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 |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
1ff5829b-f457-40c5-83c7-fd1fc31138ee | decorrelated-jet-substructure-tagging-using | 1703.03507 | null | http://arxiv.org/abs/1703.03507v1 | http://arxiv.org/pdf/1703.03507v1.pdf | Decorrelated Jet Substructure Tagging using Adversarial Neural Networks | We describe a strategy for constructing a neural network jet substructure
tagger which powerfully discriminates boosted decay signals while remaining
largely uncorrelated with the jet mass. This reduces the impact of systematic
uncertainties in background modeling while enhancing signal purity, resulting
in improved discovery significance relative to existing taggers. The network is
trained using an adversarial strategy, resulting in a tagger that learns to
balance classification accuracy with decorrelation. As a benchmark scenario, we
consider the case where large-radius jets originating from a boosted resonance
decay are discriminated from a background of nonresonant quark and gluon jets.
We show that in the presence of systematic uncertainties on the background
rate, our adversarially-trained, decorrelated tagger considerably outperforms a
conventionally trained neural network, despite having a slightly worse
signal-background separation power. We generalize the adversarial training
technique to include a parametric dependence on the signal hypothesis, training
a single network that provides optimized, interpolatable decorrelated jet
tagging across a continuous range of hypothetical resonance masses, after
training on discrete choices of the signal mass. | ['Andreas Søgaard', 'Daniel Whiteson', 'Pierre Baldi', 'Peter Sadowski', 'Edward Goul', 'Chase Shimmin', 'Edison Weik'] | 2017-03-10 | null | null | null | null | ['jet-tagging'] | ['graphs'] | [ 3.13054651e-01 1.42162636e-01 -1.32073194e-01 -4.01151776e-01
-1.07884741e+00 -9.24221337e-01 8.53089213e-01 -1.19936347e-01
-4.31810766e-01 7.34461367e-01 1.86669677e-01 -6.41520679e-01
1.10302359e-01 -8.68688703e-01 -6.54192686e-01 -9.97396588e-01
-7.87272826e-02 8.34380269e-01 5.38061142e-01 -1.30286336e-01
-3.06389511e-01 7.41933703e-01 -9.74409103e-01 2.58267850e-01
3.96412581e-01 1.07101774e+00 -6.20164573e-01 6.87335134e-01
7.27844387e-02 6.78161085e-01 -6.88910723e-01 -7.83788562e-01
7.08652079e-01 -6.77986801e-01 -3.67961198e-01 -7.15357721e-01
5.59810221e-01 1.63725361e-01 -5.58309615e-01 1.42514503e+00
7.74574459e-01 3.09573486e-02 5.55696487e-01 -9.11579072e-01
-3.88180703e-01 8.18535984e-01 1.17819048e-01 5.57889342e-01
-5.58942020e-01 -5.42419627e-02 8.96216035e-01 -5.54610848e-01
5.63709974e-01 1.13184476e+00 8.32758188e-01 3.45863312e-01
-1.88841927e+00 -9.91388977e-01 -4.72924769e-01 -5.61616898e-01
-9.37366247e-01 -6.13102198e-01 8.12092006e-01 -5.39709568e-01
5.72903693e-01 4.62414145e-01 3.78975034e-01 1.19123268e+00
4.66725916e-01 1.47517596e-03 1.03760099e+00 -5.52289784e-01
3.54188442e-01 2.40279824e-01 -1.40107274e-01 5.07779539e-01
3.25690031e-01 7.81294048e-01 -3.05336535e-01 -4.86094117e-01
7.65493393e-01 -5.57238221e-01 1.23773897e-02 -7.84681380e-01
-1.09684074e+00 1.05392325e+00 2.74449408e-01 3.90176177e-01
7.48868436e-02 2.25349069e-01 6.80546999e-01 5.28927922e-01
7.16986656e-01 9.73070920e-01 -5.89013875e-01 4.21510488e-01
-9.57181871e-01 1.95439458e-01 8.32627475e-01 6.35657251e-01
1.22779265e-01 7.55783200e-01 -4.86406326e-01 4.91613984e-01
-1.74995791e-02 7.16155171e-01 3.30431253e-01 -9.22731876e-01
1.49336636e-01 -7.33412802e-02 2.78661847e-01 -5.04256189e-01
-4.48811024e-01 -1.03959131e+00 -6.49313509e-01 5.24712503e-01
7.57076263e-01 -5.95090032e-01 -8.66568625e-01 1.78066242e+00
4.11865473e-01 -2.07899734e-01 2.06420988e-01 9.64093208e-01
6.21216357e-01 3.76742631e-01 1.65717617e-01 -1.82175279e-01
1.14697409e+00 -6.13625109e-01 -2.23672092e-01 -7.89994001e-02
1.92639738e-01 -9.31460142e-01 3.68635654e-01 4.33341444e-01
-9.19661701e-01 -5.34942269e-01 -1.00350082e+00 1.14142984e-01
-2.04219073e-01 -1.55185265e-02 1.01932669e+00 1.07041979e+00
-6.73446894e-01 7.58405626e-01 -6.74459457e-01 1.41622201e-01
-2.45857492e-01 5.06125748e-01 -3.43056470e-01 5.83003283e-01
-1.37018275e+00 8.07032764e-01 5.21628082e-01 -8.94436985e-02
-8.06052983e-01 -8.96414399e-01 -5.51843047e-01 3.73242982e-02
1.15569867e-01 -4.93346810e-01 1.54424131e+00 -1.18904591e+00
-1.35878420e+00 7.39061892e-01 6.16606057e-01 -7.06935883e-01
5.58081448e-01 3.56113076e-01 -1.03301871e+00 -4.38373834e-02
8.89214054e-02 2.67438114e-01 8.44466686e-01 -9.23031032e-01
-3.26527685e-01 -6.08302951e-02 -2.02936932e-01 -3.61809313e-01
3.62585545e-01 6.29271209e-01 -5.89151606e-02 -1.00930119e+00
1.18046388e-01 -8.53170931e-01 -3.73450428e-01 -3.63183200e-01
-1.83946207e-01 4.37889963e-01 2.66408741e-01 -4.67791021e-01
4.01977390e-01 -2.15671182e+00 -2.35167872e-02 2.40166411e-01
5.89772053e-02 3.30066800e-01 -1.04347311e-01 1.01251081e-01
-4.13272381e-01 -1.17557958e-01 1.56205848e-01 1.19078621e-01
1.62718922e-01 7.61517731e-04 -3.32121462e-01 4.82593775e-01
3.56129229e-01 7.28291333e-01 -9.99006867e-01 -3.02520618e-02
-2.35041425e-01 2.72634298e-01 -6.57273471e-01 8.59064609e-02
-1.97309181e-01 6.78968608e-01 -3.22844625e-01 4.28779155e-01
4.75831211e-01 2.69418299e-01 2.47930259e-01 -4.53616291e-01
-2.75901109e-01 2.91184813e-01 -9.21836913e-01 1.03832483e+00
-1.92100197e-01 4.89855647e-01 5.16685426e-01 -1.06717384e+00
1.40821612e+00 9.58082750e-02 3.88608456e-01 -8.63484323e-01
3.90631318e-01 2.95731008e-01 6.68514132e-01 9.26697701e-02
6.18034542e-01 -1.11855328e+00 -5.10166883e-01 9.40109491e-02
5.25351226e-01 1.44051850e-01 -3.13724279e-01 8.38392377e-02
1.23833358e+00 -1.41705677e-01 -1.29733309e-01 -3.73891115e-01
6.15899749e-02 7.88667127e-02 7.52410769e-01 9.55595493e-01
-1.92539811e-01 3.84514958e-01 6.28857315e-01 -4.37643528e-01
-1.13417315e+00 -1.35905445e+00 -4.57065016e-01 1.25923038e+00
-2.17089459e-01 8.95767808e-02 -1.54108822e-01 -7.27860153e-01
2.67124891e-01 8.28919530e-01 -4.53174442e-01 -5.40034235e-01
-7.32356906e-01 -1.20025861e+00 9.92355764e-01 6.36575401e-01
7.21700564e-02 -8.51251006e-01 7.01117292e-02 2.62936145e-01
3.74846280e-01 -7.61417150e-01 -3.81623566e-01 1.25628817e+00
-4.47723836e-01 -8.51479113e-01 -4.23214644e-01 -5.21319091e-01
2.87610501e-01 -3.84203643e-01 1.31494451e+00 -6.67081654e-01
-2.56322175e-01 3.36487778e-02 -1.77376851e-01 -4.71059740e-01
-1.18218434e+00 -4.63984102e-01 2.19453901e-01 -2.50621885e-01
2.85427779e-01 -3.21265429e-01 -7.53301904e-02 1.53549910e-01
-5.72654665e-01 -4.26804155e-01 5.42365253e-01 1.16387844e+00
4.80277568e-01 2.44159371e-01 7.08017647e-01 -7.65763760e-01
2.79093832e-01 -5.24182141e-01 -1.14117169e+00 -5.15451878e-02
-3.83338332e-01 5.69467723e-01 7.78777659e-01 -6.63154840e-01
-1.06389058e+00 -6.34657070e-02 -6.36833608e-02 -3.01184952e-01
3.89452241e-02 -1.35690928e-01 -2.99879998e-01 -6.02305233e-01
7.07257926e-01 -9.84910131e-02 -3.49490903e-02 -6.67773068e-01
5.09202540e-01 2.70098507e-01 1.15304863e+00 -9.25159812e-01
8.93645108e-01 -1.24360563e-03 3.35734487e-01 -9.53015164e-02
-6.09431267e-01 -2.46759787e-01 -4.85790730e-01 6.76705316e-02
7.41638660e-01 -8.40385199e-01 -1.37779593e-01 8.52294788e-02
-9.52963352e-01 -1.25258848e-01 -6.45644128e-01 8.58846664e-01
-3.63868505e-01 -1.56060802e-02 -5.50572634e-01 -5.36902785e-01
-2.93938875e-01 -1.04607201e+00 7.33653963e-01 -2.37400144e-01
1.60025284e-01 -8.84386241e-01 9.04166847e-02 -1.73411947e-02
4.52866256e-01 4.39778119e-01 1.21886611e+00 -1.52246892e+00
-3.33753794e-01 -5.19981027e-01 -2.62617081e-01 5.01648843e-01
-1.00769095e-01 -3.53997856e-01 -7.31183827e-01 -1.89152241e-01
6.18509650e-02 -1.62717402e-01 1.03165913e+00 2.05151677e-01
4.90372449e-01 -2.61996418e-01 -1.80579931e-01 4.94592309e-01
1.05547333e+00 4.88306910e-01 4.22754474e-02 2.32440084e-01
3.37624729e-01 1.41639441e-01 3.22334319e-01 3.25445086e-02
-4.34385657e-01 8.07120919e-01 3.30355763e-01 -2.23846257e-01
-1.40367329e-01 -1.48398787e-01 3.54606032e-01 4.89953697e-01
1.55258812e-02 -1.87972248e-01 -4.83642220e-01 1.98706433e-01
-1.27248061e+00 -1.10906518e+00 -9.86841470e-02 2.41599989e+00
1.01959002e+00 6.00161850e-01 3.51753443e-01 -2.19676673e-01
6.99370682e-01 5.25581501e-02 -4.00309473e-01 -6.35568857e-01
3.62564363e-02 9.42722023e-01 1.23570001e+00 4.66077834e-01
-1.27201343e+00 7.85403073e-01 7.00218058e+00 5.12404382e-01
-1.18891394e+00 3.72257560e-01 3.15161020e-01 -2.30239585e-01
-3.84624541e-01 -1.18315101e-01 -6.98553503e-01 4.81719136e-01
1.04092920e+00 -6.84189796e-02 3.82363409e-01 7.89776385e-01
-1.34105742e-01 3.72888625e-01 -1.07589769e+00 3.01592767e-01
-1.41603544e-01 -1.23642600e+00 -2.38761231e-01 -1.15961730e-01
4.03092414e-01 3.00171643e-01 -1.29990980e-01 7.82593012e-01
7.48226345e-01 -7.56218851e-01 1.05869842e+00 4.62756187e-01
7.82015264e-01 -9.54230666e-01 7.27698326e-01 2.08859459e-01
-7.12091148e-01 1.04710817e-01 -4.94704992e-01 2.21994415e-01
-1.18918069e-01 7.05558121e-01 -8.65168989e-01 5.00892580e-01
3.12218249e-01 -3.35902721e-01 -3.31235409e-01 8.08364630e-01
2.41670698e-01 8.64081085e-01 -9.19908583e-02 9.72821265e-02
1.47006914e-01 -2.55675972e-01 8.32107008e-01 1.46192765e+00
4.29580301e-01 4.59282957e-02 2.00876608e-01 1.02013695e+00
-1.93248063e-01 -1.62326068e-01 -5.92187583e-01 -1.66734561e-01
4.90786165e-01 1.22191262e+00 -8.44717801e-01 -1.32562488e-01
-3.04958194e-01 5.49016893e-01 1.94232780e-02 1.94750890e-01
-7.64456451e-01 -3.50137323e-01 6.59299910e-01 3.29832867e-04
6.37606978e-01 -4.67910022e-02 1.75117210e-01 -9.95405972e-01
-3.04646790e-01 -8.81206155e-01 2.58819550e-01 -4.14999872e-01
-1.30976260e+00 5.59829235e-01 -1.22336060e-01 -9.56107855e-01
-2.49496982e-01 -8.46904695e-01 -6.34267271e-01 1.10397983e+00
-8.88620615e-01 -9.47706103e-01 4.60030764e-01 -1.15103014e-02
-9.78781283e-02 -7.18612432e-01 7.78130710e-01 3.30596924e-01
-2.81479836e-01 6.89380467e-01 3.48343879e-01 3.46812516e-01
9.71659601e-01 -1.49154961e+00 4.95349586e-01 8.84056389e-01
1.96029305e-01 4.90169615e-01 1.00445890e+00 -7.78540432e-01
-1.32952595e+00 -1.58913934e+00 5.41754544e-01 -3.14422458e-01
1.06669724e+00 -6.90599740e-01 -5.81692040e-01 6.64395094e-01
-1.81151941e-01 7.61105001e-01 6.89884901e-01 1.39469773e-01
-6.42382085e-01 -1.76305428e-01 -1.37743104e+00 1.76440760e-01
7.68474460e-01 -5.13235986e-01 -6.55975044e-01 3.39212298e-01
8.22170854e-01 -3.36818516e-01 -1.07957232e+00 4.39349890e-01
4.68354583e-01 -8.16759229e-01 1.24575496e+00 -1.32339692e+00
-2.22892180e-01 -2.06563711e-01 -3.24154824e-01 -1.38067782e+00
-8.18713486e-01 -6.16567433e-01 2.87617743e-01 1.24162638e+00
2.24919438e-01 -5.78456163e-01 6.35203540e-01 3.61061513e-01
-2.81503409e-01 -7.73441121e-02 -1.25942159e+00 -1.21180296e+00
5.30113161e-01 -3.17574561e-01 4.36694652e-01 8.67512345e-01
-3.96368414e-01 2.44083226e-01 -5.22305667e-01 4.42276984e-01
5.03125250e-01 1.12560503e-01 3.43335003e-01 -1.48964810e+00
-8.76727343e-01 -4.05208766e-01 -5.41886270e-01 -4.91994858e-01
1.68367267e-01 -1.45105994e+00 1.33852556e-01 -2.90462673e-01
-1.06772408e-01 -5.72083592e-01 -5.45348346e-01 3.06781262e-01
2.96196401e-01 6.09669745e-01 -5.56002855e-02 -3.03782523e-01
-3.02700758e-01 1.34036586e-01 8.99086654e-01 -2.08937302e-01
1.02862298e-01 1.55620068e-01 -7.12724686e-01 7.72712588e-01
6.77368104e-01 -8.09605777e-01 5.55691123e-02 9.94244590e-02
2.08803803e-01 4.06857014e-01 7.45608151e-01 -1.04836750e+00
-2.53595054e-01 9.89679098e-02 9.00047481e-01 -1.62012260e-02
2.19817653e-01 -4.97297168e-01 4.23952699e-01 2.45613277e-01
-5.91784775e-01 -2.79804051e-01 3.67009252e-01 5.39225161e-01
-3.01977322e-02 -5.55819690e-01 1.04210341e+00 -1.61989868e-01
-1.01054356e-01 9.89559572e-03 -4.18433905e-01 2.05197200e-01
5.51617086e-01 4.86246079e-01 -3.65820497e-01 3.24013680e-02
-9.07694817e-01 -4.47048336e-01 3.04044008e-01 3.37125212e-01
-3.19783688e-01 -1.58318377e+00 -4.36388075e-01 5.05760908e-01
-1.36801660e-01 -6.65880322e-01 -4.44660991e-01 5.34035742e-01
-1.12232029e-01 5.20691752e-01 -1.81438819e-01 -8.97460245e-03
-1.05480599e+00 7.70521045e-01 6.66965544e-01 -1.42009646e-01
-4.26651120e-01 9.30229247e-01 -2.22731486e-01 -6.62335217e-01
3.48539837e-02 -3.93089175e-01 2.71927923e-01 -1.09355226e-01
2.31755763e-01 2.15662703e-01 3.57489139e-01 -5.62456012e-01
-1.17597088e-01 -1.62059575e-01 2.08726913e-01 9.55742076e-02
8.23602617e-01 6.87636614e-01 8.10016915e-02 1.55234396e-01
1.03369844e+00 8.07193160e-01 -9.12760258e-01 -2.73675919e-01
-3.40910479e-02 -1.61991522e-01 3.81745577e-01 -1.12425745e+00
-1.11649847e+00 3.97701353e-01 6.36359096e-01 5.49347043e-01
6.87962472e-01 2.18678668e-01 3.96017820e-01 3.71697664e-01
4.20339197e-01 -4.37528431e-01 -2.07724452e-01 4.10772860e-01
6.77279830e-01 -9.90188599e-01 -3.65927637e-01 2.43701041e-03
-3.99249166e-01 1.30621505e+00 3.71580482e-01 -1.30570248e-01
2.99190193e-01 8.47013414e-01 1.28828913e-01 -1.31339163e-01
-3.74016672e-01 -6.41540661e-02 4.97660041e-01 6.53958559e-01
3.58561367e-01 2.01557845e-01 -1.17972210e-01 1.03830600e+00
-3.05307001e-01 -4.10931945e-01 1.42106995e-01 4.12287831e-01
-5.52271605e-01 -1.39094341e+00 -6.21710002e-01 4.80125248e-01
-7.39321530e-01 -2.80350029e-01 -3.90659362e-01 9.53855753e-01
5.70184410e-01 5.96482098e-01 2.55567938e-01 -2.26986825e-01
1.29399657e-01 4.74174142e-01 6.73113346e-01 -5.81375062e-01
-6.26948833e-01 2.19344914e-01 5.55821717e-01 -2.48324588e-01
-1.53930932e-01 -7.59380937e-01 -9.51775491e-01 1.53968558e-01
-5.38113058e-01 6.09654367e-01 4.37763184e-01 7.98008442e-01
6.88282913e-03 6.71966970e-01 5.09603143e-01 -7.79778898e-01
-1.10498583e+00 -6.86263561e-01 -7.28810787e-01 -8.25783312e-02
3.95900100e-01 -8.87024701e-01 -3.97680581e-01 -3.58763874e-01] | [15.688187599182129, 2.9239747524261475] |
8f83bf4f-68c1-4256-ad6b-cb755e773675 | meta-learning-transferable-parameterized | 2206.03597 | null | https://arxiv.org/abs/2206.03597v3 | https://arxiv.org/pdf/2206.03597v3.pdf | Meta-Learning Parameterized Skills | We propose a novel parameterized skill-learning algorithm that aims to learn transferable parameterized skills and synthesize them into a new action space that supports efficient learning in long-horizon tasks. We propose to leverage off-policy Meta-RL combined with a trajectory-centric smoothness term to learn a set of parameterized skills. Our agent can use these learned skills to construct a three-level hierarchical framework that models a Temporally-extended Parameterized Action Markov Decision Process. We empirically demonstrate that the proposed algorithms enable an agent to solve a set of difficult long-horizon (obstacle-course and robot manipulation) tasks. | ['George Konidaris', 'Michael Littman', 'Saket Tiwari', 'Shangqun Yu', 'Haotian Fu'] | 2022-06-07 | null | null | null | null | ['robot-manipulation'] | ['robots'] | [ 1.18656226e-01 3.36824238e-01 -6.90790772e-01 -1.29197776e-01
-1.27204442e+00 -5.58135331e-01 7.26711512e-01 -3.11380327e-01
-5.10213554e-01 1.15214479e+00 3.12006980e-01 -4.23260689e-01
-5.19683659e-01 -4.62676346e-01 -9.53172624e-01 -6.80822194e-01
-6.64454699e-01 6.23101056e-01 3.64211649e-01 -2.49020383e-01
3.19888383e-01 4.97823715e-01 -1.36571348e+00 1.32949486e-01
1.22665811e+00 4.99342829e-01 7.29449570e-01 9.53619361e-01
2.77087688e-01 1.30875087e+00 -1.57874838e-01 4.06827301e-01
6.35056317e-01 -2.44737357e-01 -1.01336694e+00 4.23383236e-01
-3.34471688e-02 -5.55129409e-01 -3.96517009e-01 7.43058860e-01
9.29865167e-02 8.76149356e-01 5.95221460e-01 -1.15995741e+00
-1.06980070e-01 5.11538684e-01 -6.12948909e-02 4.09341194e-02
2.98252732e-01 9.39316154e-01 8.20266068e-01 -2.67157882e-01
6.29342139e-01 1.57772696e+00 2.77608484e-01 7.51371741e-01
-1.04401004e+00 -2.49707922e-01 6.31100953e-01 -5.70926862e-03
-5.97279012e-01 9.48826894e-02 3.30556571e-01 -4.51479286e-01
1.12831414e+00 -3.52201492e-01 8.79866004e-01 1.37585151e+00
7.42290497e-01 9.21572387e-01 1.52760160e+00 -6.33060746e-03
5.23310125e-01 -5.60510755e-01 -2.03166991e-01 1.04866636e+00
1.25840900e-03 8.91433895e-01 -4.67510551e-01 -2.18419269e-01
1.15518737e+00 1.30802229e-01 1.55768111e-01 -6.06208444e-01
-1.38375747e+00 6.70296729e-01 3.04930836e-01 -2.25294173e-01
-7.70188093e-01 7.93443084e-01 2.54119128e-01 6.20474279e-01
1.68069169e-01 9.63935316e-01 -7.61562884e-01 -3.94901812e-01
-4.56680655e-01 7.26870298e-01 7.89530694e-01 1.25275385e+00
9.03672099e-01 4.38045621e-01 -6.69942498e-01 2.38074392e-01
4.39532287e-02 6.96327865e-01 1.20943837e-01 -1.88994229e+00
5.50855756e-01 2.59429246e-01 9.20882046e-01 -1.42409995e-01
-5.14466584e-01 -2.22531781e-01 -1.84186235e-01 8.23066056e-01
2.70416111e-01 -5.95085919e-01 -1.17891061e+00 1.60494125e+00
2.97589421e-01 6.18693531e-01 2.79862911e-01 8.25869322e-01
-1.88454211e-01 7.98489928e-01 3.94382387e-01 -1.77762672e-01
7.22429812e-01 -1.57661259e+00 -5.56391180e-01 -4.40507561e-01
4.87465948e-01 1.04652047e-02 1.34203124e+00 4.84935075e-01
-1.44300401e+00 -4.82006639e-01 -9.48386550e-01 1.74235374e-01
7.62178451e-02 -1.73035085e-01 7.31810570e-01 -1.66995659e-01
-1.12297475e+00 1.22073996e+00 -1.27919722e+00 -3.17397974e-02
3.99872512e-01 2.99748003e-01 2.23158002e-02 1.21154949e-01
-8.38172674e-01 1.34963965e+00 6.96409106e-01 -2.60255247e-01
-2.18010974e+00 -5.53845346e-01 -8.64044189e-01 -1.20837644e-01
9.67214346e-01 -9.94362891e-01 1.95116472e+00 -9.92731988e-01
-2.37387776e+00 3.19749802e-01 1.43758938e-01 -6.16011322e-01
5.34110904e-01 -5.28895438e-01 6.77755028e-02 2.88929462e-01
1.54315710e-01 6.14997089e-01 1.28064620e+00 -1.28690445e+00
-1.24111247e+00 1.19372383e-01 6.93433762e-01 8.12154472e-01
1.50510058e-01 -1.70605823e-01 -1.34182006e-01 -4.27870363e-01
-5.46819687e-01 -1.24101639e+00 -1.00385642e+00 -2.36412123e-01
-8.91451258e-03 -2.73956686e-01 2.16316655e-01 -3.74167651e-01
6.08865380e-01 -1.57388949e+00 1.04150701e+00 -3.19263451e-02
-6.99532107e-02 -6.12039417e-02 -5.93515337e-01 5.97019553e-01
5.43868124e-01 -3.82336289e-01 -2.49106959e-01 -3.55606288e-01
9.42067653e-02 6.80850029e-01 -5.97417295e-01 2.66731083e-01
1.21528350e-01 1.07923365e+00 -1.62356818e+00 -2.40663782e-01
1.57785237e-01 -9.07878652e-02 -5.42479336e-01 7.11970925e-01
-1.10509038e+00 1.08224893e+00 -1.03897560e+00 5.31701863e-01
3.61202769e-02 4.48765010e-02 2.16713846e-01 7.19244659e-01
-4.23376918e-01 3.63543838e-01 -7.94391632e-01 2.19321132e+00
-8.10925901e-01 1.05082683e-01 2.79752344e-01 -6.09800935e-01
5.36577463e-01 1.97345048e-01 5.20069778e-01 -2.90898055e-01
-2.69777905e-02 1.11352846e-01 -2.21227974e-01 -6.74014449e-01
6.64150953e-01 -9.89239141e-02 -3.53873312e-01 3.00108343e-01
3.14187407e-01 -7.37171054e-01 3.38566210e-03 1.70891383e-03
1.26681542e+00 9.57411826e-01 2.60868728e-01 -4.61381286e-01
3.65666777e-01 5.56050181e-01 5.10529280e-01 1.07204008e+00
-2.22201020e-01 -1.52286217e-01 3.98977190e-01 -3.59252095e-01
-1.01677239e+00 -1.23754442e+00 6.37843132e-01 1.55339456e+00
1.68569461e-01 -1.84301928e-01 -4.52740610e-01 -7.87811399e-01
8.93818811e-02 8.57051075e-01 -4.92874593e-01 -1.22551650e-01
-9.18053806e-01 2.55584180e-01 1.16865411e-01 6.37036264e-01
3.11944842e-01 -1.46573246e+00 -1.19477773e+00 3.83078754e-01
2.03308105e-01 -8.28846395e-01 -7.19731390e-01 4.29664433e-01
-1.10519719e+00 -1.13666081e+00 -6.31786585e-01 -8.73510599e-01
5.90205312e-01 2.01451212e-01 9.35321033e-01 -3.80660594e-01
2.75170773e-01 1.03667343e+00 -2.93590605e-01 -2.61015236e-01
-5.64419031e-01 -6.91209659e-02 1.89830944e-01 -4.75329310e-01
-5.58871686e-01 -4.74023342e-01 -5.13710022e-01 9.76584107e-02
-5.45797050e-01 3.11612904e-01 7.28324056e-01 9.22015488e-01
6.89640403e-01 -7.32690245e-02 4.36154991e-01 -4.80329156e-01
1.00542903e+00 -4.77605730e-01 -1.10441256e+00 2.69580781e-01
-5.06518066e-01 5.04470646e-01 8.02393913e-01 -9.00163651e-01
-1.39061737e+00 2.34355763e-01 3.78185898e-01 -4.74808216e-01
-6.75955042e-02 5.07597208e-01 2.65453547e-01 -9.33379605e-02
4.75374490e-01 2.77736694e-01 -6.79043755e-02 -2.11410984e-01
8.79495680e-01 1.18675403e-01 5.55136502e-01 -1.29651463e+00
8.48732889e-01 5.24948299e-01 1.99152336e-01 -2.61242062e-01
-1.03667808e+00 -3.68619561e-01 -5.25604010e-01 -2.27449104e-01
9.77734268e-01 -1.15958476e+00 -9.79521394e-01 4.19800848e-01
-8.59151840e-01 -1.58292150e+00 -6.90750718e-01 4.89470482e-01
-1.70397544e+00 3.05415392e-02 -6.47986174e-01 -8.02362084e-01
-1.15111090e-01 -1.25018537e+00 1.00654447e+00 2.07265139e-01
1.53779283e-01 -1.08253455e+00 3.57659668e-01 -2.50499189e-01
2.64765769e-01 3.11593175e-01 5.31054437e-01 -9.21388716e-02
-9.30774629e-01 5.01697838e-01 2.86366075e-01 6.62806407e-02
5.76781444e-02 -4.74072516e-01 -2.88511544e-01 -7.11399436e-01
-1.08481012e-01 -8.26290786e-01 7.15437949e-01 4.42683220e-01
1.13900125e+00 -5.49316287e-01 -3.08588594e-01 7.34715641e-01
1.14339697e+00 2.98711330e-01 1.39102280e-01 6.61128104e-01
4.63944882e-01 4.82643872e-01 1.33143926e+00 5.89922965e-01
3.98899227e-01 4.82492507e-01 5.59848666e-01 5.78844309e-01
2.49489844e-01 -6.80200934e-01 7.24794507e-01 3.14889282e-01
-4.05595720e-01 2.28664458e-01 -7.25008905e-01 6.07907355e-01
-2.33280635e+00 -9.82248843e-01 4.87082601e-01 1.89406431e+00
9.60369647e-01 2.39005119e-01 2.85730720e-01 -7.23195493e-01
1.06045023e-01 2.36242950e-01 -1.08251059e+00 -3.98238957e-01
4.74809855e-01 2.95026451e-01 6.99542284e-01 8.14955652e-01
-1.13440681e+00 1.45174742e+00 7.29753208e+00 6.93806589e-01
-8.49453211e-01 5.10508530e-02 -1.27527162e-01 -1.21199772e-01
-2.85332888e-01 9.26947966e-02 -7.58589685e-01 6.85360506e-02
1.13139045e+00 -5.49425364e-01 9.72011149e-01 1.08730364e+00
5.98603845e-01 -1.14876889e-01 -1.19768322e+00 3.67355347e-01
-6.65108383e-01 -1.28555429e+00 -2.31986582e-01 -7.26118535e-02
1.26617634e+00 1.36924326e-01 2.99156547e-01 9.14230704e-01
1.48695862e+00 -1.15964329e+00 7.89516330e-01 6.02520645e-01
8.24290574e-01 -8.29001248e-01 -1.08706221e-01 8.16857636e-01
-1.14076936e+00 -7.68206179e-01 -2.36791089e-01 -2.79147148e-01
4.18468833e-01 -3.92762691e-01 -7.93924093e-01 4.44015890e-01
2.32195631e-01 8.89479458e-01 2.49736711e-01 8.76081169e-01
-7.55660653e-01 4.16101784e-01 1.07109532e-01 -1.67287532e-02
9.60255027e-01 -3.29537898e-01 7.60027826e-01 7.98921525e-01
2.69165516e-01 3.25274527e-01 9.59416807e-01 7.56630123e-01
3.73615146e-01 -4.43688422e-01 -6.23042524e-01 -1.20432071e-01
3.13540131e-01 9.62655365e-01 -3.13460201e-01 -3.92969429e-01
-2.97072738e-01 1.01920414e+00 7.76717484e-01 7.07791746e-01
-8.54584098e-01 1.96599275e-01 1.00794339e+00 -2.93133348e-01
3.20747197e-01 -9.20372844e-01 7.61104748e-02 -1.13686693e+00
-4.96951431e-01 -1.04897058e+00 3.09681803e-01 -6.97552919e-01
-9.54543054e-01 3.48020941e-01 3.54522049e-01 -1.20391595e+00
-6.83226883e-01 -4.59238738e-01 -5.06682038e-01 7.34175324e-01
-1.85781121e+00 -1.14542520e+00 -1.31641567e-01 8.14865351e-01
1.12735927e+00 -4.27575380e-01 6.60587072e-01 -5.52650571e-01
-1.13050498e-01 -4.08029966e-02 1.58831671e-01 -5.15380085e-01
4.44073737e-01 -1.53970301e+00 3.81266475e-01 4.94797379e-01
-4.95556563e-01 5.21247387e-01 8.35298717e-01 -9.17415500e-01
-1.63140333e+00 -1.34769464e+00 -1.50315791e-01 -4.59616989e-01
1.04932046e+00 1.34549178e-02 -5.54205835e-01 1.26779222e+00
1.88202247e-01 -1.14537515e-01 -2.04069108e-01 -1.46203071e-01
-5.16375639e-02 1.95984527e-01 -9.61957395e-01 9.36480463e-01
1.30535781e+00 -3.40001941e-01 -1.00449705e+00 4.36860710e-01
1.18489206e+00 -9.53976333e-01 -7.50590265e-01 2.94279724e-01
2.63212025e-01 -3.35611343e-01 9.98484313e-01 -1.12436700e+00
2.14852512e-01 -4.55352925e-02 2.18225285e-01 -1.91218269e+00
-5.69900572e-01 -1.24750650e+00 -5.82127273e-01 1.48233607e-01
5.71047738e-02 -6.69000268e-01 6.97748601e-01 2.40785852e-01
-6.20940506e-01 -7.77514637e-01 -7.79001772e-01 -1.23515713e+00
3.10764343e-01 -1.03509665e-01 3.35088730e-01 2.91025162e-01
1.37616888e-01 -1.37826070e-01 -8.11821997e-01 3.82238269e-01
7.00287640e-01 2.36177042e-01 7.24575698e-01 -8.09366882e-01
-5.32063663e-01 -1.29044458e-01 5.06098807e-01 -1.64768195e+00
7.72617579e-01 -6.01700783e-01 5.19389808e-01 -1.52318323e+00
-2.15509813e-02 -6.57752395e-01 -3.29180509e-01 4.98530954e-01
-1.44030347e-01 -9.07434702e-01 3.78267348e-01 2.48590961e-01
-1.09894431e+00 8.64247620e-01 1.98420668e+00 1.16032772e-01
-5.74770391e-01 3.10197711e-01 -3.23293626e-01 6.38812780e-01
9.48473096e-01 -3.17843705e-01 -8.62578154e-01 -4.35509622e-01
4.72562946e-02 6.98922098e-01 1.46509126e-01 -8.77321601e-01
1.38908163e-01 -1.30306685e+00 -3.32822740e-01 -2.79354364e-01
5.52198231e-01 -7.97480524e-01 -3.50876927e-01 9.07656372e-01
-6.70059800e-01 8.87859687e-02 1.48078755e-01 1.00883317e+00
1.41653121e-01 -1.59796979e-02 6.44274414e-01 -5.09930193e-01
-1.18550074e+00 6.44884467e-01 -7.89617658e-01 2.04081535e-01
1.49526906e+00 2.72641063e-01 -2.77207494e-01 -4.88519460e-01
-9.14322317e-01 9.96746600e-01 3.94010156e-01 3.94082159e-01
6.71194434e-01 -1.01369655e+00 -4.61223930e-01 -2.91768253e-01
-1.44677728e-01 1.16876043e-01 -3.28376330e-02 4.52008665e-01
-1.91917837e-01 5.12672186e-01 -7.00475633e-01 -2.89584935e-01
-6.88101888e-01 7.19233274e-01 4.78065044e-01 -7.93627918e-01
-8.86008382e-01 6.08940005e-01 -6.26319870e-02 -6.70563579e-01
4.07693595e-01 -6.46863580e-01 3.93107347e-02 -5.23589969e-01
3.09972435e-01 5.49572229e-01 -8.45533907e-01 1.25316843e-01
1.02934204e-01 3.20035756e-01 2.93061614e-01 -5.90806305e-01
1.33314109e+00 -1.26997173e-01 3.27461839e-01 6.06244683e-01
4.15400505e-01 -4.61415231e-01 -2.61143136e+00 -2.63350934e-01
2.45854065e-01 -4.19591218e-01 -2.42387548e-01 -7.30286896e-01
-2.71451831e-01 3.91444147e-01 1.95461120e-02 -3.85992229e-01
5.96048892e-01 -1.25883818e-01 6.54873133e-01 8.55982184e-01
1.14355600e+00 -1.29202962e+00 8.91806304e-01 1.18028116e+00
1.00159669e+00 -1.07524848e+00 -1.53142825e-01 -1.51443789e-02
-1.12872434e+00 1.03226388e+00 8.07374597e-01 -6.34717822e-01
2.72890449e-01 2.95086950e-01 -1.39806762e-01 -2.66518425e-02
-1.28974915e+00 -7.08886921e-01 1.66963920e-01 9.62908447e-01
-3.98985803e-01 1.67757168e-01 -2.34804124e-01 1.99948400e-01
1.61215916e-01 2.44802311e-01 5.26000679e-01 1.33675897e+00
-1.06003702e+00 -1.11204898e+00 -1.07872404e-01 1.76437989e-01
3.78826149e-02 3.43711495e-01 2.87964761e-01 6.49246454e-01
-1.79306507e-01 7.61509597e-01 -1.25781342e-01 -4.68731076e-02
1.96766719e-01 -1.30980080e-02 9.28731918e-01 -9.84258115e-01
-4.98149961e-01 -1.26058668e-01 1.94922730e-01 -1.23282409e+00
-3.19702476e-01 -6.31333590e-01 -1.60306036e+00 7.01202899e-02
5.33656836e-01 6.85243756e-02 3.10990930e-01 1.03643775e+00
2.38311335e-01 7.70949781e-01 6.04092419e-01 -1.21888328e+00
-1.37084758e+00 -7.39155829e-01 -4.61283356e-01 3.52729321e-01
6.77736640e-01 -9.15698230e-01 -6.38211444e-02 -1.54533714e-01] | [4.186468124389648, 1.55370032787323] |
5f35282f-c0fa-422b-9209-2d1d5150fc78 | context-matters-recovering-human-semantic | 1910.06954 | null | https://arxiv.org/abs/1910.06954v3 | https://arxiv.org/pdf/1910.06954v3.pdf | Context Matters: Recovering Human Semantic Structure from Machine Learning Analysis of Large-Scale Text Corpora | Applying machine learning algorithms to large-scale, text-based corpora (embeddings) presents a unique opportunity to investigate at scale how human semantic knowledge is organized and how people use it to judge fundamental relationships, such as similarity between concepts. However, efforts to date have shown a substantial discrepancy between algorithm predictions and empirical judgments. Here, we introduce a novel approach of generating embeddings motivated by the psychological theory that semantic context plays a critical role in human judgments. Specifically, we train state-of-the-art machine learning algorithms using contextually-constrained text corpora and show that this greatly improves predictions of similarity judgments and feature ratings. By improving the correspondence between representations derived using embeddings generated by machine learning methods and empirical measurements of human judgments, the approach we describe helps advance the use of large-scale text corpora to understand the structure of human semantic representations. | ['Cameron T. Ellis', 'Marius Cătălin Iordan', 'Tyler Giallanza', 'Nicole M. Beckage', 'Jonathan D. Cohen'] | 2019-10-15 | null | null | null | null | ['empirical-judgments'] | ['natural-language-processing'] | [ 2.33432800e-01 6.78422600e-02 -9.54270065e-02 -7.13067412e-01
-4.08067405e-01 -5.62946379e-01 9.11549211e-01 9.22575295e-01
-8.41173589e-01 2.00415537e-01 8.53892267e-01 -3.15487504e-01
-1.12602189e-01 -8.55403602e-01 -8.54506567e-02 2.87596192e-02
3.13851535e-01 7.91306078e-01 1.60115957e-01 -3.28724533e-01
