-01 5.40351197e-02 -5.47608316e-01
1.00467956e+00 -8.48469976e-03 -6.96040571e-01 6.97075188e-01
-2.33107403e-01 -8.90567824e-02 7.16454148e-01 -7.64947653e-01
-9.96082872e-02 -6.53804839e-01 -6.21511102e-01 -9.98576656e-02
2.35815167e-01 -9.38753039e-03 3.04482788e-01 -2.03591302e-01
2.98839241e-01 -7.39603758e-01 -2.86145300e-01 -2.70239022e-02
8.12383965e-02 5.99529088e-01 9.33776498e-02 -1.08785129e+00
1.28443110e+00 -1.78831482e+00 -3.39762419e-01 1.09317875e+00
2.58282453e-01 5.28378367e-01 4.59980935e-01 4.79398429e-01
-1.96257547e-01 -4.96301875e-02 -3.60453337e-01 2.36267135e-01
-1.10753313e-01 1.84071772e-02 2.71026313e-01 8.94171953e-01
-2.76652336e-01 5.74644089e-01 -5.16857564e-01 -5.77376842e-01
7.52170503e-01 5.06356001e-01 -2.89756775e-01 9.84345227e-02
4.89252537e-01 7.01303542e-01 -1.81029383e-02 2.61547089e-01
8.15994799e-01 3.09770167e-01 4.23778474e-01 -6.15712523e-01
-3.99955094e-01 -3.19090933e-01 -1.63255024e+00 1.12264085e+00
-2.20152184e-01 4.14885193e-01 -2.90383194e-02 -1.07240629e+00
1.49637008e+00 8.37113619e-01 7.40390003e-01 -8.56324852e-01
1.66534871e-01 5.37965655e-01 6.35583937e-01 -1.08361244e+00
-4.16069776e-01 -6.45312786e-01 6.98237300e-01 1.41808257e-01
-9.05666649e-02 3.74752004e-03 -1.06936395e-01 -2.43750826e-01
6.84447706e-01 2.08266392e-01 5.80837071e-01 -8.37789893e-01
9.92311776e-01 -3.71389449e-01 1.62126496e-01 9.80797350e-01
-8.17044318e-01 3.96675289e-01 3.58259022e-01 -4.00764197e-01
-9.49616432e-01 -7.74164259e-01 -8.71394515e-01 2.23405272e-01
-3.62781286e-01 3.26599181e-02 -2.64077008e-01 2.84088850e-01
4.61039096e-01 3.75193447e-01 -4.12798315e-01 1.19500682e-02
-4.84546334e-01 -9.92312551e-01 6.83945775e-01 -5.51884808e-02
3.49453211e-01 -7.28269398e-01 -1.06488097e+00 7.70924747e-01
1.68350518e-01 -7.14685082e-01 3.45341086e-01 -1.67967364e-01
-1.24082875e+00 -1.03870904e+00 -7.80137122e-01 -1.82412371e-01
1.50454581e-01 -5.20669699e-01 8.14956129e-01 -3.51637781e-01
-7.97285914e-01 4.89244312e-01 1.21645974e-02 -4.37670857e-01
-5.95829606e-01 -5.68635464e-01 6.53264672e-02 2.20868483e-01
5.14785230e-01 -1.10601330e+00 -1.12906933e+00 2.30098501e-01
-1.50964886e-01 -5.62969506e-01 5.40776491e-01 5.08021057e-01
3.59932363e-01 -2.16779366e-01 1.11418760e+00 -7.26637006e-01
9.10193205e-01 -7.34114945e-02 -6.21865153e-01 -1.66629344e-01
-1.08999336e+00 -5.07933021e-01 5.27694762e-01 -4.51701358e-02
-7.32026458e-01 -1.07563846e-01 -8.32737759e-02 1.50167882e-01
-2.66377419e-01 4.64448273e-01 2.61810124e-01 -4.34984207e-01
1.17503941e+00 3.80119264e-01 7.49102056e-01 -4.77658033e-01
-7.46058673e-02 9.07516062e-01 5.72521031e-01 -7.11901784e-01
2.74229497e-01 2.69621044e-01 6.43635154e-01 -1.20836043e+00
-3.73885632e-02 -5.56198359e-01 -7.16472626e-01 -6.52647555e-01
5.40694833e-01 -7.55091667e-01 -1.28242683e+00 4.71989125e-01
-4.11660999e-01 2.32060015e-01 -1.32533997e-01 8.73492122e-01
-5.08280098e-01 9.20152605e-01 -3.64158899e-01 -1.49489462e+00
-9.45253789e-01 -7.76562095e-01 2.78501421e-01 6.49251640e-02
-3.59366298e-01 -9.45106268e-01 2.86340415e-01 4.83091384e-01
9.55328405e-01 1.00189376e+00 5.94171286e-01 -4.44934845e-01
1.06494941e-01 -4.80277985e-01 3.95751148e-02 6.97431862e-01
1.73117414e-01 -8.61983821e-02 -1.08810115e+00 -2.76155740e-01
4.62918371e-01 6.34438619e-02 3.69883746e-01 7.07646251e-01
5.31231225e-01 -1.33562878e-01 2.36311808e-01 1.74240440e-01
1.77509594e+00 3.76572788e-01 9.97878969e-01 3.18898827e-01
5.36237322e-02 4.18068975e-01 1.71268001e-01 7.09839165e-01
-1.21689022e-01 4.87585366e-01 5.39099507e-04 -2.66951472e-01
1.42613903e-01 3.56042922e-01 -5.75576276e-02 5.92353225e-01
-6.12235606e-01 6.49995089e-01 -7.19189763e-01 1.69559523e-01
-1.44089103e+00 -1.06146550e+00 -6.99386835e-01 2.49323201e+00
9.42504525e-01 3.91351789e-01 4.40388739e-01 4.57722962e-01
5.16439021e-01 -3.63420397e-01 -2.17373595e-01 -8.98573339e-01
5.86086065e-02 8.55303705e-01 5.63471377e-01 5.15035391e-01
-8.50262940e-01 -1.74383342e-01 6.12763453e+00 -1.11061141e-01
-1.12504923e+00 -1.12403214e-01 4.28914368e-01 1.69213519e-01
4.65483099e-01 -1.51426688e-01 -3.25718760e-01 8.79230559e-01
1.61548388e+00 -3.02469373e-01 1.00086167e-01 5.38443089e-01
1.13390982e+00 -5.09375572e-01 -6.83300316e-01 7.72363663e-01
-7.43606612e-02 -1.06350851e+00 -6.54137194e-01 -1.62316561e-01
1.33283138e-01 -2.41490960e-01 -3.99462044e-01 -2.78547313e-02
-9.05339658e-01 -8.87704134e-01 1.06154904e-01 1.10754359e+00
8.88053298e-01 -1.64609358e-01 9.63214755e-01 5.23585752e-02
-7.42711961e-01 -3.78922746e-02 -1.18149452e-01 -3.59205037e-01
1.42357424e-01 7.30457067e-01 -9.25789952e-01 6.95118129e-01
3.53455395e-01 2.61713058e-01 -5.02951086e-01 1.52961624e+00
2.75580615e-01 9.74530160e-01 -6.70725107e-01 9.89072621e-02
-2.15465248e-01 -6.14328921e-01 5.89715838e-01 1.27163541e+00
9.45062637e-02 2.05270082e-01 -2.33163193e-01 8.11226666e-01