7.82599390e-01 1.61283970e-01 -1.69771266e+00 2.98667043e-01
8.07194412e-01 6.20963871e-01 1.21203170e-03 4.21743244e-01
-3.97224337e-01 5.15974939e-01 -2.31904775e-01 -5.50120831e-01
-1.08134888e-01 -3.97753984e-01 -1.01557505e+00 -2.62570113e-01
5.20029426e-01 -1.59554228e-01 -1.89139605e-01 8.20757806e-01
2.62114048e-01 4.07791764e-01 1.02685928e+00 -9.22934473e-01
-1.32649696e+00 3.13186258e-01 3.94341424e-02 4.63005245e-01
4.71115798e-01 -2.10182771e-01 1.53899443e+00 -7.96047509e-01
5.88434041e-01 1.36240733e+00 7.03302205e-01 5.94274282e-01
-1.44285524e+00 -2.21742421e-01 3.61570120e-02 4.19926822e-01
-1.01755548e+00 -2.69260675e-01 6.89390540e-01 -7.96508789e-01
1.16738582e+00 -1.26126379e-01 5.25637567e-01 9.92908597e-01
-1.78277224e-01 3.36552292e-01 1.16861260e+00 -8.88714492e-01
3.65407646e-01 3.75439674e-01 4.89079505e-01 5.51526845e-01
4.76228952e-01 1.85106277e-01 -5.32723963e-01 -3.85246366e-01
6.78364933e-01 -3.63114290e-03 7.56810904e-02 -3.49786162e-01
-1.29185545e+00 1.31869686e+00 5.03764391e-01 6.73934877e-01
-2.49071032e-01 4.64920588e-02 5.31721473e-01 1.33067816e-01
5.18944561e-01 1.07205820e+00 -5.13800740e-01 -2.87219495e-01
-5.30314386e-01 3.40838909e-01 7.10572541e-01 4.03469145e-01
8.63883972e-01 -5.12398243e-01 2.86677897e-01 1.13177431e+00
3.19028705e-01 1.71534166e-01 1.11934638e+00 -9.63958621e-01
1.54797345e-01 6.27367496e-01 2.49438614e-01 -1.36931586e+00
-3.42667162e-01 8.25147629e-02 1.61428720e-01 -7.56402034e-03
5.77854753e-01 3.67382169e-02 -3.84584129e-01 1.88557196e+00
2.19883770e-01 -3.91035289e-01 4.70746532e-02 8.83364320e-01
3.35678458e-01 3.14678878e-01 5.28120697e-01 2.36404508e-01
1.51882362e+00 -5.93314528e-01 -2.91544795e-01 -6.08453214e-01
9.97513533e-01 -6.83139086e-01 1.29761481e+00 -8.66870061e-02
-7.51893878e-01 -7.77470648e-01 -1.04633951e+00 -4.16762054e-01
-7.77267277e-01 -1.52788445e-01 1.00283754e+00 7.18828082e-01
-8.29750359e-01 6.46142960e-01 -6.61234260e-01 -1.03075564e+00
3.47934663e-01 -5.70096076e-02 -2.89623678e-01 -1.74383044e-01
-1.23428774e+00 1.43435252e+00 3.22372496e-01 -3.81853402e-01
-2.74529904e-01 -5.56267321e-01 -9.46249366e-01 1.25932351e-01
1.12279132e-01 -6.43994212e-01 1.25654137e+00 -8.79376113e-01
-8.77146542e-01 1.27391255e+00 -3.20170760e-01 -2.82834053e-01
-1.00209042e-01 -2.84666359e-01 -4.10119355e-01 2.23015726e-01
2.79772669e-01 6.92261815e-01 4.31802094e-01 -9.04599369e-01
-4.86639053e-01 -6.77117050e-01 5.94298579e-02 2.87362218e-01
-8.28232646e-01 2.31670499e-01 3.78862232e-01 -5.38892031e-01
3.36229764e-02 -8.62873733e-01 -2.26976231e-01 2.76827216e-01
3.15483809e-01 -6.67732954e-01 1.89251840e-01 -5.05441487e-01
9.86324191e-01 -2.16175461e+00 -1.24045409e-01 1.17754720e-01
3.09539288e-01 2.21974835e-01 -2.72294313e-01 8.03917050e-01
1.01829469e-01 2.22409770e-01 3.89771983e-02 -2.76386421e-02
4.92092639e-01 1.28469393e-01 -3.50735933e-01 6.79828301e-02
-4.71143425e-02 1.00568616e+00 -1.14309919e+00 -4.20847684e-01
2.41127685e-01 2.35351771e-01 -8.09617877e-01 3.44315439e-01
-1.06170237e-01 -1.99345440e-01 -6.75510645e-01 1.53889060e-01
-7.41953030e-03 -3.54010642e-01 5.41888237e-01 -9.55134332e-02
8.96232948e-02 6.45768762e-01 -5.50703406e-01 1.68824601e+00
-6.42847061e-01 1.00846767e+00 -5.58122814e-01 -1.17752743e+00
9.93874431e-01 7.40931332e-02 -4.91698533e-02 -7.73568988e-01
2.11519763e-01 -4.26977221e-03 2.46074274e-01 -6.00075305e-01
6.12331569e-01 -5.38930297e-01 -1.87534988e-01 8.72653365e-01
3.26998718e-02 -2.25561261e-01 1.01023920e-01 2.11320922e-01
8.70058298e-01 -7.35429004e-02 5.97146630e-01 -4.78827387e-01
1.49182722e-01 3.84942055e-01 2.79415816e-01 5.70889890e-01
-3.91433388e-01 3.28436106e-01 2.73640215e-01 -6.72943413e-01
-1.43558228e+00 -1.25739157e+00 -3.24093372e-01 1.57556355e+00
-4.54710275e-02 -6.27117574e-01 -7.94473171e-01 -4.24757510e-01
2.85055846e-01 1.13332450e+00 -1.09640670e+00 -3.50511879e-01
-2.33185112e-01 -5.20974398e-01 3.01753849e-01 9.76949871e-01
6.81845620e-02 -9.24621582e-01 -6.44196332e-01 2.80587286e-01
-3.71050555e-03 -1.15469682e+00 -2.67034292e-01 -1.74758270e-01
-9.28853035e-01 -1.20381331e+00 -5.53329512e-02 -8.25173616e-01
6.67173684e-01 2.65380949e-01 1.34899342e+00 2.90480942e-01
-6.29593074e-01 6.23528123e-01 -5.64533710e-01 -3.65985721e-01
-4.06873912e-01 -1.90506250e-01 2.35304803e-01 -5.06986797e-01
1.28353560e+00 -5.81866086e-01 -4.36002970e-01 4.14129019e-01
-8.66433740e-01 -3.35496306e-01 3.25663388e-01 8.63060296e-01
-4.91288537e-03 -1.90125570e-01 7.19058156e-01 -9.65363979e-01
1.11227548e+00 -4.80604649e-01 -1.28555641e-01 2.53482312e-01
-1.03451848e+00 3.80694568e-01 4.75289553e-01 -3.41665506e-01
-1.07382858e+00 -4.76198882e-01 1.18988797e-01 1.77989170e-01
-4.55931216e-01 3.13898832e-01 3.72511238e-01 3.07814062e-01
1.10894537e+00 -1.05997309e-01 1.02115653e-01 -4.00542676e-01
7.45434940e-01 8.49985182e-01 3.13351214e-01 -1.08545792e+00
5.35961747e-01 3.23578179e-01 -5.39932907e-01 -7.66128182e-01
-1.27115607e+00 -5.16197920e-01 -8.64377737e-01 3.40782285e-01
1.16495633e+00 -7.68329024e-01 -4.83844340e-01 -1.83443680e-01
-9.12031829e-01 -1.24105580e-01 -2.61652559e-01 6.63481593e-01
-5.91194630e-01 1.27420500e-01 -6.09604180e-01 -3.60017270e-01
-3.67146321e-02 -5.55677533e-01 8.08253944e-01 1.22353725e-01
-1.09081149e+00 -1.50508976e+00 5.60396671e-01 7.48688936e-01
5.90226591e-01 -1.70751572e-01 1.34184623e+00 -1.33858180e+00
2.84878779e-02 -8.91648829e-02 -4.07534480e-01 3.16078603e-01
2.06361547e-01 -2.44322076e-01 -1.06904602e+00 6.72364756e-02
-1.30546167e-01 -8.04362953e-01 6.93522573e-01 -2.40507978e-03
8.45883369e-01 1.57400325e-01 -4.13944840e-01 1.36525612e-02
1.43015981e+00 -6.01473600e-02 5.37410855e-01 3.98758441e-01
4.13185120e-01 8.29431534e-01 5.61613917e-01 2.58471698e-01
5.78942358e-01 4.40325201e-01 -4.75838602e-01 4.57717299e-01
1.06275223e-01 -7.15203166e-01 -7.86959827e-02 7.76486993e-01
6.56032786e-02 1.53838903e-01 -9.32711542e-01 7.98720419e-01
-1.69994044e+00 -1.04889989e+00 3.18489581e-01 2.19806147e+00
8.45798612e-01 2.14602038e-01 -7.19890445e-02 -3.19502950e-02
7.35642552e-01 2.58297056e-01 -3.90007108e-01 -8.14837456e-01
4.91091102e-01 4.25415635e-01 -4.64187339e-02 5.28761208e-01
-7.50196874e-01 1.09666717e+00 7.28399515e+00 3.62576485e-01
-6.01825058e-01 1.43800259e-01 3.01473737e-01 1.85540095e-01
-4.45152551e-01 3.20668891e-02 -5.65760434e-01 2.06159979e-01
1.11498737e+00 -5.10133445e-01 4.20939326e-01 8.91544640e-01
-4.67571244e-02 -4.86641228e-02 -1.63627875e+00 7.78553963e-01
4.39266205e-01 -1.02664220e+00 1.38873339e-01 -2.35049073e-02
6.07692003e-01 -1.69608995e-01 -2.67513171e-02 3.55088770e-01
5.00891089e-01 -1.03272438e+00 3.09421688e-01 2.43551776e-01
4.08521712e-01 -3.88182700e-01 6.08553946e-01 1.89060599e-01
-8.10417712e-01 -1.91173196e-01 -9.73116875e-01 -6.27182126e-01
-1.36969925e-03 4.74253178e-01 -1.01158857e+00 -2.36384496e-01
3.10243160e-01 6.77052736e-01 -7.57317126e-01 3.73005152e-01
-5.01182675e-01 3.66399497e-01 2.45008413e-02 -4.41015124e-01
5.59223071e-02 -8.09071958e-02 -6.48132265e-02 1.03092980e+00
5.92964403e-02 2.02394873e-01 2.00107601e-02 1.11189437e+00
3.38381976e-02 2.95928001e-01 -7.70609617e-01 -5.90981185e-01
7.26565719e-01 1.10207927e+00 -6.39354765e-01 -4.43202525e-01
-6.38659835e-01 8.22315514e-01 7.75677741e-01 1.53603092e-01
-3.71795326e-01 -2.91361541e-01 1.08578038e+00 1.12359583e-01
2.38545746e-01 -5.60960948e-01 -3.91674101e-01 -1.19018090e+00
-1.69051155e-01 -5.80945909e-01 3.18082243e-01 -1.04095113e+00
-1.80808246e+00 3.79124820e-01 -9.09711197e-02 -5.76661289e-01
-5.25972962e-01 -9.94005740e-01 -6.49164021e-01 7.60448337e-01
-1.20586741e+00 -6.16600275e-01 -2.60347247e-01 1.89113498e-01
5.02369463e-01 -5.25685363e-02 1.19306922e+00 -2.86487728e-01
2.00835224e-02 3.59554708e-01 1.73565894e-01 3.24914426e-01
9.65129972e-01 -1.29057181e+00 7.00642645e-01 2.25811765e-01
5.61357200e-01 1.04344404e+00 6.42836154e-01 -4.35746491e-01
-8.31996977e-01 -4.90503162e-01 1.21974134e+00 -1.06087613e+00
1.00707412e+00 -4.08565581e-01 -8.57120275e-01 7.46606469e-01
1.87432796e-01 -3.67756039e-02 1.49691987e+00 8.59754145e-01
-9.50008154e-01 1.87721118e-01 -1.16386414e+00 4.37336296e-01
1.14570451e+00 -1.19534469e+00 -1.66057181e+00 2.73810357e-01
6.94748998e-01 3.91235799e-01 -9.65885460e-01 -7.69383684e-02
7.96824634e-01 -9.39564526e-01 1.08948088e+00 -1.09513509e+00
8.43680501e-01 2.26616576e-01 -5.26218593e-01 -1.78399122e+00
-5.75900733e-01 1.00895368e-01 4.32479233e-01 9.65148389e-01
5.64365864e-01 -8.16808701e-01 7.14943945e-01 1.14996743e+00
8.73090103e-02 -4.73857820e-01 -4.02338266e-01 -5.06245792e-01
5.32113314e-01 -2.56107479e-01 4.38742965e-01 1.23175752e+00
6.31971717e-01 4.90110427e-01 4.31026310e-01 -2.09597439e-01
5.77977359e-01 1.53775871e-01 4.58224297e-01 -1.65232456e+00
-2.24357873e-01 -2.15667769e-01 -8.37785959e-01 -8.99505913e-01
4.01917845e-01 -9.27444994e-01 -1.24340832e-01 -1.48344803e+00
4.15678054e-01 -4.40929085e-01 -3.98850262e-01 1.28051057e-01
-3.34332466e-01 3.07177175e-02 2.20101565e-01 3.60828266e-02
-5.01764655e-01 4.50232387e-01 9.24322128e-01 3.09471220e-01
2.62152761e-01 -6.84665978e-01 -1.15494084e+00 9.02860045e-01
7.68927276e-01 -4.29227859e-01 -6.00637436e-01 -6.69922054e-01
3.24405700e-01 -5.38692951e-01 3.20524752e-01 -9.08528030e-01
1.30033970e-01 -2.04441205e-01 6.80500269e-01 1.79559961e-01
2.52050310e-01 -5.59724450e-01 -6.13480687e-01 1.84387460e-01
-9.87015486e-01 2.60106951e-01 3.31317969e-02 6.61335051e-01
-1.77437961e-01 -4.61687952e-01 5.70796371e-01 -3.25197995e-01
-9.51232255e-01 -1.69038892e-01 -3.25595826e-01 7.38453150e-01
7.23820508e-01 -2.67731715e-02 -5.72098136e-01 -2.58332759e-01
-7.16662705e-01 -1.76162332e-01 7.84015417e-01 5.98408222e-01
5.08301437e-01 -1.37647951e+00 -5.68038940e-01 4.21457998e-02
5.51223814e-01 -7.24331498e-01 -7.90803432e-02 8.79111812e-02
-2.77576357e-01 5.68621695e-01 -2.47681290e-01 -1.25046760e-01
-9.89293337e-01 6.96363986e-01 -1.04375295e-02 7.09981546e-02
-3.60893875e-01 8.59993875e-01 2.30902702e-01 -5.71691334e-01
-4.06820714e-01 -2.64212698e-01 -1.36959955e-01 1.42277971e-01
7.16041684e-01 2.68493593e-01 -2.57998526e-01 -3.77661347e-01
-1.68374047e-01 6.28328562e-01 -2.27112606e-01 -2.34959885e-01
1.30348015e+00 -7.67252892e-02 6.42769262e-02 5.99496365e-01
1.42913151e+00 -6.28728047e-02 -8.56679499e-01 -4.58126396e-01
2.32457548e-01 -8.67489159e-01 -1.54026493e-01 -7.10952997e-01
-1.76102161e-01 1.14191782e+00 2.88717538e-01 2.66368717e-01
4.70534474e-01 2.76919246e-01 8.90451491e-01 7.46847987e-01
4.57966894e-01 -1.45820415e+00 4.06153202e-01 2.67193913e-01
4.82756943e-01 -1.35362482e+00 4.82088216e-02 -2.14949369e-01
-6.39908135e-01 1.09418786e+00 5.64956844e-01 -4.06205177e-01
6.84851348e-01 -2.87039340e-01 5.20219579e-02 -2.27429017e-01
-9.39140797e-01 -1.89121962e-01 2.29168728e-01 7.91384876e-01
8.27872574e-01 2.04145312e-01 -3.42838764e-01 4.35445368e-01
-5.15162408e-01 -1.04577374e-02 4.06110972e-01 7.84543395e-01
-8.84452522e-01 -1.22102952e+00 -1.74819231e-01 5.05041361e-01
-2.44565085e-01 -4.69735444e-01 -5.05544841e-01 6.63217843e-01
-1.35594711e-01 1.04142022e+00 4.09899235e-01 -2.27573663e-01
9.54226404e-02 5.08533299e-01 7.08447754e-01 -9.54707146e-01
-1.87011287e-01 -6.79007649e-01 4.44374174e-01 -5.14204204e-01
-3.66063446e-01 -8.34126711e-01 -1.27932763e+00 -1.81829110e-01
-5.67993782e-02 4.07893449e-01 7.56680131e-01 1.19871175e+00
3.66633147e-01 2.84555610e-02 2.71347046e-01 -5.34755409e-01
-9.62772667e-01 -9.63546216e-01 -5.31331837e-01 1.03588557e+00
-1.92889005e-01 -7.69472539e-01 -5.74270129e-01 -4.28348780e-03] | [10.272279739379883, 8.737728118896484] |
a6e55528-c81b-4f9d-9b4e-ff25767966db | reading-text-in-the-wild-with-convolutional | 1412.1842 | null | http://arxiv.org/abs/1412.1842v1 | http://arxiv.org/pdf/1412.1842v1.pdf | Reading Text in the Wild with Convolutional Neural Networks | In this work we present an end-to-end system for text spotting -- localising
and recognising text in natural scene images -- and text based image retrieval.
This system is based on a region proposal mechanism for detection and deep
convolutional neural networks for recognition. Our pipeline uses a novel
combination of complementary proposal generation techniques to ensure high
recall, and a fast subsequent filtering stage for improving precision. For the
recognition and ranking of proposals, we train very large convolutional neural
networks to perform word recognition on the whole proposal region at the same
time, departing from the character classifier based systems of the past. These
networks are trained solely on data produced by a synthetic text generation
engine, requiring no human labelled data.
Analysing the stages of our pipeline, we show state-of-the-art performance
throughout. We perform rigorous experiments across a number of standard
end-to-end text spotting benchmarks and text-based image retrieval datasets,
showing a large improvement over all previous methods. Finally, we demonstrate
a real-world application of our text spotting system to allow thousands of
hours of news footage to be instantly searchable via a text query. | ['Andrew Zisserman', 'Karen Simonyan', 'Andrea Vedaldi', 'Max Jaderberg'] | 2014-12-04 | null | null | null | null | ['text-spotting'] | ['computer-vision'] | [ 8.49046052e-01 -2.12288067e-01 2.31494054e-01 -3.65203410e-01
-1.11731267e+00 -6.07364357e-01 1.18823564e+00 2.89429873e-01
-8.30233514e-01 1.16613291e-01 4.70382906e-02 -3.41368139e-01
2.53316462e-01 -6.35894358e-01 -6.32946610e-01 -3.32826436e-01
4.57802981e-01 9.45177138e-01 6.07967257e-01 -3.30995053e-01
8.47467124e-01 2.80472964e-01 -1.58150804e+00 9.85986948e-01
1.88323349e-01 1.00354290e+00 2.13343009e-01 1.27727270e+00
-3.88924807e-01 6.26282334e-01 -8.02392662e-01 -3.59509706e-01
3.37390542e-01 -2.51946300e-01 -7.81470180e-01 1.46105289e-01
1.05912161e+00 -5.56440055e-01 -4.46479917e-01 5.59205949e-01
7.69494593e-01 -1.25276610e-01 7.08574653e-01 -6.19301975e-01
-4.32674170e-01 4.22303081e-01 -4.63206887e-01 1.30780324e-01
5.88620424e-01 2.75514990e-01 1.14692700e+00 -1.34225023e+00
7.24471867e-01 1.23363948e+00 6.09401584e-01 3.67238760e-01
-1.05035067e+00 -3.05287153e-01 -2.78484046e-01 -1.58747941e-01
-1.33470213e+00 -8.32650006e-01 1.47759825e-01 -4.24968153e-01
1.44317520e+00 4.21549946e-01 3.35860997e-01 1.00577188e+00
2.40284711e-01 1.20554590e+00 6.10490561e-01 -9.84135926e-01
-5.22781871e-02 -8.15894082e-03 -2.38308441e-02 8.82578671e-01
-6.43017888e-02 -1.01808205e-01 -6.24609113e-01 -8.25848207e-02
5.81764042e-01 -4.14024405e-02 -3.97710018e-02 -1.86952353e-01
-1.66097498e+00 6.25798881e-01 2.07700178e-01 1.65862024e-01
-9.69449505e-02 3.82255703e-01 7.73946285e-01 3.98469687e-01
5.83871901e-01 4.47269350e-01 -1.67479753e-01 -2.60576084e-02
-1.75282180e+00 3.09348285e-01 9.10984874e-01 8.34173083e-01
4.48940337e-01 -3.57816577e-01 -7.94686556e-01 9.73017037e-01
2.42679045e-01 5.43632567e-01 7.24641860e-01 -1.99708283e-01
6.10142350e-01 6.63012564e-01 5.42903207e-02 -7.64170468e-01
-1.65192842e-01 -1.63240984e-01 -5.57297945e-01 2.74137706e-01
3.60849679e-01 1.52713701e-01 -1.50895309e+00 7.99649596e-01
-9.49597955e-02 -2.57480472e-01 -2.56736755e-01 8.35949361e-01
6.22511148e-01 8.06967080e-01 -4.05441016e-01 4.10261422e-01
1.50474381e+00 -1.21604645e+00 -2.91389465e-01 -3.81614029e-01
8.66228521e-01 -1.24680305e+00 1.13505054e+00 4.75523621e-01
-1.00320888e+00 -3.35772097e-01 -9.73443866e-01 -3.54196936e-01
-5.97322941e-01 5.45486152e-01 -3.83765996e-02 6.02357209e-01
-1.33270991e+00 3.90821993e-01 -5.53575456e-01 -8.41454685e-01
3.98053199e-01 2.50701219e-01 -2.04914927e-01 -1.07062474e-01
-7.81142771e-01 7.20648348e-01 5.46390593e-01 -3.43040489e-02
-7.52526164e-01 -2.15585157e-01 -5.44403672e-01 1.03907347e-01
2.35118225e-01 -5.41160643e-01 1.61654782e+00 -1.14246070e+00
-1.38942647e+00 1.22101331e+00 -1.14292391e-01 -6.21093094e-01
8.70823145e-01 -3.12255234e-01 -1.90760911e-01 5.53163469e-01
1.25991032e-01 1.06015825e+00 1.42717862e+00 -7.53384292e-01
-8.26011181e-01 9.56221223e-02 -4.52870935e-01 2.18270153e-01
-2.53514498e-01 4.33010221e-01 -8.14893246e-01 -8.40881467e-01
-1.38295323e-01 -5.34492314e-01 3.59060578e-02 3.87814313e-01
-5.38072526e-01 -3.03556144e-01 1.08839893e+00 -5.74648201e-01
9.86249983e-01 -1.95284438e+00 -2.88486570e-01 1.24723814e-01
2.87701726e-01 3.99913222e-01 -4.04219270e-01 7.17749953e-01
4.96427491e-02 1.87597781e-01 -2.54458606e-01 -5.75057447e-01
2.72090971e-01 -1.82885006e-01 -7.49543488e-01 4.34868574e-01
5.17904401e-01 1.24999011e+00 -7.01128781e-01 -6.63211763e-01
4.30025667e-01 3.19286436e-01 -1.87920719e-01 1.19141065e-01
-6.10983431e-01 -4.88280296e-01 -2.09375769e-01 6.34040236e-01
2.11014733e-01 -3.13032240e-01 -2.52932400e-01 8.45651552e-02
4.00273576e-02 3.81013542e-01 -9.08047497e-01 1.71396458e+00
-3.27547640e-01 1.35776162e+00 -8.80907997e-02 -8.67291808e-01
9.03672397e-01 2.11568519e-01 1.02977097e-01 -1.03188825e+00
1.61898464e-01 3.23413014e-01 -4.22159076e-01 -2.56644279e-01
1.17169321e+00 4.17756885e-01 -9.01269093e-02 1.03090012e+00
-1.18848562e-01 -2.40745276e-01 3.81109685e-01 5.22041321e-01
1.29517949e+00 1.47687539e-01 -5.12193218e-02 -1.69940725e-01
3.13606024e-01 2.27723047e-01 -5.05703866e-01 1.29374886e+00
5.62366955e-02 1.01799583e+00 3.75540525e-01 -8.87623608e-01
-1.54415011e+00 -5.54180324e-01 -6.77362606e-02 1.53278434e+00
-1.97041422e-01 -5.07747471e-01 -7.49247670e-01 -7.36237645e-01
-1.53171957e-01 6.23569250e-01 -7.44878292e-01 1.48500949e-01
-6.13433421e-01 -6.10926032e-01 1.06372702e+00 2.80806035e-01
4.45515692e-01 -1.24295568e+00 -9.86992717e-01 1.54625788e-01
1.02684528e-01 -9.68491018e-01 -6.46754026e-01 2.50187665e-01
-4.95832324e-01 -7.86856413e-01 -1.21960163e+00 -9.49704707e-01
6.86621249e-01 2.64315188e-01 1.31757140e+00 3.94591212e-01
-7.93209016e-01 5.04628479e-01 -5.00369728e-01 -2.80094922e-01
-7.23952353e-01 2.64548063e-01 -4.89849180e-01 -3.56348306e-02
3.18088055e-01 3.13174069e-01 -6.98629141e-01 4.44965124e-01
-1.32631850e+00 2.78486669e-01 8.86416197e-01 1.02696300e+00
2.20086455e-01 -3.36623758e-01 9.74224135e-02 -4.81468320e-01
9.14205313e-01 4.75845598e-02 -7.13873684e-01 5.05606532e-01
-5.30322909e-01 1.78344965e-01 3.30076545e-01 -4.00524199e-01
-6.18823946e-01 3.70477915e-01 -7.18522295e-02 -1.52862564e-01
-2.58575112e-01 3.42129171e-01 6.17750585e-01 -2.23960839e-02
1.13296747e+00 6.99451566e-01 -2.03558896e-02 -2.87735581e-01
5.19100428e-01 1.18737805e+00 4.32910174e-01 -3.46656471e-01
6.29947722e-01 4.83932108e-01 -4.85566497e-01 -9.22264934e-01
-5.70066214e-01 -7.53738165e-01 -6.95555151e-01 -1.36563331e-01
6.38958395e-01 -7.78666437e-01 -3.20895255e-01 6.36442780e-01
-1.28145778e+00 -6.30133331e-01 -1.42185345e-01 -4.22845073e-02
-3.77365589e-01 5.83493412e-01 -5.74460030e-01 -6.30769610e-01
-9.41497922e-01 -1.05351877e+00 1.93531084e+00 -1.52146444e-01
1.02035552e-02 -6.48676693e-01 1.44379601e-01 1.38269812e-01
5.91739237e-01 -3.11824143e-01 4.40873265e-01 -9.99934554e-01
-5.37630856e-01 -6.89020038e-01 -7.38450289e-01 -8.13200968e-05
-4.23951030e-01 1.86157361e-01 -9.44391131e-01 -4.20924902e-01
-6.54751658e-01 -6.53409660e-01 1.47012174e+00 -2.03283057e-02
8.64327848e-01 -2.63208717e-01 -4.13391680e-01 2.45248124e-01
1.29831183e+00 -2.65101492e-01 7.98268259e-01 5.84474504e-01
5.47319770e-01 4.73461986e-01 3.93507481e-01 3.74994248e-01
-1.82209864e-01 6.77137315e-01 2.26945952e-01 -2.98373818e-01
-1.34399861e-01 -5.15731331e-03 3.13999474e-01 2.50900120e-01
7.06792593e-01 -8.59672248e-01 -1.28305018e+00 7.62705743e-01
-1.99607491e+00 -8.14468265e-01 -1.00858837e-01 2.11555243e+00
7.16465056e-01 3.55096608e-01 2.39889279e-01 3.92297991e-02
9.02437508e-01 1.80194914e-01 -3.18478823e-01 -5.26513994e-01
-5.56565002e-02 4.21371400e-01 7.27897167e-01 2.87145317e-01
-1.24593616e+00 1.29348564e+00 6.80391026e+00 1.05272996e+00
-1.28216994e+00 -2.97855526e-01 6.98644698e-01 -2.08070546e-01
1.43274099e-01 -1.28838509e-01 -7.35838294e-01 1.09256193e-01
9.47386563e-01 1.20954923e-01 2.53285617e-01 6.97596610e-01
2.42814440e-02 -1.54398963e-01 -1.10566294e+00 1.02263486e+00
4.26935762e-01 -1.66467237e+00 4.26199228e-01 -1.80328920e-01
4.92310822e-01 5.09730399e-01 -9.51425731e-02 1.11138031e-01
2.71653742e-01 -1.20439506e+00 1.02236021e+00 4.72633153e-01
1.19852149e+00 -4.17599082e-01 5.14214754e-01 3.15193623e-01
-9.62678075e-01 1.54238582e-01 -3.86174053e-01 2.07679927e-01
-7.93084428e-02 6.41996026e-01 -1.45718801e+00 1.93701655e-01
4.26132411e-01 5.59091926e-01 -9.37860966e-01 1.15602338e+00
-4.19346392e-02 5.11831701e-01 -6.02588713e-01 -5.29028952e-01
6.13214493e-01 5.44097364e-01 4.33602333e-01 2.05931401e+00
1.35932118e-01 -4.72555131e-01 8.28029588e-02 8.73178482e-01
-2.79712558e-01 2.95188904e-01 -5.19202471e-01 -2.32911319e-01
1.37503833e-01 1.39107454e+00 -1.29638863e+00 -7.16416776e-01
-4.86983471e-02 1.43517470e+00 5.95380887e-02 1.22231349e-01
-5.37472308e-01 -1.06405735e+00 -2.21276149e-01 6.30821139e-02
5.67622840e-01 -2.89408237e-01 -1.00987606e-01 -1.12414634e+00
1.86623439e-01 -1.03663254e+00 1.65099546e-01 -1.07360876e+00
-9.85006988e-01 5.16907513e-01 -3.39221984e-01 -8.94151092e-01
-3.67919087e-01 -8.09456825e-01 -6.72179878e-01 9.59379792e-01
-1.27876186e+00 -1.27703047e+00 -1.62387162e-01 3.89953017e-01
1.07203233e+00 -2.10442171e-01 6.70652866e-01 6.15986548e-02
-4.25103664e-01 7.68033147e-01 3.67912740e-01 5.94305694e-01
1.03632534e+00 -1.19910610e+00 1.39677310e+00 9.34745073e-01
5.33995330e-01 3.85144919e-01 3.18850935e-01 -6.78043604e-01
-1.35949695e+00 -1.07238758e+00 8.92107606e-01 -6.19177461e-01
6.11115217e-01 -8.61605942e-01 -7.62538254e-01 3.62502992e-01
4.35262561e-01 -2.19412774e-01 1.27517190e-02 -2.93064207e-01
-4.85578299e-01 3.09633493e-01 -7.67752290e-01 6.41005695e-01
5.88897824e-01 -6.36391342e-01 -7.34977007e-01 7.80561328e-01
5.49722254e-01 -4.26031202e-01 -1.19011968e-01 -1.25621215e-01
9.28500950e-01 -8.55108619e-01 7.34423697e-01 -2.81286329e-01
6.74861550e-01 -2.31845334e-01 9.46207196e-02 -6.35888815e-01
-5.49087599e-02 -7.17121184e-01 3.18164527e-01 7.52182424e-01
6.62393808e-01 -3.70400310e-01 8.52993667e-01 1.11080147e-01
-9.63888094e-02 -4.78258431e-01 -9.83954132e-01 -4.36505258e-01
-1.08653389e-01 -4.56063628e-01 1.43560037e-01 6.80129707e-01
-7.73667917e-02 5.26501954e-01 -2.36292973e-01 -3.49913508e-01
2.74326175e-01 -3.26877460e-02 9.41822886e-01 -9.27892745e-01
-1.88596219e-01 -8.78679574e-01 -4.53387678e-01 -1.51592445e+00
-2.29383245e-01 -8.86431158e-01 5.61211884e-01 -1.67169583e+00
2.22592443e-01 -4.98740897e-02 7.24430159e-02 6.28612936e-01
-7.34099224e-02 5.96824348e-01 1.00679040e-01 5.29255807e-01
-1.13266277e+00 1.94357976e-01 8.03555846e-01 -4.00046557e-01
1.16951829e-02 -3.30944628e-01 -1.94721520e-01 3.00131232e-01
6.55931115e-01 -4.82562274e-01 8.37700907e-03 -7.76556492e-01
6.48699105e-01 -2.99071521e-01 5.86872220e-01 -9.04330373e-01
5.26457250e-01 2.76702553e-01 5.39235651e-01 -9.82596636e-01
6.45653531e-02 -4.52557504e-01 -6.80807531e-01 4.25871760e-01
-8.09412360e-01 5.93309253e-02 2.82914549e-01 5.57251096e-01
1.21125184e-01 -3.53281945e-01 6.74173594e-01 -1.47015825e-01
-6.42949581e-01 -3.80192585e-02 -7.77692914e-01 -7.25560486e-02
5.39408386e-01 -4.00705636e-01 -5.47990084e-01 -3.45878929e-01
-1.49128810e-01 4.74964380e-02 6.33525550e-01 5.16220391e-01
7.95431137e-01 -7.16378570e-01 -1.07012701e+00 3.13409418e-01
4.91385341e-01 1.97968329e-03 -4.74733561e-01 6.00268364e-01
-1.11062396e+00 6.72257662e-01 7.28044584e-02 -9.97186542e-01
-1.20747912e+00 3.13842893e-01 4.62898254e-01 -3.77565503e-01
-8.92082512e-01 7.36357272e-01 -1.04526356e-01 -4.44573492e-01
3.94667745e-01 -4.42478418e-01 2.31153145e-01 -9.36792493e-02
7.99698532e-01 -1.24417711e-03 4.55692351e-01 -3.60493690e-01
-2.65425414e-01 4.64546502e-01 -6.05821252e-01 -5.87289393e-01
9.58705723e-01 -6.93088537e-03 4.51947376e-02 9.32364091e-02
1.02748561e+00 -1.51660427e-01 -9.11537945e-01 -2.90797800e-01
2.32652321e-01 -3.08018357e-01 2.71197796e-01 -1.21084297e+00
-4.44369972e-01 1.01595259e+00 6.98326707e-01 3.34233075e-01
8.33523750e-01 -2.32738987e-01 8.14148664e-01 1.16901660e+00
-1.72200978e-01 -1.40385044e+00 3.56838077e-01 8.01928997e-01
8.54572713e-01 -1.32196522e+00 1.33520842e-01 2.28641510e-01
-2.10628435e-01 1.58733356e+00 5.86689264e-02 -1.35679320e-01
1.05787247e-01 2.81275064e-01 1.56504035e-01 -4.13446933e-01
-9.06976581e-01 -1.41488001e-01 4.60420012e-01 1.47601694e-01
4.99244750e-01 -3.44322085e-01 1.12776488e-01 -3.63820016e-01
1.19091049e-01 -1.95003405e-01 4.70691413e-01 1.23189187e+00
-9.20388162e-01 -1.02615058e+00 -5.34675241e-01 8.29070866e-01
-5.95256448e-01 -6.62038624e-01 -1.04830968e+00 5.58547139e-01
-5.61461210e-01 7.23945856e-01 2.03610405e-01 -1.48161054e-01
1.82029277e-01 3.35596710e-01 1.63153067e-01 -7.20184445e-01
-8.63751411e-01 3.29813838e-01 1.32827774e-01 -4.41885769e-01
-1.13954090e-01 -3.54317039e-01 -8.66352737e-01 -1.34107962e-01
-6.35564804e-01 -2.11929440e-01 1.06545007e+00 9.00267303e-01
5.29238880e-01 2.34007746e-01 4.43032920e-01 -1.12793839e+00
-4.73759234e-01 -1.19060528e+00 -1.16188921e-01 2.39487112e-01
3.76479268e-01 5.76495603e-02 -1.38095543e-01 3.41519386e-01] | [11.811220169067383, 2.3506577014923096] |
52ff4c6b-f937-4006-afa5-f3c733587c16 | optimizing-pessimism-in-dynamic-treatment | 2210.14420 | null | https://arxiv.org/abs/2210.14420v2 | https://arxiv.org/pdf/2210.14420v2.pdf | Optimizing Pessimism in Dynamic Treatment Regimes: A Bayesian Learning Approach | In this article, we propose a novel pessimism-based Bayesian learning method for optimal dynamic treatment regimes in the offline setting. When the coverage condition does not hold, which is common for offline data, the existing solutions would produce sub-optimal policies. The pessimism principle addresses this issue by discouraging recommendation of actions that are less explored conditioning on the state. However, nearly all pessimism-based methods rely on a key hyper-parameter that quantifies the degree of pessimism, and the performance of the methods can be highly sensitive to the choice of this parameter. We propose to integrate the pessimism principle with Thompson sampling and Bayesian machine learning for optimizing the degree of pessimism. We derive a credible set whose boundary uniformly lower bounds the optimal Q-function, and thus we do not require additional tuning of the degree of pessimism. We develop a general Bayesian learning method that works with a range of models, from Bayesian linear basis model to Bayesian neural network model. We develop the computational algorithm based on variational inference, which is highly efficient and scalable. We establish the theoretical guarantees of the proposed method, and show empirically that it outperforms the existing state-of-the-art solutions through both simulations and a real data example. | ['Lexin Li', 'Chengchun Shi', 'Zhengling Qi', 'Yunzhe Zhou'] | 2022-10-26 | null | null | null | null | ['thompson-sampling'] | ['methodology'] | [-1.03309706e-01 2.21322730e-01 -1.05132079e+00 -3.23793679e-01
-8.44322145e-01 -3.70969027e-01 2.51877218e-01 1.94186121e-01
-4.14793909e-01 9.37927246e-01 3.91089767e-02 -6.27307236e-01
-5.73982000e-01 -7.41875708e-01 -5.80357909e-01 -9.43527341e-01
4.47101295e-02 6.58849001e-01 2.24437311e-01 6.22159392e-02
3.93245250e-01 1.53180793e-01 -9.03863668e-01 -1.13626800e-01
1.09680223e+00 1.13650167e+00 9.52686369e-02 1.72404096e-01
5.15879929e-01 5.32724917e-01 -3.67825121e-01 -2.48445496e-01
2.13918597e-01 -1.06779903e-01 -3.68066460e-01 -1.04228199e-01
-2.83034503e-01 -6.01123035e-01 -1.80446446e-01 1.42812550e+00
3.46376359e-01 6.55616969e-02 8.89874160e-01 -1.18456006e+00
-1.04626670e-01 8.98651004e-01 -6.91866636e-01 2.32331336e-01
-1.34438574e-01 1.77133858e-01 9.70632255e-01 1.48948224e-03
1.29647538e-01 1.51415193e+00 4.17109996e-01 3.54829311e-01
-1.24605715e+00 -7.51742303e-01 4.50612158e-01 3.17488074e-01
-1.44038439e+00 -2.54265100e-01 6.77624404e-01 -4.36018527e-01
2.87732691e-01 -5.70628457e-02 6.22323275e-01 1.16321671e+00
6.41962826e-01 7.35887110e-01 1.43886971e+00 -4.94550347e-01
8.60677361e-01 1.80063665e-01 6.73121214e-02 4.58279610e-01
6.49361908e-01 5.79194248e-01 -1.35790884e-01 -6.28248394e-01
7.08983421e-01 3.28420192e-01 -8.34418163e-02 -5.23638487e-01
-3.42349708e-01 1.24534929e+00 -2.37128995e-02 -4.05997247e-01
-4.35887843e-01 1.77852362e-01 2.25313812e-01 7.90016428e-02
3.19899619e-01 1.49474204e-01 -3.58896464e-01 -1.51288942e-01
-9.39275622e-01 5.07899165e-01 7.39802837e-01 5.34023225e-01
2.70019829e-01 -2.35078871e-01 -6.87128305e-01 5.02297938e-01
6.65563524e-01 5.16925037e-01 8.52447599e-02 -1.35455072e+00