9.72282767e-01 5.35479724e-01 -7.87182003e-02 3.89336526e-01
3.19102257e-01 1.08784676e+00 -3.96986753e-01 -3.83234382e-01
-5.06058335e-01 1.31540328e-01 -6.03012562e-01 1.00199610e-01
-6.23690069e-01 -5.28729677e-01 6.53510094e-02 5.35043001e-01
-8.26966539e-02 3.20892900e-01 -6.88087881e-01 -5.75225413e-01
2.10303023e-01 -9.08357441e-01 5.24352670e-01 -3.90474766e-01
-1.10648763e+00 3.32175225e-01 1.84441522e-01 -1.22255111e+00
-1.58753302e-02 -4.78363454e-01 -5.00440300e-01 1.65830564e+00
-1.27128541e+00 -7.04515100e-01 -3.22663754e-01 2.23829746e-01
-2.12395161e-01 -5.88534363e-02 1.07653284e+00 7.28814244e-01
-3.66343111e-01 3.44703019e-01 7.55858943e-02 -3.80778730e-01
5.57526290e-01 -1.28959119e+00 -4.17190015e-01 6.35354042e-01
-9.15058851e-01 7.81384230e-01 1.06562161e+00 -4.98438001e-01
-1.12211883e+00 -6.45355225e-01 7.72928834e-01 -3.94575261e-02
3.84253502e-01 5.43318950e-02 -6.97034836e-01 7.47194290e-02
-9.62909237e-02 1.51735932e-01 8.54951203e-01 -3.81177336e-01
2.99243063e-01 -4.52311754e-01 -1.71665227e+00 6.05883710e-02
2.32935652e-01 -1.59033835e-01 -7.11982012e-01 1.45385385e-01
-2.74584889e-01 -4.73614544e-01 -1.78563821e+00 5.15698731e-01
1.26737928e+00 -1.10892022e+00 8.58573496e-01 -4.49362725e-01
1.49322316e-01 -3.16520482e-01 1.18267335e-01 -8.32709253e-01
-2.75932729e-01 -9.69480515e-01 -1.04653098e-01 1.04684210e+00
3.30646813e-01 -1.16921043e+00 5.67435801e-01 1.13218558e+00
-5.37869446e-02 -7.80353904e-01 -1.29124808e+00 -4.99258816e-01
-3.46246511e-01 -5.21089025e-02 -9.53057036e-02 6.05522037e-01
1.47214919e-01 3.25932391e-02 -5.39816678e-01 3.64703611e-02
9.51073349e-01 6.16471237e-03 6.71009779e-01 -1.48074687e+00
-5.58389068e-01 -1.80813894e-01 -8.76594484e-01 -3.26685384e-02
-7.03689992e-01 -6.38962448e-01 -4.18167919e-01 -1.28225553e+00
-6.06083237e-02 -3.69809180e-01 -5.71383178e-01 8.55711922e-02
-1.35098025e-01 3.37168813e-01 -1.39713421e-01 2.47482234e-03
4.91151333e-01 -8.23776945e-02 1.05658817e+00 3.11603427e-01
-8.18415463e-01 4.35129106e-01 -5.84298611e-01 7.88272843e-02
8.64587069e-01 -4.35068071e-01 -3.71195763e-01 7.69706488e-01
-1.22688331e-01 2.86908537e-01 5.91013253e-01 -1.12903881e+00
-1.21215530e-01 2.61425257e-01 6.35664821e-01 -2.77872562e-01
1.05274260e-01 -1.05006218e+00 8.47400606e-01 9.65169370e-01
-1.62183926e-01 -3.21326554e-01 4.12615210e-01 2.46722206e-01
1.05367534e-01 -2.42963493e-01 1.20413077e+00 -4.61522460e-01
-7.16803595e-02 -3.65881503e-01 -3.05661827e-01 -1.88815817e-01
8.80131900e-01 -8.29841435e-01 -1.96472079e-01 -1.60967708e-01
-1.18568289e+00 4.82722092e-03 -2.85371989e-01 -1.75594136e-01
2.47580647e-01 -8.47658038e-01 -1.14128447e+00 4.67214882e-01
-5.15311547e-02 -6.09752476e-01 6.75031006e-01 1.67946804e+00
-9.29756403e-01 4.35770601e-01 -4.98218894e-01 -8.47555876e-01
-1.35042632e+00 6.05480112e-02 7.95969725e-01 1.53201163e-01
-1.09286141e+00 2.36881167e-01 -8.22353423e-01 8.82782415e-02
9.26503092e-02 -1.66112870e-01 -3.21641356e-01 -5.10418080e-02
4.23677266e-01 7.96464086e-01 4.23724771e-01 -4.17665958e-01
-3.03309411e-01 5.04946768e-01 3.78771782e-01 -6.41535595e-02
1.28961563e+00 -3.01396847e-01 -2.38091543e-01 4.15348291e-01
8.99430692e-01 1.14652783e-01 -5.88458717e-01 -1.48447976e-02
-6.46080822e-02 -5.93529463e-01 7.12398887e-02 -1.29296923e+00
-6.86338484e-01 6.39710248e-01 1.38529480e+00 2.42058635e-01
1.13958800e+00 -7.67766118e-01 1.79839030e-01 -5.09111397e-02
-2.90347785e-02 -1.18475914e+00 -6.61254883e-01 -5.28600574e-01
8.26888919e-01 -7.26326168e-01 5.59016943e-01 -2.35530704e-01
-3.63413066e-01 1.35260832e+00 2.48156413e-01 -5.35745844e-02
8.75111520e-01 -6.81084022e-02 4.44877505e-01 -2.06681848e-01
-1.97553650e-01 1.37997359e-01 1.30663276e-01 7.22977698e-01
5.95140100e-01 2.15571702e-01 -1.27800477e+00 4.37659442e-01
-6.45422116e-02 4.94991004e-01 6.29262507e-01 7.74562478e-01
-6.38455570e-01 -7.87024200e-01 -3.41855854e-01 8.39378417e-01
-7.65700102e-01 1.62205398e-02 1.91309944e-01 8.67322087e-01
6.16226569e-02 9.78939533e-01 -5.63800275e-01 1.28673017e-01
8.13700914e-01 4.72997069e-01 4.43289101e-01 -1.16312921e-01
-9.52718019e-01 3.84778708e-01 1.82258874e-01 -3.55760485e-01
-8.08828592e-01 -6.47737682e-01 -6.56750321e-01 7.45308995e-02
-1.75533444e-01 2.89741576e-01 9.13671911e-01 6.77536607e-01
5.00446022e-01 4.53523248e-01 7.56260812e-01 -4.05300230e-01
-8.83942127e-01 -1.26100540e+00 -8.95723462e-01 3.27491254e-01
2.04712987e-01 -4.11670655e-01 -5.12112260e-01 2.64745168e-02] | [14.05484676361084, 2.9198548793792725] |
b3b8c179-70d9-4819-9af2-7124a6c3eef2 | deep-learning-for-skin-lesion-classification | 1703.04364 | null | http://arxiv.org/abs/1703.04364v1 | http://arxiv.org/pdf/1703.04364v1.pdf | Deep Learning for Skin Lesion Classification | Melanoma, a malignant form of skin cancer is very threatening to life.