3.26998711e-01 1.06006891e-01 8.34708869e-01 -8.06818724e-01
-1.13291308e-01 -3.55730534e-01 -5.76356888e-01 6.14897497e-02
4.74041879e-01 -4.78330433e-01 -6.10177279e-01 2.03479385e+00
5.29065192e-01 1.50801122e-01 -1.90285280e-01 7.74588525e-01
-1.01806864e-01 4.89634722e-01 3.29717733e-02 -9.90899086e-01
1.11675835e+00 -4.23058838e-01 -1.06877637e+00 -8.59492123e-02
3.34655762e-01 -3.46361160e-01 8.95694375e-01 7.20755637e-01
-1.19988608e+00 9.78749394e-02 -1.08157861e+00 7.58237600e-01
2.83260077e-01 -1.44731686e-01 4.86199170e-01 1.07307923e+00
-5.90478897e-01 7.95560837e-01 -1.04693747e+00 -3.12300399e-02
4.38430816e-01 1.21417239e-01 5.78612208e-01 1.66200444e-01
-1.22377264e+00 1.06696498e+00 4.76169527e-01 3.11653018e-02
-1.50599849e+00 -4.16554600e-01 -4.34187561e-01 2.90058106e-01
1.21418858e+00 -5.36604345e-01 1.64595222e+00 -5.45436025e-01
-1.76036966e+00 9.34889093e-02 7.15096816e-02 -6.33613110e-01
7.03822970e-01 -7.11982250e-02 -7.63536766e-02 1.16371922e-01
-1.12691075e-01 2.94530150e-02 9.11842763e-01 -1.12432480e+00
-5.37799656e-01 -3.35197479e-01 3.48318487e-01 1.87699065e-01
-2.37117767e-01 -3.83726768e-02 -2.35503674e-01 -2.68858671e-01
-4.84972894e-02 -9.71800983e-01 -6.23748779e-01 -3.06228548e-01
-4.74862546e-01 -4.07488644e-01 2.08109751e-01 -1.81109533e-01
1.60652483e+00 -1.87259972e+00 -2.67105997e-01 5.02789795e-01
-1.76522315e-01 -5.89433052e-02 5.23733914e-01 4.10107940e-01
4.36663926e-01 1.19928665e-01 -1.28125235e-01 9.35005024e-02
3.07181507e-01 3.37642998e-01 -4.91901368e-01 7.71516085e-01
-4.32890415e-01 3.07756215e-01 -9.07613337e-01 -6.54706299e-01
-1.86799318e-02 -1.12903237e-01 -7.99840569e-01 2.15374932e-01
-3.13463241e-01 1.78317651e-01 -8.82200241e-01 4.09168333e-01
6.75365448e-01 -4.10494387e-01 7.54592478e-01 -4.49676067e-02
-2.93875374e-02 2.30704978e-01 -1.47738087e+00 1.07643974e+00
-2.17029944e-01 -2.46388137e-01 1.59589574e-01 -1.43901896e+00
3.56750250e-01 2.03559309e-01 4.21779484e-01 -1.94424540e-01
3.50671351e-01 -1.27999276e-01 -7.58896247e-02 -5.50178170e-01
1.82746761e-02 -5.03054857e-01 -3.09648931e-01 5.85615873e-01
-4.08759415e-01 -1.66921780e-01 -8.33678469e-02 1.33286595e-01
7.95045614e-01 1.14633113e-01 6.90238059e-01 -6.09000981e-01
3.16033065e-02 -3.20132971e-01 9.81519997e-01 1.25253689e+00
-6.28284752e-01 -8.81433189e-02 1.05594361e+00 1.69445965e-02
-6.19488657e-01 -1.12650597e+00 -4.13976252e-01 9.68486369e-01
2.26606533e-01 -1.84227750e-01 -9.33277667e-01 -7.20279157e-01
1.23727523e-01 1.01087070e+00 -6.54737175e-01 -3.62558633e-01
9.88475159e-02 -1.14814055e+00 6.78488314e-02 4.24941540e-01
1.69118688e-01 -4.17158365e-01 -7.49808550e-01 8.05036798e-02
-1.06605113e-01 -7.51227379e-01 -3.68512630e-01 2.75267363e-02
-9.17579472e-01 -9.60427582e-01 -2.82167137e-01 2.60464251e-01
4.02079731e-01 -2.24580932e-02 8.07889163e-01 -3.46155941e-01
3.58941942e-01 2.83397168e-01 7.18956813e-02 -5.36919415e-01
-4.02169853e-01 -2.11618677e-01 4.56626832e-01 -1.22372001e-01
1.36682868e-01 -3.82025748e-01 -9.58373189e-01 2.88763165e-01
-6.80535138e-01 -2.33377814e-01 5.64327836e-01 8.00527096e-01
7.18722224e-01 3.29298019e-01 8.03270042e-01 -9.63426709e-01
7.08743393e-01 -7.01100409e-01 -1.14407694e+00 4.22296256e-01
-1.03122485e+00 2.13991717e-01 2.87930131e-01 -5.14715135e-01
-1.37476349e+00 -3.10373843e-01 1.60860732e-01 -3.89040828e-01
2.23905846e-01 6.31363690e-01 -2.13780954e-01 4.37549740e-01
4.60006654e-01 -2.85594255e-01 -8.73201415e-02 -3.92023236e-01
2.48897627e-01 8.66639435e-01 1.48120433e-01 -1.03428113e+00
1.80653438e-01 6.42290413e-01 1.26628056e-01 -2.44842395e-01
-1.52576804e+00 -5.28892241e-02 -1.32877296e-02 -2.56919891e-01
4.28099006e-01 -7.63182521e-01 -1.40495920e+00 -1.15857035e-01
-7.08975375e-01 -4.30679977e-01 -2.47296924e-03 6.08794332e-01
-1.06660604e+00 5.52289665e-01 -4.44859236e-01 -1.45327854e+00
-6.70823008e-02 -1.29202616e+00 5.07670820e-01 2.47357830e-01
-1.40625173e-02 -9.23615575e-01 2.81320274e-01 3.02564144e-01
1.04367658e-01 2.20983475e-01 8.42583179e-01 -4.54109222e-01
-3.97519171e-01 -6.78399354e-02 8.99229944e-02 9.94906425e-02
-1.07956842e-01 -9.37348008e-02 -6.55435741e-01 -6.35079324e-01
4.34842348e-01 -3.93792182e-01 7.88918793e-01 1.01721692e+00
1.49418151e+00 -9.23294365e-01 -5.36736190e-01 2.40559444e-01
1.45645440e+00 4.26755667e-01 3.01958144e-01 2.63340265e-01
4.81905527e-02 5.39395094e-01 9.59177971e-01 1.07096446e+00
4.83529270e-01 7.19117939e-01 6.49476528e-01 5.43054283e-01
8.42772365e-01 -3.45747173e-01 4.37383652e-01 1.53400779e-01
1.25493258e-01 -5.00595458e-02 -6.08540833e-01 4.35154647e-01
-2.20673656e+00 -1.02189314e+00 4.02249694e-01 2.80909252e+00
1.33295298e+00 5.07190347e-01 3.12943757e-01 -1.47522718e-01
6.74597561e-01 -9.84713361e-02 -8.97070348e-01 -5.97623527e-01
4.98887658e-01 -2.18452901e-01 7.92724967e-01 6.20291173e-01
-1.00137734e+00 6.48138940e-01 7.02586651e+00 1.27589548e+00
-5.92569351e-01 3.86093795e-01 8.45514417e-01 -2.96711266e-01
-2.80446172e-01 2.35431984e-01 -1.01823485e+00 8.48904729e-01
1.29174030e+00 -2.64385551e-01 3.99759263e-01 9.54755127e-01
8.19844544e-01 -4.69124794e-01 -1.01325369e+00 6.54282808e-01
-4.78828460e-01 -1.07058263e+00 -5.02765238e-01 4.90649343e-02
9.30305541e-01 -1.32835269e-01 -7.52413692e-03 3.16299081e-01
9.29554939e-01 -8.57320547e-01 7.68862486e-01 7.96807230e-01
4.36360151e-01 -1.08053052e+00 6.28473639e-01 7.89663792e-01
-3.84643167e-01 -6.23233378e-01 -5.22888064e-01 -1.10702470e-01
2.61668533e-01 9.07779992e-01 -4.54917461e-01 2.24516541e-01
6.50031507e-01 2.29786322e-01 1.46275029e-01 1.03887153e+00
-4.27001953e-01 7.99229503e-01 -4.59956497e-01 -1.56888917e-01
2.49810159e-01 -3.09229404e-01 3.99415493e-01 7.84776270e-01
2.14807764e-01 1.97770640e-01 4.86151010e-01 8.51980984e-01
2.91433364e-01 -2.50397354e-01 -2.92697072e-01 8.27857256e-02
9.78036642e-01 9.25159395e-01 -5.46710968e-01 -4.23025757e-01
-9.54936296e-02 6.16637580e-02 2.15221331e-01 3.93957108e-01
-9.06257272e-01 1.23331591e-01 4.85798478e-01 -3.86557020e-02
3.34605902e-01 1.48124635e-01 -4.31105137e-01 -8.52853239e-01
-2.59573668e-01 -9.06919003e-01 7.99857557e-01 -1.94332272e-01
-1.31924164e+00 -2.28037536e-01 6.84566736e-01 -8.95862341e-01
-5.98960638e-01 -4.52062458e-01 -5.57934105e-01 5.46997607e-01
-1.31636655e+00 -4.96243834e-01 4.87679988e-01 3.61400336e-01
1.36116803e-01 3.32228124e-01 3.75852346e-01 -1.84741899e-01
-8.26870084e-01 5.00348210e-01 5.58539748e-01 -5.00536323e-01
6.05535865e-01 -1.05448365e+00 -6.28043532e-01 5.90638638e-01
-6.47037983e-01 5.62526941e-01 1.25176430e+00 -6.20479167e-01
-1.21075678e+00 -7.76037455e-01 1.89842761e-01 -2.02328548e-01
7.38015234e-01 -3.73067148e-02 -6.00040734e-01 7.15404570e-01
-6.59955740e-02 -3.37436527e-01 5.93594611e-01 2.84290761e-01
-8.90292898e-02 -2.77763128e-01 -1.37570369e+00 6.93233967e-01
6.39609993e-01 -5.58929443e-02 -5.25790513e-01 5.94091892e-01
5.38511634e-01 -2.37024143e-01 -7.61595905e-01 5.80346823e-01
6.32398963e-01 -8.78375351e-01 8.54300916e-01 -8.71729612e-01
1.52962744e-01 7.18953162e-02 -2.65236974e-01 -1.22905111e+00
-2.58108407e-01 -9.63572204e-01 -6.69667900e-01 7.84709990e-01
1.53998867e-01 -6.04324996e-01 5.64791083e-01 6.32672608e-01
1.68407008e-01 -1.08141780e+00 -1.19966829e+00 -9.76629853e-01
4.28876460e-01 -4.80781198e-01 4.74646002e-01 6.58360481e-01
3.54656875e-01 -5.67270704e-02 -6.07012331e-01 3.74644578e-01
1.13091695e+00 2.62206286e-01 3.13612223e-01 -1.02905226e+00
-7.79912829e-01 -5.09477258e-01 3.14535171e-01 -1.00557816e+00
1.53905466e-01 -1.36368394e-01 2.17701122e-01 -1.19598091e+00
7.79681623e-01 -3.84754449e-01 -5.98386109e-01 4.00569290e-01
-2.33970970e-01 -5.85446835e-01 -1.61074385e-01 1.60095245e-01
-8.10986936e-01 7.14750290e-01 1.09840250e+00 2.80176681e-02
-2.67893672e-01 4.33977604e-01 -9.15211737e-01 9.78707254e-01
7.33393908e-01 -7.08502710e-01 -6.26499236e-01 2.37302974e-01
4.09420729e-01 6.43407941e-01 1.53496474e-01 -3.20162684e-01
-4.49733175e-02 -1.12945449e+00 -9.22754630e-02 -6.57236278e-01
1.47739798e-01 -6.04161739e-01 -6.28985465e-02 8.37481558e-01
-5.63501358e-01 -4.17204976e-01 -2.41775483e-01 1.07165051e+00
5.26098251e-01 -5.92090368e-01 1.12402213e+00 -1.06334962e-01
-8.85933451e-03 3.13230097e-01 -4.37698275e-01 1.12729900e-01
8.97795975e-01 3.02952856e-01 -1.96883634e-01 -6.84931040e-01
-6.52647018e-01 5.46842575e-01 1.64470151e-01 -8.37441087e-02
2.50422150e-01 -1.20470285e+00 -4.51460779e-01 -3.60022634e-01
-2.05265388e-01 -5.00480831e-01 3.73755306e-01 1.08348989e+00
2.54932851e-01 4.37327325e-01 2.49708802e-01 -5.76150179e-01
-6.44815862e-01 8.06827903e-01 4.87898946e-01 -5.89093089e-01
-2.03753233e-01 1.82375520e-01 1.03043303e-01 3.26951593e-02
4.02285337e-01 -1.47691250e-01 -2.08113834e-01 -1.25666425e-01
5.30435920e-01 6.48103118e-01 -2.47286662e-01 -4.45711091e-02
-1.84552878e-01 1.21886162e-02 -4.61093895e-02 -4.20130521e-01
9.93178308e-01 -4.40296561e-01 1.73153222e-01 4.65959907e-01
5.27844369e-01 -3.46095949e-01 -1.78724599e+00 -3.32141101e-01
-2.07512341e-02 -6.03074849e-01 5.47794580e-01 -8.15868676e-01
-7.85502911e-01 6.98007166e-01 7.04008758e-01 4.04996753e-01
9.25349116e-01 -7.76562616e-02 2.22051591e-01 3.85142595e-01
5.39343059e-01 -1.66002762e+00 -6.41601756e-02 1.82472572e-01
4.69278067e-01 -1.26371348e+00 3.23391616e-01 -5.23817688e-02
-6.60418451e-01 5.43614984e-01 3.97296965e-01 -6.56201914e-02
9.76157129e-01 1.77733630e-01 -3.25927794e-01 -4.03647646e-02
-1.04123878e+00 1.40368789e-01 8.01049024e-02 2.68644452e-01
-2.16079950e-02 5.76905906e-01 -8.66393566e-01 1.07872486e+00
8.97469521e-02 1.68696597e-01 6.00969970e-01 8.78325880e-01
-7.37838447e-01 -8.17291141e-01 -5.93749225e-01 4.51333791e-01
-8.14646304e-01 2.24988490e-01 2.04880446e-01 5.49731553e-01
-2.46881321e-01 1.19504011e+00 3.98138016e-02 2.70044178e-01
-1.83647349e-02 -9.63743404e-02 6.38469636e-01 -3.34852397e-01
2.44762272e-01 5.62328398e-01 4.96247485e-02 -6.30295813e-01
-3.84631366e-01 -7.33083069e-01 -9.02487159e-01 -4.40187037e-01
-4.89245802e-01 2.10641339e-01 3.31678301e-01 1.31631458e+00
1.14172257e-01 9.34119821e-02 9.50444937e-01 -5.41790307e-01
-1.78010380e+00 -9.46893394e-01 -7.91994810e-01 -1.21091446e-03
1.73680916e-01 -1.23956037e+00 -4.03829068e-01 -5.45669854e-01] | [4.54005241394043, 3.1942732334136963] |
c812904e-28b4-4ed6-abdf-e7ebf02633d1 | facial-expression-retargeting-from-human-to | 2008.05110 | null | https://arxiv.org/abs/2008.05110v1 | https://arxiv.org/pdf/2008.05110v1.pdf | Facial Expression Retargeting from Human to Avatar Made Easy | Facial expression retargeting from humans to virtual characters is a useful technique in computer graphics and animation. Traditional methods use markers or blendshapes to construct a mapping between the human and avatar faces. However, these approaches require a tedious 3D modeling process, and the performance relies on the modelers' experience. In this paper, we propose a brand-new solution to this cross-domain expression transfer problem via nonlinear expression embedding and expression domain translation. We first build low-dimensional latent spaces for the human and avatar facial expressions with variational autoencoder. Then we construct correspondences between the two latent spaces guided by geometric and perceptual constraints. Specifically, we design geometric correspondences to reflect geometric matching and utilize a triplet data structure to express users' perceptual preference of avatar expressions. A user-friendly method is proposed to automatically generate triplets for a system allowing users to easily and efficiently annotate the correspondences. Using both geometric and perceptual correspondences, we trained a network for expression domain translation from human to avatar. Extensive experimental results and user studies demonstrate that even nonprofessional users can apply our method to generate high-quality facial expression retargeting results with less time and effort. | ['Juyong Zhang', 'Keyu Chen', 'Jianmin Zheng'] | 2020-08-12 | null | null | null | null | ['geometric-matching'] | ['computer-vision'] | [-1.59731403e-01 -2.04843655e-03 -4.23710793e-02 -5.88468552e-01
-4.37332094e-01 -5.53131998e-01 4.66776967e-01 -6.20922267e-01
-8.84529203e-02 3.44335169e-01 1.32496521e-01 2.82685280e-01
5.72241366e-01 -5.93190849e-01 -3.01025420e-01 -5.03578961e-01
4.38558519e-01 3.92804950e-01 -2.90517181e-01 -5.06737113e-01
-2.20269114e-01 6.49558365e-01 -1.49316525e+00 2.02136964e-01
7.96598911e-01 9.45412338e-01 -1.95296943e-01 4.07429397e-01
-4.08503830e-01 5.37868738e-01 -4.93433833e-01 -6.23342335e-01
2.93598264e-01 -5.66910207e-01 -4.77673531e-01 3.99848789e-01
3.47828388e-01 -4.63566959e-01 -2.72766687e-02 1.12752426e+00
5.37185669e-01 3.63129139e-01 7.73899496e-01 -1.60887468e+00
-1.13316309e+00 1.14804946e-01 -8.14407587e-01 -6.88483238e-01
7.12655783e-01 1.66953616e-02 8.32796991e-01 -1.12517929e+00
7.23463476e-01 1.47512209e+00 6.66333735e-01 1.05177987e+00
-1.37952912e+00 -9.88193810e-01 1.17511876e-01 7.27648586e-02
-1.73546219e+00 -5.20992100e-01 1.32458735e+00 -7.50954926e-01
4.09014434e-01 4.09156322e-01 1.07369387e+00 1.44208515e+00
-3.10303450e-01 7.54131258e-01 8.34204137e-01 -3.28840733e-01
-5.70862330e-02 4.99380171e-01 -3.69261295e-01 8.16400349e-01
-6.30726159e-01 -3.74789573e-02 -3.57106805e-01 -4.36261356e-01
1.53050148e+00 -1.14271343e-01 -2.20678627e-01 -6.86296999e-01
-9.35636759e-01 1.04740882e+00 3.24611336e-01 1.53019696e-01
-5.61875701e-01 1.11961253e-01 3.39586526e-01 1.94358919e-02
5.54679036e-01 3.48734766e-01 -1.68420985e-01 -2.40219146e-01
-5.89877963e-01 2.91079521e-01 5.59646368e-01 1.04900956e+00
7.75141776e-01 3.95544022e-01 -1.43730700e-01 1.25668156e+00
5.43621898e-01 4.50182438e-01 4.10893053e-01 -1.23114276e+00
-1.94456160e-01 5.26064992e-01 2.84319520e-01 -1.47820330e+00
-2.29606461e-02 3.57389808e-01 -7.10317552e-01 3.37242901e-01
4.11935262e-02 -2.06187963e-01 -6.30001545e-01 1.98527396e+00
6.03867650e-01 3.16297024e-01 -2.61967033e-02 1.26692212e+00
8.34322810e-01 7.64328301e-01 2.80435801e-01 -2.23426178e-01
1.34246349e+00 -8.79283249e-01 -1.07935691e+00 2.19449833e-01
6.33339703e-01 -7.73367703e-01 1.52001095e+00 -1.47339432e-02
-1.12500894e+00 -6.83116734e-01 -8.30184221e-01 -1.61629871e-01
8.75730291e-02 2.50364900e-01 6.87004328e-01 5.88577688e-01
-9.88376558e-01 3.87981772e-01 -6.17104530e-01 -1.95438877e-01
1.62810698e-01 4.09567982e-01 -4.64351952e-01 6.71543479e-01
-1.29039454e+00 8.47948253e-01 -1.77374527e-01 1.37950495e-01
-4.58698153e-01 -6.73458457e-01 -1.32951462e+00 -1.22314468e-01
-9.63390991e-02 -8.47630322e-01 1.49929595e+00 -1.71341717e+00
-2.35803270e+00 1.03056967e+00 -6.17475025e-02 3.76173586e-01
4.58265990e-01 -4.94048782e-02 -3.83160651e-01 7.37963570e-03
-8.92109238e-03 8.17071974e-01 1.03644490e+00 -1.49311531e+00
-1.88859701e-01 -3.80012542e-01 -8.43368694e-02 3.90359223e-01
-4.67401564e-01 2.77694702e-01 -6.39914453e-01 -7.38421321e-01
-2.45515238e-02 -1.09955549e+00 -1.49929971e-01 6.67555630e-01
-1.14518322e-01 -2.80720383e-01 1.08874106e+00 -5.93542874e-01
8.46816778e-01 -2.28870845e+00 4.19807583e-01 3.86936694e-01
3.46708030e-01 9.96171013e-02 -1.79001719e-01 -7.78454915e-02
-2.78679341e-01 -9.84784439e-02 7.37028243e-03 -6.75987840e-01
2.33303726e-01 1.90252006e-01 -2.55583823e-01 4.31569576e-01
7.96135291e-02 9.51671422e-01 -7.60325909e-01 -7.20004618e-01
2.52604544e-01 9.96863544e-01 -6.61255777e-01 6.26668096e-01
-1.51467934e-01 6.98633671e-01 -5.99450171e-01 4.07481313e-01
5.99212348e-01 -6.93628490e-02 1.18979283e-01 -4.80811864e-01
4.03240882e-02 -3.22196037e-01 -1.04932928e+00 1.84587026e+00
-7.22247899e-01 5.70622265e-01 2.36155242e-01 -4.72162098e-01
1.33570278e+00 4.39088583e-01 4.99936610e-01 -5.78833520e-01
5.24779439e-01 -6.60990030e-02 -5.99467278e-01 -6.39564574e-01
5.66670954e-01 -6.19750738e-01 -2.23757207e-01 5.55080712e-01
-1.54212922e-01 -3.01056117e-01 -4.81946409e-01 -4.42373678e-02
1.44312203e-01 5.86457193e-01 1.06397480e-01 1.46255180e-01
4.79654342e-01 -3.94748926e-01 6.53341651e-01 -2.52680749e-01
-1.80132970e-01 5.60843468e-01 4.81993675e-01 -5.26252091e-01
-1.13897097e+00 -9.84892607e-01 2.51195729e-01 1.32812798e+00
1.95544317e-01 -3.05846810e-01 -9.52903569e-01 -3.49834502e-01
-1.76513880e-01 6.64504051e-01 -8.72465491e-01 -1.65284395e-01
-5.09206057e-01 -1.57380328e-01 5.34254789e-01 6.59968495e-01
3.14598382e-01 -9.92458224e-01 -3.68291885e-01 -4.35757916e-03
-2.15585992e-01 -1.09605420e+00 -9.09130156e-01 -8.09123397e-01
-4.84584332e-01 -4.86565858e-01 -1.03943741e+00 -1.01359153e+00
8.99099588e-01 -1.26789901e-02 7.50225365e-01 1.55560647e-05
-1.25313535e-01 4.88574177e-01 -2.81954437e-01 -1.32520333e-01
-5.60134292e-01 -5.17207980e-01 4.21013474e-01 5.57514787e-01
2.19457805e-01 -7.13469803e-01 -4.62690741e-01 5.31704247e-01
-6.71715498e-01 4.41220313e-01 1.41976386e-01 7.80072331e-01
5.53569317e-01 -8.74684989e-01 1.14215486e-01 -6.80343628e-01
9.46853340e-01 -1.89022664e-02 -5.60384691e-01 8.62343237e-02
-2.25668728e-01 1.25098769e-02 4.85538751e-01 -9.49876368e-01
-1.23385382e+00 3.62106055e-01 -1.62664920e-01 -1.16767466e+00
-7.05848495e-03 1.85605794e-01 -4.38528985e-01 -2.48030528e-01
7.31934309e-01 -4.41500209e-02 3.77568334e-01 -2.46140137e-01
8.91332805e-01 6.62916958e-01 6.33313596e-01 -8.38277817e-01
7.29702890e-01 4.98935908e-01 -3.44589978e-01 -8.09685528e-01
-3.41780722e-01 5.04174642e-02 -7.38124073e-01 -4.87369776e-01
1.04269016e+00 -8.17710578e-01 -1.11801445e+00 3.01450580e-01
-1.45105314e+00 -3.06148946e-01 -2.06067234e-01 3.49533707e-01
-8.62520456e-01 4.99850959e-01 -5.51762819e-01 -6.28354847e-01
-2.76597947e-01 -1.27975583e+00 1.33373225e+00 2.05640525e-01
-6.90112233e-01 -1.05954993e+00 2.81687379e-01 1.56380430e-01
2.52354950e-01 6.63640261e-01 7.20669806e-01 5.99729642e-02
-1.70291681e-02 -1.95872828e-01 -2.04219431e-01 1.52039602e-01
3.69882643e-01 3.84606063e-01 -9.05671954e-01 -2.67193653e-02
-2.30127126e-01 -2.64193565e-01 -7.30726644e-02 2.03588784e-01
1.11910486e+00 -4.88166451e-01 -2.22735763e-01 8.87873411e-01
7.73689687e-01 2.53936797e-01 5.00753880e-01 -1.66069046e-02
1.12092924e+00 8.64276826e-01 4.39360946e-01 6.14613056e-01
5.00667274e-01 1.19763112e+00 -1.72326751e-02 -4.64066774e-01
2.47283712e-01 -6.28775775e-01 3.71261775e-01 8.41903508e-01
-2.18485653e-01 3.29649508e-01 -5.93289316e-01 2.87048340e-01
-1.78741300e+00 -8.33019316e-01 2.95790911e-01 1.78023779e+00
1.02051032e+00 -5.23179948e-01 2.14389309e-01 -3.95141274e-01
7.88403749e-01 1.63735487e-02 -4.01902258e-01 -7.39997506e-01
1.60169974e-01 4.19305488e-02 -1.89987496e-01 6.20628953e-01
-7.05356359e-01 1.32450628e+00 5.92127943e+00 6.51034594e-01
-1.37393045e+00 1.19589761e-01 3.81246030e-01 -9.25484896e-02
-6.36917710e-01 -1.97989523e-01 -2.92477071e-01 1.84020519e-01
4.47209656e-01 -2.70042807e-01 2.07203344e-01 1.00755537e+00
4.26251531e-01 6.85007274e-01 -1.08105850e+00 1.67180479e+00
1.59377456e-01 -1.29309571e+00 2.82090485e-01 -7.31220767e-02
6.75182283e-01 -9.19384420e-01 3.83474201e-01 3.71149451e-01
2.65285969e-01 -1.04419827e+00 7.47961223e-01 6.59023106e-01
1.23712957e+00 -7.77887821e-01 1.94309890e-01 1.09555703e-02
-1.19841671e+00 3.83439392e-01 -7.87114277e-02 -4.05463763e-02
4.04752731e-01 -3.41698110e-01 -6.96180642e-01 -1.19686825e-03
3.95504087e-01 4.79339421e-01 1.10977674e-02 2.27006495e-01
-2.23517701e-01 -2.73964070e-02 -8.95987004e-02 -1.54502720e-01
-1.91503316e-02 -5.83758116e-01 4.14618939e-01 1.03529251e+00
3.52160215e-01 4.09925401e-01 1.91497996e-01 1.26177227e+00
-5.21418080e-02 4.87848073e-01 -7.84073472e-01 -4.29470055e-02
3.07541668e-01 1.40505540e+00 -1.75686091e-01 -2.74705946e-01
-2.17534408e-01 1.53078640e+00 3.34534198e-01 5.45517027e-01
-1.17276978e+00 -2.38426611e-01 1.21541071e+00 1.01301804e-01
-2.04550713e-01 -3.50440234e-01 -1.68999195e-01 -1.22853661e+00
-1.81808919e-01 -8.68545711e-01 -5.45457490e-02 -1.17041326e+00
-1.19303751e+00 8.83641601e-01 -3.86265479e-02 -1.26183295e+00
-5.38591504e-01 -4.42510128e-01 -5.07706404e-01 9.55027103e-01
-7.63749838e-01 -1.55657542e+00 -6.04151249e-01 8.68879378e-01
3.15496266e-01 -1.68495685e-01 1.10653067e+00 2.58356541e-01
-4.61205363e-01 1.03300405e+00 -5.38570940e-01 3.54324400e-01
6.40963912e-01 -8.52846920e-01 4.05298859e-01 1.87896550e-01
3.13975722e-01 7.14083612e-01 7.81529784e-01 -3.90236557e-01
-1.17007017e+00 -7.91236937e-01 4.62103873e-01 -4.34318960e-01
4.57883239e-01 -3.91376317e-01 -9.86486614e-01 8.02175581e-01
1.77536890e-01 -4.32522818e-02 1.13702857e+00 9.76394415e-02
-5.37804604e-01 2.52595335e-01 -1.09227133e+00 1.15133011e+00
9.20054913e-01 -7.91575313e-01 -3.44479352e-01 -1.94312353e-02
7.17058897e-01 -7.25061893e-01 -9.09280419e-01 2.63849229e-01
8.55701923e-01 -6.67183518e-01 7.91899860e-01 -8.10789287e-01
2.35792726e-01 -3.66415411e-01 -1.04047656e-01 -1.26172948e+00
-3.50699276e-01 -1.03592288e+00 1.85940221e-01 1.09530056e+00
2.20501646e-01 -2.79818118e-01 9.34680879e-01 1.16552246e+00
1.41774848e-01 -5.98156869e-01 -6.80038571e-01 -3.59809071e-01
-4.20786142e-02 -3.47945005e-01 8.97208750e-01 1.46624863e+00
1.12332426e-01 4.82445091e-01 -7.64139593e-01 8.27225000e-02
2.77525246e-01 2.42816374e-01 1.22806466e+00 -1.18758047e+00
-2.63527423e-01 -5.49946129e-01 -4.59246367e-01 -1.19047129e+00
7.44471312e-01 -7.62646556e-01 1.55759891e-02 -1.04528868e+00
3.74799520e-02 -2.94760615e-01 1.85005277e-01 5.29794276e-01
9.40057710e-02 3.52481514e-01 2.28769436e-01 2.62824595e-01
-2.91561961e-01 1.25676572e+00 1.51672566e+00 5.86506873e-02
-5.79228938e-01 -1.86000109e-01 -6.82600617e-01 9.70378101e-01
5.07447124e-01 -1.28668621e-01 -4.09553677e-01 -4.69363630e-01
9.63576138e-02 2.68440098e-01 4.61444527e-01 -3.93489450e-01
-7.85189867e-02 -4.63065237e-01 4.44331080e-01 1.06207766e-02
8.97569358e-01 -7.79826164e-01 3.98199826e-01 -1.16895325e-01
-2.94833809e-01 1.90865368e-01 3.33372951e-01 1.87595770e-01
-1.88429296e-01 2.17871755e-01 9.51250553e-01 8.71628746e-02
-7.56138742e-01 6.03556514e-01 -3.31999213e-01 -4.09531355e-01
1.20353413e+00 -3.76062244e-01 3.60595137e-01 -9.18293834e-01
-1.13597941e+00 3.08866799e-02 6.74672782e-01 7.00540066e-01
9.24287081e-01 -2.04705381e+00 -6.82679653e-01 4.54386830e-01
2.43314803e-01 -2.82003582e-01 4.47517782e-01 4.50736403e-01
-5.75671732e-01 -2.73494214e-01 -6.53644264e-01 -5.92815220e-01
-1.58605671e+00 3.86776924e-01 5.49660444e-01 4.46685612e-01
-5.64777851e-01 1.05141902e+00 4.39758122e-01 -7.59962916e-01
-5.75618818e-02 1.03561610e-01 -2.40154132e-01 1.27173632e-01
4.00005996e-01 1.00475490e-01 -6.47860229e-01 -1.33409250e+00
-8.77981409e-02 8.99033368e-01 1.52731776e-01 -4.48339522e-01
9.44024324e-01 -1.01484098e-01 -3.22150439e-02 4.27454412e-01
1.55573583e+00 -3.17486003e-02 -1.40343058e+00 -2.54546821e-01
-5.12683511e-01 -7.74774134e-01 -2.16127083e-01 -9.87842903e-02
-1.14131379e+00 9.41499293e-01 6.34691358e-01 -2.08446100e-01
1.00893283e+00 -6.78053647e-02 9.26188827e-01 4.83307764e-02
1.70936301e-01 -1.09583879e+00 4.80974823e-01 1.91439405e-01
1.23965955e+00 -9.51189339e-01 -3.49373728e-01 -4.37540591e-01
-1.22394955e+00 9.74861801e-01 8.97785962e-01 -3.64044905e-02
4.87648547e-01 5.68445511e-02 6.63495779e-01 -1.96986377e-01
-4.61378038e-01 2.09260374e-01 3.13400894e-01 7.34525204e-01
5.41979790e-01 2.82509774e-01 5.65264337e-02 7.23119676e-01
-4.64296252e-01 -9.49441940e-02 2.82877296e-01 5.08925378e-01
-1.67468920e-01 -1.25768471e+00 -3.26186031e-01 -2.54034787e-01
-1.00024780e-02 3.33791286e-01 -4.53185022e-01 6.22118115e-01
-1.95350070e-02 4.33843970e-01 2.09033757e-01 -6.16616368e-01
6.83998764e-01 -3.70315872e-02 5.99051595e-01 -4.98031050e-01
-2.40426511e-01 2.65531778e-01 -2.07542881e-01 -6.18265331e-01
-3.01922947e-01 -4.82395500e-01 -1.36461699e+00 -3.85550708e-01
-1.06659286e-01 1.58098117e-01 5.48306286e-01 5.32802701e-01
5.20096242e-01 1.52288944e-01 8.16411853e-01 -1.20828795e+00
-3.35929781e-01 -7.97402978e-01 -5.91051936e-01 7.93848693e-01
7.13207647e-02 -9.05436516e-01 7.25163892e-02 2.77298808e-01] | [12.934860229492188, -0.38018229603767395] |
63cca976-f7e3-4740-a913-c4e12a42e691 | component-aware-anomaly-detection-framework | 2305.08509 | null | https://arxiv.org/abs/2305.08509v1 | https://arxiv.org/pdf/2305.08509v1.pdf | Component-aware anomaly detection framework for adjustable and logical industrial visual inspection | Industrial visual inspection aims at detecting surface defects in products during the manufacturing process. Although existing anomaly detection models have shown great performance on many public benchmarks, their limited adjustability and ability to detect logical anomalies hinder their broader use in real-world settings. To this end, in this paper, we propose a novel component-aware anomaly detection framework (ComAD) which can simultaneously achieve adjustable and logical anomaly detection for industrial scenarios. Specifically, we propose to segment images into multiple components based on a lightweight and nearly training-free unsupervised semantic segmentation model. Then, we design an interpretable logical anomaly detection model through modeling the metrological features of each component and their relationships. Despite its simplicity, our framework achieves state-of-the-art performance on image-level logical anomaly detection. Meanwhile, segmenting a product image into multiple components provides a novel perspective for industrial visual inspection, demonstrating great potential in model customization, noise resistance, and anomaly classification. The code will be available at https://github.com/liutongkun/ComAD. | ['Zhuo Zhao', 'Liuyi Jin', 'Xiao Jin', 'Bingke Jiang', 'Xiao Du', 'Bing Li', 'Tongkun Liu'] | 2023-05-15 | null | null | null | null | ['unsupervised-semantic-segmentation', 'anomaly-classification'] | ['computer-vision', 'computer-vision'] | [ 1.33746892e-01 -7.31799081e-02 2.09738478e-01 -3.71199995e-01
-3.48376393e-01 -4.41742301e-01 2.51637727e-01 5.66383481e-01
5.29746175e-01 -2.62827277e-01 -7.67463803e-01 -4.30269927e-01
-2.56488919e-01 -8.97861779e-01 -4.58706349e-01 -5.91068268e-01
1.37049317e-01 3.59357893e-01 1.47152677e-01 -1.94514692e-02
1.77637160e-01 7.99368262e-01 -1.77418005e+00 5.90130314e-02
1.18830335e+00 1.37129617e+00 5.78133352e-02 3.89064878e-01
5.80929313e-03 2.80977070e-01 -6.80498540e-01 -1.20516419e-01
4.33903128e-01 -3.17237616e-01 -4.92377639e-01 7.86420524e-01
2.88739234e-01 -1.76092654e-01 -5.84755372e-03 1.29827225e+00
4.91605625e-02 -7.91134164e-02 5.38540661e-01 -1.58401108e+00
-7.13624656e-01 9.35341269e-02 -7.50062525e-01 -6.74171001e-02
1.85081050e-01 6.39347970e-01 9.21961427e-01 -6.40090108e-01
2.02468842e-01 9.23579991e-01 5.14581442e-01 1.13996796e-01
-1.19609201e+00 -3.97874475e-01 5.77450514e-01 2.83206373e-01
-1.04533136e+00 8.55633244e-02 1.07319486e+00 -5.86196125e-01
8.00455749e-01 3.52948934e-01 6.25636339e-01 6.22261286e-01
3.39239925e-01 7.62648046e-01 7.10945725e-01 -2.40345120e-01
3.19129169e-01 -1.84782162e-01 1.83747828e-01 7.52755105e-01
6.48652494e-01 -3.05031389e-01 -2.65323162e-01 4.52343151e-02
6.49228275e-01 4.34408098e-01 1.00853264e-01 -4.92253304e-01
-8.95408452e-01 5.14981449e-01 2.38306269e-01 -4.81846593e-02
-5.52904367e-01 -1.28796786e-01 5.38932979e-01 4.98456866e-01
4.87932980e-01 6.62540197e-01 -5.19553125e-01 1.81682929e-01
-5.32187521e-01 -3.84666212e-02 3.34318757e-01 8.71825576e-01
8.02398264e-01 2.77782142e-01 -6.90980032e-02 8.40899348e-01
5.43009341e-01 5.23049593e-01 7.10921884e-02 -9.51710284e-01
-9.41709504e-02 1.08501363e+00 -1.59065604e-01 -1.02395427e+00
-3.38843107e-01 -3.08043808e-01 -7.20161021e-01 7.17232049e-01
1.83808118e-01 2.66761363e-01 -1.11500561e+00 9.46165442e-01
5.34143269e-01 1.87988430e-02 -9.18495208e-02 7.50615716e-01
2.70540923e-01 3.78170073e-01 8.85715634e-02 5.18736281e-02
1.41911113e+00 -9.47708488e-01 -7.97049105e-01 -3.56245071e-01
5.99949121e-01 -8.10568452e-01 1.30857813e+00 8.47976208e-01
-7.59341657e-01 -5.29122591e-01 -1.17463160e+00 2.83204377e-01
-4.32651818e-01 1.81962982e-01 7.15200365e-01 4.73620027e-01
-3.82750839e-01 4.71230358e-01 -1.22980905e+00 -4.03057754e-01
4.95625347e-01 1.05420358e-01 -2.53865272e-01 -1.59470379e-01
-5.00197709e-01 3.98552984e-01 2.17891172e-01 3.79308820e-01
-9.10492539e-01 -6.81326628e-01 -1.07152224e+00 -1.82773441e-01
7.06017733e-01 -4.19021547e-01 1.26976526e+00 -6.15420938e-01
-1.32227898e+00 7.35929310e-01 8.43220875e-02 -3.06512058e-01
3.93032491e-01 -4.08699989e-01 -7.70340621e-01 1.78834766e-01
4.94051687e-02 4.06316184e-02 9.02710795e-01 -1.53237593e+00
-6.67833388e-01 -4.41952884e-01 4.27810336e-03 -3.30805004e-01
-3.14987481e-01 -1.70151427e-01 -4.51304764e-01 -6.51223898e-01