Diagnosis of melanoma at an earlier stage is highly needed as it has a very
high cure rate. Benign and malignant forms of skin cancer can be detected by
analyzing the lesions present on the surface of the skin using dermoscopic
images. In this work, an automated skin lesion detection system has been
developed which learns the representation of the image using Google's
pretrained CNN model known as Inception-v3 \cite{cnn}. After obtaining the
representation vector for our input dermoscopic images we have trained two
layer feed forward neural network to classify the images as malignant or
benign. The system also classifies the images based on the cause of the cancer
either due to melanocytic or non-melanocytic cells using a different neural
network. These classification tasks are part of the challenge organized by
International Skin Imaging Collaboration (ISIC) 2017. Our system learns to
classify the images based on the model built using the training images given in
the challenge and the experimental results were evaluated using validation and
test sets. Our system has achieved an overall accuracy of 65.8\% for the
validation set. | ['P. Mirunalini', 'Aravindan Chandrabose', 'S. M. Jaisakthi', 'Vignesh Gokul'] | 2017-03-13 | null | null | null | null | ['skin-lesion-classification'] | ['medical'] | [ 6.86040878e-01 1.71059012e-01 -2.79803902e-01 -1.32611558e-01
-4.30418938e-01 -3.01442623e-01 5.68357110e-01 2.09451348e-01
-5.03771901e-01 4.70599055e-01 -1.06092110e-01 -3.69479477e-01
3.04757953e-01 -8.87952805e-01 -2.90703207e-01 -7.18307912e-01
2.39865348e-01 -5.43028265e-02 1.87843040e-01 -1.43327981e-01
9.72888395e-02 4.70446974e-01 -1.25811672e+00 8.75032842e-01
7.58225977e-01 1.09118497e+00 -1.75270349e-01 1.31988752e+00
-7.19927251e-02 9.20854032e-01 -4.50672060e-01 -2.63129711e-01
3.19577843e-01 -2.83576727e-01 -9.98686790e-01 9.26166400e-02
7.84459770e-01 -2.60640681e-01 -1.75567791e-01 1.06304324e+00
2.71801233e-01 -5.35347879e-01 8.96998107e-01 -6.86829388e-01
-3.27728659e-01 -1.94134921e-01 -6.27925277e-01 2.68459260e-01
2.67070502e-01 6.07429333e-02 2.51948923e-01 -5.87502360e-01
7.27401197e-01 6.42717779e-01 8.92598212e-01 1.14831400e+00
-7.44327843e-01 -2.26355195e-01 -4.25668865e-01 2.72867739e-01
-1.33526373e+00 -7.78234676e-02 1.16245627e-01 -4.80032027e-01
8.51426244e-01 8.35924506e-01 8.38265061e-01 9.27714288e-01
4.59444433e-01 7.62706935e-01 1.45733428e+00 -6.86581016e-01
1.41093358e-01 4.73021328e-01 4.00341451e-02 9.61620092e-01
2.08736718e-01 8.29435959e-02 -2.54195388e-02 5.06182667e-03
8.36216450e-01 -1.44566856e-02 -8.00665319e-02 3.72441232e-01
-4.46794719e-01 5.77845514e-01 8.79460454e-01 1.90227315e-01
-6.36378348e-01 2.40365282e-01 6.00338936e-01 2.09061131e-01
4.79117185e-01 1.51831120e-01 -6.34806082e-02 3.20228964e-01
-7.91981339e-01 -3.23922127e-01 8.30767632e-01 -4.49196100e-02
2.74853110e-01 -2.88068861e-01 4.81724814e-02 9.91189539e-01
2.62249321e-01 2.28223741e-01 6.29059911e-01 -1.19255960e-01
-7.99395218e-02 1.03050518e+00 -3.56697053e-01 -4.19551939e-01
-3.46387923e-01 -1.41269490e-01 -9.87432301e-01 5.87828875e-01
4.64329839e-01 -3.27214718e-01 -1.72018480e+00 8.32601845e-01
4.13571924e-01 1.23379901e-01 1.98005989e-01 8.82476330e-01
1.06024384e+00 3.40197831e-01 3.81937087e-01 4.49497193e-01
1.28415990e+00 -8.48160088e-01 -4.05953825e-01 -2.36336999e-02
5.75096607e-01 -6.45211160e-01 2.63714820e-01 5.04954934e-01
-6.79383993e-01 -2.33259439e-01 -1.06033075e+00 2.49651633e-02
-7.58212507e-01 5.71982741e-01 7.35561013e-01 8.72553706e-01
-1.36946285e+00 3.53450298e-01 -8.20196450e-01 -1.14270961e+00
5.73097825e-01 5.34864426e-01 -7.35185146e-01 -2.58780539e-01
-8.45639586e-01 1.18590724e+00 2.36674115e-01 8.13923702e-02
-9.18419182e-01 -4.70765561e-01 -6.89965546e-01 -5.95346928e-01
-1.22299910e-01 -5.18945038e-01 9.62762237e-01 -1.66735518e+00
-1.18962288e+00 1.37328005e+00 -9.63876620e-02 -6.05391085e-01
5.23504496e-01 2.13291004e-01 -7.45238066e-01 5.11721492e-01
-2.31621727e-01 6.32980883e-01 8.44309986e-01 -1.01867926e+00
-9.54857707e-01 -3.68537962e-01 4.97603863e-02 1.55910224e-01
-3.13340932e-01 -1.84365824e-01 -3.61340523e-01 -2.30880752e-01
-2.84756839e-01 -9.82446909e-01 -3.62324744e-01 4.84623343e-01
-7.75340557e-01 -4.83164527e-02 7.85945415e-01 -1.11213231e+00
8.62917662e-01 -1.95146894e+00 -2.78501451e-01 6.20117068e-01
3.35143000e-01 7.94897795e-01 -2.12785959e-01 4.81551677e-01
-3.23219299e-01 3.05239946e-01 1.59690231e-02 1.39361963e-01
-8.12805831e-01 -8.80899280e-02 4.54403549e-01 4.77796823e-01
4.52104092e-01 6.79267704e-01 -6.70135260e-01 -2.97758609e-01
5.24155378e-01 8.10060203e-01 4.33874905e-01 7.70970285e-02
-1.02721704e-02 -8.20119902e-02 -1.50358930e-01 9.58969235e-01
5.70009947e-01 -2.33935595e-01 2.13366762e-01 -2.71374106e-01
2.67727733e-01 -3.64142209e-01 -6.46650255e-01 1.22554183e+00
-5.65052092e-01 8.69701087e-01 7.88553432e-02 -4.18425828e-01
6.21209979e-01 4.62055683e-01 1.79565027e-01 -3.37373585e-01
1.90881863e-01 2.60306656e-01 1.43425584e-01 -1.02130830e+00
5.78121468e-03 -1.65628478e-01 5.74821532e-01 4.37552966e-02
-1.36832399e-02 2.57026196e-01 2.21150324e-01 4.16361131e-02
1.27169776e+00 -1.52720943e-01 6.25334918e-01 1.44537911e-02
6.99011624e-01 3.33711535e-01 -1.35014981e-01 2.29916200e-01
-2.47865230e-01 4.78453875e-01 3.07243317e-01 -7.97789037e-01
-7.55491495e-01 -9.37645257e-01 -3.02603483e-01 5.23101151e-01
-4.02671136e-02 -1.34592235e-01 -9.81219292e-01 -9.18988585e-01
-5.68972901e-03 1.47187099e-01 -1.42102551e+00 -6.85275793e-02
-2.24572737e-02 -8.43421221e-01 6.22463107e-01 4.01879787e-01
6.67155087e-01 -9.74685252e-01 -4.40421849e-01 -1.89488798e-01
3.37532699e-01 -5.35559475e-01 1.53712377e-01 1.54886410e-01
-4.70468909e-01 -1.83477283e+00 -9.31303561e-01 -9.74717379e-01
1.33899534e+00 8.76612514e-02 4.72983211e-01 5.18166542e-01
-1.41644001e+00 3.39927614e-01 -1.88625172e-01 -7.65402317e-01
-6.09098852e-01 1.88806653e-02 -3.62530053e-01 4.39978898e-01
7.17337012e-01 3.11646670e-01 -7.88708270e-01 -2.24625647e-01
-1.12709236e+00 1.22983411e-01 8.67219448e-01 6.99518085e-01
5.25493741e-01 1.73829630e-01 2.41084144e-01 -1.21635294e+00
6.99791431e-01 -6.96419716e-01 7.69722015e-02 4.18582678e-01
-2.25479066e-01 -4.81618553e-01 4.60325122e-01 -2.24444270e-01
-8.24115098e-01 4.07866269e-01 -3.88205677e-01 -1.03952117e-01
-4.26170617e-01 4.83315796e-01 5.30601203e-01 -5.53123295e-01
1.11042094e+00 1.76841289e-01 3.95970821e-01 -6.39202818e-02
-1.56671256e-01 1.09376633e+00 3.68651569e-01 4.57878411e-01
4.56945807e-01 6.99606717e-01 1.37249917e-01 -1.14202726e+00
-6.90202177e-01 -8.75244439e-01 -4.93310899e-01 -6.88936949e-01
9.86125350e-01 -8.53803217e-01 -1.67736873e-01 1.07911968e+00
-6.68447018e-01 -4.27073628e-01 -9.91885588e-02 1.17267847e-01
1.03044719e-01 2.46102318e-01 -7.00945735e-01 -9.47694659e-01
-7.99291134e-01 -6.29610777e-01 8.23648334e-01 7.33941078e-01
-4.16838557e-01 -1.56826282e+00 2.19419599e-01 3.88172805e-01
5.82324803e-01 7.31313050e-01 7.12400734e-01 -5.61837792e-01
1.20239623e-01 -9.41008329e-01 -2.55281299e-01 6.50631607e-01
5.81307411e-01 4.82994229e-01 -1.18695486e+00 -2.11466759e-01
-5.26530564e-01 -3.45111102e-01 1.25606000e+00 3.50616187e-01
1.29724646e+00 -1.57959118e-01 -5.61704457e-01 5.58695436e-01
2.02130461e+00 -2.99040861e-02 7.42487013e-01 1.34278372e-01
5.18768847e-01 6.73846781e-01 1.47337481e-01 -1.38480783e-01
1.04278713e-01 -2.08166048e-01 8.14346194e-01 -7.40990996e-01
-5.91147780e-01 -7.85104409e-02 1.21390164e-01 8.18673000e-02
-4.67325419e-01 1.04164882e-02 -9.60415840e-01 7.46675014e-01
-1.16865814e+00 -6.60019815e-01 -4.20174718e-01 2.19522071e+00
7.64852643e-01 -8.72698352e-02 1.45534143e-01 1.68216899e-01
9.05797064e-01 -2.77654678e-01 -5.65222859e-01 -8.13621759e-01
1.24834046e-01 6.56222463e-01 7.43573725e-01 3.70433927e-01
-1.30426133e+00 6.83550715e-01 6.60374546e+00 6.95474088e-01
-2.00391507e+00 -2.29639232e-01 9.20834839e-01 5.49194627e-02
1.04422405e-01 -6.22265816e-01 -4.97461379e-01 2.74129152e-01
9.48368013e-01 1.34183049e-01 2.56514758e-01 5.62443137e-01
-1.42760187e-01 -6.64857626e-01 -7.94327438e-01 7.55690396e-01
3.48332942e-01 -1.48292220e+00 6.90994263e-02 1.56792954e-01
7.15185106e-01 1.36386305e-01 2.03213722e-01 -4.32591736e-02
2.64445730e-02 -1.67440784e+00 -2.55119324e-01 1.03764939e+00
1.59297216e+00 -4.14950371e-01 1.09522784e+00 2.88145542e-01
-8.01689267e-01 -3.63706524e-04 -2.32722193e-01 1.86995268e-01
-5.60190320e-01 2.67976761e-01 -1.52366614e+00 2.04744160e-01
4.95613784e-01 7.46762931e-01 -1.08515084e+00 1.28152514e+00
-1.46513149e-01 7.82430947e-01 -2.59069234e-01 -3.38877559e-01
1.46693021e-01 2.97755420e-01 8.33791718e-02 1.23193467e+00
-3.25195561e-03 -2.85040766e-01 -2.93812215e-01 3.72006536e-01
4.52528670e-02 1.92362636e-01 -5.56023300e-01 -1.93512812e-01
-8.63610879e-02 1.79502904e+00 -4.75154072e-01 -2.12800324e-01
-1.66153610e-01 1.07268691e+00 6.81369081e-02 1.43757284e-01
-1.06814399e-01 -3.98851514e-01 1.74236789e-01 4.72658873e-01
-2.42554292e-01 6.33443058e-01 -2.51274228e-01 -7.11141169e-01