6.39312267e-01 -6.21445179e-01 -3.52479875e-01 2.04736322e-01
-8.17366481e-01 -2.75526613e-01 1.12636268e+00 -6.44834042e-01
1.00884664e+00 -2.44193196e+00 -3.87992114e-01 5.81474423e-01
2.28569806e-01 2.74180293e-01 -6.09784015e-03 2.33475998e-01
6.07475685e-03 -2.37189885e-02 -6.88915193e-01 -2.71438807e-01
2.12652802e-01 2.23007768e-01 4.97629307e-02 5.01843095e-01
6.96180761e-01 6.52451456e-01 -8.44049633e-01 -3.08248490e-01
9.16794837e-01 -8.09958577e-03 -4.17975336e-01 1.98253110e-01
-3.62557411e-01 4.40035373e-01 -6.10007226e-01 1.38929141e+00
8.98278534e-01 -1.97639614e-01 1.16723776e-01 -2.87363172e-01
6.03901893e-02 -1.91305786e-01 -1.12253571e+00 1.52671540e+00
-3.12796921e-01 2.40996510e-01 2.17454463e-01 -1.03042960e+00
1.24495423e+00 1.13829933e-01 7.82917678e-01 -7.20215440e-01
4.97836340e-03 2.59420514e-01 -1.24328189e-01 -6.61597371e-01
2.95624524e-01 2.52998173e-01 -1.26213759e-01 1.70635715e-01
-9.75009948e-02 -4.19681638e-01 3.36623251e-01 -1.50003955e-01
1.08993208e+00 7.90242627e-02 5.68348654e-02 -2.02168196e-01
6.18776798e-01 1.49188548e-01 6.19274318e-01 5.45735657e-01
-3.14561188e-01 6.74755514e-01 5.67645490e-01 -3.63060385e-01
-8.41336548e-01 -1.50053120e+00 -1.76822320e-01 4.64314878e-01
5.19791126e-01 -2.22622961e-01 -7.02868223e-01 -8.67966831e-01
3.88419479e-01 8.00545037e-01 -3.43251467e-01 -4.62007612e-01
-2.65466154e-01 -8.03483248e-01 2.01083168e-01 4.77978051e-01
4.98330265e-01 -9.87740874e-01 -4.20235574e-01 5.44595569e-02
6.75191358e-02 -1.10292661e+00 7.34846815e-02 5.72073385e-02
-8.76907945e-01 -1.43745399e+00 9.97036323e-02 -6.16350353e-01
9.14367974e-01 2.13816196e-01 1.05234408e+00 9.26110521e-02
-8.14663708e-01 5.74001849e-01 -3.25384229e-01 -7.22317278e-01
-5.44183612e-01 -3.19878221e-01 9.23766196e-02 1.80170357e-01
6.29983604e-01 -4.30891812e-01 -7.18356371e-01 5.05252838e-01
-1.13044775e+00 -5.07001102e-01 6.14957809e-01 4.41806376e-01
9.73371804e-01 3.17624867e-01 4.10933375e-01 -7.09095359e-01
4.53516960e-01 -5.60193777e-01 -7.65816212e-01 9.23996568e-02
-1.04436111e+00 -3.67290556e-01 5.80145240e-01 -1.37399137e-02
-1.07183182e+00 1.35777602e-02 -2.56764948e-01 -4.84879643e-01
-8.55739713e-01 4.04883981e-01 -3.46427679e-01 1.66073576e-01
3.64696264e-01 -1.28273770e-01 2.22421721e-01 -8.55672419e-01
9.47120860e-02 5.88909507e-01 6.88846171e-01 -5.05447865e-01
1.09029686e+00 4.78894740e-01 -5.53534292e-02 -8.62199664e-01
-6.59978688e-01 -6.42253101e-01 -5.35760581e-01 -3.58403116e-01
8.59191298e-01 -7.39681482e-01 -6.72835827e-01 7.91463733e-01
-7.91640341e-01 -2.75764525e-01 -6.37131453e-01 2.72826076e-01
-2.60123253e-01 8.22386682e-01 -6.60003185e-01 -8.15836787e-01
-2.08934724e-01 -1.10826075e+00 1.22952437e+00 1.48808956e-02
-3.08960944e-01 -7.94857025e-01 -3.23894918e-01 5.29215991e-01
1.26008868e-01 5.24531603e-01 9.89411235e-01 -7.29081810e-01
-5.45923412e-01 -5.42549074e-01 -1.95674032e-01 7.94384062e-01
6.57670975e-01 4.62975532e-01 -1.00922918e+00 -2.38183394e-01
5.98175153e-02 1.18433878e-01 5.79137444e-01 3.06685716e-01
1.64236832e+00 1.41509414e-01 -2.61040121e-01 4.36512381e-01
1.31524837e+00 2.71010727e-01 5.76249361e-01 5.65819979e-01
8.16983044e-01 6.24592483e-01 1.14291310e+00 5.70575953e-01
-1.20958909e-02 4.78529811e-01 1.13132310e+00 -3.76245856e-01
7.06383400e-03 7.61081576e-02 2.97340006e-01 7.55917013e-01
4.26891670e-02 -3.06840628e-01 -9.26022828e-01 6.16728187e-01
-1.78703964e+00 -5.01101494e-01 -5.17901599e-01 2.05371141e+00
7.45470151e-02 3.86736363e-01 -1.20496690e-01 3.65196973e-01
5.99483848e-01 -1.79426391e-02 -7.67564952e-01 -5.77198207e-01
6.75392672e-02 -1.24159135e-01 2.40990728e-01 7.11993277e-02
-1.13934422e+00 5.35238564e-01 5.84787416e+00 6.77783370e-01
-8.36929560e-01 -5.93158714e-02 6.32629037e-01 8.44023451e-02
-1.58017412e-01 -3.53006840e-01 -3.19549978e-01 4.20306891e-01
6.05020344e-01 2.06856087e-01 -5.93334362e-02 1.06435168e+00
3.41267169e-01 -3.47907059e-02 -9.45484161e-01 7.52207279e-01
-8.36657733e-02 -7.50154018e-01 1.32602960e-01 1.10654756e-01
6.65279508e-01 -2.13622749e-01 -9.95202828e-03 -1.58069864e-01
1.49149433e-01 -7.57488608e-01 6.65634096e-01 3.37332934e-01
4.09658730e-01 -6.57759249e-01 9.76563096e-01 -2.24292204e-01
-1.18810058e+00 -2.96242654e-01 -1.49957359e-01 6.69972450e-02
3.01636100e-01 1.07704115e+00 -4.73784268e-01 1.06313515e+00
9.18478370e-01 7.24507868e-01 -7.56238759e-01 9.87921655e-01
-2.32307538e-01 7.05290258e-01 -2.74351984e-01 6.57020509e-01
1.16641343e-01 -5.31146646e-01 5.72298765e-01 1.01583147e+00
4.94313687e-01 -5.94029486e-01 4.74429756e-01 1.05960083e+00
3.22488904e-01 -1.30533859e-01 -7.08134115e-01 -6.34578541e-02
2.66045660e-01 1.38224697e+00 -9.93358433e-01 8.42149630e-02
-5.23677826e-01 1.09149516e+00 -3.15162063e-01 3.06876391e-01
-8.93581629e-01 -5.29930651e-01 1.33371890e+00 1.54919267e-01
3.19438845e-01 -2.60482639e-01 -6.07209086e-01 -7.47144639e-01
4.52822506e-01 -9.10582364e-01 4.55156803e-01 -5.74282646e-01
-1.71498203e+00 3.57907027e-01 -6.59716949e-02 -1.57628632e+00
1.25014499e-01 -1.25169873e+00 -7.95973182e-01 3.92287910e-01
-1.27921259e+00 -1.30446064e+00 -7.40916550e-01 3.45377862e-01
8.08930278e-01 -2.06531703e-01 6.35937095e-01 2.53590822e-01
-1.03994441e+00 4.00331318e-01 5.13968952e-02 4.37502824e-02
4.88380760e-01 -1.42886341e+00 5.76011896e-01 1.32573521e+00
-1.53888613e-01 3.12570363e-01 8.45175683e-01 -6.56596482e-01
-1.28562760e+00 -1.27322412e+00 -1.56145832e-02 -4.55899537e-01
8.17734540e-01 -2.33757794e-01 -1.16575634e+00 7.30001032e-01
-8.93768296e-02 1.41183704e-01 5.05487800e-01 -1.11544114e-02
-1.68535739e-01 -3.83718520e-01 -1.31107259e+00 4.87022072e-01
8.86733353e-01 -4.38807368e-01 -1.71607718e-01 4.83021289e-01
6.94501519e-01 -2.86176145e-01 -1.09253621e+00 7.09205389e-01
1.45140976e-01 -9.40027714e-01 1.01738751e+00 -2.13856399e-01
3.77804458e-01 -8.19504678e-01 6.26830012e-02 -1.12246346e+00
-3.43169600e-01 -3.23541641e-01 -3.35658997e-01 1.43279660e+00
2.87435859e-01 -1.00777793e+00 6.65621400e-01 3.70546639e-01
-7.76413560e-01 -6.91688359e-01 -5.76264322e-01 -1.10808289e+00
-3.41231883e-01 -8.14450264e-01 8.24952066e-01 8.30705106e-01
-2.18270585e-01 -2.43289724e-01 1.32983595e-01 9.93842959e-01
6.41935527e-01 1.72226697e-01 6.12718284e-01 -1.45530570e+00
-1.17211029e-01 -4.70675051e-01 -8.92895520e-01 -6.48191214e-01
-1.61616519e-01 -6.02687657e-01 2.60664314e-01 -1.55351663e+00
-2.47832686e-01 -4.21200484e-01 -6.94417894e-01 5.47987878e-01
-1.51230991e-01 4.51620907e-01 -1.95618078e-01 7.68815130e-02
-5.26139498e-01 3.90284836e-01 1.08986592e+00 -4.99186814e-01
-1.08560301e-01 5.36384881e-02 -5.19121587e-01 7.63099849e-01
1.30636466e+00 -1.78224713e-01 -3.12980413e-01 -3.29134703e-01
-1.03786021e-01 -7.09190547e-01 6.43706381e-01 -1.19511580e+00
-2.10639119e-01 -1.16126917e-01 1.55613631e-01 -5.79675436e-01
6.00320585e-02 -1.04028618e+00 8.49056840e-02 5.33362925e-01
2.73512304e-01 1.96200430e-01 4.48456168e-01 7.62273908e-01
-4.70472753e-01 -2.89295077e-01 4.92009193e-01 8.97634551e-02
-9.95418370e-01 3.60574752e-01 -5.87208390e-01 -4.96522993e-01
1.23496747e+00 -3.31166357e-01 -2.74204165e-01 6.38402104e-02
-5.32137871e-01 3.78630817e-01 8.55478942e-01 5.92264771e-01
5.32328427e-01 -1.05415952e+00 -3.53157252e-01 5.85929394e-01
7.20238268e-01 4.25287426e-01 5.01017094e-01 8.63422453e-01
-8.28205466e-01 4.13630828e-02 -1.53012827e-01 -9.94732678e-01
-1.12147355e+00 7.05341578e-01 3.32529575e-01 -1.47058800e-01
-7.72260487e-01 3.96841288e-01 2.29294196e-01 -5.72417915e-01
-9.81278718e-02 -5.73843718e-01 2.18186140e-01 -3.38510245e-01
1.86702013e-01 6.16744936e-01 5.12244523e-01 -2.41765991e-01
-3.18397492e-01 5.75828016e-01 5.18240295e-02 6.76771462e-01
1.10617745e+00 -2.49450937e-01 -2.68324375e-01 6.09374762e-01
6.89414561e-01 5.91068789e-02 -1.48997676e+00 1.52773067e-01
2.94405997e-01 -5.04900455e-01 -9.65354294e-02 -6.84019983e-01
-1.38013721e+00 7.09972203e-01 9.39617932e-01 7.18943477e-01
1.37849391e+00 1.48331001e-01 8.24852109e-01 3.96019906e-01
1.08572662e-01 -1.12172699e+00 2.00450733e-01 1.15011178e-01
5.84317863e-01 -1.50346875e+00 5.38962819e-02 -9.63474035e-01
-5.56261778e-01 1.02810872e+00 9.31200027e-01 4.42938954e-02
4.93522972e-01 2.43564621e-01 4.27091837e-01 -7.35701144e-01
-1.84144378e-01 -1.24527581e-01 2.58127481e-01 9.07887161e-01
1.23482116e-01 1.17368057e-01 3.58036995e-01 2.35327259e-01
2.20912814e-01 -4.39073563e-01 1.87479556e-01 9.65635061e-01
-4.09386873e-01 -9.59685266e-01 -3.17449927e-01 7.31399477e-01
-4.14010912e-01 3.16901833e-01 -7.04530552e-02 7.09778428e-01
1.97002351e-01 1.23248041e+00 3.16951871e-01 -3.65755945e-01
7.83210516e-01 8.02532583e-02 8.64327624e-02 -6.05567992e-01
-2.56210625e-01 -1.39214307e-01 -3.87843326e-03 -8.62051964e-01
-5.24156988e-02 -8.43379796e-01 -1.25620389e+00 -1.26695186e-01
-2.58793086e-01 -7.07576424e-02 5.97281039e-01 8.56989324e-01
5.25536001e-01 1.10839677e+00 6.45373344e-01 -6.55377984e-01
-2.81734169e-01 -8.04479957e-01 -9.46833134e-01 7.10037589e-01
2.61707008e-01 -7.79302001e-01 -4.01686698e-01 1.48843497e-01] | [7.535694122314453, 1.9798685312271118] |
0d28c572-6c93-4705-92d4-f0f985efd20d | learning-robust-speech-representation-with-an | 2104.03204 | null | https://arxiv.org/abs/2104.03204v1 | https://arxiv.org/pdf/2104.03204v1.pdf | Learning robust speech representation with an articulatory-regularized variational autoencoder | It is increasingly considered that human speech perception and production both rely on articulatory representations. In this paper, we investigate whether this type of representation could improve the performances of a deep generative model (here a variational autoencoder) trained to encode and decode acoustic speech features. First we develop an articulatory model able to associate articulatory parameters describing the jaw, tongue, lips and velum configurations with vocal tract shapes and spectral features. Then we incorporate these articulatory parameters into a variational autoencoder applied on spectral features by using a regularization technique that constraints part of the latent space to follow articulatory trajectories. We show that this articulatory constraint improves model training by decreasing time to convergence and reconstruction loss at convergence, and yields better performance in a speech denoising task. | ['Thomas Hueber', 'Jean-Luc Schwartz', 'Laurent Girin', 'Marc-Antoine Georges'] | 2021-04-07 | null | null | null | null | ['speech-denoising'] | ['speech'] | [-4.79588434e-02 2.25295633e-01 2.14076057e-01 -1.32436201e-01
-4.21112210e-01 -6.44082487e-01 8.57171416e-01 -3.74066144e-01
-1.97565094e-01 4.73638535e-01 5.35941482e-01 -9.15137008e-02
-1.17988713e-01 -6.58207059e-01 -8.17334890e-01 -8.95170867e-01
3.04648191e-01 3.89470220e-01 -2.17836484e-01 -2.95022111e-02
-3.42133194e-01 5.31139076e-01 -1.82232642e+00 1.26754805e-01
6.40992761e-01 7.40933061e-01 4.64371324e-01 9.05705214e-01
-5.50526194e-03 3.26657027e-01 -4.42052096e-01 -2.59044826e-01
3.90787572e-02 -5.50261557e-01 -3.72145414e-01 1.87389180e-01
2.80930176e-02 -2.02624559e-01 -2.15730071e-01 1.02935338e+00
4.28461283e-01 4.95220244e-01 1.19062603e+00 -7.32681155e-01
-9.69371200e-01 5.72233856e-01 9.56149772e-02 -2.21350770e-02
-5.43933138e-02 -2.51697190e-02 1.15099525e+00 -9.19767737e-01
6.49540186e-01 1.32756793e+00 6.13197386e-01 6.86815381e-01
-1.46847439e+00 -9.74760205e-02 -3.37245077e-01 9.00106952e-02
-1.11941838e+00 -8.64243031e-01 1.07964611e+00 -4.64341700e-01
1.05121732e+00 1.34590328e-01 4.95022982e-01 1.30783165e+00
7.42135048e-02 7.47197330e-01 5.97669244e-01 -6.55148923e-01
2.83660948e-01 6.42777234e-02 -2.37452209e-01 5.46000421e-01
-5.21983683e-01 5.11602640e-01 -2.68894136e-01 -1.51865780e-01
8.01328421e-01 -2.84445822e-01 -2.66898662e-01 -3.82010221e-01
-6.75174952e-01 1.30702710e+00 1.56787366e-01 5.74485123e-01
-6.90435052e-01 4.83320475e-01 2.41366237e-01 3.17499757e-01
6.61325157e-01 1.13764569e-01 -3.69347155e-01 -1.90901220e-01
-8.37499917e-01 -2.99950447e-02 8.16684365e-01 2.99283415e-01
6.82343245e-01 6.69452190e-01 6.17515948e-03 1.48958528e+00
8.52463782e-01 7.42828488e-01 6.67359114e-01 -1.17494178e+00
-1.36192575e-01 -1.09770730e-01 -3.02573651e-01 -5.81193864e-01
-2.29834870e-01 -2.45440185e-01 -6.27752006e-01 2.22043186e-01
1.37085915e-01 -2.44190902e-01 -1.12355125e+00 1.96161044e+00
1.57686770e-01 3.77523154e-01 2.02484712e-01 9.36557770e-01
6.16737068e-01 1.08581865e+00 1.31903803e-02 -4.69100058e-01
1.15435529e+00 -6.62824094e-01 -1.08690715e+00 -4.65949327e-02
6.79386407e-02 -1.08374035e+00 7.65257120e-01 3.76449436e-01
-1.38041794e+00 -9.87236381e-01 -8.77128184e-01 -2.27811649e-01
-1.98931247e-01 3.12106878e-01 4.22536701e-01 7.83817589e-01
-1.11151540e+00 8.06835711e-01 -1.14497316e+00 1.18946023e-01
-8.43928531e-02 2.69918352e-01 -3.06003988e-01 3.97422731e-01
-1.11580861e+00 8.90165687e-01 6.82056025e-02 6.15404062e-02
-1.09878969e+00 -6.19989872e-01 -1.08659589e+00 1.46821573e-01
-2.24160969e-01 -9.15862143e-01 1.33847594e+00 -7.61556447e-01
-2.32364035e+00 5.09435117e-01 -3.10544461e-01 -4.52951998e-01
1.88650891e-01 -1.65919125e-01 -3.30199957e-01 6.98640049e-02
-5.06499290e-01 5.53376496e-01 1.59042978e+00 -1.40450418e+00
4.25856262e-02 -1.84810147e-01 -6.35504484e-01 -7.42087737e-02
-2.21350297e-01 -2.74042934e-01 -1.78225011e-01 -7.84929693e-01
7.53592476e-02 -9.80373681e-01 -3.94450780e-03 -2.47520134e-01
-1.39199108e-01 -4.57946926e-01 7.23851144e-01 -7.53103852e-01
8.70518029e-01 -2.25514460e+00 8.91310751e-01 1.87247336e-01
-9.07771662e-02 4.90342766e-01 -2.36985624e-01 5.79778433e-01
-1.69099346e-01 -3.95998433e-02 -3.35202157e-01 -9.46046770e-01
3.12935352e-01 9.66133773e-01 -5.68768382e-01 5.04071653e-01
2.01528832e-01 8.02797794e-01 -5.26228845e-01 -5.56463860e-02
6.53137088e-01 1.29698265e+00 -9.21700060e-01 2.59651750e-01
-3.70302290e-01 3.07340235e-01 -5.80705702e-02 6.67051226e-02
6.12689495e-01 4.94268209e-01 9.13274139e-02 -1.20972119e-01
-1.83106586e-01 5.32425404e-01 -9.13658381e-01 1.69192326e+00
-9.43064451e-01 7.75637746e-01 5.70035100e-01 -1.19334245e+00
1.06838834e+00 7.58723974e-01 3.50767106e-01 -1.65384680e-01
3.98809582e-01 1.52938962e-01 1.33878469e-01 -5.58744848e-01
1.87141269e-01 -4.95372146e-01 3.13825697e-01 1.68988824e-01
6.34711802e-01 -5.73197365e-01 -1.64873213e-01 -2.85244226e-01
4.59287226e-01 1.76756069e-01 -4.22603339e-02 -1.18978292e-01
7.22560167e-01 -6.83368385e-01 1.61195889e-01 1.91223741e-01
1.51192307e-01 6.57700181e-01 1.95487827e-01 7.72677660e-02
-1.29445589e+00 -1.57209885e+00 -4.20251966e-01 1.22216487e+00
-6.73135579e-01 -2.23342717e-01 -8.08248878e-01 6.02527261e-02
6.09615929e-02 1.05059564e+00 -6.46007776e-01 -3.05418640e-01
-5.49126625e-01 -3.90721619e-01 3.41805220e-01 4.27629292e-01
-4.71510202e-01 -1.38075161e+00 -2.14247972e-01 3.97177935e-01
-5.74528724e-02 -9.36992109e-01 -3.19237351e-01 5.24256229e-01
-8.64136815e-01 -4.25454408e-01 -8.26249421e-01 -8.05575371e-01
2.42311493e-01 -4.86858040e-01 8.16738665e-01 -3.33201885e-01
-2.10540473e-01 5.11743486e-01 -3.47055703e-01 -4.32744831e-01
-1.16341317e+00 -2.03155577e-01 3.27198952e-01 8.48500952e-02
-1.58601757e-02 -1.03394711e+00 -2.85668999e-01 -2.77491175e-02
-1.03753293e+00 -4.86315340e-01 1.95015535e-01 9.04151499e-01
5.64232528e-01 -2.86580324e-01 5.66434681e-01 -3.09071153e-01
9.02156532e-01 -3.97334456e-01 -5.49388826e-01 -2.02181727e-01
-3.21278870e-01 4.06145662e-01 6.46204770e-01 -4.63186741e-01
-9.99789417e-01 2.24448413e-01 -9.18387711e-01 -9.79683876e-01
-2.52021044e-01 4.27917123e-01 -1.06816916e-02 3.71908307e-01
5.54100931e-01 4.11126584e-01 3.80489290e-01 -8.28678370e-01
8.48102570e-01 6.32203519e-01 7.40259588e-01 -4.54046935e-01
7.22753584e-01 3.20288420e-01 -1.12644389e-01 -1.31247520e+00
-3.44355881e-01 -4.11924630e-01 -7.16649115e-01 -5.41524701e-02
1.03548741e+00 -6.91633880e-01 -7.73817122e-01 1.71244770e-01
-1.34213555e+00 -1.20083965e-01 -6.60406768e-01 7.09284246e-01
-9.67476130e-01 4.66105253e-01 -6.75283074e-01 -1.15804839e+00
-3.42027038e-01 -1.22427380e+00 1.02920508e+00 -5.49102239e-02
-2.02276167e-02 -1.21108365e+00 5.56472361e-01 1.75390784e-02
4.63098437e-01 -1.10353902e-01 8.32857609e-01 -2.74657220e-01
-1.11464545e-01 1.34606495e-01 4.77624416e-01 9.59530234e-01
2.19237566e-01 4.42814946e-01 -1.42486966e+00 -1.03902511e-01
2.28734612e-01 -2.99427025e-02 1.24891818e+00 1.03430331e+00
8.37723255e-01 -2.17049584e-01 1.88843217e-02 8.84782553e-01
1.10253382e+00 2.12172627e-01 5.89978039e-01 -4.71466273e-01
3.88509780e-01 6.80065513e-01 -1.85995489e-01 4.08268154e-01
-1.78702772e-01 6.66806936e-01 5.15509665e-01 2.56743461e-01
-4.47826058e-01 -2.53742576e-01 7.84708798e-01 1.53936529e+00
-3.51602614e-01 -1.59465834e-01 -5.29052615e-01 8.33468437e-01
-1.57524896e+00 -8.63925099e-01 4.24571661e-03 1.74850714e+00
8.13604474e-01 -1.94180086e-01 3.42993364e-02 3.17133099e-01
3.98530275e-01 2.08260909e-01 -2.64273316e-01 -9.66471434e-01
5.13805896e-02 8.80047619e-01 -7.89134651e-02 9.45140541e-01
-8.31261992e-01 9.26868856e-01 6.83282280e+00 7.37177312e-01
-1.19051874e+00 3.31219971e-01 -1.55510455e-01 1.49469361e-01
-5.55262089e-01 -4.41612244e-01 -3.93248469e-01 2.92086601e-01
1.27129424e+00 1.43716425e-01 8.19818735e-01 8.66510332e-01
2.63946027e-01 5.38640499e-01 -1.04538500e+00 6.01795733e-01
-5.36372475e-02 -1.25102890e+00 -6.85736910e-02 1.27235666e-01
5.57638466e-01 1.40788481e-01 4.69084918e-01 2.56003797e-01
2.38139927e-01 -1.17144740e+00 6.53171957e-01 6.36116564e-01
6.13229573e-01 -8.11061621e-01 3.53264064e-01 4.41853583e-01
-9.32181478e-01 2.47250274e-02 -5.96954942e-01 1.09537557e-01
4.16296244e-01 3.32906127e-01 -1.13219905e+00 3.39204639e-01
2.41581723e-01 3.88930529e-01 1.25273183e-01 7.10492253e-01
-4.11875069e-01 8.98701310e-01 -4.12417382e-01 3.15930769e-02
2.24957079e-01 -4.74684685e-01 8.58673275e-01 1.09846389e+00
5.48596263e-01 -1.24247804e-01 -2.82720655e-01 1.07562888e+00
5.76976053e-02 1.65192746e-02 -5.94042957e-01 -1.46438420e-01
2.23575622e-01 9.48000193e-01 -2.05286533e-01 -8.10736790e-02
-7.85162002e-02 8.64161134e-01 -8.93433637e-04 4.86910820e-01
-6.91104889e-01 1.91309035e-01 1.16143167e+00 -9.92736444e-02
7.07783639e-01 -4.58625942e-01 1.20861597e-01 -9.44573283e-01
-4.07609671e-01 -4.63680536e-01 -9.49820727e-02 -8.36019218e-01
-1.17387199e+00 6.79775834e-01 -2.12629914e-01 -7.59475291e-01
-1.10038042e+00 -8.40411782e-01 -6.63306475e-01 1.09804404e+00
-1.49811018e+00 -1.04566932e+00 4.13248062e-01 6.51160657e-01
6.11799598e-01 -1.86945215e-01 1.15926516e+00 1.62714392e-01
-3.03626090e-01 3.80657524e-01 5.54425478e-01 -8.63824561e-02
3.99499744e-01 -1.38678885e+00 3.82219523e-01 5.24760067e-01
7.45841622e-01 7.68953204e-01 1.06845176e+00 -1.41002759e-01
-1.19332802e+00 -6.81184530e-01 7.08522677e-01 -4.01919216e-01
6.80114329e-01 -3.26588482e-01 -9.99309599e-01 6.16397560e-01
4.31311041e-01 6.70288950e-02 8.97906184e-01 2.64518976e-01
-1.41691059e-01 1.80181563e-01 -7.98981786e-01 1.78816453e-01
7.39533186e-01 -1.00039637e+00 -1.04603696e+00 1.45908609e-01
7.08911061e-01 -1.56994402e-01 -1.01407170e+00 1.25507757e-01
5.22878528e-01 -7.13997424e-01 1.32401145e+00 -5.26941001e-01
2.96016902e-01 -3.33908387e-02 -1.73420772e-01 -1.85086989e+00
-2.18928307e-01 -9.36049044e-01 -4.08101678e-01 1.13382351e+00
2.64136881e-01 -4.44094837e-01 4.62483406e-01 -2.16030836e-01
-6.53479576e-01 -2.94331670e-01 -1.31829381e+00 -5.63621461e-01
5.39692879e-01 -7.17537045e-01 1.71637431e-01 5.94976008e-01
-2.28404582e-01 2.99114168e-01 -4.99926448e-01 2.63495266e-01
3.26222241e-01 -5.73124960e-02 4.34916139e-01 -1.29260135e+00
-5.30988932e-01 -5.54249346e-01 -1.16697297e-01 -1.03849351e+00
4.52732027e-01 -1.01067758e+00 2.40929261e-01 -1.36911094e+00
-5.34585595e-01 -3.79808038e-03 -3.44834149e-01 2.21728712e-01
6.94194958e-02 2.34082080e-02 2.55151212e-01 1.01611475e-02
3.04343045e-01 9.63686228e-01 1.34357190e+00 8.44122097e-03
-3.48655641e-01 2.67363280e-01 4.11568731e-02 9.14607108e-01
5.07295251e-01 -4.48032796e-01 -3.89125645e-01 -3.95475477e-01
-1.03779398e-02 3.31309468e-01 3.33034933e-01 -7.30939984e-01
8.98130015e-02 2.40690783e-01 1.77213475e-01 -4.62895751e-01
9.76330638e-01 -7.41649151e-01 1.52965903e-01 5.35508931e-01
-2.65548140e-01 -3.50317150e-01 3.06228012e-01 4.83069152e-01
-4.14110601e-01 -5.83586633e-01 7.46682167e-01 1.87204912e-01
-2.89878786e-01 -1.11336760e-01 -8.70195746e-01 -4.65491116e-01
5.34413218e-01 1.66480154e-01 2.14094341e-01 -5.02968848e-01
-1.42011189e+00 -4.34541762e-01 -2.90636625e-02 5.09042442e-01
5.44648409e-01 -1.21473742e+00 -8.64550471e-01 7.40084291e-01
-4.70574945e-01 -2.69550830e-01 2.88041741e-01 5.67055881e-01
-3.13689053e-01 4.16671753e-01 -1.36270598e-01 -8.27695191e-01
-1.19992983e+00 5.60962915e-01 5.67787707e-01 7.61394650e-02
-3.76854867e-01 1.18558013e+00 7.08275512e-02 -6.07756674e-01
1.53289720e-01 -7.25010276e-01 -3.19647700e-01 3.12931031e-01
2.22065262e-02 1.97043687e-01 -3.58998515e-02 -9.43782687e-01
-2.95863539e-01 7.34982014e-01 4.84539300e-01 -4.30221170e-01
1.62298942e+00 -6.82838559e-02 1.33776456e-01 3.98016542e-01
1.46161294e+00 4.78743523e-01 -1.48642170e+00 1.11383973e-02
-4.06529039e-01 -1.36305159e-02 5.44941545e-01 -5.20080745e-01
-9.86760795e-01 1.28485024e+00 5.14118195e-01 6.52173758e-01
9.40730751e-01 1.68737963e-01 8.48484933e-01 9.17924345e-02
-3.47189367e-01 -1.00860739e+00 7.20013157e-02 6.87513471e-01
1.17725909e+00 -9.45298016e-01 -4.25371319e-01 -3.05564344e-01
-5.06509006e-01 1.30211365e+00 -2.73555845e-01 -3.44178438e-01
1.01642394e+00 3.98995787e-01 9.50651467e-02 -2.56463550e-02
-8.40755224e-01 -5.29918432e-01 7.62365580e-01 7.89054334e-01
4.61926162e-01 1.78862974e-01 -2.13490367e-01 5.31444550e-01
-5.49723089e-01 -3.50755990e-01 2.01043889e-01 1.25397652e-01
-5.05384922e-01 -1.38576949e+00 -3.42399895e-01 -1.32901773e-01
-5.04971981e-01 -1.04250655e-01 -1.59308791e-01 3.57418418e-01
3.80104691e-01 9.53450441e-01 3.86307687e-01 -1.37525350e-01
2.47311398e-01 4.96325374e-01 5.91098666e-01 -6.34528697e-01
-4.73913908e-01 7.86965430e-01 -4.23046537e-02 -1.47672400e-01
-3.82713467e-01 -9.46462989e-01 -1.24265552e+00 6.75062165e-02
-4.97444212e-01 2.20518395e-01 1.24102652e+00 9.46437716e-01
9.99363288e-02 9.17308569e-01 6.13101900e-01 -9.05343831e-01
-8.87616336e-01 -1.20885313e+00 -4.92506742e-01 3.15626174e-01
7.26643324e-01 -7.26253808e-01 -6.20129287e-01 5.41834116e-01] | [15.026107788085938, 6.338521957397461] |
e95baa59-e1c3-4a5e-8539-7e433c79fa5b | spoken-conversational-search-for-general | 1909.11980 | null | https://arxiv.org/abs/1909.11980v1 | https://arxiv.org/pdf/1909.11980v1.pdf | Spoken Conversational Search for General Knowledge | We present a spoken conversational question answering proof of concept that is able to answer questions about general knowledge from Wikidata. The dialogue component does not only orchestrate various components but also solve coreferences and ellipsis. | ['Frédéric Herledan', 'Géraldine Damnati', 'Olivier Le-Blouch', 'Martinho Dos-Santos', 'Lina M. Rojas-Barahona', 'Pascal Bellec', 'Benoit Besset', 'Johannes Heinecke', 'Jean Y. Lancien', 'Munshi Asadullah', 'Emmanuel Mory'] | 2019-09-26 | spoken-conversational-search-for-general-1 | https://aclanthology.org/W19-5914 | https://aclanthology.org/W19-5914.pdf | ws-2019-9 | ['conversational-search'] | ['natural-language-processing'] | [-2.84441054e-01 1.30680966e+00 3.98986489e-01 -2.98340559e-01
-5.77657461e-01 -8.10447216e-01 8.80274832e-01 5.00170767e-01
-2.22540170e-01 1.08338666e+00 5.69438279e-01 -6.16415262e-01
-7.24344611e-01 -8.10403466e-01 -2.20234748e-02 -2.91442615e-03
-3.34796086e-02 9.51573193e-01 8.34369063e-01 -1.19469380e+00
2.79525936e-01 -2.85898179e-01 -1.29039550e+00 7.60544598e-01
8.39975417e-01 5.05764484e-01 2.22665414e-01 1.34535003e+00
-1.11834812e+00 2.18995595e+00 -1.02765572e+00 -7.11567163e-01
-3.57917696e-01 -5.46366751e-01 -2.04258919e+00 -4.70742404e-01
5.33469021e-01 -4.38736707e-01 1.55720517e-01 1.00547588e+00
3.51213545e-01 9.76047218e-02 1.68412820e-01 -1.54848254e+00
-1.02960512e-01 1.17027891e+00 5.82611978e-01 4.80420202e-01
1.33354092e+00 -2.96714693e-01 1.38331163e+00 -2.04206035e-01
8.54020000e-01 1.42354763e+00 6.66917205e-01 1.18712735e+00
-4.57000911e-01 -2.19554797e-01 2.54062731e-02 6.53692842e-01
-8.02872598e-01 -4.49472845e-01 3.57357949e-01 -2.49431774e-01
1.34683239e+00 7.49468029e-01 3.59763026e-01 4.20248806e-01
-1.95613235e-01 5.77092946e-01 7.35896409e-01 -7.89585471e-01
1.41861275e-01 4.82767195e-01 7.33607590e-01 8.75812173e-01
-2.80988514e-01 -7.80700505e-01 -7.26858079e-01 -6.41990006e-01
4.58740801e-01 -8.92322302e-01 -5.22306085e-01 8.04127567e-03
-8.56257260e-01 8.41843247e-01 -4.62860197e-01 5.10880888e-01
-3.43268722e-01 -2.42786808e-03 5.89361370e-01 8.82417321e-01
-1.27552241e-01 1.17000556e+00 -9.82685924e-01 -6.98579729e-01
2.07271680e-01 7.00816035e-01 2.33362317e+00 7.47001827e-01
4.38178241e-01 -5.50879538e-01 -8.80656298e-03 6.43655837e-01
2.22520575e-01 5.13921142e-01 4.64545459e-01 -1.89035845e+00
3.26289594e-01 8.75456631e-01 5.84601343e-01 -8.91277671e-01
-6.28936708e-01 6.21311665e-01 2.67865181e-01 -2.15811044e-01
1.03310001e+00 -7.38590300e-01 4.20939773e-01 1.30548441e+00
8.20282817e-01 -2.56753087e-01 8.94155443e-01 7.76650906e-01
1.96859789e+00 3.90191704e-01 3.65380645e-02 -2.56976098e-01
1.87422085e+00 -8.96323979e-01 -1.24347079e+00 1.65191740e-01
9.53084767e-01 -5.50948977e-01 6.58636212e-01 3.86988521e-01
-1.24999201e+00 1.59244016e-01 -6.25139236e-01 -4.46382731e-01
-3.46723497e-01 -4.29484397e-01 8.39202881e-01 2.60741204e-01
-1.12438214e+00 2.62904372e-02 -3.76180820e-02 -6.21226430e-01
-3.45515102e-01 5.44439554e-02 -3.76708269e-01 3.10715139e-01
-1.62313724e+00 1.01378512e+00 4.76163536e-01 -1.56690165e-01
-1.96964249e-01 -7.36402869e-01 -8.43099535e-01 1.45735219e-01
9.06951427e-01 -5.49813449e-01 2.14462376e+00 -4.86008197e-01
-2.16754103e+00 9.08100367e-01 1.38198048e-01 -6.21268451e-01
6.58656806e-02 -2.84959316e-01 -4.15213674e-01 7.93972194e-01
2.91388065e-01 3.33897173e-01 -7.62641802e-02 -8.27287734e-01
-1.16464376e+00 -5.14076278e-02 1.01820111e+00 4.99332607e-01
6.32093623e-02 5.48177600e-01 -1.02805756e-01 2.22212732e-01
-2.49315143e-01 -1.08389027e-01 6.37012646e-02 -7.11009428e-02
-2.30845809e-01 -9.32491481e-01 5.27883351e-01 -7.32026637e-01
1.16039670e+00 -1.34357226e+00 6.08983561e-02 -2.59110808e-01
4.34577823e-01 3.41598123e-01 -7.82242566e-02 1.02031302e+00
1.42416090e-01 -1.57106519e-01 3.85729820e-01 5.16404152e-01
5.05481243e-01 5.49204886e-01 -3.62630099e-01 -4.74263906e-01
-1.81588963e-01 8.27465534e-01 -1.27758646e+00 -6.40135407e-01
2.56208122e-01 -3.58732224e-01 -5.24389923e-01 5.56904197e-01
-1.24240708e+00 -1.19134955e-01 -7.82305658e-01 2.62314469e-01
3.25291127e-01 -3.14318210e-01 7.49936223e-01 6.90838337e-01
-3.51634681e-01 1.11747670e+00 -1.12624872e+00 1.78828359e+00
-1.66595295e-01 6.25163138e-01 9.85989273e-01 -9.30166900e-01
7.07152665e-01 8.50946844e-01 -8.43250975e-02 -6.63661599e-01
1.15229771e-01 1.92222401e-01 3.43796648e-02 -1.43511510e+00
4.10257876e-01 -3.01127136e-01 -1.82222798e-01 1.03343999e+00
2.20664933e-01 -4.93014336e-01 5.43975607e-02 9.35856342e-01
1.40590072e+00 -4.59120758e-02 7.10828185e-01 -5.12701094e-01
1.22280741e+00 8.39417815e-01 4.19539474e-02 6.66417658e-01
-8.71135220e-02 -2.52588600e-01 1.06882131e+00 -5.01416266e-01
-6.30147755e-02 -8.28779638e-01 3.30935925e-01 1.17962849e+00
9.35847983e-02 -8.60548437e-01 -9.32237267e-01 -8.14174294e-01
-1.40246511e-01 9.77549851e-01 -3.17757934e-01 5.36153734e-01
-3.31901163e-01 4.13507551e-01 8.83931279e-01 -1.74299162e-02
6.03252888e-01 -1.43589222e+00 -6.65939987e-01 2.46485323e-01
-8.75282049e-01 -1.16145146e+00 1.57345444e-01 -1.88312717e-02
-3.50589633e-01 -2.12673259e+00 -1.24543525e-01 -9.88434732e-01
1.50326505e-01 1.56876475e-01 1.33075762e+00 7.22950041e-01
2.51837552e-01 1.19336712e+00 -6.45841002e-01 -6.63856685e-01
-8.40063334e-01 1.27118289e-01 -4.05002564e-01 -5.40358186e-01
8.45617115e-01 -6.11732960e-01 4.05919552e-02 1.99303672e-01
-4.36346889e-01 -1.97556719e-01 -5.21093965e-01 3.78900707e-01
-6.15335286e-01 -1.78034782e-01 7.71784604e-01 -1.08082509e+00
1.40536201e+00 -5.38261890e-01 -4.48693573e-01 8.26240778e-01
-1.69762984e-01 2.38232035e-02 5.68171203e-01 5.06258532e-02
-1.63393939e+00 -5.53808033e-01 -1.95896432e-01 7.41887629e-01
-4.54671949e-01 6.33939624e-01 -5.79271605e-03 9.84878838e-02
9.19171035e-01 -6.93661766e-03 4.96430606e-01 -5.16807497e-01
5.81989527e-01 8.85874867e-01 8.94238830e-01 -1.09623921e+00
4.44258571e-01 9.39181969e-02 -5.91404736e-01 -1.17249417e+00
-8.03618968e-01 -7.94065654e-01 -2.19618261e-01 -4.96195555e-01