-2.36032113e-01 -5.83893955e-01 4.66341674e-01 -5.35891593e-01
-1.51779926e+00 7.44385183e-01 -5.79108894e-01 -9.12126958e-01
-1.69161305e-01 -1.12858391e+00 -1.26222086e+00 1.21760917e+00
-1.61854541e+00 -1.61752975e+00 -8.56848419e-01 5.20577967e-01
4.96538967e-01 -3.01342905e-01 1.37481332e+00 -3.24943244e-01
-6.78773642e-01 4.51464713e-01 -1.62682906e-01 3.84503692e-01
7.88899779e-01 -1.68861580e+00 5.56392930e-02 3.80358189e-01
-4.43811476e-01 4.58024740e-01 6.07898906e-02 -7.41066158e-01
-1.02238643e+00 -1.48929656e+00 7.01037526e-01 -2.57313311e-01
6.18151426e-01 1.51286349e-01 -5.37678063e-01 2.36381054e-01
5.23036003e-01 1.88779503e-01 1.39480746e+00 -2.06373528e-01
-3.26227307e-01 -4.68939506e-02 -1.66438997e+00 3.54684711e-01
1.43930912e-01 -4.89068776e-01 9.59175080e-03 7.69897938e-01
-2.87508398e-01 -2.80712754e-01 -1.00899613e+00 2.05870673e-01
6.60561860e-01 -6.88777089e-01 6.74706697e-01 -7.25465298e-01
9.49611425e-01 -4.08085436e-02 1.42537504e-01 -1.38431776e+00
-4.10250574e-02 -9.16973967e-03 -1.79917589e-02 5.82934141e-01
5.59864283e-01 -4.28194225e-01 1.18039179e+00 3.45544279e-01
5.32159328e-01 -1.35782087e+00 -6.55433178e-01 -1.16012476e-01
1.53041512e-01 2.05776885e-01 2.69595645e-02 8.00035238e-01
-1.99572816e-01 1.87159032e-02 2.07217578e-02 -1.21642705e-02
5.95120668e-01 -5.54460824e-01 4.68881279e-01 -1.06235468e+00
1.45931825e-01 -1.49347723e-01 -7.90687084e-01 7.56600464e-04
-4.43942070e-01 -9.81906831e-01 -3.56744826e-01 -1.95269859e+00
3.29317540e-01 -3.65159869e-01 -4.09888864e-01 7.30025709e-01
-1.98168859e-01 8.12995076e-01 -3.98908071e-02 2.91214380e-02
-2.30626389e-01 -5.98081946e-01 1.21546400e+00 -4.51201707e-01
1.80376753e-01 2.36985624e-01 -6.11251891e-01 7.37597942e-01
1.07936609e+00 2.81220585e-01 -8.82858187e-02 -2.01836377e-01
1.58610836e-01 -7.19972029e-02 6.09797478e-01 -1.24261642e+00
4.09967214e-01 -6.14451468e-02 9.82888520e-01 -4.53316808e-01
5.19130111e-01 -6.64091706e-01 7.79488236e-02 9.46948171e-01
-5.17304063e-01 -7.14041054e-01 1.66195139e-01 5.00826001e-01
-1.09717704e-01 -4.80281085e-01 9.84155536e-01 -5.17379940e-01
-1.01710641e+00 3.45035076e-01 -5.73332548e-01 -7.66388714e-01
1.54531157e+00 -5.14440536e-01 -4.87916619e-01 -1.37251243e-01
-9.82651949e-01 -7.79557899e-02 5.08519232e-01 2.14781836e-01
7.67273068e-01 -1.07849681e+00 -9.24477756e-01 1.41959906e-01
4.33119744e-01 -3.33494663e-01 4.64450896e-01 7.84171462e-01
-1.18479204e+00 2.30320454e-01 -5.60263455e-01 -4.78751481e-01
-1.72742486e+00 2.42806990e-02 9.59927440e-01 -1.68575943e-01
-1.16868541e-01 1.19620800e+00 -3.69328111e-01 -3.34892392e-01
1.05050728e-01 4.95755039e-02 -6.65473878e-01 -3.36251885e-01
8.41626167e-01 2.13959977e-01 1.61294773e-01 -5.54409921e-01
-1.22253954e-01 3.88923645e-01 -5.93872726e-01 3.80813718e-01
1.05282891e+00 5.39236307e-01 -3.18473428e-01 1.69609249e-01
1.24496734e+00 -1.93042874e-01 -9.30795670e-01 6.35441765e-02
-4.24826324e-01 -3.53915215e-01 2.08986625e-01 -1.54908860e+00
-1.17561769e+00 9.71533895e-01 1.20757508e+00 3.51787210e-01
1.14229202e+00 -4.38896298e-01 6.63004398e-01 1.11728512e-01
1.10665138e-03 -1.30001974e+00 -2.36671597e-01 2.02560076e-03
7.07017064e-01 -1.45725441e+00 1.85646396e-02 -7.74798095e-01
-6.48699760e-01 1.54562664e+00 8.04920137e-01 -6.47553861e-01
7.34339237e-01 1.72483861e-01 6.75444186e-01 -2.86182374e-01
-7.25025296e-01 -2.83843070e-01 4.68186885e-01 9.75828767e-01
5.29812157e-01 4.39181894e-01 -3.78643513e-01 -2.30615847e-02
2.40440339e-01 2.20922649e-01 5.50657153e-01 8.44589889e-01
-3.61277014e-01 -1.08539951e+00 -2.13272214e-01 1.24111366e+00
-8.39022994e-01 -2.08154600e-03 -1.20714128e+00 8.19268227e-01
3.73617053e-01 8.10699463e-01 -2.73402613e-02 -5.37619889e-01
-1.88072398e-01 -2.14273244e-01 5.47241867e-01 -8.71926725e-01
-1.05116856e+00 -3.28727126e-01 3.42091769e-01 -2.52246261e-01
-3.77012521e-01 -2.68119127e-01 -1.00248539e+00 8.54263455e-02
-1.57374069e-01 -3.09755594e-01 1.33809531e+00 7.16225326e-01
-6.99376017e-02 5.18457949e-01 5.35186648e-01 -2.38219127e-01
-4.50538367e-01 -1.11420274e+00 -6.57083094e-01 3.67429197e-01
2.88551480e-01 -3.67345735e-02 -5.52300140e-02 2.41083220e-01] | [15.669272422790527, -3.0046756267547607] |