8.47919941e-01 -7.24336863e-01 -1.91124845e+00 3.99191007e-02
-1.56867182e+00 -7.21615076e-01 -5.45660377e-01 -1.12917749e-02
-6.46186531e-01 5.96372545e-01 -7.41104603e-01 -9.51088607e-01
-5.74159980e-01 1.39453590e-01 3.98462951e-01 4.10680026e-01
-8.33738387e-01 -1.15942419e+00 5.51424444e-01 7.21117854e-01
4.55757767e-01 -2.51207173e-01 9.13281739e-01 -1.34541178e+00
-2.61256639e-02 -4.95913550e-02 -1.71641722e-01 -3.56624685e-02
-1.28937103e-02 -3.53487104e-01 -5.65768719e-01 5.06806731e-01
1.87629201e-02 -7.43650377e-01 -3.32234830e-01 -2.75531977e-01
-6.46587536e-02 -8.89630795e-01 7.29580745e-02 -5.24168849e-01
6.75650060e-01 2.51523200e-02 6.92729354e-01 5.53573310e-01
-1.42364204e-01 1.43814242e+00 6.36948884e-01 1.95698991e-01
1.18601286e+00 5.16750038e-01 1.25684530e-01 6.38635397e-01
1.20950108e-02 -2.09171414e-01 8.62246379e-02 5.43098450e-01
1.49044663e-01 6.61568716e-02 -1.10532737e+00 5.82793415e-01
-2.34101939e+00 -1.20800269e+00 -7.46607661e-01 1.00858259e+00
1.29938173e+00 -5.66658914e-01 7.69223198e-02 -7.59231439e-03
4.37377751e-01 -1.31924823e-01 4.44477499e-02 -7.51463711e-01
4.40819236e-03 3.71772438e-01 -4.88043278e-01 1.29043567e+00
-3.61823052e-01 1.13632274e+00 7.19812298e+00 2.41604239e-01
-3.06382567e-01 6.20954148e-02 -6.00247562e-01 3.53681654e-01
-3.79931867e-01 2.38164052e-01 -3.33957583e-01 -2.70632535e-01
8.00222993e-01 -6.05953574e-01 5.40964901e-01 6.73358858e-01
-2.56466419e-01 -7.27836967e-01 -9.20440316e-01 2.61857927e-01
2.10440800e-01 -1.79946828e+00 3.67581584e-02 -6.71875656e-01
-3.16743143e-02 -2.95811087e-01 -1.17225516e+00 6.06537044e-01
8.00779819e-01 -6.02372885e-01 4.06572640e-01 6.18755281e-01
-1.27340073e-03 -2.96815395e-01 8.43845427e-01 9.02021170e-01
-5.39083600e-01 -1.98140979e-01 -7.62467831e-02 -9.08717453e-01
2.77603835e-01 -1.38527066e-01 -1.02347767e+00 5.01337528e-01
6.60386026e-01 -5.14040701e-02 -1.38178825e-01 7.50530601e-01
-7.97134042e-01 3.47472697e-01 -3.87539238e-01 -6.70031846e-01
-2.18903825e-01 -1.34106651e-01 7.18433857e-01 1.01930094e+00
-3.83480519e-01 1.33790851e+00 -4.78750058e-02 5.10000885e-01
2.13122353e-01 3.84659529e-01 -2.14190394e-01 9.36902016e-02
4.91652817e-01 1.08116412e+00 -1.58250168e-01 -5.52002549e-01
-2.74566889e-01 6.89673483e-01 3.35336506e-01 2.77896583e-01
2.00368818e-02 -7.09054351e-01 5.83884299e-01 1.76261470e-01
2.38303378e-01 -8.51802081e-02 5.41660070e-01 -1.06594121e+00
7.64597580e-02 -1.24843419e+00 8.99047375e-01 -1.17557049e+00
-1.03771091e+00 3.08864415e-01 -6.60311664e-03 -1.60877794e-01
-6.67464852e-01 -6.39505446e-01 -1.17305064e+00 6.47438824e-01
-1.41245818e+00 -6.35862768e-01 -3.74519587e-01 7.19169557e-01
1.61300063e-01 -7.64540723e-03 1.74816430e+00 -8.60827193e-02
1.59358624e-02 1.38895670e-02 -8.61209869e-01 1.77962348e-01
5.50983965e-01 -1.36193609e+00 -1.17577784e-01 7.33040497e-02
-2.04247862e-01 7.34830439e-01 1.00528073e+00 -2.88990110e-01
-1.39239931e+00 -1.04135998e-01 1.66362894e+00 -7.34120190e-01
1.21370935e+00 2.08560210e-02 -9.64101195e-01 5.50207257e-01
7.81975150e-01 -5.65791607e-01 9.58863199e-01 4.13419724e-01
-6.20217204e-01 9.56683233e-02 -1.14550829e+00 5.08454144e-01
7.36308038e-01 -8.93007696e-01 -1.73496175e+00 5.28574586e-01
9.33589756e-01 -5.01795173e-01 -8.92284453e-01 -1.96119681e-01
3.66439581e-01 -9.23988283e-01 5.90691090e-01 -1.23144150e+00
3.74178797e-01 -3.54877234e-01 -1.38695881e-01 -8.28642726e-01
5.60047507e-01 -1.36631346e+00 -3.92356873e-01 1.28961158e+00
8.36638570e-01 -1.04462707e+00 6.18384600e-01 1.47929931e+00
1.15214158e-02 1.84097439e-01 -1.12122679e+00 -2.77881265e-01
3.12911943e-02 -4.91761506e-01 2.98092633e-01 1.21563733e+00
1.78472519e+00 1.05270636e+00 1.96935609e-01 4.95618880e-02
2.00286329e-01 7.55819529e-02 9.79087830e-01 -1.43207383e+00
-2.65751123e-01 -2.43354306e-01 1.51318997e-01 -1.23595154e+00
2.71843463e-01 -7.87722290e-01 3.36551815e-01 -1.82475030e+00
-2.26206720e-01 -9.65750962e-02 7.89406061e-01 6.08510137e-01
1.56068638e-01 -6.93151116e-01 -1.11381426e-01 -1.71178296e-01
-1.35722637e+00 1.22020066e-01 1.21181667e+00 6.84146360e-02
-1.51509762e-01 -1.79824203e-01 -1.00453305e+00 7.04551756e-01
7.18276083e-01 -3.09316605e-01 -3.50268245e-01 -1.98195085e-01
9.87449825e-01 6.33922577e-01 2.96941310e-01 -4.73555803e-01
1.05393672e+00 -2.04824403e-01 -9.41517055e-01 1.28477901e-01
-2.30680883e-01 -5.40593565e-01 -3.96154404e-01 3.04622531e-01
-9.07187104e-01 -9.22426432e-02 4.65379387e-01 3.18016291e-01
-4.49478924e-01 -9.36482787e-01 -2.53541887e-01 -4.45425451e-01
-6.77064657e-01 -6.51004791e-01 -1.15194774e+00 5.94735444e-01
7.54067123e-01 1.62776843e-01 -8.39941621e-01 -1.12296665e+00
-8.10859025e-01 1.06761932e+00 -7.78029710e-02 3.93190801e-01
2.59402007e-01 -6.53970659e-01 -7.32637465e-01 -5.97826719e-01
2.76184618e-01 2.72686891e-02 8.71816650e-02 3.36800605e-01
-9.46285903e-01 7.38819957e-01 9.77009907e-02 3.70684601e-02
-1.40368319e+00 1.86021745e-01 6.68908298e-01 -4.27912958e-02
-6.88498437e-01 7.98372507e-01 -5.29173672e-01 -1.21765578e+00
6.29111826e-01 -1.47168115e-01 -8.67104828e-01 7.15612099e-02
1.37357295e+00 2.39617392e-01 -2.74133503e-01 1.12024341e-02
-4.08461481e-01 1.16676137e-01 1.57915443e-01 -4.14404154e-01
1.14965558e+00 -3.76966983e-01 -7.19253361e-01 3.62869352e-01
4.87949878e-01 -1.67195089e-02 -3.05895090e-01 -4.11679924e-01
4.94509161e-01 1.88999906e-01 -4.88307834e-01 -1.42772722e+00
-1.97665039e-02 2.81775296e-01 -5.14806926e-01 1.14253163e+00
1.97555155e-01 4.33394134e-01 7.78268158e-01 1.40108275e+00
3.32870871e-01 -1.23355234e+00 4.47155349e-02 9.91013646e-01
1.21384847e+00 -9.94083226e-01 -1.47013620e-01 -8.03610921e-01
-5.12920856e-01 1.48629475e+00 7.78113842e-01 3.42573643e-01
5.32726645e-01 4.64396358e-01 7.21208811e-01 -1.21803904e+00
-1.37553000e+00 -2.65095264e-01 -4.76918191e-01 1.01225543e+00
2.11837441e-01 -2.89134234e-01 -6.68860495e-01 9.46197033e-01
-3.79082471e-01 8.69504809e-02 8.22953582e-01 9.43187296e-01
-6.91622019e-01 -9.84013081e-01 -3.42417538e-01 -1.92403361e-01
-3.21620077e-01 -1.20276660e-01 -1.22665918e+00 9.57397640e-01
-4.28231031e-01 1.70880091e+00 1.20872870e-01 -1.21355638e-01
6.07616067e-01 3.25974405e-01 4.17325467e-01 -7.87783742e-01
-1.17564762e+00 -9.92048502e-01 1.42338479e+00 -4.37486947e-01
-6.37345910e-01 -4.64111537e-01 -2.03141093e+00 -4.48941618e-01
-5.31561911e-01 1.38292277e+00 3.83076221e-01 1.25058937e+00
1.28345847e-01 -3.87748368e-02 1.55386506e-02 1.68557853e-01
-4.89431143e-01 -1.03601527e+00 -1.38489515e-01 1.08044863e-01
2.19572023e-01 3.78538743e-02 -6.03219211e-01 -2.18768269e-01] | [12.441058158874512, 8.052445411682129] |
6f69d7b0-a232-489e-82f8-20e2905d79fc | scalable-learning-of-latent-language | 2305.20018 | null | https://arxiv.org/abs/2305.20018v1 | https://arxiv.org/pdf/2305.20018v1.pdf | Scalable Learning of Latent Language Structure With Logical Offline Cycle Consistency | We introduce Logical Offline Cycle Consistency Optimization (LOCCO), a scalable, semi-supervised method for training a neural semantic parser. Conceptually, LOCCO can be viewed as a form of self-learning where the semantic parser being trained is used to generate annotations for unlabeled text that are then used as new supervision. To increase the quality of annotations, our method utilizes a count-based prior over valid formal meaning representations and a cycle-consistency score produced by a neural text generation model as additional signals. Both the prior and semantic parser are updated in an alternate fashion from full passes over the training data, which can be seen as approximating the marginalization of latent structures through stochastic variational inference. The use of a count-based prior, frozen text generation model, and offline annotation process yields an approach with negligible complexity and latency increases as compared to conventional self-learning. As an added bonus, the annotations produced by LOCCO can be trivially repurposed to train a neural text generation model. We demonstrate the utility of LOCCO on the well-known WebNLG benchmark where we obtain an improvement of 2 points against a self-learning parser under equivalent conditions, an improvement of 1.3 points against the previous state-of-the-art parser, and competitive text generation performance in terms of BLEU score. | ['Alexander Gray', 'Salim Roukos', 'Pavan Kapanipathi', 'Subhajit Chaudhury', 'Tahira Naseem', 'Ramon Astudillo', 'Maxwell Crouse'] | 2023-05-31 | null | null | null | null | ['self-learning'] | ['natural-language-processing'] | [ 5.29803872e-01 9.91238236e-01 -1.75687209e-01 -4.50369388e-01
-1.27671814e+00 -7.88684249e-01 8.94500613e-01 2.35830605e-01
-3.72926742e-01 7.31175959e-01 3.76998663e-01 -2.55230278e-01
5.43591678e-01 -6.82475448e-01 -1.20685136e+00 -5.98376751e-01
3.79154891e-01 8.27911854e-01 3.00809860e-01 8.50183442e-02
-9.71928984e-02 -2.22857490e-01 -1.35793293e+00 2.32952744e-01
9.32604134e-01 7.49409616e-01 3.67020160e-01 6.94890678e-01
-5.65196872e-01 8.40927124e-01 -4.80740577e-01 -6.37415171e-01
4.46295142e-02 -6.86045527e-01 -1.04902554e+00 1.08531274e-01
3.40204775e-01 -1.90116763e-01 -8.28535110e-02 9.88321602e-01
1.44896790e-01 5.89464121e-02 3.85705203e-01 -9.90557671e-01
-4.57144886e-01 1.33326805e+00 -2.03500509e-01 -2.17728153e-01
1.88639700e-01 6.10866770e-02 1.37689507e+00 -7.03133583e-01
7.80186832e-01 1.34642959e+00 4.64132994e-01 8.64016771e-01
-1.54918873e+00 -4.73241001e-01 1.93740413e-01 -2.89269298e-01
-8.40237141e-01 -6.42965138e-01 5.75156510e-01 -5.97464517e-02
1.13618422e+00 -1.22550979e-01 3.81569952e-01 1.15713167e+00
5.48390709e-02 1.07297707e+00 7.87666321e-01 -7.62255549e-01
6.67243838e-01 7.64347091e-02 1.79287598e-01 1.07188880e+00
9.70533565e-02 -2.97914110e-02 -8.51806223e-01 -1.71160758e-01
3.36098075e-01 -4.92749006e-01 1.72913820e-01 -3.32351506e-01
-8.53477955e-01 8.83091927e-01 4.66084033e-02 1.00962734e-02
-7.81787783e-02 5.98445356e-01 4.84807283e-01 1.36029273e-01
6.86816454e-01 2.83769161e-01 -5.94777763e-01 -2.63406724e-01
-1.23149753e+00 3.91922146e-02 1.14283013e+00 1.13018537e+00
8.29371274e-01 1.56994671e-01 -2.59102076e-01 6.00707471e-01
4.85941350e-01 3.84998888e-01 8.03750455e-01 -1.22650993e+00
5.79415143e-01 3.94839078e-01 -4.93198596e-02 1.52983181e-02
1.04926117e-02 -3.20173562e-01 -2.63250947e-01 -9.57059711e-02
4.16485876e-01 -2.99154520e-01 -1.07065868e+00 2.08343029e+00
3.51756483e-01 1.95293456e-01 2.95789659e-01 2.41776839e-01
3.21838319e-01 7.67880023e-01 1.77237242e-01 -3.39941889e-01
1.39455760e+00 -1.26010597e+00 -6.19678378e-01 -4.92093533e-01
9.38946605e-01 -4.55856651e-01 1.11041629e+00 4.59398568e-01
-1.44417882e+00 -3.46402884e-01 -1.07941592e+00 -2.49893367e-01
-9.20769945e-02 7.55761564e-02 7.08308697e-01 7.41776049e-01
-1.33356917e+00 7.62296319e-01 -1.45708239e+00 -2.69581318e-01
4.33568597e-01 2.55896360e-01 -2.20154613e-01 3.90436500e-02
-8.89756680e-01 6.31233096e-01 8.00841808e-01 -2.94477820e-01
-1.14985466e+00 -6.14366174e-01 -1.09201431e+00 3.30478847e-01
6.76604509e-01 -6.71088934e-01 1.74415612e+00 -1.08105433e+00
-1.90070224e+00 7.53513694e-01 -6.78573370e-01 -7.02005208e-01
4.22668219e-01 -2.14243740e-01 3.98140401e-02 1.31417871e-01
3.50745022e-01 7.58005977e-01 8.44215512e-01 -1.11354113e+00
-7.14959800e-01 -1.55302450e-01 -1.23370280e-02 7.14129135e-02
-1.77242875e-01 -1.12442113e-01 -7.33140767e-01 -4.72723067e-01
2.36690983e-01 -1.01668704e+00 -2.00660065e-01 -3.36858988e-01
-5.10927916e-01 -4.89053786e-01 5.67831218e-01 -7.73278117e-01
1.02585018e+00 -1.91226900e+00 1.89667474e-02 -1.12780713e-01
-2.92134285e-02 -1.95042826e-02 -1.48839712e-01 3.25164706e-01
1.14487052e-01 3.30385536e-01 -6.30997658e-01 -1.10145295e+00
2.31810078e-01 3.08466315e-01 -4.46701527e-01 -9.72258002e-02
5.06594241e-01 1.04026377e+00 -1.27813113e+00 -5.64292192e-01
-1.30932570e-01 1.81467205e-01 -7.00288832e-01 2.36035094e-01
-9.17410314e-01 1.35354057e-01 -3.28028113e-01 2.84487963e-01
3.15898001e-01 -4.85209912e-01 6.53635085e-01 1.55661702e-01
1.65600866e-01 6.02044165e-01 -9.23705935e-01 2.19608259e+00
-7.83407390e-01 4.20025200e-01 -1.85634300e-01 -8.69867265e-01
6.31241560e-01 5.17237782e-01 4.75708917e-02 -3.16809118e-01
-1.21206656e-01 2.32956544e-01 -5.26450872e-01 -1.00606643e-01
5.81016183e-01 -1.85425997e-01 -2.51144826e-01 8.88605058e-01
7.20325708e-01 -2.06444919e-01 4.93056089e-01 5.56508422e-01
1.26932251e+00 8.53381276e-01 1.76666126e-01 -9.67994109e-02
2.65885085e-01 9.67423394e-02 5.30348718e-01 7.82670915e-01
2.27800012e-01 3.23066205e-01 7.89451122e-01 -4.64860685e-02
-1.15894091e+00 -1.20789242e+00 2.18972728e-01 1.29005206e+00
-4.19742286e-01 -7.24565446e-01 -1.24002635e+00 -1.12967038e+00
-2.88147688e-01 1.06924832e+00 -4.17904735e-01 -9.68223065e-02
-6.87929213e-01 -6.81621790e-01 6.20214760e-01 8.24483275e-01
3.25969696e-01 -1.06238043e+00 -4.67478573e-01 4.76581246e-01
-1.56335771e-01 -1.27033937e+00 -4.28647429e-01 5.80087602e-01
-1.16675591e+00 -6.54226482e-01 -2.57359266e-01 -7.98109829e-01
7.76426911e-01 -3.33272517e-01 1.24595404e+00 -2.37770334e-01
6.60070553e-02 1.33346930e-01 -1.83248580e-01 -2.28809461e-01
-1.07744753e+00 2.45119959e-01 -2.11421713e-01 -1.71661317e-01
5.14128096e-02 -6.03354454e-01 -2.43301138e-01 -3.81297201e-01
-9.88681912e-01 2.62653857e-01 4.91580039e-01 1.11451745e+00
5.74893355e-01 -1.53696284e-01 4.58903104e-01 -1.36040723e+00
2.64737368e-01 -4.68112051e-01 -9.51444626e-01 2.95904428e-01
-8.64545941e-01 9.00121748e-01 8.06553364e-01 -1.07478790e-01
-1.62414968e+00 5.07645488e-01 -2.21874509e-02 -1.04128353e-01
-3.90789285e-02 4.69907463e-01 -2.60670662e-01 5.38229644e-01
5.64409733e-01 2.40937755e-01 -8.31169412e-02 -4.20670211e-01
8.39860916e-01 5.55542350e-01 7.10988045e-01 -8.14183295e-01
7.40110397e-01 4.26474124e-01 -7.92819262e-02 -3.73272419e-01
-1.35271680e+00 -1.72491506e-01 -5.63311577e-01 2.98605710e-01
9.47164536e-01 -1.05532157e+00 -3.64090651e-01 1.96812153e-01
-1.29221296e+00 -7.09190130e-01 -5.92535615e-01 1.23245185e-02
-7.01038957e-01 4.82013524e-01 -6.41592741e-01 -9.13170576e-01
-5.36110163e-01 -8.57949793e-01 1.23650765e+00 1.54866830e-01
-1.93330482e-01 -1.32077932e+00 5.50961718e-02 5.28700829e-01
1.62130982e-01 4.01794538e-02 9.75184679e-01 -8.83780420e-01
-6.65111899e-01 -1.01100877e-01 2.49660239e-02 4.92695212e-01
-4.54000756e-02 -2.69650042e-01 -1.26531637e+00 -2.09002867e-01
1.45109312e-03 -7.10159361e-01 9.51506615e-01 5.30448332e-02
8.94707918e-01 -4.26721811e-01 -3.66502970e-01 4.37958062e-01
1.45648026e+00 -9.61879492e-02 2.37768292e-01 -6.99138716e-02
6.44838512e-01 3.62198830e-01 2.35335529e-01 2.47771502e-01
5.07445753e-01 3.44998658e-01 1.80327594e-01 2.78097302e-01
-1.35352969e-01 -8.84872556e-01 6.17911816e-01 9.71176982e-01
3.42372715e-01 -3.72673720e-01 -8.18575442e-01 5.29213548e-01
-1.96405625e+00 -8.02294552e-01 1.68300897e-01 2.19581938e+00
1.25080550e+00 3.15946162e-01 -1.37039438e-01 -4.48666140e-02
5.24926841e-01 7.15745986e-02 -6.05911732e-01 -2.93739051e-01
3.83363068e-02 7.15267241e-01 5.60509324e-01 7.83850193e-01
-9.16077256e-01 1.45861983e+00 6.41381550e+00 6.72259867e-01
-7.68434525e-01 5.70109904e-01 6.44824743e-01 -1.97630227e-01
-4.99334395e-01 5.66045225e-01 -9.09343302e-01 4.32053089e-01
1.54693079e+00 -7.46306628e-02 5.00192225e-01 8.93560946e-01
1.59055680e-01 -1.87101394e-01 -1.32579434e+00 4.61190879e-01
1.75384700e-01 -1.38747919e+00 -3.96042280e-02 -1.17972128e-01
8.87399733e-01 1.90759659e-01 -4.00456399e-01 4.97573048e-01
1.08003104e+00 -6.10991180e-01 1.00284922e+00 2.49153376e-01
7.66870141e-01 -5.31969249e-01 6.00271404e-01 4.50771779e-01
-1.01487088e+00 5.05608432e-02 -2.62199759e-01 1.11162126e-01
3.71414602e-01 5.45379996e-01 -9.92276371e-01 6.20913208e-01
3.40489268e-01 3.37066859e-01 -3.42948645e-01 4.18051273e-01
-7.60797560e-01 1.24198413e+00 -3.55995655e-01 -6.24107104e-03
2.73442358e-01 8.97398219e-02 2.68700451e-01 1.14836860e+00
1.21434763e-01 -2.41259411e-01 1.83253750e-01 1.16573584e+00
-4.02023554e-01 3.10128536e-02 -8.99071544e-02 -3.80944014e-01
5.28441250e-01 1.21726823e+00 -8.02058101e-01 -6.66458964e-01
-2.39114866e-01 1.38288283e+00 5.66189110e-01 2.66914308e-01
-6.51300728e-01 -1.40955865e-01 -7.28038177e-02 -1.74527690e-01
5.24932086e-01 2.72028372e-02 -3.71430486e-01 -1.23861706e+00
1.68154344e-01 -5.33543885e-01 3.84620160e-01 -6.79454982e-01
-9.87081826e-01 4.03941602e-01 -4.03149128e-02 -5.67355752e-01
-7.26629674e-01 -4.49857354e-01 -6.69944167e-01 8.84551525e-01
-1.40488684e+00 -1.35276699e+00 -5.72856814e-02 1.00419223e-01
7.38616109e-01 -6.00994602e-02 1.04209328e+00 -4.07823086e-01
-4.94932920e-01 7.05517173e-01 1.05875373e-01 7.00749038e-03
4.74538743e-01 -1.71149576e+00 8.49368989e-01 1.18273354e+00
4.73719120e-01 5.02116323e-01 5.69752872e-01 -6.95012033e-01
-1.41141403e+00 -1.22836566e+00 1.23045838e+00 -5.69430590e-01
7.88658500e-01 -7.22969532e-01 -7.52682209e-01 1.07731378e+00
3.42805445e-01 -4.32649702e-02 5.02948761e-01 1.76594973e-01
-4.17862922e-01 7.61769339e-02 -8.80878985e-01 4.51948643e-01
1.09803450e+00 -5.41955888e-01 -7.98721194e-01 4.39915955e-01
1.11102629e+00 -5.17145216e-01 -6.57893181e-01 -2.75161237e-01
3.79375905e-01 -6.31357729e-01 2.76027381e-01 -4.69562620e-01
5.76791108e-01 -8.31826404e-02 -7.82591850e-03 -1.15879285e+00
6.40886277e-02 -1.22096944e+00 -4.06293511e-01 1.49124122e+00
6.62707329e-01 -4.79240417e-01 1.15988159e+00 5.76063514e-01
-3.44738334e-01 -4.53750849e-01 -8.94705117e-01 -7.35021412e-01
1.06035374e-01 -6.41560614e-01 4.73574102e-01 5.69087684e-01
2.72047997e-01 7.23300815e-01 -4.54899156e-03 6.51403666e-02
8.54601443e-01 -1.96710378e-01 5.94117939e-01 -1.15970504e+00
-7.81133115e-01 -5.69324531e-02 1.92251042e-01 -1.18720639e+00
6.83357835e-01 -1.40066922e+00 4.16235566e-01 -1.55296123e+00
2.98121721e-01 -3.09066385e-01 -9.59483460e-02 7.78951705e-01
-2.82488883e-01 -1.72149315e-01 2.10543245e-01 6.01665601e-02
-6.38806760e-01 4.39180493e-01 6.86062872e-01 4.21482623e-02
-2.36590698e-01 -1.67000532e-01 -7.61654258e-01 6.72120392e-01
6.26800001e-01 -7.12669790e-01 -6.50874197e-01 -5.22104383e-01
3.82723778e-01 2.71088064e-01 1.37678102e-01 -8.08179915e-01
3.51065159e-01 2.17240199e-01 1.62816644e-02 -3.21512878e-01
1.90427631e-01 -2.75564492e-01 -1.20474674e-01 3.37715089e-01
-5.66334486e-01 -2.26140201e-01 1.52703539e-01 6.58405423e-01
-6.49697930e-02 -6.32313609e-01 4.77209449e-01 -4.31941062e-01
-3.77406269e-01 1.87254865e-02 -2.17739537e-01 3.47918421e-01
5.38327217e-01 1.70094930e-02 -8.66001621e-02 -2.78082341e-01
-8.02500665e-01 1.94927275e-01 4.35715407e-01 1.94029868e-01
2.11542100e-01 -9.51550126e-01 -4.94467944e-01 1.92051440e-01
-5.43006659e-02 4.16445941e-01 -8.96107629e-02 3.56389880e-01
-1.98174134e-01 4.86089408e-01 3.45655680e-01 -5.12070775e-01
-8.27686191e-01 1.51635110e-01 1.20392330e-01 -8.00802648e-01
-4.59759325e-01 8.82441223e-01 6.04137853e-02 -4.88921821e-01
1.22581311e-01 -4.52420503e-01 4.93856728e-01 -2.31625050e-01
3.25973094e-01 1.00932285e-01 9.14278775e-02 4.20087390e-03
-3.34665813e-02 -1.28965527e-01 -4.84617352e-02 -7.65174270e-01
1.27317905e+00 4.90510315e-02 -7.06853271e-02 5.84375203e-01
8.65735710e-01 1.88152899e-03 -1.63738441e+00 -3.87391806e-01
3.71024758e-01 2.71586090e-01 1.14558659e-01 -1.01101983e+00
-6.92799330e-01 6.73383832e-01 2.03148171e-01 7.85782561e-03
8.34874213e-01 3.16450536e-01 8.63092124e-01 4.72316623e-01
3.31356168e-01 -1.33500004e+00 2.39744455e-01 6.51844323e-01
3.99259627e-01 -1.10942364e+00 -4.22834873e-01 -3.59354705e-01
-6.65387392e-01 9.83718157e-01 3.25947553e-01 -6.62164539e-02
4.03948247e-01 6.49991632e-01 -1.50110587e-01 1.59999639e-01
-1.15151787e+00 3.81003655e-02 -8.88156742e-02 3.40755999e-01
5.86809099e-01 -6.33786321e-02 -7.12031424e-02 7.92900681e-01
-4.32819009e-01 6.46261573e-02 3.70972365e-01 9.65710580e-01
-4.83281344e-01 -1.48535359e+00 7.49648735e-02 3.08669508e-01
-5.21357894e-01 -4.44035351e-01 -6.34732991e-02 4.09683943e-01
-1.08882569e-01 9.02405918e-01 2.21316099e-01 3.04213539e-02
-4.29629907e-02 7.90591598e-01 4.43429500e-01 -1.03631103e+00
-5.12314856e-01 6.98171034e-02 3.75134617e-01 -5.99564731e-01
-2.12223694e-01 -7.05724537e-01 -1.74320292e+00 2.28813499e-01
-4.20432746e-01 3.25248986e-01 8.73579383e-01 1.28829432e+00
3.56016517e-01 4.24279213e-01 4.69549507e-01 -5.83665907e-01
-8.76820862e-01 -1.09733236e+00 -1.55107215e-01 2.85008043e-01
-4.79628220e-02 -2.07882047e-01 -3.68408829e-01 6.40280485e-01] | [10.576879501342773, 9.288956642150879] |
9ee07213-b15a-4c5a-99e8-a71909a9060f | learning-to-fix-build-errors-with-graph2diff | 1911.01205 | null | https://arxiv.org/abs/1911.01205v1 | https://arxiv.org/pdf/1911.01205v1.pdf | Learning to Fix Build Errors with Graph2Diff Neural Networks | Professional software developers spend a significant amount of time fixing builds, but this has received little attention as a problem in automatic program repair. We present a new deep learning architecture, called Graph2Diff, for automatically localizing and fixing build errors. We represent source code, build configuration files, and compiler diagnostic messages as a graph, and then use a Graph Neural Network model to predict a diff. A diff specifies how to modify the code's abstract syntax tree, represented in the neural network as a sequence of tokens and of pointers to code locations. Our network is an instance of a more general abstraction that we call Graph2Tocopo, which is potentially useful in any development tool for predicting source code changes. We evaluate the model on a dataset of over 500k real build errors and their resolutions from professional developers. Compared to the approach of DeepDelta (Mesbah et al., 2019), our approach tackles the harder task of predicting a more precise diff but still achieves over double the accuracy. | ['Pierre-Antoine Manzagol', 'Zimin Chen', 'Subhodeep Moitra', 'Daniel Tarlow', 'Charles Sutton', 'Edward Aftandilian', 'Andrew Rice'] | 2019-11-04 | null | null | null | null | ['program-repair', 'program-repair'] | ['computer-code', 'reasoning'] | [-9.91768483e-03 3.93523425e-01 -3.04303449e-02 -4.17580187e-01
-6.81346953e-01 -6.52781785e-01 3.65460999e-02 6.15676582e-01
2.21655712e-01 2.31059358e-01 2.69770380e-02 -9.04888690e-01
1.85379669e-01 -8.03710759e-01 -1.09334099e+00 1.50242567e-01
-9.30727422e-02 3.87484320e-02 2.83502758e-01 -2.09001824e-01
7.38643885e-01 -4.57119346e-02 -1.31029296e+00 5.66030741e-01
7.45094955e-01 4.04887110e-01 9.50183645e-02 8.65587771e-01
-4.36156690e-01 1.27788329e+00 -9.70420361e-01 -7.69882202e-01
-2.45534349e-02 -4.33329016e-01 -1.20987558e+00 -5.14972746e-01
7.91586459e-01 -7.47530386e-02 1.64666593e-01 1.37010002e+00
4.44430888e-01 -4.17373478e-01 2.82909513e-01 -1.03918087e+00
-1.03126514e+00 1.42977250e+00 -5.36809385e-01 3.58825207e-01
5.22604585e-01 5.97009435e-02 1.12266266e+00 -6.37258053e-01
6.81734443e-01 8.98379743e-01 1.55120802e+00 3.38283062e-01
-1.25433958e+00 -1.26085147e-01 -7.81139880e-02 3.44233334e-01
-1.19483852e+00 -1.52178735e-01 5.92157543e-01 -1.09561396e+00
1.65300322e+00 1.59393072e-01 4.68456566e-01 7.67687023e-01
5.94428897e-01 3.58766526e-01 3.46019983e-01 -7.17374444e-01
2.61553228e-01 -2.42273495e-01 6.61020100e-01 1.26803398e+00
2.07447052e-01 -1.86114997e-01 -4.95068319e-02 -3.18860829e-01
3.71164650e-01 -1.59286156e-01 -3.14886421e-01 -7.75634870e-02
-9.59324718e-01 7.20071912e-01 4.48772132e-01 5.65415621e-01
-6.97548315e-02 9.25832689e-01 6.82623327e-01 5.67280471e-01
3.36656421e-01 7.84293056e-01 -6.71801567e-01 -5.79369187e-01
-9.52024281e-01 3.01862895e-01 1.10565794e+00 9.64279950e-01
1.04882693e+00 3.12955856e-01 6.99278852e-03 7.53494859e-01
2.11365178e-01 -1.21421553e-01 5.36756694e-01 -7.92166710e-01
7.20522106e-01 1.05159461e+00 -2.21129999e-01 -1.23018074e+00
-4.82378453e-01 -4.78119969e-01 -4.05347437e-01 6.12969041e-01
2.13963971e-01 -1.59193084e-01 -5.43588817e-01 1.34402645e+00
-1.52246699e-01 8.14899355e-02 -4.17946875e-01 2.69518793e-01
7.07334638e-01 3.35492909e-01 -3.79991949e-01 2.87793785e-01
8.33586812e-01 -9.82247353e-01 -4.05420184e-01 -3.51334453e-01
1.31849456e+00 -5.61865151e-01 9.35513198e-01 4.99189943e-01
-1.18990016e+00 -5.92210710e-01 -1.20204663e+00 -2.48669013e-01
-4.93365973e-01 2.42375061e-01 5.17066896e-01 7.36158192e-01
-1.67997289e+00 1.08855081e+00 -8.64497602e-01 -1.82126999e-01
1.89524144e-01 1.84114978e-01 -2.52864599e-01 1.85957029e-01
-7.13927925e-01 9.07991469e-01 3.52446824e-01 1.01119071e-01
-8.06706846e-01 -8.90557110e-01 -1.05349863e+00 3.64983886e-01
1.64006069e-01 -5.24748385e-01 1.61223531e+00 -9.72453415e-01
-1.06196523e+00 8.16033602e-01 1.94600567e-01 -5.47428668e-01
1.95122108e-01 -1.64270774e-01 -2.92134017e-01 -8.01963210e-01
-4.67440933e-02 1.31383343e-02 7.97947824e-01 -1.00586009e+00
-5.07040560e-01 -1.71700284e-01 4.30653304e-01 -6.77003026e-01
9.23764929e-02 4.15650040e-01 -1.84538573e-01 -5.63195109e-01
-2.06557468e-01 -7.54546046e-01 -3.05625737e-01 -4.00535196e-01
-7.30932653e-01 -3.62539113e-01 2.21338391e-01 -1.31794071e+00
1.95796192e+00 -1.95761204e+00 5.25765896e-01 6.67084232e-02
7.15011656e-01 2.53527254e-01 -2.56289333e-01 5.93763471e-01
-5.60582459e-01 5.91213703e-01 -5.93260586e-01 -2.70387471e-01
2.67742455e-01 7.14567257e-03 -5.30831404e-02 2.92528629e-01
1.54767394e-01 1.02035511e+00 -9.00729179e-01 1.01192065e-01
-3.27518374e-01 1.43282831e-01 -9.13541377e-01 1.23781107e-01
-6.28954291e-01 -4.30527776e-01 -1.55830696e-01 6.03099108e-01
4.03916895e-01 -3.18147928e-01 2.93469906e-01 1.43808782e-01
-3.01553994e-01 4.52745467e-01 -9.72970843e-01 1.81171763e+00
-6.03820920e-01 1.04001379e+00 -3.23248625e-01 -1.05856729e+00
1.08966339e+00 4.77532484e-02 -8.66882205e-02 -6.03746295e-01
-1.47841796e-01 6.48382157e-02 7.82601237e-02 -8.99318933e-01
7.68436968e-01 4.34474170e-01 -4.67593282e-01 7.72191048e-01
2.24029664e-02 -6.16059825e-02 3.11759204e-01 2.09818378e-01
1.81397843e+00 3.12646896e-01 4.10207868e-01 -1.12199567e-01
3.02062750e-01 -1.77735120e-01 4.57965434e-01 9.21844780e-01
3.70112777e-01 6.08919919e-01 1.24213374e+00 -8.26999962e-01
-9.68579829e-01 -6.51107311e-01 4.09345508e-01 1.18972957e+00
-6.49262249e-01 -1.10494256e+00 -1.25341797e+00 -1.04997361e+00
-3.40903969e-03 8.43375981e-01 -8.27688456e-01 -2.99333751e-01
-1.09972990e+00 -4.36533391e-01 9.33328748e-01 7.40745246e-01
-1.14648297e-01 -1.20161486e+00 -5.83448648e-01 4.33945864e-01
7.33678713e-02 -2.96854943e-01 -3.31066877e-01 4.34269279e-01
-4.95572597e-01 -1.41688502e+00 -1.54934242e-01 -9.26900625e-01
6.00872934e-01 -5.31498253e-01 1.77764189e+00 8.58613729e-01
-4.64423656e-01 1.98275492e-01 -5.07955968e-01 -2.76603967e-01
-1.16252685e+00 3.58259022e-01 -4.32270229e-01 -5.93092084e-01
2.51414359e-01 -6.00252211e-01 9.27307531e-02 -2.71163523e-01
-7.24258721e-01 -2.74538308e-01 1.83549345e-01 7.28472531e-01
9.78768691e-02 -5.97428158e-02 2.69858748e-01 -1.32141209e+00
8.72516394e-01 -7.09230006e-01 -1.05952621e+00 3.98006976e-01
-7.25772917e-01 3.83153349e-01 7.84861565e-01 2.08886564e-01
-7.07067609e-01 -7.82388672e-02 -5.82503080e-01 2.56928871e-03
-7.78780058e-02 1.13307571e+00 1.76571488e-01 -4.08008963e-01
1.15779293e+00 -9.55386832e-02 -3.29594404e-01 -9.81155694e-01
3.78675699e-01 4.96364027e-01 5.95638752e-01 -6.73393548e-01
7.48732865e-01 -5.47869325e-01 -3.06209952e-01 -1.75069675e-01
-4.15688843e-01 1.06579460e-01 -8.27756226e-01 8.37808847e-02
5.99215806e-01 -3.76493424e-01 -4.17532533e-01 5.69489300e-01
-1.65771592e+00 -6.59460306e-01 -2.01870486e-01 -3.14492017e-01
-4.45839673e-01 5.47615647e-01 -7.46409535e-01 -2.25486606e-01
-1.60325006e-01 -1.39156890e+00 8.00808311e-01 -2.24430874e-01
-3.43228430e-01 -1.12931275e+00 5.63673854e-01 -1.11108124e-01
5.79690456e-01 5.48937321e-01 1.64405787e+00 -4.97043997e-01
-8.14812005e-01 -1.22552738e-01 -7.33133554e-02 6.29181802e-01
5.79395108e-02 4.87373829e-01 -5.97331345e-01 -1.55301824e-01
-2.16826245e-01 9.14115906e-02 6.50790215e-01 1.01152167e-01
1.52093792e+00 -5.31054854e-01 -1.71597317e-01 7.30051756e-01
1.69475269e+00 1.77211016e-01 7.40768909e-01 4.38051939e-01
1.04151797e+00 1.65672466e-01 -1.30311288e-02 4.76518333e-01
5.77148914e-01 6.37475789e-01 9.48203444e-01 2.16856271e-01
-1.68527082e-01 -3.77275608e-02 3.62236023e-01 1.10503435e+00
-1.62625745e-01 -2.76567698e-01 -1.58430874e+00 6.99783385e-01
-1.73605537e+00 -8.92219365e-01 -5.71057200e-01 1.94753051e+00
9.49038029e-01 -4.52058874e-02 -7.25533813e-02 2.69695789e-01
7.06138790e-01 5.10529950e-02 -2.29484111e-01 -1.04696774e+00
5.08444250e-01 3.95653009e-01 2.39602163e-01 5.18401265e-01
-6.69855893e-01 6.73566461e-01 6.95218897e+00 2.71351457e-01
-1.01561713e+00 1.38572067e-01 1.92087218e-01 3.06492925e-01
-4.13467407e-01 1.66630909e-01 -5.68985224e-01 6.96395934e-01
1.34186482e+00 -1.29745290e-01 7.15006888e-01 1.14684618e+00
-3.21955889e-01 1.32661760e-01 -1.68584430e+00 6.64199769e-01
2.13885665e-01 -1.73011625e+00 -4.34727043e-01 -2.64195174e-01
6.18980944e-01 2.16087937e-01 -2.41337121e-01 5.25405705e-01
7.08006918e-01 -9.90191817e-01 9.62273419e-01 6.33423626e-01
7.38859594e-01 -6.07111633e-01 7.92614818e-01 2.51857996e-01
-8.02835822e-01 -2.54844785e-01 -5.19157529e-01 -3.33680362e-01
-1.50304809e-01 6.29372418e-01 -1.13427567e+00 4.42588121e-01
8.06346655e-01 1.05027902e+00 -1.23316967e+00 1.21301699e+00
-4.00431812e-01 7.23467052e-01 3.30478907e-01 4.72483225e-02
-1.25227809e-01 2.17764512e-01 1.29958600e-01 1.69993234e+00
5.75924397e-01 -5.50554335e-01 -2.09388241e-01 1.14421248e+00
-3.17141771e-01 -2.18210593e-01 -5.43987870e-01 -2.09423482e-01
3.79689336e-01 8.11311185e-01 -3.28941375e-01 -2.91967541e-01
-3.05352241e-01 6.62368178e-01 9.07869101e-01 2.86916614e-01
-8.10057819e-01 -8.47262740e-01 4.77225244e-01 -1.16475101e-03
5.01346469e-01 3.14791985e-02 -3.47698480e-01 -1.04034603e+00
3.54223967e-01 -1.15374959e+00 1.90915599e-01 -9.43867385e-01
-9.18021262e-01 6.32671595e-01 -2.36648843e-01 -9.61877406e-01
-6.08221531e-01 -5.55803478e-01 -1.01155877e+00 8.48249257e-01
-1.09966898e+00 -7.96080410e-01 -3.34157556e-01 -3.37657146e-02
3.17778409e-01 -2.07812369e-01 7.32160747e-01 4.16247696e-01
-7.19178438e-01 9.62478280e-01 5.95833473e-02 3.37173164e-01
2.57830828e-01 -1.79673541e+00 1.42649722e+00 1.22945058e+00
7.98392517e-04 9.67360914e-01 6.46046937e-01 -7.17312992e-01
-1.32510900e+00 -1.28596950e+00 1.10742784e+00 -8.67330313e-01
8.71004283e-01 -3.20177555e-01 -1.27002823e+00 1.23433423e+00
4.04293656e-01 -4.59093004e-02 3.78790438e-01 2.68265545e-01
-7.49434769e-01 2.24578127e-01 -8.81511927e-01 1.65642500e-01
1.21083665e+00 -6.16006613e-01 -7.23480999e-01 4.74483043e-01
9.19221699e-01 -7.87072718e-01 -9.92534995e-01 -1.53816476e-01
8.75000283e-02 -1.39964807e+00 7.08712697e-01 -6.67431831e-01
1.02376437e+00 -3.33011717e-01 4.28244025e-02 -1.69444680e+00
-6.48532033e-01 -5.80833316e-01 -2.55751103e-01 1.38836384e+00
4.81873840e-01 -4.80049223e-01 4.57077920e-01 3.06897253e-01
-9.40195441e-01 -6.63761497e-01 -6.18350506e-01 -5.33609390e-01
2.13115603e-01 -5.83322823e-01 9.93838489e-01 1.24939680e+00
1.91381887e-01 3.68204527e-03 -1.53774410e-01 1.55861065e-01
1.63256511e-01 1.05153739e-01 6.88856244e-01 -1.23901772e+00
-7.95513391e-01 -7.03715086e-01 -7.02111363e-01 -5.02805233e-01
3.95373106e-01 -1.28736711e+00 2.82331347e-01 -1.66251981e+00
-1.03126585e-01 -4.88534540e-01 -1.99531600e-01 9.50310946e-01
-4.29747254e-02 -1.12832241e-01 -1.03463247e-01 -2.12035820e-01
-4.78598356e-01 -2.16730848e-01 1.66543514e-01 -3.27745974e-01
4.76143733e-02 -3.87746058e-02 -7.53169715e-01 7.57367432e-01
5.82964897e-01 -8.78879309e-01 -6.16150734e-04 -1.19167435e+00
1.07443202e+00 3.42652112e-01 3.55090499e-01 -9.96389508e-01
1.31087735e-01 1.69317469e-01 -1.23414360e-01 -4.16363895e-01
-5.17814875e-01 -3.84279817e-01 4.66580808e-01 5.13658702e-01
-4.21457291e-01 7.48927474e-01 8.36177841e-02 2.97695547e-01
-9.92676169e-02 -8.83792937e-01 4.25633520e-01 -2.98220247e-01
-8.87367904e-01 6.44274196e-03 -3.67946506e-01 8.19458291e-02
5.83285511e-01 -8.29685777e-02 -8.14564705e-01 3.41307312e-01
-6.05946660e-01 -2.61005938e-01 8.17871869e-01 6.33502424e-01
3.45176280e-01 -1.08832014e+00 -5.26486099e-01 3.06823313e-01
2.30205804e-01 -8.75973031e-02 -2.30043661e-02 6.02347851e-01
-1.01142645e+00 8.00632834e-02 -1.43521965e-01 -3.47284943e-01
-1.11395299e+00 6.64569080e-01 6.40816987e-01 -2.77166903e-01
-7.49586940e-01 1.12646937e+00 -2.38053441e-01 -6.29473209e-01
-1.04301879e-02 -1.00135899e+00 -2.02651862e-02 -1.68145865e-01
6.34175718e-01 5.00447392e-01 6.88697577e-01 -1.58280328e-01
-3.86530548e-01 5.76783061e-01 -1.24902435e-01 6.86716318e-01
1.61105239e+00 4.13703591e-01 -8.13096881e-01 3.72092575e-01
1.29155767e+00 1.16196513e-01 -8.61961663e-01 -4.01756391e-02
5.61228275e-01 -3.65552872e-01 -2.70391196e-01 -1.00737834e+00
-1.15092587e+00 9.18162048e-01 3.69220495e-01 5.77841759e-01
6.90493703e-01 9.08754840e-02 4.76655543e-01 6.08440995e-01
5.15451193e-01 -6.01084292e-01 -5.27790338e-02 8.38356256e-01
9.26849842e-01 -9.72400606e-01 -2.70112157e-01 7.75567070e-02
3.50829475e-02 1.42583597e+00 6.37730777e-01 -2.64578730e-01
2.59398103e-01 7.09770501e-01 -1.16231009e-01 -3.85663420e-01
-7.15514779e-01 4.35131758e-01 6.53865188e-02 8.99699032e-01
6.97508574e-01 -1.54078066e-01 1.63904671e-02 6.60338461e-01
-2.77618498e-01 1.38410583e-01 1.24490857e+00 1.08883393e+00
-3.71777564e-01 -1.21172464e+00 -1.39058411e-01 6.68424070e-01
-4.93318260e-01 -4.60583001e-01 -3.58280182e-01 5.57216704e-01
3.50663066e-01 5.53281844e-01 -7.65984878e-02 -7.05653369e-01
4.87181157e-01 1.28047854e-01 4.42806691e-01 -1.16927814e+00
-1.27329659e+00 -8.21094036e-01 2.30181530e-01 -8.35494399e-01
1.58320561e-01 -5.94345093e-01 -1.06331515e+00 -2.51431227e-01
-1.17572762e-01 1.34501364e-02 8.21479797e-01 7.81159341e-01
5.68578124e-01 9.98745799e-01 2.20152989e-01 -6.07891142e-01
-5.51367581e-01 -7.76889741e-01 -2.92150259e-01 1.79103807e-01
7.31330931e-01 -2.39884689e-01 -3.07820231e-01 4.40399170e-01] | [7.60707950592041, 7.763955593109131] |
45c8f5cb-17f8-47cb-93f3-10a38e04e6d8 | semi-supervised-monaural-singing-voice | 1812.06087 | null | https://arxiv.org/abs/1812.06087v3 | https://arxiv.org/pdf/1812.06087v3.pdf | Semi-Supervised Monaural Singing Voice Separation With a Masking Network Trained on Synthetic Mixtures | We study the problem of semi-supervised singing voice separation, in which the training data contains a set of samples of mixed music (singing and instrumental) and an unmatched set of instrumental music. Our solution employs a single mapping function g, which, applied to a mixed sample, recovers the underlying instrumental music, and, applied to an instrumental sample, returns the same sample. The network g is trained using purely instrumental samples, as well as on synthetic mixed samples that are created by mixing reconstructed singing voices with random instrumental samples. Our results indicate that we are on a par with or better than fully supervised methods, which are also provided with training samples of unmixed singing voices, and are better than other recent semi-supervised methods. | ['Michael Michelashvili', 'Sagie Benaim', 'Lior Wolf'] | 2018-12-14 | null | null | null | null | ['music-source-separation'] | ['music'] | [ 7.23967016e-01 2.81676441e-01 -4.95108888e-02 2.20972183e-03
-1.19264591e+00 -8.48797739e-01 5.62494099e-01 -8.00448477e-01
-1.13133363e-01 8.59726071e-01 4.97039407e-01 1.77867234e-01
-2.28030413e-01 -4.03262496e-01 -6.83209956e-01 -9.00363982e-01
-4.60762009e-02 7.20651686e-01 -2.89919019e-01 -1.27875179e-01
-2.71379769e-01 1.21976301e-01 -1.76408994e+00 3.70017558e-01
8.38066339e-01 5.09908438e-01 1.47395566e-01 1.09534216e+00
-9.21591595e-02 9.28809285e-01 -6.95528150e-01 -2.97463890e-02
4.27691698e-01 -1.24458826e+00 -6.82354271e-01 4.09024715e-01
5.88837862e-01 -2.14825526e-01 -2.47129306e-01 9.77314472e-01
5.14682412e-01 2.87168264e-01 4.47907180e-01 -9.94231701e-01
-2.47365296e-01 1.26228380e+00 -1.30611122e-01 -5.37108958e-01
4.55787748e-01 -7.55260661e-02 1.25904262e+00 -9.24441457e-01
6.49993777e-01 1.03426147e+00 8.55097771e-01 8.71093869e-01
-1.65797389e+00 -6.99702919e-01 -6.85193479e-01 -3.56030792e-01
-1.09725451e+00 -1.20227373e+00 1.20338714e+00 -2.56086737e-01
3.08527231e-01 4.89879429e-01 6.71669662e-01 9.65567887e-01
-7.78571904e-01 1.04690111e+00 1.04949093e+00 -9.02305663e-01
7.59498402e-02 1.92799106e-01 -1.38101041e-01 3.22003722e-01
-3.26472700e-01 4.62044001e-01 -7.58218229e-01 -4.08187330e-01
5.16319394e-01 -4.70255822e-01 -5.32963157e-01 -2.16379508e-01
-1.52095151e+00 4.46200043e-01 -1.01797082e-01 6.09086514e-01
-6.47031069e-01 -1.69121251e-02 1.50089785e-01 6.17568672e-01
2.19231382e-01 6.72614515e-01 -1.41794086e-01 -9.47695673e-02
-1.86815512e+00 3.57275635e-01 1.21055067e+00 8.11381996e-01
6.11420155e-01 7.01903820e-01 9.40358490e-02 1.19445181e+00
4.15415391e-02 6.73735201e-01 6.66092753e-01 -1.45741630e+00
2.86329687e-01 -1.35577187e-01 2.62713462e-01 -1.86997309e-01
1.71317399e-01 -5.80157638e-01 -7.30591536e-01 3.25222343e-01
7.00890243e-01 -1.67384431e-01 -6.30615413e-01 2.20283008e+00
-3.27582145e-03 5.81466973e-01 3.24306399e-01 6.65098429e-01
8.45095575e-01 6.32143080e-01 -6.94608390e-01 -7.99842894e-01
6.43327057e-01 -1.42033076e+00 -1.03156722e+00 -1.90903932e-01
-1.99368075e-01 -1.09317958e+00 1.07740676e+00 6.56704962e-01
-1.66609037e+00 -1.04308879e+00 -1.15662277e+00 3.10503393e-01
1.79958150e-01 4.41156328e-01 1.05993357e-02 6.49032116e-01
-1.13667953e+00 1.29230011e+00 -3.29669088e-01 -1.08335555e-01
2.64278287e-03 4.36180860e-01 -4.69842464e-01 2.97642589e-01
-8.86408985e-01 2.88809389e-01 2.54974872e-01 2.34969646e-01
-1.21102786e+00 -4.45016086e-01 -6.30397141e-01 -2.45430842e-01
3.86114001e-01 -4.14747685e-01 1.56432807e+00 -1.51107109e+00
-1.79022479e+00 7.48111606e-01 -5.25331609e-02 -2.73074239e-01
4.78701711e-01 -8.17579776e-02 -8.24450612e-01 2.32714757e-01
2.13200644e-01 1.89185232e-01 1.50951982e+00 -1.62093961e+00
-2.10752293e-01 -1.17993400e-01 -5.65361202e-01 1.63582250e-01
-5.70933782e-02 -4.51615267e-03 -2.79315561e-01 -7.56977558e-01
2.36990169e-01 -9.86546159e-01 8.36735293e-02 -5.53660929e-01
-8.87070179e-01 4.95043129e-01 5.74185312e-01 -9.71993923e-01
1.20873964e+00 -2.48442435e+00 5.33019066e-01 3.23898762e-01
1.45387769e-01 2.11087227e-01 -7.00021684e-01 6.16172969e-01
-5.42431295e-01 -2.54101604e-01 -7.52125561e-01 -8.40949237e-01
8.51450190e-02 1.18943729e-01 -7.40632951e-01 2.14932457e-01
-4.06277925e-02 6.96795046e-01 -1.09194136e+00 -2.28527769e-01
2.49809418e-02 7.53642321e-02 -5.00506043e-01 6.51165247e-01
-1.32520333e-01 6.97583020e-01 3.34597491e-02 4.29936022e-01
3.22743475e-01 3.08986992e-01 4.84150261e-01 -5.20122014e-02
-2.48021111e-02 5.12313068e-01 -1.66035211e+00 1.82633150e+00
-1.83587059e-01 4.97144520e-01 6.75822079e-01 -8.90205145e-01
1.00894618e+00 9.01078582e-01 3.76014769e-01 6.73324615e-02
-1.14011668e-01 7.64031768e-01 7.74140880e-02 -5.03022909e-01
5.83607495e-01 -7.78459549e-01 1.81007043e-01 8.43508065e-01
5.37831843e-01 -5.29505312e-01 1.01234645e-01 -2.36579981e-02
9.74571407e-01 6.59423888e-01 2.26051077e-01 2.51290351e-02
6.15269542e-01 -3.25456150e-02 4.60922748e-01 8.28617871e-01
1.95106566e-01 8.14634562e-01 3.30426723e-01 1.75705016e-01
-1.25413430e+00 -1.19224811e+00 1.52997911e-01 9.27942395e-01
-5.12113154e-01 -4.02518988e-01 -1.04308474e+00 -3.71950388e-01
-1.05663933e-01 8.62804115e-01 -3.40677708e-01 -4.37548347e-02
-8.92740011e-01 -2.25904301e-01 9.17401373e-01 3.16094279e-01
1.90554246e-01 -1.49099827e+00 2.70120770e-01 5.51349640e-01
-1.89879417e-01 -6.79147959e-01 -9.53425765e-01 5.11239529e-01
-9.38114345e-01 -1.17809904e+00 -7.30309844e-01 -1.05864143e+00
3.72611880e-01 1.69819608e-01 1.14332151e+00 -2.13898774e-02
1.92464888e-01 2.21247137e-01 -1.37632132e-01 -7.31824432e-03
-1.22348702e+00 -2.10440889e-01 4.41984683e-01 4.75352526e-01
-2.27497369e-01 -1.21587062e+00 1.54830113e-01 2.65855640e-01
-1.02606988e+00 7.93920010e-02 3.35358351e-01 1.26491773e+00
5.33929944e-01 1.99101821e-01 8.51714671e-01 -1.10850871e+00
4.98831451e-01 -1.78990588e-01 -8.13391656e-02 -9.08777043e-02
-1.35283872e-01 2.74872966e-02 1.02575397e+00 -8.46973300e-01
-9.22642350e-01 1.30506113e-01 -1.73818678e-01 -8.41678977e-01
-2.56363541e-01 1.76066011e-01 -3.33487868e-01 3.18407536e-01
8.04046631e-01 4.36571449e-01 4.84103322e-01 -9.25415218e-01
7.05659449e-01 1.01674998e+00 1.33212292e+00 -5.77551723e-01
1.35223484e+00 2.84022570e-01 -4.06162024e-01 -1.05652773e+00
-6.59252465e-01 -3.40440184e-01 -8.94378066e-01 -8.16786736e-02
4.89218712e-01 -6.75760806e-01 -5.47870295e-03 6.62513971e-01
-7.67967463e-01 -6.55418217e-01 -1.18052649e+00 7.95461476e-01
-9.74530399e-01 3.54722977e-01 -6.44469917e-01 -1.05070889e+00
-1.41711503e-01 -6.87169909e-01 8.96648943e-01 -2.02461004e-01
-4.43500906e-01 -8.46027076e-01 4.97788787e-01 4.28862780e-01
1.01450287e-01 1.20000042e-01 6.79189861e-01 -8.58609259e-01
-1.18621960e-01 -3.77794772e-01 6.03980184e-01 1.03596044e+00
4.96484548e-01 6.84827194e-02 -1.42529881e+00 -3.72371316e-01
3.10980588e-01 -5.49370229e-01 9.12013412e-01 3.77456218e-01
4.32863086e-01 -5.13952732e-01 3.24155539e-02 5.82960784e-01
7.91063190e-01 1.35668680e-01 3.84809613e-01 -3.01203668e-01
7.05481172e-01 7.90573299e-01 3.07094336e-01 2.57831588e-02
-3.36315542e-01 5.23048818e-01 2.68203206e-03 -1.35672972e-01
-5.55271685e-01 -7.18002141e-01 5.83550870e-01 1.50548577e+00
-3.63189012e-01 1.60383448e-01 -2.76201874e-01 6.09228969e-01
-1.74768305e+00 -1.37469387e+00 -1.42923864e-02 2.49775839e+00
1.07587898e+00 -1.61195949e-01 7.04233587e-01 1.00491381e+00
9.46967304e-01 3.60364437e-01 -5.50543070e-01 7.23949373e-02
-2.46458918e-01 8.78780901e-01 -1.34890586e-01 7.53200531e-01
-1.09725988e+00 8.03562820e-01 7.35356092e+00 8.52116823e-01
-8.56701195e-01 -8.15705396e-03 -2.20964015e-01 -3.32429707e-01
-5.07844090e-01 3.36493179e-02 -7.38451779e-02 1.87964335e-01
1.08088303e+00 4.70802821e-02 1.55089569e+00 4.20194149e-01
1.63659483e-01 4.67912257e-01 -1.36522961e+00 8.05612504e-01
1.81634426e-01 -1.06279540e+00 -2.26543695e-01 -8.69415402e-02
8.80908251e-01 -2.72483289e-01 -1.86121941e-01 3.53953630e-01
2.39803284e-01 -9.88983750e-01 1.09065247e+00 5.24184465e-01
9.99310851e-01 -4.70184714e-01 3.04814577e-01 7.31736898e-01
-1.10372496e+00 3.90163362e-02 1.38433725e-01 -7.01914448e-03
6.49261996e-02 2.50838667e-01 -6.56526923e-01 6.69173717e-01
1.51554257e-01 8.79522562e-01 1.00393698e-03 6.61346912e-01
-3.73118043e-01 1.03426635e+00 -1.04247451e-01 4.56556618e-01
-1.17451362e-01 -6.09339654e-01 1.03944278e+00 8.73279035e-01
4.18029040e-01 -3.88991714e-01 -3.85693386e-02 1.09657550e+00
-7.57403448e-02 1.54111937e-01 -8.20280015e-01 -6.20006204e-01
4.97955620e-01 1.32911932e+00 -1.81732774e-01 -6.86302722e-01
-1.68020338e-01 9.67615068e-01 -6.54071718e-02 6.30139410e-01
-2.37075701e-01 -3.78937304e-01 2.84363478e-01 5.05198203e-02
1.58648774e-01 -1.77848823e-02 9.58002284e-02 -1.14018941e+00
-1.65632233e-01 -1.36420417e+00 6.21077418e-02 -9.94692147e-01
-1.24653697e+00 7.97961652e-01 -3.80282372e-01 -1.70423365e+00
-8.72812927e-01 -1.48430303e-01 -9.79962945e-01 1.03758681e+00
-9.18120742e-01 -9.10236835e-01 3.21107358e-03 5.69918752e-01
3.84346575e-01 -7.06915319e-01 1.06489193e+00 3.17309499e-01
-2.72142142e-01 1.43219978e-01 2.44456917e-01 1.37520939e-01
7.59858489e-01 -1.47203410e+00 2.46126011e-01 5.33820391e-01
9.36610579e-01 5.86464107e-01 7.63434887e-01 -2.60479152e-01
-1.47559929e+00 -8.54971707e-01 7.92899728e-01 -2.92607993e-01
5.22033095e-01 -2.85039872e-01 -9.73554850e-01 5.72824836e-01
2.06271574e-01 -2.36664370e-01 9.79500473e-01 -1.43912598e-01
-1.47270590e-01 -1.96997263e-02 -1.01327670e+00 3.58869970e-01
1.01708877e+00 -1.05540860e+00 -1.24554837e+00 9.25512388e-02
5.06606162e-01 -2.28818700e-01 -6.62055790e-01 1.59293815e-01
7.65442371e-01 -9.97955322e-01 9.84299779e-01 -8.82313251e-01
3.29443246e-01 -5.62072754e-01 -3.33712667e-01 -1.53620529e+00
-1.90690234e-01 -1.10485935e+00 -2.17611156e-02 1.35338318e+00
2.96634108e-01 -2.98064947e-01 5.23738563e-01 -2.02917740e-01
-2.79200822e-01 -1.13906059e-02 -7.38642573e-01 -9.63597000e-01
-4.95976172e-02 -5.71807086e-01 6.99360490e-01 1.07101047e+00
-1.61092058e-01 5.96883714e-01 -8.40199769e-01 -1.92173958e-01
9.16626155e-01 4.51257139e-01 1.04646242e+00 -1.26029360e+00
-9.72125888e-01 -3.37621897e-01 3.76117975e-01 -6.62783027e-01
5.32542825e-01 -1.22046554e+00 3.14342141e-01 -1.12511611e+00
-3.31795029e-02 -1.59824923e-01 -2.22575113e-01 4.95445848e-01
-2.17332523e-02 7.59651661e-01 1.94704816e-01 5.52070737e-01
1.47160115e-02 4.40621465e-01 1.38128936e+00 -1.37062386e-01
-7.40767062e-01 4.14493024e-01 -6.66946352e-01 1.05096233e+00
4.54047501e-01 -4.80804950e-01 -1.83246776e-01 2.69948035e-01
-2.72815585e-01 7.06419826e-01 3.53811979e-01 -1.07521021e+00
-8.57235491e-03 2.54709601e-01 1.28335312e-01 -4.68701720e-01
5.73166668e-01 -3.97899538e-01 6.99465454e-01 2.06247002e-01
-6.01998448e-01 -7.71726787e-01 -1.48272902e-01 2.91174948e-01
-4.88740116e-01 -6.70687973e-01 6.97871804e-01 -1.07987083e-01
-2.29590246e-03 -1.05966829e-01 -5.04186191e-02 1.08491138e-01
1.75202459e-01 -2.96623141e-01 2.53469020e-01 -6.70463085e-01
-1.25483370e+00 -4.05309647e-01 3.93805236e-01 1.31400913e-01
2.90752232e-01 -1.74831390e+00 -8.42305720e-01 7.45473862e-01
-2.99379557e-01 -1.80605099e-01 1.16574690e-02 7.97841609e-01
9.67048109e-02 -5.46367727e-02 4.86678723e-03 -2.07843721e-01
-1.20860934e+00 4.98328328e-01 2.04284832e-01 1.26674846e-01
-5.45104980e-01 4.85773534e-01 -2.11075783e-01 -1.07069850e+00
2.56900281e-01 3.89014035e-02 -4.50676046e-02 2.16225594e-01
4.48847890e-01 4.58180249e-01 -3.78439367e-01 -8.42012703e-01
6.56599402e-02 6.28337145e-01 7.14253366e-01 -7.30794549e-01
1.36309171e+00 3.52748334e-01 -3.34644198e-01 1.10302854e+00
9.75175858e-01 9.99873757e-01 -1.01658845e+00 -5.54847658e-01
-3.42522040e-02 -2.88490087e-01 -2.44027108e-01 -6.20352745e-01
-8.23018074e-01 5.68355381e-01 4.61854301e-02 6.22453451e-01
1.09180164e+00 3.83332856e-02 6.08698547e-01 3.98043722e-01
1.22074261e-01 -9.04843986e-01 3.08111906e-02 1.55780122e-01
9.60171878e-01 -6.93687797e-01 -4.21646804e-01 -7.52695724e-02
-5.28260708e-01 1.13197792e+00 -2.07589343e-01 -5.02220988e-01
3.73502851e-01 5.34590006e-01 2.68598884e-01 2.11867690e-01
-5.74383676e-01 -4.16099638e-01 5.61448812e-01 5.83343565e-01
5.34197450e-01 1.75863296e-01 1.19961768e-01 8.65949035e-01
-7.19244003e-01 1.50537103e-01 2.03473866e-01 4.99544978e-01
-4.18368965e-01 -1.48814356e+00 -8.33667815e-01 5.18317938e-01
-1.93020850e-01 -1.53359413e-01 -6.32030666e-01 5.75055063e-01
3.13479125e-01 1.06236970e+00 -1.59904227e-01 -6.65523350e-01
3.36105704e-01 6.45017743e-01 5.51808476e-01 -7.55482197e-01
-6.04254544e-01 8.05309296e-01 5.13183653e-01 -1.37644023e-01
-7.17625678e-01 -7.14980602e-01 -1.10558724e+00 1.14415161e-01
-3.66571724e-01 7.10542262e-01 4.26744998e-01 8.31817031e-01
-3.81560147e-01 6.89167976e-01 1.16663301e+00 -1.30144322e+00
-8.23656857e-01 -1.29178035e+00 -1.33810341e+00 4.58419651e-01
7.14470208e-01 -1.31400630e-01 -9.00075495e-01 2.79366046e-01] | [15.550756454467773, 5.6529340744018555] |
1cacacad-2760-46eb-9c06-6083999462a4 | curriculum-learning-a-regularization-method | 2108.06084 | null | https://arxiv.org/abs/2108.06084v4 | https://arxiv.org/pdf/2108.06084v4.pdf | The Stability-Efficiency Dilemma: Investigating Sequence Length Warmup for Training GPT Models | Recent works have demonstrated great success in pre-training large-scale autoregressive language models on massive GPUs. To reduce the wall-clock training time, a common practice is to increase the batch size and learning rate. However, such practice is often brittle and leads to a so-called stability-efficiency dilemma: increasing the batch sizes and learning rates leads to better training efficiency but can also result in training instability, leading to poor generalization accuracy or failed runs. To better understand this phenomenon, we conduct an in-depth analysis on large-scale pre-training experiments replicating the GPT-2 model. We find that there is a strong correlation between training instability and extreme values of gradient variance, and that samples with long sequence lengths contribute to these extreme gradient variance values, especially at the beginning of the training, indicating that long sequence length can be a main source of training instability. Based on the analysis, we present a Sequence Length Warmup method that aims to solve the training stability-efficiency dilemma. Experiments replicating GPT-2 models show that our approach enables stable training with 8x larger batch size and 4x larger learning rate, whereas the baseline approach struggles with training instability. To achieve the same or better zero-shot evaluation results, our method reduces the required number of training tokens and wall clock time by up to 2.2x and 3.7x, respectively. Experiments replicating GPT-3 model (125M) show that our approach enables stable training with 8x larger batch size and 40x larger learning rate, and retains 99% of the zero-shot accuracy on 11 tasks using 10x less data and 17x less time compared to the original GPT-3 training recipe, while the baseline diverges under the same settings and only retain 95% of accuracy under lower learning rate. | ['Yuxiong He', 'Minjia Zhang', 'Conglong Li'] | 2021-08-13 | curriculum-learning-a-regularization-method-1 | https://openreview.net/forum?id=rhDaUTtfsqs | https://openreview.net/pdf?id=rhDaUTtfsqs | null | ['lambada'] | ['natural-language-processing'] | [-5.18288603e-03 -5.07640064e-01 -7.29574412e-02 -8.18674937e-02
-7.54239976e-01 -5.15734613e-01 2.62605548e-01 8.01365077e-02
-7.27785945e-01 4.81622219e-01 -3.68567854e-01 -8.56633186e-01
2.64672816e-01 -6.05765700e-01 -7.85827577e-01 -7.14848220e-01
-8.61776844e-02 3.94402504e-01 4.71780598e-01 -2.48806849e-01
2.49066755e-01 1.74429446e-01 -1.67855978e+00 1.70369267e-01
8.52116406e-01 7.55383193e-01 3.04530740e-01 9.12275314e-01
-1.55710086e-01 6.56543851e-01 -7.65355229e-01 -1.45979226e-01
2.96628565e-01 -3.32756668e-01 -7.03955472e-01 -1.25178739e-01
4.44168270e-01 -1.75654843e-01 3.23609896e-02 9.65222836e-01
7.37257600e-01 2.10069641e-01 1.45605400e-01 -1.02852023e+00
-1.78419366e-01 5.49292266e-01 -7.53549278e-01 3.67106408e-01
-2.24340215e-01 3.47792745e-01 5.35480142e-01 -7.65641391e-01
2.17287406e-01 1.14879811e+00 9.29443121e-01 5.31206012e-01
-9.96503651e-01 -7.17165411e-01 6.48381114e-02 -1.81121618e-01
-1.31825244e+00 -3.72804374e-01 1.89646631e-02 -2.93409646e-01
1.33847594e+00 2.97152638e-01 4.73809153e-01 1.00052643e+00
3.54612887e-01 6.01539254e-01 1.06844413e+00 -5.50380588e-01
3.81094277e-01 1.17772177e-01 4.59016681e-01 7.25310624e-01
2.48817891e-01 7.02191070e-02 -3.32145214e-01 -2.13721946e-01
8.00432503e-01 -1.61772504e-01 -4.19001728e-02 2.64050215e-01
-9.67851937e-01 7.87477076e-01 9.58441421e-02 4.44825739e-01
-7.52060339e-02 1.26194686e-01 8.08896601e-01 5.51145375e-01
5.76759338e-01 5.14706016e-01 -5.65630198e-01 -7.86587119e-01
-8.97760510e-01 1.65330499e-01 7.18992472e-01 8.22022140e-01
7.00797915e-01 2.48499572e-01 -1.71916202e-01 9.72640514e-01
-1.13384239e-01 3.71046364e-01 9.87516522e-01 -7.55437136e-01
6.41183496e-01 5.42827725e-01 -1.22479908e-01 -6.12047911e-01
-4.85033631e-01 -7.51611650e-01 -9.68156457e-01 1.48692772e-01
6.54430866e-01 -4.68279660e-01 -9.94806647e-01 1.69563174e+00
2.62951702e-01 4.53222334e-01 2.08607744e-02 7.86170423e-01
3.24088544e-01 7.19572425e-01 8.83006155e-02 -2.39192531e-01
1.59137249e+00 -1.21215177e+00 -3.28631550e-01 -5.47719121e-01
1.37741590e+00 -8.47543418e-01 1.73125398e+00 2.81585574e-01
-9.68799889e-01 -6.44743860e-01 -1.08854580e+00 7.56840780e-02
-6.17599376e-02 2.20890254e-01 5.41945040e-01 1.02756250e+00
-9.89465415e-01 8.96099329e-01 -1.07977951e+00 -2.14710712e-01
1.17714308e-01 3.87829512e-01 -2.12476067e-02 -7.42825642e-02
-1.03153801e+00 6.21313453e-01 4.89598334e-01 -1.05786016e-02
-3.92267555e-01 -1.02098298e+00 -5.87061048e-01 2.60581195e-01
4.12904978e-01 -5.07878184e-01 1.31330967e+00 -8.52301180e-01
-1.80893254e+00 6.17348433e-01 -1.54580817e-01 -6.03740752e-01
6.60608232e-01 -5.76188803e-01 -1.98799506e-01 -4.56525981e-01
-2.54204363e-01 3.68822098e-01 6.80738389e-01 -8.05771470e-01
-4.78122264e-01 -3.21652919e-01 -3.50231767e-01 1.40817448e-01
-6.26412511e-01 1.57683134e-01 -6.88733459e-01 -5.00208497e-01
-4.88476641e-02 -1.34369814e+00 -3.62627774e-01 -8.22546124e-01
-8.41664225e-02 -1.28408313e-01 6.30785823e-01 -2.91803479e-01
1.49578857e+00 -2.13242674e+00 -3.03636938e-01 8.71116817e-02
1.05084881e-01 6.83183730e-01 -2.39784256e-01 1.44630656e-01
4.79687713e-02 2.99438149e-01 -6.46289289e-02 -4.95260358e-01
-2.37164095e-01 3.81308109e-01 -4.38592076e-01 2.30707034e-01
9.78545547e-02 7.07241654e-01 -9.96001780e-01 -2.93832302e-01
-5.20304143e-02 3.15787822e-01 -6.72285676e-01 3.22282352e-02
-2.93018355e-04 3.41463387e-01 -2.31446922e-01 2.07258850e-01
6.62265360e-01 -5.22415161e-01 2.61589676e-01 3.71929258e-01
-1.69373035e-01 3.21183622e-01 -1.17468357e+00 1.35333431e+00
-6.86078429e-01 7.01321185e-01 -2.45971307e-01 -7.50390053e-01
8.39084268e-01 1.75891161e-01 7.02234656e-02 -7.12378919e-01
2.81987339e-01 4.21376258e-01 1.76437199e-01 -2.65215158e-01
7.96434700e-01 -2.51179878e-02 5.34987114e-02 5.06679654e-01
-2.79526651e-01 1.25836814e-02 2.34075531e-01 8.20041001e-02
1.19474745e+00 -5.95524050e-02 -1.82019547e-02 -3.01416546e-01
3.63654763e-01 -9.21965837e-02 5.20971954e-01 9.13880885e-01
6.60073459e-02 3.16554666e-01 6.44002974e-01 -5.84088802e-01
-1.33878970e+00 -4.46751565e-01 -1.34899616e-01 1.56303036e+00
-2.26949126e-01 -7.25386322e-01 -7.71435440e-01 -4.24008012e-01
-2.22399577e-01 6.17769957e-01 -3.76967967e-01 -1.66707933e-01
-8.95377994e-01 -1.13407671e+00 8.49994540e-01 5.41429102e-01
6.03756845e-01 -9.03091013e-01 -8.59576344e-01 1.44162044e-01
-6.53349087e-02 -1.18363941e+00 -4.70539659e-01 1.99471384e-01
-1.22165132e+00 -5.18308997e-01 -6.53343081e-01 -6.64417267e-01
7.72567987e-01 2.10562304e-01 1.34130740e+00 6.08905673e-01
-1.76263958e-01 -2.94777244e-01 -2.92253375e-01 -3.23547006e-01
-7.03202367e-01 4.91257876e-01 2.41013199e-01 -4.65141535e-01
8.42037275e-02 -4.26161230e-01 -4.31277543e-01 4.85915512e-01
-7.36154974e-01 1.33578166e-01 4.23825413e-01 1.11252725e+00
3.42577696e-01 -1.26206428e-02 3.68976235e-01 -9.61462080e-01
6.98743582e-01 -3.44384491e-01 -7.99528301e-01 1.74993128e-01
-8.81719470e-01 2.95046717e-01 9.30594921e-01 -8.99646640e-01
-8.63684893e-01 -2.02701733e-01 -4.00849491e-01 -6.04885876e-01
2.07514331e-01 4.21841145e-01 4.05694574e-01 3.69361378e-02
9.30951774e-01 2.04106271e-01 -4.54347879e-02 -4.78927702e-01
8.31338391e-02 5.40206552e-01 2.26024136e-01 -6.78316593e-01
3.47768664e-01 -6.24057429e-05 -2.02082455e-01 -8.23034942e-01
-5.38656473e-01 -3.88572812e-01 -2.14726686e-01 1.86961368e-01
2.81449258e-01 -1.00131476e+00 -9.44536507e-01 7.40400493e-01
-7.97778666e-01 -9.32932615e-01 -1.13421259e-02 4.97530460e-01
-1.52590871e-01 4.44074631e-01 -9.21083629e-01 -8.85696411e-01
-7.68251598e-01 -1.31527889e+00 1.12436092e+00 4.93357442e-02
-1.37765333e-01 -8.13096523e-01 4.29951623e-02 5.09913005e-02
5.83343744e-01 -2.38185510e-01 8.07337761e-01 -4.68739063e-01
-2.28236347e-01 -1.65091336e-01 -1.34571850e-01 3.32032442e-01
-2.90533155e-01 -3.74933146e-02 -7.77859092e-01 -7.25642681e-01
8.15676004e-02 -3.14318269e-01 7.05511808e-01 1.58119127e-01
1.07742846e+00 -1.23626985e-01 -1.81147143e-01 7.03327000e-01
1.48234630e+00 -1.91819388e-02 7.21825123e-01 3.75783056e-01
6.18615866e-01 1.97315663e-01 6.17335141e-01 5.22734702e-01
-4.74198684e-02 8.37730467e-01 2.79139820e-02 -3.90890166e-02
1.05446622e-01 -1.08861357e-01 6.79861128e-01 1.08559585e+00
-7.89513141e-02 -2.03026101e-01 -1.20452070e+00 2.61202574e-01
-1.81486201e+00 -6.75647855e-01 -3.48240286e-01 2.70825338e+00
1.05675256e+00 3.84256065e-01 2.47019529e-01 1.20685063e-01
4.14138258e-01 -1.87592939e-01 -4.45269376e-01 -5.87937176e-01
1.23280905e-01 1.51919529e-01 7.58570969e-01 6.07041538e-01
-6.67449892e-01 1.20644784e+00 6.32719946e+00 1.19294322e+00
-1.64573574e+00 1.53314590e-01 8.54097784e-01 -3.95364374e-01
2.12594032e-01 -7.36138746e-02 -1.17517853e+00 6.89867914e-01
1.54930818e+00 -2.02135772e-01 4.07656699e-01 1.00850141e+00
1.01572409e-01 -1.92677677e-01 -8.40536177e-01 1.10424984e+00
-1.93943247e-01 -1.15489459e+00 -3.30805272e-01 1.39659524e-01
7.28952110e-01 4.24183458e-01 -7.26711750e-02 7.81680048e-01
3.84019375e-01 -8.87657106e-01 4.67008680e-01 -2.02869862e-01
8.40107203e-01 -7.40352154e-01 8.65839601e-01 7.20834196e-01
-1.12128079e+00 -1.09727703e-01 -6.82357371e-01 -4.17187721e-01
-2.50321515e-02 7.56690025e-01 -9.41536069e-01 2.77914375e-01
8.18025470e-01 2.04395607e-01 -6.62957788e-01 8.95655036e-01
1.84711963e-02 9.41398561e-01 -6.32615805e-01 -1.57867387e-01
2.49713644e-01 -1.85121521e-01 2.66329497e-01 1.30971527e+00
5.10102391e-01 9.22270566e-02 1.10928848e-01 3.11075538e-01
-2.00523958e-01 2.74244666e-01 -2.52553821e-01 8.41601416e-02
5.20570576e-01 1.18744767e+00 -9.59156036e-01 -7.65653968e-01
-8.43641609e-02 9.38329816e-01 5.41091919e-01 2.20260248e-01
-1.01370406e+00 -1.12546436e-01 5.47524035e-01 -1.67660024e-02
1.81520119e-01 -4.50405419e-01 -5.46844721e-01 -1.20634246e+00