41cb829b-fb7a-4669-aea3-822a316044e7 | gender-bias-in-transformer-models-a | 2306.10530 | null | https://arxiv.org/abs/2306.10530v1 | https://arxiv.org/pdf/2306.10530v1.pdf | Gender Bias in Transformer Models: A comprehensive survey | Gender bias in artificial intelligence (AI) has emerged as a pressing concern with profound implications for individuals' lives. This paper presents a comprehensive survey that explores gender bias in Transformer models from a linguistic perspective. While the existence of gender bias in language models has been acknowledged in previous studies, there remains a lack of consensus on how to effectively measure and evaluate this bias. Our survey critically examines the existing literature on gender bias in Transformers, shedding light on the diverse methodologies and metrics employed to assess bias. Several limitations in current approaches to measuring gender bias in Transformers are identified, encompassing the utilization of incomplete or flawed metrics, inadequate dataset sizes, and a dearth of standardization in evaluation methods. Furthermore, our survey delves into the potential ramifications of gender bias in Transformers for downstream applications, including dialogue systems and machine translation. We underscore the importance of fostering equity and fairness in these systems by emphasizing the need for heightened awareness and accountability in developing and deploying language technologies. This paper serves as a comprehensive overview of gender bias in Transformer models, providing novel insights and offering valuable directions for future research in this critical domain. | ['Farhana Ferdousi Liza', 'Palla Vijay', 'Yericherla Deepak Joel', 'Praneeth Nemani'] | 2023-06-18 | null | null | null | null | ['machine-translation'] | ['natural-language-processing'] | [ 5.15702739e-02 4.20856446e-01 -7.32692063e-01 -7.08016336e-01
-3.79238814e-01 -7.66505241e-01 8.51194203e-01 3.14762235e-01
-6.87705100e-01 6.92154706e-01 5.35141647e-01 -6.45787239e-01
-1.31445065e-01 -4.68599409e-01 -2.09395289e-02 -3.17924887e-01
3.47766608e-01 7.14461088e-01 -7.67810345e-01 -3.40125263e-01
7.67605305e-01 2.13113949e-01 -1.45925236e+00 -1.50520727e-01
1.16004777e+00 5.66389024e-01 -4.11996096e-01 2.57904470e-01
-2.13916853e-01 8.12946737e-01 -1.04192150e+00 -1.29981363e+00
-2.77097058e-03 -3.67680788e-01 -9.28252518e-01 -2.13313028e-01
6.60704911e-01 -5.46148241e-01 4.82118055e-02 9.67521846e-01
8.35175216e-01 -2.58866370e-01 7.02519715e-01 -1.61729789e+00
-8.55456114e-01 9.58504021e-01 -2.93626010e-01 4.25173193e-01
1.29740655e-01 2.23241642e-01 1.25982928e+00 -6.24374330e-01
4.51840818e-01 1.65798962e+00 7.13079989e-01 7.33366787e-01
-1.17269289e+00 -1.13922417e+00 2.90349394e-01 -7.94010088e-02
-1.09642637e+00 -1.02637303e+00 4.62981969e-01 -7.09017754e-01
8.13135147e-01 5.24217248e-01 9.25389230e-01 1.13233984e+00
1.66637093e-01 6.06691480e-01 1.23669815e+00 -4.74709332e-01
-3.26822288e-02 6.28642261e-01 2.54385054e-01 4.03719842e-01
6.68347895e-01 1.26405612e-01 -9.08839107e-01 -4.72619832e-01
4.55969274e-01 -6.07883990e-01 3.01955849e-01 -2.60423958e-01
-1.22341597e+00 1.09144604e+00 -1.17756777e-01 4.55643415e-01
-1.19524859e-01 -1.19447168e-02 8.18116307e-01 8.18160027e-02
8.12936604e-01 6.77826583e-01 -2.33038947e-01 -6.94165468e-01
-8.60536158e-01 6.59611166e-01 8.26421022e-01 5.43478012e-01
5.52794747e-02 2.72925586e-01 -3.16367656e-01 1.12897420e+00
5.09987056e-01 7.14222908e-01 2.39267275e-01 -1.25742698e+00
5.32271862e-01 5.75755060e-01 2.44856730e-01 -1.36964679e+00
-2.48485699e-01 -3.58420998e-01 -3.90762925e-01 3.91767807e-02
6.54748440e-01 -9.68300700e-02 -3.35979909e-01 2.08124590e+00
-6.42483309e-03 -8.70396912e-01 -1.05171375e-01 8.60466838e-01
8.70256543e-01 -9.72970128e-02 7.42003560e-01 8.18690807e-02
1.45010352e+00 -3.43966007e-01 -7.70357847e-01 -7.30788291e-01
5.49447119e-01 -8.55337322e-01 1.07814562e+00 1.64077505e-01
-1.25086260e+00 6.29846305e-02 -7.51008928e-01 -3.93754423e-01
-3.77684861e-01 8.60569254e-02 1.22260571e+00 1.73810267e+00
-8.40073764e-01 1.52648225e-01 -5.47918022e-01 -8.19351077e-01
5.88511825e-01 3.86396140e-01 -6.12346493e-02 2.69739747e-01
-1.31820965e+00 1.22305119e+00 -2.33172432e-01 1.22082457e-01
-4.22417909e-01 -8.28120530e-01 -9.36268449e-01 -4.57832098e-01
-1.03817455e-01 -8.55362296e-01 1.45488226e+00 -1.14375520e+00
-1.13474214e+00 1.42829919e+00 -3.66666555e-01 -1.50851101e-01
5.64015567e-01 1.18394710e-01 -2.92088747e-01 -2.78289527e-01
6.14674807e-01 7.03458667e-01 2.40552738e-01 -9.67830956e-01
-7.20823586e-01 -6.88528180e-01 2.39402801e-01 6.17153168e-01
-5.55577934e-01 7.00165808e-01 3.02396715e-01 -7.17481852e-01
-1.55165598e-01 -6.80413723e-01 9.53335539e-02 2.35093813e-02
-1.17397510e-01 -5.03351629e-01 3.11535656e-01 -6.31448030e-01