8.29078332e-02 -1.08122718e+00 1.75462484e-01 -3.43103290e-01
-8.88765991e-01 7.18627036e-01 -2.74438798e-01 -1.11699152e+00
-4.14846897e-01 -4.87804443e-01 -7.21323729e-01 9.06100571e-01
-1.21991038e+00 -5.15398860e-01 -2.63585865e-01 2.35855997e-01
7.15404630e-01 -1.15141300e-02 8.30866992e-01 4.90447998e-01
-8.69424164e-01 1.10599077e+00 3.38945746e-01 -1.11629531e-01
6.10861897e-01 -9.71987665e-01 9.29556608e-01 8.42379689e-01
-7.76973218e-02 9.16934729e-01 7.87156403e-01 -5.43761492e-01
-1.42862010e+00 -1.06459069e+00 9.25764918e-01 -4.67563182e-01
6.17418408e-01 -4.61651444e-01 -1.28069794e+00 4.59457189e-01
-1.81934103e-01 2.65136454e-02 5.74383199e-01 4.36033428e-01
-2.52366364e-01 1.78452283e-01 -7.38612175e-01 8.12906325e-01
1.02067769e+00 -3.10381681e-01 -9.88817289e-02 4.07007158e-01
7.91363001e-01 -8.36674869e-01 -8.90583456e-01 2.79093862e-01
5.54316759e-01 -7.54391253e-01 7.67080009e-01 -4.87811267e-01
2.73938626e-01 3.58382165e-02 2.76843548e-01 -1.15642178e+00
-2.06497714e-01 -7.47030556e-01 5.43268546e-02 9.60851967e-01
4.59561229e-01 -8.23320568e-01 9.02919114e-01 5.38025737e-01
-1.40558630e-01 -1.03261125e+00 -8.67177129e-01 -1.13081336e+00
4.12841558e-01 -6.11771107e-01 4.94285017e-01 8.63023758e-01
-1.46055564e-01 5.74746132e-01 -4.58969861e-01 6.38683289e-02
1.06193542e-01 -1.25084072e-01 9.25676167e-01 -9.08044279e-01
-4.80920732e-01 -5.83373964e-01 3.77604738e-02 -1.30278385e+00
-1.40076935e-01 -4.99598652e-01 1.37991488e-01 -8.96679759e-01
-1.32240141e-02 -8.58869731e-01 -9.35210586e-02 7.71306157e-01
-6.03168726e-01 2.66103536e-01 3.99954647e-01 3.80965114e-01
-4.72698361e-01 2.94832170e-01 8.65539730e-01 3.69112045e-01
-5.97307801e-01 2.49526650e-02 -3.21423501e-01 5.30297697e-01
8.17174554e-01 -4.64905947e-01 -4.84084994e-01 -6.31746233e-01
5.28155804e-01 -3.69384736e-02 -7.12380782e-02 -1.20293105e+00
1.95450276e-01 9.50928032e-02 1.11497320e-01 -2.29471818e-01
9.02832448e-02 -4.10740077e-01 -3.83763690e-04 6.38917685e-01
-7.59852305e-02 4.47500288e-01 7.65306950e-01 2.02944413e-01
7.37252831e-02 -2.80738205e-01 6.85327053e-01 1.14793211e-01
-5.29692948e-01 5.24244457e-02 -4.01381612e-01 5.53153902e-02
6.54437482e-01 -1.89292848e-01 -3.97380054e-01 1.66227501e-02
-3.68200570e-01 3.51911724e-01 5.40459216e-01 5.73911905e-01
1.77019954e-01 -1.09843159e+00 -7.28070736e-01 5.01267672e-01
-1.74534991e-01 -7.22400025e-02 4.96488094e-01 1.05369461e+00
-5.43994486e-01 2.99291372e-01 4.89963628e-02 -8.74678016e-01
-1.66956127e+00 5.06561518e-01 2.96982557e-01 -7.52104640e-01
-6.92479908e-01 9.99862611e-01 2.12049168e-02 -3.90527993e-01
1.57547057e-01 -5.12457013e-01 1.30980268e-01 -2.17914358e-01
6.70299768e-01 4.35007155e-01 4.63412255e-01 -1.71562657e-01
-2.89773673e-01 5.60477853e-01 -2.72413135e-01 2.69756429e-02
1.06288886e+00 1.74143821e-01 1.18842542e-01 4.27347869e-01
1.23426533e+00 -1.63755473e-02 -1.20272613e+00 -3.43830854e-01
-1.90129667e-03 -5.34912109e-01 5.14191240e-02 -4.41631585e-01
-8.92346680e-01 9.60456312e-01 7.63487697e-01 2.18141198e-01
1.02430928e+00 -4.24154788e-01 9.93341565e-01 4.09355134e-01
3.86149734e-01 -1.06082392e+00 2.98328120e-02 1.07470095e+00
4.02413517e-01 -1.25821674e+00 -4.31140736e-02 -3.12281936e-01
-6.90884948e-01 9.98688757e-01 7.45728195e-01 2.13868581e-02
1.73090503e-01 4.76381749e-01 2.28389129e-02 5.84726455e-03
-1.05146635e+00 1.69741035e-01 -4.59256442e-03 -7.94061199e-02
6.52138948e-01 3.23613063e-02 -3.86447728e-01 4.50449049e-01
-5.60562193e-01 7.01561794e-02 2.76256472e-01 9.73507285e-01
-3.32633466e-01 -1.19985735e+00 -4.27740365e-01 3.25617641e-01
-5.42639077e-01 -3.59103888e-01 2.26401538e-01 5.35558283e-01
-2.02595353e-01 8.17597091e-01 3.73504549e-01 -5.43769777e-01
2.97688961e-01 2.53136605e-01 3.27383697e-01 -4.95795429e-01
-1.01313198e+00 1.07273914e-01 1.61564946e-02 -3.56927991e-01
1.72584608e-01 -2.29061350e-01 -1.41708553e+00 -8.62188876e-01
-5.29820621e-01 1.88632816e-01 6.86806142e-01 9.01390612e-01
7.05281854e-01 6.05097651e-01 4.22556132e-01 -7.40097940e-01
-8.43461752e-01 -1.07819986e+00 -2.71290839e-01 3.62204760e-01
6.93288669e-02 -2.14769483e-01 -3.09510708e-01 -1.26414239e-01] | [8.638585090637207, 3.4745960235595703] |
21b95924-652f-4b44-8427-fe58ef649e41 | unsupervised-feature-learning-by-cross-level | 2008.03813 | null | https://arxiv.org/abs/2008.03813v5 | https://arxiv.org/pdf/2008.03813v5.pdf | Unsupervised Feature Learning by Cross-Level Instance-Group Discrimination | Unsupervised feature learning has made great strides with contrastive learning based on instance discrimination and invariant mapping, as benchmarked on curated class-balanced datasets. However, natural data could be highly correlated and long-tail distributed. Natural between-instance similarity conflicts with the presumed instance distinction, causing unstable training and poor performance. Our idea is to discover and integrate between-instance similarity into contrastive learning, not directly by instance grouping, but by cross-level discrimination (CLD) between instances and local instance groups. While invariant mapping of each instance is imposed by attraction within its augmented views, between-instance similarity could emerge from common repulsion against instance groups. Our batch-wise and cross-view comparisons also greatly improve the positive/negative sample ratio of contrastive learning and achieve better invariant mapping. To effect both grouping and discrimination objectives, we impose them on features separately derived from a shared representation. In addition, we propose normalized projection heads and unsupervised hyper-parameter tuning for the first time. Our extensive experimentation demonstrates that CLD is a lean and powerful add-on to existing methods such as NPID, MoCo, InfoMin, and BYOL on highly correlated, long-tail, or balanced datasets. It not only achieves new state-of-the-art on self-supervision, semi-supervision, and transfer learning benchmarks, but also beats MoCo v2 and SimCLR on every reported performance attained with a much larger compute. CLD effectively brings unsupervised learning closer to natural data and real-world applications. Our code is publicly available at: https://github.com/frank-xwang/CLD-UnsupervisedLearning. | ['Xudong Wang', 'Ziwei Liu', 'Stella X. Yu'] | 2020-08-09 | null | http://openaccess.thecvf.com//content/CVPR2021/html/Wang_Unsupervised_Feature_Learning_by_Cross-Level_Instance-Group_Discrimination_CVPR_2021_paper.html | http://openaccess.thecvf.com//content/CVPR2021/papers/Wang_Unsupervised_Feature_Learning_by_Cross-Level_Instance-Group_Discrimination_CVPR_2021_paper.pdf | cvpr-2021-1 | ['unsupervised-image-classification'] | ['computer-vision'] | [ 2.44712397e-01 2.24999581e-02 -6.58495784e-01 -6.42514765e-01
-8.16241324e-01 -5.83943129e-01 8.82330060e-01 1.31222591e-01
-2.77163655e-01 7.75408506e-01 1.92650720e-01 9.86729562e-02
-4.83619362e-01 -6.17599308e-01 -5.60375690e-01 -8.20075095e-01
-1.30127341e-01 6.09236419e-01 1.35768265e-01 -1.52727395e-01
4.97187600e-02 3.31090808e-01 -1.71380711e+00 4.14804220e-01
7.94338703e-01 9.30833340e-01 2.88311928e-03 1.40176952e-01
5.64192887e-03 7.38552272e-01 -3.18063736e-01 -3.38910908e-01
4.09299433e-01 -3.88285398e-01 -7.97678709e-01 2.28648633e-01
8.29198837e-01 2.10161597e-01 6.78486675e-02 9.68044937e-01
6.58824444e-01 6.91275746e-02 8.48265231e-01 -1.54586864e+00
-6.91179156e-01 7.04502285e-01 -8.48176837e-01 2.55958766e-01
6.46005720e-02 3.44016612e-01 1.25639379e+00 -1.21476924e+00
6.85955584e-01 9.73180950e-01 6.01611435e-01 4.36963022e-01
-1.53744256e+00 -8.81589532e-01 2.43986949e-01 5.42835966e-02
-1.27902985e+00 -3.87060970e-01 7.31955588e-01 -4.29410666e-01
8.70971203e-01 4.60209072e-01 3.56940895e-01 1.22792685e+00
-1.72220156e-01 9.74967599e-01 1.37184167e+00 -2.75879383e-01
1.52510032e-01 3.52650017e-01 2.50994861e-01 6.39160037e-01
1.72884300e-01 3.45837265e-01 -7.13584721e-01 -2.22812686e-02
4.41690356e-01 4.02789474e-01 -3.41893345e-01 -7.38020241e-01
-1.23486841e+00 5.77239513e-01 5.38111031e-01 1.87464282e-01
-3.40182856e-02 -4.47636098e-01 3.95261407e-01 5.38186669e-01
5.22374213e-01 6.65357053e-01 -7.96156585e-01 -6.69497102e-02
-9.97821808e-01 6.64423639e-03 5.41084826e-01 9.70968127e-01
1.05318999e+00 -1.81357577e-01 -2.18818307e-01 1.05785239e+00
-1.19697683e-01 5.48278213e-01 7.24791884e-01 -7.12236702e-01
5.24793327e-01 8.80366564e-01 -4.52531993e-01 -8.23380351e-01
-4.41406071e-01 -9.57137525e-01 -1.05322230e+00 1.01357073e-01
3.72193873e-01 9.34541076e-02 -9.22429025e-01 1.86604548e+00
3.34533334e-01 1.33279637e-01 1.41283780e-01 7.81716943e-01
9.13402200e-01 2.10513845e-01 -1.52744859e-01 -1.08476959e-01
1.15648508e+00 -1.15760207e+00 -3.71633381e-01 -3.25680017e-01
7.54987061e-01 -4.50704575e-01 1.37879109e+00 3.04633290e-01
-8.99769187e-01 -5.96515179e-01 -9.28470612e-01 1.28252760e-01
-5.13488293e-01 -1.15859829e-01 6.94790006e-01 3.94029528e-01
-5.85590959e-01 5.85843027e-01 -6.44817710e-01 -1.70440122e-01
6.86223686e-01 3.11231881e-01 -9.09943342e-01 -1.26742959e-01
-1.00854695e+00 5.58461845e-01 3.19243014e-01 -2.55677462e-01
-7.08152592e-01 -1.09039199e+00 -8.04051697e-01 5.27246110e-02
5.40877223e-01 -6.19837821e-01 8.08193028e-01 -1.14626062e+00
-1.13763714e+00 1.12310028e+00 1.76729247e-01 -4.28721964e-01
5.13514996e-01 -2.30341882e-01 -4.79301512e-01 -8.94310847e-02
3.51184130e-01 8.70010197e-01 6.49363220e-01 -1.30597079e+00
-5.75703323e-01 -5.51730990e-01 -1.54959232e-01 3.86667699e-01
-5.73720634e-01 -2.52273381e-01 -3.38762343e-01 -6.21114671e-01
1.36014983e-01 -9.11973476e-01 -8.10409933e-02 -1.45202413e-01
-4.58348721e-01 -1.49912685e-01 6.78031564e-01 4.83431183e-02
1.11476362e+00 -2.36343050e+00 -5.95959881e-03 2.40973219e-01
3.52481246e-01 1.73171639e-01 -2.07535058e-01 3.36988866e-01
-4.95061785e-01 -7.09816255e-03 -5.18429339e-01 -2.15302378e-01
-6.39774874e-02 1.93735391e-01 -3.43389422e-01 4.68746811e-01
5.02344847e-01 8.28761399e-01 -1.09279513e+00 -4.82817501e-01
-3.75064053e-02 2.06090838e-01 -6.06172085e-01 2.46827722e-01
1.26000673e-01 3.19764912e-01 -1.25303073e-02 8.25505137e-01
6.11420631e-01 -6.51023984e-01 3.13378930e-01 -1.64132535e-01
1.39943734e-01 2.89893568e-01 -1.22076142e+00 1.53474534e+00
-2.85308391e-01 3.47469419e-01 -3.20175707e-01 -1.19547629e+00
9.63308811e-01 -7.55633786e-02 3.75471622e-01 -9.37474668e-01
-2.37030983e-01 3.15928131e-01 5.03205918e-02 -2.41749987e-01
2.23708183e-01 -1.12749793e-01 4.36137393e-02 3.19331586e-01
4.32481140e-01 1.12622805e-01 3.53920758e-01 3.39026272e-01
9.88142312e-01 1.11068293e-01 4.10829455e-01 -4.73110348e-01
3.61705631e-01 -1.96558908e-01 7.29587853e-01 6.68693244e-01
-5.67454621e-02 8.04693758e-01 4.86427754e-01 -2.84821332e-01
-5.76609135e-01 -1.37321842e+00 -5.99188149e-01 1.29334748e+00
2.03897178e-01 -4.55222815e-01 -2.50217587e-01 -1.29096949e+00
2.24024042e-01 4.67595726e-01 -8.36063921e-01 -3.26224476e-01
-1.72297865e-01 -9.94345546e-01 3.69014621e-01 7.01788485e-01
4.07043040e-01 -1.05249095e+00 -9.49488729e-02 -1.40499413e-01
9.10671055e-02 -8.88468504e-01 -4.65567827e-01 9.85278904e-01
-8.37535739e-01 -1.17150748e+00 -4.75922436e-01 -6.82980359e-01
8.82291079e-01 3.88633102e-01 1.52973592e+00 -1.25794858e-01
-3.22379529e-01 2.02132180e-01 -3.37369949e-01 -2.94795573e-01
1.22986317e-01 2.38765880e-01 3.40101242e-01 8.47070366e-02
4.10735220e-01 -8.46468985e-01 -5.99739671e-01 8.00694227e-01
-9.93797660e-01 -6.31456226e-02 6.29586458e-01 1.28714180e+00
8.67180586e-01 -1.70074299e-01 7.09742069e-01 -1.29522729e+00
1.28988430e-01 -6.96316540e-01 -3.23355287e-01 1.67319968e-01
-9.64771390e-01 1.65697664e-01 6.38082802e-01 -4.32138503e-01
-8.87676239e-01 1.85051356e-02 3.52237672e-01 -5.42462826e-01
-2.28932425e-01 4.47058260e-01 -4.02215511e-01 3.17141265e-01
1.08410132e+00 1.24239743e-01 1.12303540e-01 -3.81512791e-01
4.45427746e-01 5.93998611e-01 4.90575790e-01 -7.55962014e-01
9.05296206e-01 5.85603416e-01 -2.28798509e-01 -5.49833715e-01
-1.19220185e+00 -6.74116969e-01 -8.30678403e-01 1.17764778e-01
3.44039023e-01 -1.15495229e+00 -1.80444360e-01 2.05916911e-01
-2.56366283e-01 -4.93503809e-01 -6.52041256e-01 4.52207416e-01
-3.25780332e-01 1.93545073e-01 -1.85269400e-01 -2.28297412e-01
-1.48633614e-01 -8.93388033e-01 1.00191998e+00 1.88298792e-01
-2.25639895e-01 -1.03169250e+00 2.82643020e-01 4.84739155e-01
1.68157756e-01 3.57137807e-02 5.59876144e-01 -1.09284651e+00
-2.57007062e-01 -8.95382315e-02 -2.69930422e-01 4.72135782e-01
5.24753690e-01 2.04331987e-02 -1.10416591e+00 -4.64746982e-01
-3.02696377e-01 -7.67618656e-01 1.12950110e+00 -1.09170191e-02
1.22269917e+00 -1.96180224e-01 -2.74145275e-01 9.84074533e-01
1.27334428e+00 -2.82497078e-01 3.40173572e-01 4.54566598e-01
7.20475972e-01 6.89216614e-01 7.94575810e-01 3.68819118e-01
2.64359057e-01 5.51936090e-01 4.21379358e-01 -3.56285989e-01
-1.91055939e-01 -1.24197975e-01 3.92507792e-01 7.77160645e-01
-4.66923192e-02 2.00039688e-02 -9.58730042e-01 5.75343430e-01
-1.94564855e+00 -1.02820528e+00 1.67392530e-02 2.34821105e+00
1.11079431e+00 3.38805586e-01 2.66480356e-01 8.80582556e-02
6.08772755e-01 1.54962271e-01 -5.33515692e-01 6.49834052e-02
-4.63810742e-01 2.10798174e-01 3.89208198e-01 1.19458370e-01
-1.28931046e+00 6.97784722e-01 5.03926325e+00 1.02448869e+00
-1.30077183e+00 7.06633702e-02 9.19546485e-01 -4.46044385e-01
-3.13169152e-01 -1.60678253e-01 -9.62418079e-01 4.57978100e-01
5.91012716e-01 5.50257303e-02 1.82644695e-01 1.08135962e+00
-3.48556459e-01 1.86261386e-01 -1.45585036e+00 9.82141018e-01
5.62923737e-02 -1.25293100e+00 -2.96300910e-02 5.23351356e-02
1.14463210e+00 3.13276410e-01 3.68951291e-01 5.35754979e-01
2.52174109e-01 -1.03587091e+00 3.93454790e-01 1.61016107e-01
9.11209643e-01 -7.34108984e-01 7.43373454e-01 3.48118722e-01
-1.07738471e+00 -1.08090378e-01 -3.83271217e-01 4.39656340e-02
-3.83665413e-01 8.28155339e-01 -5.78366935e-01 7.11843312e-01
8.38309348e-01 1.10291946e+00 -8.24815571e-01 7.36339211e-01
-1.73737124e-01 6.55104339e-01 -2.55665064e-01 3.12932342e-01
2.52642393e-01 -6.80499896e-02 4.10789907e-01 1.35844564e+00
-7.99063891e-02 -2.49518901e-01 4.71262425e-01 6.67314351e-01
-1.27542853e-01 1.00748025e-01 -6.66407824e-01 2.43549302e-01
4.23600614e-01 1.55595505e+00 -5.76332569e-01 -4.12458360e-01
-4.87805784e-01 5.83481848e-01 6.15651906e-01 1.21869244e-01
-9.07735527e-01 -2.26174131e-01 6.46432698e-01 2.01645494e-01
3.55806291e-01 2.75412709e-01 -2.75185645e-01 -1.38325500e+00
2.04334438e-01 -1.14362657e+00 7.22491026e-01 -2.52761215e-01
-1.95157838e+00 6.12786293e-01 6.15456775e-02 -1.69827998e+00
-1.15069762e-01 -6.63448572e-01 -5.32623291e-01 5.74956357e-01
-1.64576542e+00 -1.08465838e+00 -4.10963327e-01 8.01582515e-01
4.68807131e-01 -4.52149719e-01 7.97937930e-01 3.18827778e-01
-7.33027756e-01 1.05465293e+00 1.52628481e-01 1.82843000e-01
1.31386352e+00 -1.60002172e+00 -2.92819515e-02 5.55372894e-01
6.04421914e-01 7.12205052e-01 2.64496177e-01 -4.37508434e-01
-1.12225282e+00 -1.28526378e+00 4.94427681e-01 -7.37545013e-01
7.73762465e-01 -4.54855055e-01 -1.11319065e+00 5.96218407e-01
1.18346058e-01 6.47933245e-01 1.02952111e+00 5.77485919e-01
-9.12039161e-01 -4.84896898e-01 -9.42010164e-01 4.11120057e-01
1.23724163e+00 -6.93410277e-01 -6.68025851e-01 4.56133574e-01
3.26582193e-01 -2.10151196e-01 -9.86950994e-01 7.38858640e-01
3.94865185e-01 -1.13323534e+00 9.63367283e-01 -8.54801476e-01
7.34593809e-01 -3.29214364e-01 -3.14553976e-01 -1.32268643e+00
-5.09702921e-01 -4.06341016e-01 -4.04589921e-02 1.47707343e+00
8.45350444e-01 -6.97429359e-01 8.82628858e-01 2.81268060e-01
-1.81754529e-01 -1.00932205e+00 -5.30136585e-01 -1.17918158e+00
6.86508343e-02 -1.67963058e-01 4.55719054e-01 1.59303248e+00
1.31444842e-01 5.72812378e-01 -1.70113534e-01 1.56975374e-01
6.70961976e-01 4.21927720e-01 9.23584163e-01 -1.22480798e+00
-5.53280056e-01 -5.00729442e-01 -3.93075764e-01 -7.67378926e-01
3.96844633e-02 -1.43584621e+00 -1.22068658e-01 -8.58167648e-01
6.29771709e-01 -6.39538348e-01 -8.59764457e-01 8.69430900e-01
-4.84010756e-01 4.81713355e-01 1.09938726e-01 3.81591946e-01
-8.06443453e-01 5.80554008e-01 9.96287763e-01 -2.44612172e-01
-2.65010744e-01 -9.86207798e-02 -8.22585344e-01 6.60145879e-01
8.68012667e-01 -5.10799110e-01 -4.84542757e-01 -8.37206319e-02
2.09940314e-01 -3.31891119e-01 3.33463520e-01 -9.95574057e-01
1.36854097e-01 -5.88339344e-02 5.57733059e-01 -3.66338253e-01
5.91855496e-02 -7.71831393e-01 -9.74906310e-02 1.34767056e-01
-5.13763785e-01 -2.00928356e-02 4.95651215e-02 5.35206616e-01
-4.94739026e-01 1.24198757e-01 8.02260816e-01 -1.30923092e-01
-5.84009290e-01 3.72188836e-01 4.15389240e-01 6.24073744e-01
7.82545388e-01 -2.32439861e-01 -6.72462165e-01 -2.00801603e-02
-7.01263309e-01 3.52273405e-01 4.88714397e-01 5.48766255e-01
3.50250334e-01 -1.40033031e+00 -7.83000767e-01 4.23929542e-01
6.49786532e-01 1.76216960e-01 1.71284407e-01 1.04321909e+00
1.06848523e-01 4.91351783e-02 -4.04150277e-01 -1.06922448e+00
-1.27787995e+00 6.24111891e-01 8.13102350e-02 -5.08781672e-01
-4.67385888e-01 9.97176647e-01 6.79082334e-01 -1.02017188e+00
2.28966892e-01 5.41457795e-02 -7.16714635e-02 3.68648171e-01
4.35409367e-01 1.98494464e-01 1.31273523e-01 -3.58857870e-01
-5.45440376e-01 4.87173259e-01 -3.36619109e-01 2.96300113e-01
1.48168170e+00 2.40705714e-01 3.14723030e-02 5.54854333e-01
1.30294299e+00 2.39439324e-01 -1.36693943e+00 -4.28922892e-01
5.89447357e-02 -4.28204983e-01 -1.91284224e-01 -1.03244376e+00
-1.27349448e+00 7.60245860e-01 5.38918793e-01 2.24124163e-01
1.09368682e+00 1.99092031e-01 2.42125288e-01 4.57593322e-01
1.68664455e-01 -9.75223720e-01 3.91888767e-01 4.18447852e-01
7.80680180e-01 -1.71190679e+00 1.94709510e-01 -3.25471729e-01
-8.20862353e-01 7.83264875e-01 1.06689239e+00 -1.05303198e-01
5.71559608e-01 3.52857769e-01 5.67452461e-02 -2.94928193e-01
-9.95896578e-01 -3.20489019e-01 5.60661435e-01 4.42966998e-01
5.68600416e-01 9.92117822e-02 6.14612848e-02 5.64257741e-01
-2.41510615e-01 -3.59850347e-01 5.81449382e-02 8.28435719e-01
-1.26471281e-01 -1.11055851e+00 -1.13718212e-01 5.75005651e-01
-3.85634512e-01 -3.41101408e-01 -4.98389393e-01 1.05604684e+00
1.50528222e-01 4.54980403e-01 2.23877817e-01 -4.52033103e-01
2.52265245e-01 7.55030662e-02 3.11250061e-01 -7.63440967e-01
-8.20514023e-01 -5.98000959e-02 8.00563172e-02 -5.79310894e-01
-4.72396612e-01 -6.05532885e-01 -1.09016037e+00 8.34170282e-02
-4.96783942e-01 1.36539757e-01 1.46902770e-01 7.65848339e-01
5.52828252e-01 4.25153375e-01 1.06442773e+00 -6.63664222e-01
-7.40692973e-01 -8.68585944e-01 -5.34162045e-01 7.39601076e-01
2.13118985e-01 -7.29582310e-01 -7.00549066e-01 -1.24358930e-01] | [9.572707176208496, 3.0152173042297363] |
5db2d13b-b48a-42b6-9863-323dcd156d75 | paraamr-a-large-scale-syntactically-diverse | 2305.16585 | null | https://arxiv.org/abs/2305.16585v1 | https://arxiv.org/pdf/2305.16585v1.pdf | ParaAMR: A Large-Scale Syntactically Diverse Paraphrase Dataset by AMR Back-Translation | Paraphrase generation is a long-standing task in natural language processing (NLP). Supervised paraphrase generation models, which rely on human-annotated paraphrase pairs, are cost-inefficient and hard to scale up. On the other hand, automatically annotated paraphrase pairs (e.g., by machine back-translation), usually suffer from the lack of syntactic diversity -- the generated paraphrase sentences are very similar to the source sentences in terms of syntax. In this work, we present ParaAMR, a large-scale syntactically diverse paraphrase dataset created by abstract meaning representation back-translation. Our quantitative analysis, qualitative examples, and human evaluation demonstrate that the paraphrases of ParaAMR are syntactically more diverse compared to existing large-scale paraphrase datasets while preserving good semantic similarity. In addition, we show that ParaAMR can be used to improve on three NLP tasks: learning sentence embeddings, syntactically controlled paraphrase generation, and data augmentation for few-shot learning. Our results thus showcase the potential of ParaAMR for improving various NLP applications. | ['Aram Galstyan', 'Kai-Wei Chang', 'Anoop Kumar', 'I-Hung Hsu', 'Varun Iyer', 'Kuan-Hao Huang'] | 2023-05-26 | null | null | null | null | ['paraphrase-generation', 'sentence-embeddings', 'sentence-embeddings', 'semantic-textual-similarity', 'semantic-similarity', 'paraphrase-generation'] | ['computer-code', 'methodology', 'natural-language-processing', 'natural-language-processing', 'natural-language-processing', 'natural-language-processing'] | [ 3.57105225e-01 6.23348281e-02 -3.23554635e-01 -4.60416079e-01
-1.16597641e+00 -6.66485012e-01 6.06040418e-01 4.81288791e-01
-1.01184301e-01 8.73455763e-01 8.21786046e-01 -2.17648104e-01
1.86852917e-01 -7.96132445e-01 -7.95139074e-01 -1.72927126e-01
6.47368968e-01 5.79884887e-01 -8.40917155e-02 -6.87307954e-01
5.23133516e-01 1.01642311e-01 -1.42000020e+00 7.82324731e-01
1.17340994e+00 4.73650903e-01 2.31724977e-01 3.45222235e-01
-7.19799995e-01 8.31895769e-01 -6.16106212e-01 -6.41856432e-01
3.72207128e-02 -8.92538488e-01 -1.17610204e+00 -2.84668297e-01
5.76689720e-01 8.27631950e-02 -1.00952551e-01 1.10552847e+00
4.16390806e-01 7.16153681e-02 5.16197741e-01 -1.20859849e+00
-1.56976879e+00 7.53689170e-01 -2.61259258e-01 1.44095346e-01
9.94499683e-01 2.57343240e-02 1.46245480e+00 -1.55610669e+00
7.85200834e-01 1.27843213e+00 6.92751110e-01 8.70545387e-01
-1.36339033e+00 -4.57886070e-01 -4.60167497e-01 3.79516274e-01
-9.94283915e-01 -5.12724400e-01 1.00595367e+00 -1.01212561e-01
1.37300563e+00 2.87292391e-01 5.65061748e-01 1.44598579e+00
2.89212912e-01 7.35762417e-01 9.87798214e-01 -7.78073013e-01
4.24405158e-01 1.74543396e-01 4.26987022e-01 5.61047733e-01
3.47073436e-01 -3.79896760e-01 -7.11322784e-01 -2.93255091e-01
2.18740389e-01 1.67111486e-01 -2.50958890e-01 -3.72399181e-01
-1.13595104e+00 1.21534157e+00 4.31452632e-01 4.17365134e-01
-1.01213917e-01 -8.93280804e-02 7.13161469e-01 8.30312431e-01
5.45751393e-01 1.30433607e+00 -2.50755727e-01 -1.25745207e-01
-9.12064850e-01 3.88343602e-01 9.66062486e-01 1.21375716e+00
7.57854223e-01 -2.15586483e-01 -4.40415531e-01 1.42906439e+00
-1.21331893e-01 6.25423729e-01 1.35972250e+00 -9.56878662e-01
8.18331778e-01 7.35746324e-01 5.65176383e-02 -8.58813941e-01
1.17459163e-01 1.35516062e-01 -7.21589983e-01 -3.61457467e-01
1.20530548e-02 9.30135548e-02 -5.77976644e-01 1.64790988e+00
-2.65532017e-01 -2.22699866e-01 5.15062094e-01 5.94896317e-01
8.83668065e-01 8.81009877e-01 -4.66295145e-02 -2.32295439e-01
1.30653858e+00 -1.40845501e+00 -5.32012284e-01 -7.02968121e-01
8.12892616e-01 -8.90061736e-01 1.80052841e+00 -2.02105969e-01
-1.28053260e+00 -5.74746430e-01 -9.13463652e-01 -3.99574965e-01
-3.17353100e-01 -1.63118139e-01 4.32756662e-01 2.44005576e-01
-8.92986476e-01 8.13994825e-01 -3.27100933e-01 -6.78747654e-01
4.45214957e-01 -2.48424038e-01 -4.71735895e-01 -4.68147397e-01
-1.34356225e+00 1.17168438e+00 4.02813584e-01 -3.99836242e-01
-2.67712712e-01 -1.02284443e+00 -1.21771359e+00 2.62012750e-01
-3.36724184e-02 -1.07021403e+00 1.48232651e+00 -9.55263734e-01
-1.40573227e+00 1.17229664e+00 -4.83801156e-01 -7.38972247e-01
8.04321021e-02 -3.70515227e-01 -2.20824435e-01 1.91979021e-01
5.16647279e-01 6.37555957e-01 6.61622703e-01 -7.26596832e-01
-1.55488938e-01 -1.12348132e-01 -6.59909248e-02 4.11474258e-01
-5.49319029e-01 2.18404040e-01 2.67127246e-01 -8.80750775e-01
-2.06644759e-01 -6.42071724e-01 -3.35827738e-01 -8.42293426e-02
-1.74264356e-01 -3.31784755e-01 4.16771829e-01 -6.47401154e-01
1.13730192e+00 -1.96014309e+00 4.25080091e-01 -4.14077163e-01
-1.55431941e-01 5.22987783e-01 -4.88012582e-01 9.97569680e-01
-2.58169889e-01 3.17609012e-01 -4.14842814e-01 -3.74257475e-01
8.15232564e-03 3.79570067e-01 -9.35244083e-01 -1.52439073e-01
2.91676670e-01 1.47350347e+00 -1.37942755e+00 -5.60283840e-01
9.32765976e-02 -3.15065533e-01 -4.33094054e-01 3.62888128e-01
-2.27240026e-01 -1.93044409e-01 -3.04442883e-01 4.41127300e-01
1.67688340e-01 -2.05408171e-01 -2.08203778e-01 4.78722714e-02
3.41204286e-01 5.79925299e-01 -1.68952808e-01 2.22880793e+00
-1.12218583e+00 6.58576846e-01 -5.29302716e-01 -9.35415506e-01
1.11293650e+00 1.44477025e-01 -3.08041517e-02 -7.74938166e-01
-1.60260141e-01 4.91842568e-01 -2.21596003e-01 -5.64152837e-01
8.13903630e-01 -6.54129028e-01 -4.69324529e-01 7.04795599e-01
8.54341462e-02 -8.68794441e-01 5.04915237e-01 4.28649366e-01
1.19792807e+00 -1.51856676e-01 8.55773687e-01 -1.96419626e-01
5.13198078e-01 5.75258315e-01 2.35453427e-01 8.57822776e-01
-1.73913628e-01 9.29672003e-01 2.91618764e-01 -2.90788382e-01
-1.50942039e+00 -1.28879917e+00 1.30937889e-01 8.68096292e-01
1.04973599e-01 -5.53573430e-01 -6.28572285e-01 -6.23088360e-01
8.25544968e-02 1.20415497e+00 -4.04038727e-01 -6.08087063e-01
-7.69114733e-01 -3.41939121e-01 5.85824370e-01 6.92007482e-01
2.28284597e-01 -1.46379101e+00 -1.42188966e-01 4.08281237e-01
-4.13011044e-01 -1.07525778e+00 -6.59130275e-01 -2.08122775e-01
-9.17239845e-01 -7.82059729e-01 -8.85322332e-01 -1.29630971e+00
5.91340125e-01 6.33348942e-01 1.52986014e+00 -2.58248538e-01
-2.13477433e-01 1.94009990e-02 -7.85103679e-01 -2.31162116e-01
-1.11029768e+00 7.76025206e-02 6.19839644e-03 -5.86730957e-01
7.06322968e-01 -8.74834061e-01 -3.18226993e-01 -7.60457739e-02
-6.84834421e-01 3.20996493e-01 6.11570239e-01 1.20183110e+00
4.84632313e-01 -6.82106197e-01 1.08166742e+00 -9.11326528e-01
1.31621408e+00 -6.04365468e-01 1.46236002e-01 6.37100935e-01
-4.79587436e-01 2.15471447e-01 1.24799883e+00 -2.91446120e-01
-8.56643915e-01 -3.12083989e-01 -1.62889704e-01 -4.93405759e-01
-5.12481034e-02 4.09552038e-01 2.26820916e-01 3.26832682e-01
1.14759672e+00 6.50701880e-01 2.10240379e-01 -3.93956721e-01
8.04632187e-01 9.07489717e-01 4.43846941e-01 -5.13286769e-01
8.94746244e-01 1.61756620e-01 -2.28253305e-01 -8.67191017e-01
-1.43543351e+00 -4.89545405e-01 -3.27964693e-01 4.22169417e-01
5.65678835e-01 -9.08589423e-01 4.64310020e-01 1.29958317e-02
-1.48778641e+00 4.55249026e-02 -1.02622890e+00 2.08770041e-04
-6.66432559e-01 7.46590555e-01 -6.95954800e-01 -1.46650702e-01
-9.62850451e-01 -5.74983716e-01 1.18990445e+00 1.05322137e-01
-8.62890542e-01 -9.23879623e-01 4.70587939e-01 6.21324480e-01
5.58409691e-01 -1.72091186e-01 1.18053687e+00 -1.01113999e+00
5.20450957e-02 -1.74271956e-01 -3.39069247e-01 5.34573197e-01
4.38176006e-01 -4.35234994e-01 -5.37846684e-01 -1.21272594e-01
1.71970099e-01 -9.19274032e-01 7.00563014e-01 -1.22244202e-01
8.89129460e-01 -5.98254204e-01 1.01825707e-02 2.62249976e-01
1.14267325e+00 -3.68024558e-01 4.27671254e-01 1.09259583e-01
5.01989245e-01 6.02678776e-01 7.64656126e-01 1.08921759e-01
-3.56382802e-02 5.76739669e-01 -2.95299441e-01 2.77928144e-01
-5.52343011e-01 -8.53271663e-01 5.59422135e-01 1.27400029e+00
6.29336298e-01 -3.49594392e-02 -7.91997910e-01 7.63469100e-01
-1.91985726e+00 -1.13422084e+00 -5.47314584e-02 1.73823822e+00
1.36908948e+00 -1.06646791e-01 -2.80477941e-01 -5.45695126e-02
6.42170846e-01 2.92708874e-01 -3.81003112e-01 -1.01959717e+00
-2.02971235e-01 6.85926437e-01 -1.43299222e-01 3.76096815e-01
-5.50289750e-01 1.11663055e+00 5.77203417e+00 1.00612128e+00
-3.97395939e-01 2.63434201e-01 1.14484645e-01 -1.72016725e-01
-6.35728061e-01 1.74752101e-01 -5.30274928e-01 6.04612052e-01
9.87599969e-01 -8.62008810e-01 4.52108353e-01 1.17451143e+00
1.52146772e-01 2.80963957e-01 -1.49186456e+00 1.07879138e+00
6.45510674e-01 -1.69607830e+00 6.94749177e-01 -6.09833360e-01
8.09980333e-01 -8.41426253e-02 -3.42403561e-01 6.98746860e-01
3.87240320e-01 -1.00740683e+00 3.86191219e-01 -9.54642240e-03
6.99070930e-01 -6.23242259e-01 7.97964394e-01 3.59896958e-01
-9.25074458e-01 -1.22488044e-01 -8.33087385e-01 -3.37948322e-01
5.60243964e-01 5.80493271e-01 -6.83266699e-01 4.42988247e-01
2.63163775e-01 9.75246370e-01 -8.29171538e-01 6.49588168e-01
-7.15683699e-01 3.95196766e-01 1.47438169e-01 -5.27722597e-01
1.45520374e-01 -1.51426554e-01 5.38937867e-01 1.29151535e+00
4.22492683e-01 -2.25090295e-01 -5.36568724e-02 1.13160658e+00
-3.18102747e-01 3.96202624e-01 -1.16472805e+00 -3.11356127e-01
6.12086594e-01 9.16418910e-01 3.41151794e-03 -6.58692241e-01
-4.22406971e-01 1.48185253e+00 6.58083975e-01 1.14406846e-01
-6.47405565e-01 -8.23181629e-01 6.66857481e-01 -1.34927705e-01
-6.41309991e-02 4.33478095e-02 -4.85144675e-01 -1.73877394e+00
3.07792038e-01 -8.13865304e-01 4.88408394e-02 -1.22047615e+00
-2.01394176e+00 5.29465079e-01 -4.68372665e-02 -1.20719957e+00
-5.24313331e-01 -3.86813432e-01 -1.01782143e+00 9.42907870e-01
-1.31431365e+00 -1.21024656e+00 -1.33257836e-01 1.72786206e-01
1.43089604e+00 -3.35692793e-01 1.12353039e+00 -7.93354511e-02
-3.56299788e-01 5.81270635e-01 3.55931222e-02 2.96691190e-02
6.64827347e-01 -1.25060749e+00 8.99379671e-01 8.18975151e-01