1.48665929e+00 -1.81174207e+00 -2.40516379e-01 -6.20178953e-02
6.23621225e-01 -3.68610844e-02 3.38794857e-01 5.21899879e-01
3.40393752e-01 5.99567175e-01 -1.70679800e-02 -3.97775471e-01
4.01411951e-01 9.19647068e-02 -2.63992131e-01 5.53146601e-01
-8.59185606e-02 8.15053999e-01 -9.73217189e-01 -7.45204806e-01
1.89107195e-01 3.97953570e-01 -5.00037014e-01 -2.66883284e-01
3.84149641e-01 2.26411909e-01 -2.28367716e-01 9.77869213e-01
6.21433258e-01 2.70836771e-01 3.22716504e-01 1.54789627e-01
-5.95586181e-01 7.75561750e-01 -5.59503078e-01 1.06776273e+00
-1.42897785e-01 8.23203504e-01 3.61111462e-01 -8.28598440e-01
9.46847260e-01 1.72736540e-01 1.28696829e-01 -8.14795315e-01
2.57007986e-01 5.59788704e-01 4.36541438e-01 -2.65958220e-01
9.30257082e-01 -5.67523956e-01 -4.11114186e-01 7.15325177e-01
-2.33986676e-01 -4.33863759e-01 2.92486399e-01 6.27568737e-02
1.51441768e-01 -1.46760896e-01 1.49851978e-01 -7.47151911e-01
2.44191110e-01 8.95482749e-02 7.49936521e-01 5.06575525e-01
-8.27837050e-01 1.44170195e-01 9.00097191e-01 -3.56772989e-01
-8.04552436e-01 -6.78838015e-01 -3.83700639e-01 1.43719482e+00
-1.92747533e-01 -3.93631011e-01 -7.60730505e-01 -5.69328368e-01
3.08385819e-01 1.07980609e+00 -8.24810684e-01 -2.41779432e-01
-1.93527207e-01 -1.25416231e+00 1.14338160e+00 4.24543500e-01
4.88512844e-01 -6.13320351e-01 -1.04933095e+00 -3.19572657e-01
-5.43953359e-01 -8.08742225e-01 -1.83863834e-01 -3.55830431e-01
-9.77418482e-01 -1.04480445e+00 -5.05357087e-01 -6.73847795e-01
4.21471477e-01 3.95459756e-02 1.46552110e+00 2.52466332e-02
7.65912533e-02 5.24956048e-01 2.68997937e-01 -1.01741838e+00
-4.68159020e-01 1.62835672e-01 6.75972775e-02 -6.58197880e-01
1.02150226e+00 1.35281729e-02 -5.72316706e-01 3.03309947e-01
-2.95423478e-01 -3.60595137e-02 2.90482521e-01 7.47493207e-01
-3.38385850e-01 -4.23959136e-01 8.08247745e-01 -1.08005381e+00
1.06165051e+00 -5.10987699e-01 -1.65454641e-01 7.95456693e-02
-1.28550959e+00 -5.44862449e-01 3.42496345e-03 1.08048826e-01
-1.24034536e+00 -9.32129204e-01 2.11606070e-01 4.63889837e-01
1.76488981e-01 4.10503060e-01 -7.79768229e-02 -7.63928592e-02
7.13799775e-01 -4.88794863e-01 5.17077863e-01 -1.18341506e-01
1.34398222e-01 9.76184309e-01 1.12195686e-01 -1.04400837e+00
2.75860697e-01 3.07611316e-01 -4.42358732e-01 -7.80779243e-01
-3.81510824e-01 1.13537185e-01 -2.63152540e-01 -3.79326820e-01
4.42767471e-01 -8.66132200e-01 -7.95958519e-01 4.68750894e-01
-6.26108408e-01 -2.72582203e-01 2.69527454e-02 5.49090385e-01
-2.83400387e-01 1.92770287e-01 -5.00027716e-01 -1.07383072e+00
-5.74985206e-01 -1.25218368e+00 8.31504464e-01 2.12881163e-01
-1.15301192e+00 -1.16428137e+00 -1.52646288e-01 9.93733943e-01
4.58761573e-01 -1.34658977e-01 1.19943213e+00 -6.71681941e-01
9.35265794e-02 6.71013072e-03 -1.77600250e-01 -1.11147679e-01
1.83423340e-01 2.82965630e-01 -1.07348299e+00 3.31356339e-02
-1.23585880e-01 -4.69168216e-01 2.34769985e-01 4.55331773e-01
6.01344585e-01 -2.17883974e-01 -3.05454165e-01 7.39484429e-02
7.97519147e-01 3.26306224e-01 8.30854252e-02 5.49910903e-01
5.05771816e-01 1.20964193e+00 8.55020881e-01 4.46655244e-01
1.17541957e+00 4.02316093e-01 -8.66186544e-02 3.53556350e-02
1.81581482e-01 -1.21065557e-01 8.72322097e-02 4.56682533e-01
-1.84070572e-01 1.17867701e-01 -1.32996011e+00 8.27102244e-01
-1.48558593e+00 -7.99627483e-01 4.79272902e-02 2.35129905e+00
7.29507267e-01 1.17830224e-01 5.08694708e-01 2.32248992e-01
7.08360553e-01 2.54949361e-01 -5.17566025e-01 -1.02305341e+00
-2.56402820e-01 -2.34555021e-01 3.35365832e-01 5.16515613e-01
-7.83855796e-01 8.84494007e-01 8.04922295e+00 2.72217005e-01
-1.07362056e+00 -1.95674613e-01 1.21812594e+00 -1.37993515e-01
-7.91848660e-01 -2.08142679e-02 -3.75239938e-01 2.23666787e-01
6.79946065e-01 -9.11877990e-01 2.70952940e-01 5.47168553e-01
3.61655146e-01 -2.82021761e-01 -1.10563052e+00 8.09718013e-01
3.34709674e-01 -7.22250640e-01 -1.27302587e-01 1.98416650e-01
3.41917306e-01 -2.43801102e-01 5.57491064e-01 2.39904478e-01
3.43324304e-01 -1.25210881e+00 1.13796175e+00 -1.34827256e-01
8.98399115e-01 -8.98103118e-01 6.75692797e-01 -1.88244596e-01
-5.31616390e-01 -1.64978072e-01 -1.67079329e-01 -6.39972210e-01
-1.28080130e-01 5.46706855e-01 -7.71596432e-01 1.00174643e-01
9.22849298e-01 5.78314066e-01 -4.51008439e-01 2.55873442e-01
1.57838196e-01 5.72692513e-01 4.38748673e-03 -1.73382998e-01
-9.28625166e-02 -3.84469151e-01 4.52375829e-01 1.20498204e+00
-7.14187473e-02 -1.77260607e-01 -3.90992910e-01 8.35081041e-01
6.68276697e-02 1.96971059e-01 -8.35969627e-01 -8.79656911e-01
9.01377499e-01 1.18340266e+00 -7.22396791e-01 -2.03811064e-01
-6.39102161e-01 2.71657676e-01 -1.28624542e-02 1.19553342e-01
-4.44665670e-01 -1.88474119e-01 1.11976266e+00 -5.12196273e-02
-6.85463667e-01 -8.38859752e-02 -1.24355114e+00 -8.27101827e-01
-3.30089360e-01 -1.46244192e+00 6.14410937e-01 -2.23625809e-01
-9.96973932e-01 2.23870855e-02 2.49696493e-01 -4.96573418e-01
-4.09227729e-01 -4.46760297e-01 -1.65332392e-01 1.10012662e+00
-8.60443771e-01 -1.30492306e+00 -1.04878724e-01 -2.96556372e-02
2.22522289e-01 -3.93042147e-01 9.11542654e-01 3.03215057e-01
-7.45057583e-01 1.12999487e+00 -4.31791186e-01 -2.96104848e-02
1.17678118e+00 -1.19382465e+00 5.64997971e-01 4.92461324e-01
-5.54097593e-01 1.27862978e+00 8.98473501e-01 -6.67144179e-01
-1.32790470e+00 -3.80134076e-01 1.54403234e+00 -8.43002439e-01
5.49641132e-01 -4.13669914e-01 -1.17882654e-01 1.00612521e+00
2.63449490e-01 -9.15489316e-01 1.32289791e+00 6.03505194e-01
-3.10090125e-01 -6.44515380e-02 -1.61680794e+00 9.96974468e-01
1.10843277e+00 -7.70294309e-01 -5.21933675e-01 -1.85867056e-01
1.55205011e-01 -3.38951409e-01 -9.70455885e-01 2.49331161e-01
1.08111453e+00 -1.09231532e+00 7.92021334e-01 -3.87978137e-01
6.00110590e-01 2.76953161e-01 -1.08486630e-01 -1.12902272e+00
-3.20070297e-01 -4.57536489e-01 5.87006450e-01 1.63171721e+00
2.44607449e-01 -1.11700022e+00 7.82402992e-01 1.52320147e+00
7.40181804e-02 -6.67763114e-01 -8.02164078e-01 3.81756574e-02
7.90774882e-01 -4.03905869e-01 9.31572378e-01 1.24389327e+00
5.44504166e-01 3.60010892e-01 -1.31711299e-02 -3.73835117e-01
6.58360839e-01 -6.74853697e-02 7.60560870e-01 -1.19342530e+00
4.73314106e-01 -9.41981614e-01 -3.59210908e-01 -2.41063178e-01
3.09112221e-01 -7.51987100e-01 -4.98405129e-01 -1.43982041e+00
4.70105261e-01 -5.39518595e-01 1.80866435e-01 1.86517239e-01
-9.49888676e-02 3.52383137e-01 1.37498975e-01 4.36141454e-02
2.27545202e-03 3.63289177e-01 9.20287132e-01 -1.73810124e-02
2.05710024e-01 -9.44356471e-02 -1.85687971e+00 7.87526429e-01
9.73457575e-01 -1.48054272e-01 -6.27475202e-01 -6.44322515e-01
5.25132775e-01 -4.43620890e-01 3.19763005e-01 -2.69005090e-01
-6.30075485e-03 -5.61049640e-01 3.19429845e-01 -2.16434807e-01
2.33515695e-01 -5.09503305e-01 -6.56386986e-02 5.09447634e-01
-3.10399652e-01 4.40140516e-01 4.36531723e-01 -3.13920885e-01
8.53899866e-02 -5.88002354e-02 4.37418431e-01 -1.85246825e-01
-3.56520921e-01 -1.77827641e-01 -7.60457516e-01 4.43496168e-01
6.19683564e-01 -3.91646832e-01 -5.20672679e-01 -4.23538625e-01
5.48497364e-02 4.71416116e-01 7.14420259e-01 6.55231595e-01
1.28302455e-01 -1.21576619e+00 -6.56500638e-01 3.80997732e-02
3.85673434e-01 -4.90031481e-01 -2.19192076e-02 7.62515008e-01
-3.05315822e-01 6.07941508e-01 -3.59065890e-01 -1.86070755e-01
-1.37631249e+00 -7.76533559e-02 4.01328474e-01 2.02273116e-01
2.21699290e-02 8.41031432e-01 1.76955804e-01 -8.73980761e-01
2.36487478e-01 -6.33410141e-02 -3.93166363e-01 4.83343571e-01
2.54002512e-01 9.24711347e-01 -1.40483573e-01 -9.09727633e-01
-6.67952955e-01 2.98751146e-01 -6.75534159e-02 -6.53957665e-01
8.17633212e-01 -4.09294218e-01 -7.39195406e-01 8.19363296e-01
5.38956344e-01 2.39316523e-01 -4.17117715e-01 3.39253873e-01
-5.31120636e-02 -7.98880637e-01 -1.75044313e-01 -1.03032339e+00
-8.14296246e-01 7.18938589e-01 3.78859222e-01 -1.54604763e-02
8.00549865e-01 -2.80915558e-01 7.30710328e-02 -8.50889236e-02
4.02552545e-01 -1.53585804e+00 -5.00523686e-01 4.18492079e-01
9.31555986e-01 -1.12425041e+00 1.62807465e-01 -3.43160182e-01
-9.58618999e-01 7.06478119e-01 7.61409819e-01 6.05502665e-01
3.72615546e-01 1.05034262e-01 6.06924534e-01 -2.44498789e-01
-6.58667684e-01 3.96439165e-01 -7.06537962e-02 8.33507955e-01
1.36106205e+00 4.21398491e-01 -1.18718636e+00 7.37707138e-01
-1.07837784e+00 1.34222209e-02 3.84189308e-01 8.68636787e-01
3.23062874e-02 -1.42918372e+00 -5.67362070e-01 8.29090595e-01
-8.38147163e-01 -1.08412266e-01 -1.01346672e+00 7.97635436e-01
1.09240234e-01 1.39422417e+00 2.09162325e-01 -3.81245106e-01
1.43232971e-01 2.04155341e-01 4.22518760e-01 -4.11721647e-01
-1.03536022e+00 -3.07382822e-01 8.89133394e-01 -2.51663864e-01
-4.26465124e-01 -1.27442682e+00 -5.95159709e-01 -8.83242249e-01
1.22299641e-02 1.95916548e-01 6.00476563e-01 7.29850531e-01
2.97891200e-01 9.10727382e-02 9.97820795e-02 -2.93204457e-01
-4.46306229e-01 -9.41760182e-01 -5.42293191e-01 8.73292536e-02
1.11327767e-01 -7.67126739e-01 -3.05842072e-01 -3.66620809e-01] | [9.379997253417969, 10.131121635437012] |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.