4.48882639e-01 8.59231412e-01 8.08666289e-01 -4.48201060e-01
-1.27995527e+00 -1.35658324e+00 1.59184229e+00 -5.08659065e-01
1.06637490e+00 -4.15156305e-01 -1.08813334e+00 4.44249898e-01
5.30839801e-01 -3.12049687e-01 8.13075960e-01 9.38897058e-02
-7.08844364e-01 1.12978570e-01 -1.13739657e+00 9.87708926e-01
1.28704274e+00 -8.57653379e-01 -1.78023386e+00 7.43510842e-01
1.20143604e+00 2.97114756e-02 -4.97686863e-01 6.28626794e-02
2.28924066e-01 -7.61028051e-01 1.08432102e+00 -1.09200597e+00
1.32295454e+00 2.73265749e-01 -1.90434918e-01 -1.72044683e+00
-3.02126139e-01 -5.26550949e-01 -1.09405063e-01 1.39254427e+00
4.68488455e-01 -4.36551988e-01 7.21625209e-01 4.99598950e-01
-2.81100333e-01 -7.10277498e-01 -9.52610314e-01 -1.22449315e+00
5.24708450e-01 4.13690619e-02 4.38895226e-01 9.51268673e-01
8.17263722e-01 1.09005296e+00 -1.72205538e-01 -5.56796491e-01
5.05533636e-01 6.50274277e-01 6.60848022e-01 -7.90934384e-01
-4.61316884e-01 -4.12889481e-01 -1.39457300e-01 -9.54249918e-01
7.49266505e-01 -1.46517479e+00 3.51396874e-02 -1.75704432e+00
6.36350513e-01 4.43441793e-02 1.51304409e-01 4.37332273e-01
-2.89016843e-01 8.40366483e-02 1.52199119e-01 3.68385017e-01
-3.25014949e-01 7.88013041e-01 1.13376927e+00 -2.49564499e-01
-1.21663146e-01 -3.02103400e-01 -8.71912301e-01 7.02012181e-01
1.00138772e+00 -5.59383392e-01 -4.97306794e-01 -5.46150982e-01
2.56619930e-01 -3.50855254e-02 2.28046626e-01 -7.19355464e-01
-4.74182591e-02 -4.14576083e-01 -1.72193810e-01 -1.29993513e-01
3.15187186e-01 -3.78490895e-01 -3.57221782e-01 5.19876480e-01
-7.94440746e-01 4.29942578e-01 -4.84253354e-02 5.38549364e-01
-4.19994831e-01 -8.52161646e-01 7.90162563e-01 -4.98568177e-01
-5.96138477e-01 -2.44631127e-01 -2.67896235e-01 7.57528841e-01
8.89390826e-01 -2.58710146e-01 -4.62521970e-01 -2.46526748e-01
-1.69939354e-01 1.18326284e-01 7.16205657e-01 7.36214936e-01
9.37152028e-01 -1.56501496e+00 -8.06296885e-01 3.53812762e-02
5.87468505e-01 -2.24851862e-01 3.57171372e-02 1.73516497e-01
-4.23884511e-01 3.69360983e-01 -2.21501738e-01 -8.92815515e-02
-1.18994045e+00 7.47395039e-01 -1.58259496e-01 -4.03274804e-01
-7.20785618e-01 7.08369553e-01 -7.89643750e-02 -4.84271288e-01
-2.39095911e-01 -5.20434380e-01 7.04416186e-02 4.52063419e-03
6.06112838e-01 1.07228272e-01 -6.20046929e-02 -2.83520073e-01
-1.10845357e-01 4.92722213e-01 -3.38678539e-01 -9.79379788e-02
1.28714395e+00 -5.77528067e-02 -2.99097419e-01 3.63998950e-01
1.44856083e+00 -1.80972710e-01 -2.62192935e-01 -3.42375666e-01
1.55462906e-01 -5.58560610e-01 -6.53793097e-01 -6.07904613e-01
-3.31751108e-01 9.94246721e-01 -2.24586517e-01 5.80207333e-02
7.31171191e-01 1.87591150e-01 1.45499098e+00 9.36302245e-01
5.08789718e-01 -9.87811983e-01 5.75018942e-01 6.98114753e-01
1.33943474e+00 -1.12490118e+00 -1.96375340e-01 -3.34116340e-01
-8.31652105e-01 9.60946858e-01 5.48478305e-01 -4.16012645e-01
6.87603578e-02 -2.63933003e-01 -1.28466979e-01 4.38637473e-02
-9.27438557e-01 2.07394511e-01 -4.16883230e-02 5.84225357e-01
3.53439718e-01 -2.38370240e-01 -6.09986544e-01 6.36645675e-01
-6.38345301e-01 1.11437097e-01 8.16084564e-01 1.00643027e+00
-6.55568480e-01 -1.37703085e+00 6.96682632e-02 5.35516620e-01
-9.86124203e-02 -5.51820338e-01 -7.89256155e-01 2.79148519e-01
-5.24560809e-01 9.67961550e-01 -1.00468911e-01 -7.66490353e-03
4.08231467e-01 4.23183352e-01 5.63169777e-01 -1.41329122e+00
-5.81452072e-01 -9.82626498e-01 3.82625252e-01 -3.92208666e-01
-2.22907569e-02 -3.36840481e-01 -1.01805544e+00 -1.95670262e-01
-3.80008481e-02 4.72574502e-01 3.37398291e-01 8.36356819e-01
6.73194468e-01 6.43011332e-02 6.52789235e-01 -5.89315176e-01
-1.31111324e+00 -1.10046709e+00 -2.96614408e-01 1.13323021e+00
-2.71991752e-02 -7.46406242e-02 -5.20547211e-01 -2.61691008e-02] | [11.63687515258789, 9.250875473022461] |
4f2b654d-5d68-4743-863b-21e67c27492c | open-set-semantic-segmentation-for-point | null | null | http://openaccess.thecvf.com//content/CVPR2023/html/Li_Open-Set_Semantic_Segmentation_for_Point_Clouds_via_Adversarial_Prototype_Framework_CVPR_2023_paper.html | http://openaccess.thecvf.com//content/CVPR2023/papers/Li_Open-Set_Semantic_Segmentation_for_Point_Clouds_via_Adversarial_Prototype_Framework_CVPR_2023_paper.pdf | Open-Set Semantic Segmentation for Point Clouds via Adversarial Prototype Framework | Recently, point cloud semantic segmentation has attracted much attention in computer vision. Most of the existing works in literature assume that the training and testing point clouds have the same object classes, but they are generally invalid in many real-world scenarios for identifying the 3D objects whose classes are not seen in the training set. To address this problem, we propose an Adversarial Prototype Framework (APF) for handling the open-set 3D semantic segmentation task, which aims to identify 3D unseen-class points while maintaining the segmentation performance on seen-class points. The proposed APF consists of a feature extraction module for extracting point features, a prototypical constraint module, and a feature adversarial module. The prototypical constraint module is designed to learn prototypes for each seen class from point features. The feature adversarial module utilizes generative adversarial networks to estimate the distribution of unseen-class features implicitly, and the synthetic unseen-class features are utilized to prompt the model to learn more effective point features and prototypes for discriminating unseen-class samples from the seen-class ones. Experimental results on two public datasets demonstrate that the proposed APF outperforms the comparative methods by a large margin in most cases. | ['Qiulei Dong', 'Jianan Li'] | 2023-01-01 | null | null | null | cvpr-2023-1 | ['3d-semantic-segmentation'] | ['computer-vision'] | [ 1.25961989e-01 8.82293284e-02 -4.21952493e-02 -4.42328215e-01
-6.84863925e-01 -7.04640388e-01 5.73852777e-01 -1.74179435e-01
-1.37189806e-01 2.45705500e-01 -6.60387397e-01 -5.03043123e-02
3.71695049e-02 -1.04821169e+00 -9.45075810e-01 -8.75542760e-01
2.15724885e-01 9.36785221e-01 5.68824649e-01 -9.19998682e-04
3.58102769e-01 9.18115020e-01 -1.77367806e+00 3.63976732e-02
1.12717044e+00 1.21265388e+00 1.10101245e-01 2.86156893e-01
-4.55577433e-01 8.80542863e-03 -8.57472360e-01 -1.87134251e-01
7.88701355e-01 -2.33436488e-02 -5.04210651e-01 4.46580470e-01
5.27765393e-01 -2.39882872e-01 -1.19990677e-01 1.32760322e+00
3.25198501e-01 2.75648028e-01 1.10716593e+00 -1.67681515e+00
-7.34731972e-01 -6.46294281e-02 -5.69727063e-01 -4.87753861e-02
2.38368437e-01 2.10285768e-01 6.02823913e-01 -9.78829801e-01
3.24268043e-01 1.45602393e+00 5.75997651e-01 7.28583813e-01
-9.74083543e-01 -8.19933057e-01 2.33559027e-01 4.57750261e-02
-1.51917601e+00 1.23416306e-02 1.28207040e+00 -4.50167239e-01
4.64410752e-01 3.94557893e-01 7.11880505e-01 1.01489174e+00
-1.34541988e-02 9.00295675e-01 9.23243165e-01 -7.84811005e-02
3.92143101e-01 4.90110934e-01 2.39953488e-01 2.60493010e-01
2.05953345e-01 3.06155801e-01 2.51100689e-01 -3.89819384e-01
8.43184173e-01 3.63595307e-01 -2.02252656e-01 -7.31476724e-01
-7.76852727e-01 8.93903792e-01 7.51910567e-01 8.55620652e-02
-3.38496476e-01 -2.60821253e-01 1.26828015e-01 1.10858858e-01
5.87651610e-01 2.42234498e-01 -3.47947538e-01 5.24992049e-01
-7.83398151e-01 5.40349126e-01 6.63112998e-01 1.30492508e+00
8.93573344e-01 -6.62727579e-02 4.48098360e-03 7.51739264e-01
5.12985289e-01 8.63402069e-01 5.05718112e-01 -6.93216562e-01
3.31810802e-01 1.03763139e+00 1.30425230e-01 -1.05056977e+00
9.08756033e-02 -3.64770412e-01 -5.14824688e-01 3.84813309e-01
2.97188044e-01 1.56285957e-01 -1.34869206e+00 1.26672864e+00
8.94211590e-01 5.68232775e-01 1.53569847e-01 1.16076267e+00
9.68940616e-01 8.47951233e-01 -1.31834462e-01 2.40639791e-01
9.51971114e-01 -6.56864882e-01 -8.57790038e-02 -3.10875863e-01
8.38308558e-02 -6.66883349e-01 9.40289080e-01 -1.88825410e-02
-7.38130510e-01 -7.31242001e-01 -1.13135374e+00 2.54207879e-01
-4.98778433e-01 -1.84921294e-01 3.45022023e-01 5.83010077e-01
-5.38404942e-01 4.70460683e-01 -7.06551433e-01 -1.71327040e-01
9.01962757e-01 4.71695960e-01 -2.35348701e-01 -1.49283677e-01
-7.85383105e-01 4.83397007e-01 8.04987311e-01 2.79878285e-02
-1.11328828e+00 -7.34368861e-01 -8.15885603e-01 -7.54394606e-02
2.61092722e-01 -6.28267586e-01 8.96118045e-01 -1.21583581e+00
-1.06091344e+00 1.16917956e+00 1.65167943e-01 -7.99193904e-02
6.63984895e-01 2.52613463e-02 -2.23101094e-01 1.75925359e-01
2.17025474e-01 6.61994219e-01 1.33689451e+00 -1.74471617e+00
-7.50470996e-01 -7.59692729e-01 -8.89370590e-02 2.61809468e-01
1.22566797e-01 -5.09421110e-01 -4.79041129e-01 -5.69711506e-01
6.36428595e-01 -8.99098039e-01 -2.00879291e-01 2.04008654e-01
-6.71524048e-01 -3.82180959e-01 1.42035735e+00 -2.73460168e-02
1.69302791e-01 -2.24142814e+00 -3.24393203e-03 4.27192181e-01
6.84270114e-02 3.14747751e-01 9.79038235e-03 -4.77237394e-03
-1.65332928e-01 -1.27041359e-02 -3.97750854e-01 -9.26659405e-02
1.81513336e-02 3.97719860e-01 -5.70317030e-01 6.24378026e-01
6.47898734e-01 7.92500496e-01 -9.12917852e-01 -5.12358785e-01
6.22005105e-01 4.10409123e-01 -3.38297725e-01 6.10820591e-01
-3.86624545e-01 6.11543894e-01 -1.09442067e+00 9.21540320e-01
1.13852668e+00 3.03637125e-02 -5.62658966e-01 7.45251253e-02
3.71456534e-01 -4.70258564e-01 -1.27602160e+00 1.38920343e+00
7.18280450e-02 -4.10385691e-02 -1.61999971e-01 -1.01470351e+00
1.19098222e+00 -3.64979543e-02 3.19982260e-01 -1.74098879e-01
3.48878711e-01 3.76522720e-01 -1.34544134e-01 -5.58724821e-01
2.76170969e-01 -2.69644767e-01 -3.48138884e-02 4.35979329e-02
-1.27171174e-01 -8.59150887e-01 -3.85304153e-01 -1.58614755e-01
6.62540138e-01 1.77257910e-01 -5.79250194e-02 -9.83619243e-02
6.66646361e-01 3.09305310e-01 7.41560638e-01 6.36486292e-01
-4.36775774e-01 8.37163985e-01 2.38367692e-02 -4.62112427e-01
-9.96460497e-01 -1.38869333e+00 -4.51707631e-01 4.00683194e-01
9.81945217e-01 5.06148994e-01 -8.66690338e-01 -1.18467116e+00
3.89012009e-01 7.95960784e-01 -7.07437456e-01 -3.43951166e-01
-3.26745629e-01 -3.04178894e-01 3.15081179e-01 4.62301731e-01
5.40112376e-01 -1.07184041e+00 -5.07360816e-01 -1.55110374e-01
2.60359377e-01 -9.48342562e-01 -2.54091352e-01 2.21657995e-02
-8.77382815e-01 -1.29034722e+00 -8.60078633e-01 -1.03733790e+00
1.09749031e+00 3.64700884e-01 8.08509350e-01 1.41162351e-01
-1.68612346e-01 2.63420939e-01 -5.19064784e-01 -8.22710276e-01
-4.83238608e-01 -1.77050978e-01 -3.98940183e-02 2.34118536e-01
7.27474391e-01 -4.55978870e-01 -5.18995702e-01 5.67665100e-01
-9.75264370e-01 -3.14942926e-01 5.17371893e-01 7.87302375e-01
9.09429908e-01 6.18984811e-02 5.64462602e-01 -9.99921978e-01
8.28517303e-02 -7.05006838e-01 -5.42215645e-01 3.44543755e-02
-3.57650012e-01 -2.27672949e-01 6.92102551e-01 -6.21778667e-01
-6.95904732e-01 4.50130887e-02 -1.81491882e-01 -1.18261540e+00
-6.40312076e-01 -1.64447546e-01 -6.31130755e-01 -1.90249518e-01
5.60443699e-01 3.84734869e-01 -7.05371425e-02 -3.29486042e-01
2.29540929e-01 7.48049438e-01 5.61379611e-01 -7.48378992e-01
1.54506993e+00 7.37773836e-01 -1.66535806e-02 -7.64937878e-01
-7.03854799e-01 -7.12381661e-01 -7.10788190e-01 -2.25589693e-01
9.43692088e-01 -7.51973927e-01 -7.59751424e-02 5.71690202e-01
-1.08685625e+00 1.38743505e-01 -6.99441016e-01 2.56651610e-01
-7.14462757e-01 2.78638273e-01 -5.52238077e-02 -7.70875990e-01
-2.21659213e-01 -1.47477353e+00 1.52998900e+00 5.96346259e-01
1.67838290e-01 -8.08918774e-01 -2.29192197e-01 3.94726545e-01
-1.96387678e-01 7.74590611e-01 9.38456118e-01 -1.11414635e+00
-7.93155551e-01 -7.27233112e-01 -8.18344951e-02 6.23113394e-01
9.29547250e-02 -2.33938862e-02 -1.25799870e+00 -3.88847232e-01
3.20173085e-01 -3.17543149e-01 4.23008204e-01 2.80357361e-01
1.39602768e+00 5.86130656e-02 -6.22597992e-01 8.08675051e-01
1.32943797e+00 4.97407645e-01 5.03826678e-01 1.12953827e-01
7.92220175e-01 3.84010434e-01 9.98427868e-01 3.86420563e-02
-1.06291652e-01 3.32634538e-01 9.02853906e-01 -6.82035238e-02
2.14035437e-01 -3.84936094e-01 -2.06890225e-01 2.99796760e-01
2.61535108e-01 -3.30706120e-01 -8.27125490e-01 6.43083811e-01
-1.62956655e+00 -7.60024905e-01 1.23282745e-01 2.10765815e+00
3.06129098e-01 3.25804979e-01 -2.09854506e-02 3.84835035e-01
1.09025574e+00 4.07922044e-02 -9.86060858e-01 -4.96572368e-02
1.41098462e-02 2.08161533e-01 3.03137988e-01 -7.99138751e-03
-1.22258675e+00 7.76992917e-01 4.76786709e+00 9.54974949e-01
-1.01619053e+00 -9.67385471e-02 5.07615864e-01 3.68308842e-01
-3.64372939e-01 6.13286644e-02 -6.25985622e-01 5.10699272e-01
2.21737489e-01 6.05313387e-03 -5.46490364e-02 1.33820188e+00
-1.03352532e-01 3.63392949e-01 -1.24673891e+00 1.11180258e+00
9.17494744e-02 -1.01791072e+00 4.71755952e-01 8.84501487e-02
6.93132937e-01 -5.79059720e-02 1.74158514e-01 3.75793666e-01
-1.55115038e-01 -8.78388584e-01 7.26931214e-01 5.29120803e-01
6.54968798e-01 -9.63368177e-01 7.04755008e-01 7.67922699e-01
-9.29250836e-01 -3.74915302e-02 -6.31680250e-01 3.79206926e-01
-2.10777223e-01 4.89768177e-01 -9.24392104e-01 6.79587543e-01
6.93823516e-01 7.45499909e-01 -4.05453742e-01 1.24782956e+00
-1.39238447e-01 3.73704970e-01 -4.17651832e-01 8.84926040e-03
5.02162814e-01 -4.83325958e-01 1.06230462e+00 5.04338384e-01
1.18956663e-01 7.58666396e-02 4.43331808e-01 1.20368135e+00
-5.44623546e-02 -8.13083164e-03 -7.70833492e-01 1.08950764e-01
6.68422401e-01 1.02219677e+00 -9.10173237e-01 -2.01161772e-01
-1.44276261e-01 9.49989200e-01 5.44122644e-02 2.69026011e-01
-8.37758601e-01 -6.84113026e-01 6.77892983e-01 1.43639147e-01
3.54952067e-01 1.17314264e-01 -1.92164838e-01 -1.04082370e+00
1.85312614e-01 -5.92404783e-01 2.27389798e-01 -7.73983061e-01
-1.70953965e+00 5.68411648e-01 -4.90617268e-02 -1.67406678e+00
-1.34292886e-01 -6.80080891e-01 -8.80645156e-01 1.08510423e+00
-1.53491926e+00 -1.36045432e+00 -5.40986001e-01 6.62849188e-01
8.16006184e-01 -1.51655614e-01 6.19123518e-01 5.73299974e-02
-3.83969218e-01 4.16385949e-01 1.62296385e-01 3.44260156e-01
2.25690424e-01 -1.28720915e+00 3.51877421e-01 6.50684297e-01
-9.37766880e-02 3.69318157e-01 5.59485555e-01 -6.90080106e-01
-1.21541429e+00 -1.46182406e+00 1.39684707e-01 -6.79241478e-01
1.85166508e-01 -4.29154694e-01 -1.08157945e+00 5.86750090e-01
-4.70094204e-01 4.51644123e-01 3.94662470e-01 -4.29914951e-01
-1.90571293e-01 7.69039467e-02 -1.76985872e+00 2.95904547e-01
8.53086829e-01 -2.24981219e-01 -1.09126103e+00 4.13331836e-01
8.59188020e-01 -7.75259376e-01 -7.12598443e-01 6.02377832e-01
1.60842121e-01 -7.11223841e-01 1.17517316e+00 -6.22237802e-01
3.15806597e-01 -6.29938781e-01 -3.02689224e-01 -1.24345517e+00
2.00449359e-02 -5.94149344e-02 4.65458333e-02 1.19471753e+00
-5.73169300e-03 -7.81610191e-01 1.13639784e+00 4.67983246e-01
-4.43761021e-01 -8.16360116e-01 -1.07562876e+00 -9.99185503e-01
2.38342062e-01 -3.91768247e-01 9.99691308e-01 1.05652666e+00
-9.79628265e-01 9.56667289e-02 3.42753112e-01 6.66475236e-01
9.11077619e-01 6.19393647e-01 1.05343020e+00 -1.50181508e+00
-6.03732951e-02 -2.27048442e-01 -1.04343259e+00 -9.65803564e-01
3.98584396e-01 -1.02666426e+00 6.39505833e-02 -1.27611375e+00
3.63129377e-03 -1.01166809e+00 -8.77822191e-02 1.10299595e-01
-3.32061499e-01 1.15533866e-01 6.58234879e-02 4.17726249e-01
-8.18028674e-02 7.78485060e-01 1.49845386e+00 -5.81356108e-01
-2.82935590e-01 4.97213662e-01 -4.70912784e-01 8.93114209e-01
4.76701647e-01 -5.67498922e-01 -6.91944063e-01 -3.06461513e-01
-5.82801282e-01 -2.86663294e-01 7.83288717e-01 -1.09824038e+00
-1.66698694e-01 -3.39173257e-01 7.32445300e-01 -1.05051363e+00
4.83885080e-01 -1.38792121e+00 2.28003487e-02 3.02004397e-01
1.09302960e-01 -4.93059814e-01 2.02884451e-01 8.22063148e-01
-3.79406437e-02 -4.37030733e-01 1.00441039e+00 -1.26780435e-01
-7.61086524e-01 7.33373046e-01 3.55554670e-01 3.09310466e-01
1.65579987e+00 -8.12638462e-01 -5.54450527e-02 1.09018043e-01
-6.30228937e-01 3.76192838e-01 6.99535787e-01 6.04878426e-01
1.02212358e+00 -1.23219895e+00 -6.03013217e-01 6.65379524e-01
3.57362032e-01 9.72737551e-01 1.59999564e-01 -5.18493960e-03
-5.71804881e-01 -1.57999918e-01 -6.89521059e-02 -1.16996038e+00
-8.34302962e-01 8.12408566e-01 5.66995084e-01 3.92873228e-01
-6.01800025e-01 8.71600032e-01 3.81296068e-01 -7.37903059e-01
6.92786872e-02 -2.36344799e-01 4.38208021e-02 -2.95992166e-01
1.32795423e-01 2.01617301e-01 -1.19766600e-01 -8.68018985e-01
-2.14359924e-01 7.93267012e-01 -1.51916385e-01 4.13596094e-01
1.24115551e+00 1.98429465e-01 1.78244665e-01 6.08895898e-01
1.27569056e+00 -2.51179338e-01 -1.29876888e+00 -2.38053247e-01
-3.56652498e-01 -7.66336620e-01 -2.12763742e-01 -5.07374525e-01
-1.04269695e+00 8.26745927e-01 8.67603362e-01 3.80689919e-01
9.93662238e-01 2.21408248e-01 8.84023070e-01 1.77016482e-01
6.44556403e-01 -7.65374839e-01 -1.50876224e-01 2.88631678e-01
7.45077550e-01 -1.29699802e+00 -2.94618577e-01 -8.27930510e-01
-3.26829344e-01 9.36858594e-01 8.33482981e-01 -7.33269095e-01
7.63723969e-01 -3.05767208e-01 1.31736055e-01 -3.88558358e-01
3.68797518e-02 1.82208437e-02 4.47122157e-01 9.51005459e-01
-5.68120420e-01 6.70015439e-02 2.17083350e-01 6.77305996e-01
-2.02010423e-01 -3.72798502e-01 2.17876844e-02 9.07742560e-01
-4.62876737e-01 -7.59174407e-01 -6.68456495e-01 5.65628290e-01
-2.70669702e-02 3.96657735e-01 -4.35565889e-01 1.09870183e+00
4.96940672e-01 6.71235681e-01 4.17509615e-01 -4.64430958e-01
5.40747762e-01 -7.03223143e-03 3.12101603e-01 -7.31969893e-01
-4.92967814e-01 -2.16587871e-01 -5.76969504e-01 -2.87002176e-01
-2.73568451e-01 -5.92776060e-01 -1.46108055e+00 8.40075314e-02
-5.93807399e-01 3.10672194e-01 6.36152148e-01 8.45004499e-01
1.64802790e-01 3.13111037e-01 1.15416181e+00 -1.07774353e+00
-8.76344502e-01 -6.45490408e-01 -7.41097510e-01 8.54384243e-01
3.41969728e-01 -9.88091528e-01 -6.85344636e-01 -2.39823490e-01] | [8.02165412902832, -3.268444776535034] |
027363d7-baea-4146-9ae3-349a66525841 | unifying-architectures-tasks-and-modalities | 2202.03052 | null | https://arxiv.org/abs/2202.03052v2 | https://arxiv.org/pdf/2202.03052v2.pdf | OFA: Unifying Architectures, Tasks, and Modalities Through a Simple Sequence-to-Sequence Learning Framework | In this work, we pursue a unified paradigm for multimodal pretraining to break the scaffolds of complex task/modality-specific customization. We propose OFA, a Task-Agnostic and Modality-Agnostic framework that supports Task Comprehensiveness. OFA unifies a diverse set of cross-modal and unimodal tasks, including image generation, visual grounding, image captioning, image classification, language modeling, etc., in a simple sequence-to-sequence learning framework. OFA follows the instruction-based learning in both pretraining and finetuning stages, requiring no extra task-specific layers for downstream tasks. In comparison with the recent state-of-the-art vision & language models that rely on extremely large cross-modal datasets, OFA is pretrained on only 20M publicly available image-text pairs. Despite its simplicity and relatively small-scale training data, OFA achieves new SOTAs in a series of cross-modal tasks while attaining highly competitive performances on uni-modal tasks. Our further analysis indicates that OFA can also effectively transfer to unseen tasks and unseen domains. Our code and models are publicly available at https://github.com/OFA-Sys/OFA. | ['Hongxia Yang', 'Jingren Zhou', 'Chang Zhou', 'Jianxin Ma', 'Zhikang Li', 'Shuai Bai', 'Junyang Lin', 'Rui Men', 'An Yang', 'Peng Wang'] | 2022-02-07 | null | null | null | null | ['self-supervised-image-classification', 'object-categorization', 'visual-entailment'] | ['computer-vision', 'computer-vision', 'reasoning'] | [ 4.23612535e-01 -1.17586657e-01 -1.25291035e-01 -4.97214764e-01
-1.10159707e+00 -9.21062231e-01 1.02433753e+00 -1.65344536e-01
-6.59820855e-01 4.83597040e-01 7.16859549e-02 -5.87997735e-01
2.60744184e-01 -2.89827913e-01 -9.05878544e-01 -4.06398624e-01
5.38170993e-01 5.09673178e-01 3.12872529e-02 -3.48776042e-01
6.37668073e-02 1.10224947e-01 -1.62456083e+00 9.49963212e-01
5.80819666e-01 9.87373650e-01 4.82606769e-01 7.93784261e-01
-2.42494792e-01 4.40604329e-01 -1.92948058e-01 -6.41685605e-01
-1.53290248e-02 -2.93991268e-01 -1.16506112e+00 2.09532082e-01
9.34423745e-01 -1.88900486e-01 -5.54987431e-01 7.07657397e-01
6.97134793e-01 6.06532805e-02 7.79883444e-01 -1.46027851e+00
-1.17310834e+00 4.84166294e-01 -4.42227304e-01 -1.44144055e-02
3.25387985e-01 6.56485379e-01 8.38326514e-01 -1.25439799e+00
5.47173142e-01 1.23885334e+00 4.08804685e-01 1.13175225e+00
-1.28351212e+00 -7.06884384e-01 3.19773048e-01 7.16819540e-02
-1.23703516e+00 -7.38443077e-01 4.66847390e-01 -6.02079272e-01
1.04681575e+00 1.11789048e-01 1.83106571e-01 1.83936965e+00
-1.94729552e-01 1.08577836e+00 1.21623552e+00 -5.45041502e-01
-1.21372119e-01 5.24112098e-02 9.47752818e-02 6.88710690e-01
-8.74491408e-02 -1.10993776e-02 -7.72739172e-01 1.92273602e-01
6.55816495e-01 -1.84208974e-01 -2.45443270e-01 -3.41269284e-01
-1.75746584e+00 6.08893514e-01 3.37222010e-01 2.70079434e-01
-3.04441378e-02 1.82572514e-01 6.86068177e-01 3.60218912e-01
3.09874918e-02 3.32449168e-01 -5.94815969e-01 6.25390261e-02
-8.41403484e-01 8.90041818e-04 3.84005219e-01 1.12806094e+00
6.95939660e-01 1.31326884e-01 -6.50985539e-01 9.51449931e-01
2.98285753e-01 6.71299875e-01 7.26726294e-01 -7.55233228e-01
7.40655720e-01 3.42305422e-01 -1.33559674e-01 -2.99597710e-01
-3.47303450e-01 -2.27854088e-01 -8.71024191e-01 1.40371859e-01
5.40398359e-01 -1.12477429e-01 -1.37161469e+00 2.16674376e+00
-9.54650640e-02 -5.59654981e-02 2.19554052e-01 8.57625961e-01
1.11840248e+00 6.05446100e-01 6.04683340e-01 1.58959538e-01
1.59141517e+00 -1.25174463e+00 -4.30712312e-01 -5.77861130e-01
4.72336113e-01 -9.42950249e-01 1.67372811e+00 2.53109694e-01
-1.12511790e+00 -9.49589014e-01 -6.75829828e-01 -4.69429314e-01
-7.09161699e-01 3.69797260e-01 7.14018643e-01 2.94898599e-01
-1.41060555e+00 -7.12688863e-02 -4.39251542e-01 -7.14605331e-01
5.68268359e-01 1.15717582e-01 -7.68262208e-01 -3.24722052e-01
-1.04611039e+00 8.77724349e-01 7.10256040e-01 -1.05287852e-02
-1.44664717e+00 -7.82051861e-01 -1.17075634e+00 -1.23401649e-01
3.66655827e-01 -1.24184072e+00 1.50395656e+00 -1.03904700e+00
-1.41663241e+00 1.50761855e+00 -2.07765460e-01 -2.94630915e-01
4.14127678e-01 -2.22083077e-01 -4.43275541e-01 -1.07953018e-02
6.74449727e-02 1.45771646e+00 1.00318933e+00 -1.45587254e+00
-4.69044060e-01 -1.54120252e-01 2.14005530e-01 2.66665459e-01
-4.52390581e-01 -1.19794458e-01 -8.19734633e-01 -6.09856248e-01
-3.39312673e-01 -8.44933748e-01 1.86977237e-02 -3.45934480e-02
-5.15966773e-01 -2.97318161e-01 4.85897452e-01 -3.68777394e-01
9.90417480e-01 -2.16719079e+00 2.77404755e-01 -2.43352473e-01
7.32794404e-02 3.37376237e-01 -8.12519729e-01 6.81125164e-01
-2.41449907e-01 9.35277727e-04 -3.44663113e-01 -5.56130052e-01
2.82178044e-01 2.41624974e-02 -3.47358078e-01 1.04673885e-01
3.47475260e-01 1.30734384e+00 -8.85814965e-01 -5.43326199e-01
2.87114352e-01 3.55086654e-01 -3.77046108e-01 3.73636216e-01
-4.52055097e-01 4.66537207e-01 -1.90383226e-01 7.89159000e-01
4.88458157e-01 -5.43149352e-01 7.16388747e-02 -4.66165721e-01
7.20669776e-02 2.51745293e-03 -6.36924684e-01 2.46011257e+00
-7.82108128e-01 7.46021330e-01 1.10754482e-01 -1.08295941e+00
4.33253795e-01 4.40638155e-01 1.04670927e-01 -9.27100062e-01
1.71440288e-01 1.36315048e-01 -6.56968430e-02 -6.88152134e-01
4.96921927e-01 1.61240567e-02 -3.48756999e-01 3.38133425e-01
7.19674051e-01 8.78485069e-02 3.76041859e-01 4.33581024e-01
7.24947870e-01 3.43183339e-01 1.43280089e-01 -2.29749143e-01
6.80352449e-01 8.44402686e-02 -1.11703984e-01 8.31517220e-01
-2.72389323e-01 8.08756649e-01 1.10067353e-02 -8.63338560e-02
-9.87960994e-01 -1.29462361e+00 -1.63473904e-01 1.76829243e+00
-6.61302432e-02 -3.24189126e-01 -6.98892415e-01 -7.12584078e-01
1.45354401e-02 7.04436541e-01 -7.52768934e-01 -1.30341828e-01
-1.44042581e-01 -5.25813282e-01 6.69601858e-01 4.42148834e-01
4.70073193e-01 -1.32789254e+00 -7.24378675e-02 -1.79961503e-01
-2.70757467e-01 -1.53956330e+00 -7.00084388e-01 4.58321050e-02
-4.59117949e-01 -9.71605301e-01 -1.03769147e+00 -1.20453024e+00
7.01321900e-01 3.70205462e-01 1.55052829e+00 -2.07426816e-01
-1.48532182e-01 9.87727761e-01 -1.94390386e-01 -2.48065323e-01
-4.38843757e-01 1.69236600e-01 -2.04249009e-01 8.03964362e-02
2.10698098e-01 -3.41231346e-01 -6.38432920e-01 1.84439391e-01
-1.01566482e+00 4.23100591e-01 8.19528282e-01 1.00807607e+00
6.08462691e-01 -6.88238680e-01 6.18879020e-01 -8.18646133e-01
6.63718998e-01 -5.16475677e-01 -3.48350137e-01 5.38483620e-01
-2.88218707e-01 -8.83334130e-02 4.59700227e-01 -5.98773479e-01
-1.16245472e+00 2.49782741e-01 -1.55876562e-01 -6.21562898e-01
-7.87858129e-01 4.59449977e-01 -2.11291566e-01 4.04644608e-02
7.14903295e-01 6.32699430e-01 2.62875799e-02 -3.84123325e-01
9.60488975e-01 5.13670385e-01 9.57674742e-01 -7.18064368e-01
7.83632994e-01 3.96063298e-01 -3.10187280e-01 -6.59663439e-01
-9.49941576e-01 -6.20510519e-01 -7.16142952e-01 -2.63643473e-01
1.07846093e+00 -1.26794136e+00 -7.50779629e-01 6.30563319e-01
-1.20434856e+00 -6.84297442e-01 -3.75814699e-02 2.75173992e-01
-7.50139058e-01 1.91322908e-01 -4.96797025e-01 -4.05766457e-01
-2.98528552e-01 -1.22966564e+00 1.29584074e+00 1.67317912e-01
-7.17412680e-02 -1.09678543e+00 3.95230986e-02 6.78942144e-01
3.88675898e-01 -8.42658430e-02 7.45163739e-01 -6.36489809e-01
-4.60828185e-01 1.18549436e-01 -4.39783186e-01 3.87842089e-01
-4.61132526e-02 -2.35958412e-01 -1.28798115e+00 -4.98164624e-01
-5.26421428e-01 -9.80316818e-01 1.11194110e+00 1.22207746e-01
1.32405508e+00 -7.10358098e-02 -2.70279914e-01 7.90613770e-01
1.29365766e+00 -1.29391178e-01 5.25269568e-01 3.37175697e-01
9.03988779e-01 6.10804319e-01 4.71210361e-01 1.02471597e-01
6.55224442e-01 5.36304832e-01 4.65175420e-01 -3.00226748e-01
-4.69476551e-01 -2.49944642e-01 5.54224372e-01 7.43470490e-01
2.79601991e-01 -4.23259139e-01 -1.16746688e+00 9.36447501e-01
-1.84140193e+00 -9.70136702e-01 -3.04621235e-02 1.96253812e+00
1.17007542e+00 -1.81155249e-01 2.10568562e-01 -3.87720674e-01
5.43255746e-01 8.32806006e-02 -4.70181495e-01 -3.78182501e-01
-2.21539721e-01 3.27835828e-02 2.42836937e-01 4.51316923e-01
-1.33713520e+00 1.18633294e+00 6.26039982e+00 9.90379453e-01
-1.18048048e+00 2.29377553e-01 6.26625240e-01 2.47441046e-02
-5.15994966e-01 -3.66486013e-01 -7.71436632e-01 3.53879035e-01
9.20133233e-01 4.26965626e-03 4.04866427e-01 7.29397893e-01
-2.08871946e-01 1.49542928e-01 -1.37427521e+00 1.24627101e+00
2.89973855e-01 -1.46811962e+00 5.73242307e-01 -3.14928830e-01
7.27193236e-01 3.51905793e-01 4.85267222e-01 6.24512970e-01
2.14597717e-01 -1.17154276e+00 8.66173863e-01 4.48027670e-01
1.22192681e+00 -3.50911051e-01 4.19930816e-01 2.42668986e-01
-8.55713427e-01 7.03247935e-02 1.36404708e-02 1.63869902e-01
1.36173397e-01 4.01657522e-02 -6.00323975e-01 5.01469254e-01
7.07150936e-01 5.77935398e-01 -8.36170018e-01 8.00096810e-01
-8.95980969e-02 4.51975912e-01 6.09868057e-02 4.34562832e-01
4.83855993e-01 1.49761036e-01 2.26284847e-01 1.70731854e+00
1.20445833e-01 -2.10209206e-01 2.20923409e-01 7.48120844e-01
-3.08595300e-01 7.34754726e-02 -7.40677536e-01 -2.11399332e-01
3.31184655e-01 1.48248959e+00 -3.63444477e-01 -4.49938715e-01
-7.88079500e-01 1.21760488e+00 4.04796153e-01 7.44227469e-01
-8.88991594e-01 -1.67222053e-01 4.54210639e-01 -5.08366637e-02
3.20204496e-01 -1.81757525e-01 -1.87685117e-01 -1.36146522e+00
-2.11701676e-01 -1.07077289e+00 5.84312260e-01 -1.01780570e+00
-1.71024930e+00 7.44354963e-01 9.35829952e-02 -1.16580617e+00
-2.96717614e-01 -9.88227725e-01 -6.19129717e-01 9.07383382e-01
-1.70272887e+00 -1.82797813e+00 -4.05921608e-01 1.17667115e+00
7.46609151e-01 -4.47141558e-01 8.25925052e-01 2.99829334e-01
-4.97778386e-01 8.41971338e-01 -2.61753589e-01 2.48173311e-01
1.11552739e+00 -1.14603353e+00 4.01746064e-01 8.24915051e-01
2.17905596e-01 5.63042521e-01 4.17790741e-01 -2.01843545e-01
-1.53552532e+00 -1.09282303e+00 8.59911144e-01 -7.31347322e-01
9.89343286e-01 -6.28294587e-01 -8.82765055e-01 8.50752354e-01
8.53131771e-01 1.06696254e-02 6.97234035e-01 1.99477956e-01
-7.88639545e-01 5.28056137e-02 -6.53231978e-01 6.54471755e-01
1.15211022e+00 -9.98253822e-01 -5.27630746e-01 5.35081685e-01
6.40287638e-01 -3.58668923e-01 -6.97375119e-01 4.21840727e-01
4.07787770e-01 -7.55181313e-01 1.21020699e+00 -9.07567739e-01
5.28007865e-01 -2.16910854e-01 -2.10060075e-01 -1.03765917e+00
-3.83592427e-01 -6.21129155e-01 7.66720548e-02 1.34024894e+00
6.38481915e-01 -3.96848112e-01 3.57855737e-01 1.89977676e-01
-5.04305422e-01 -4.40681428e-01 -7.77847469e-01 -7.55161464e-01
3.13654602e-01 -5.43434381e-01 1.10771790e-01 1.10567379e+00
-1.29643485e-01 7.82725036e-01 -3.72376412e-01 -1.50372565e-03
6.78305387e-01 2.08225295e-01 8.46136808e-01 -8.33855450e-01
-3.06003153e-01 -6.46460533e-01 8.60141292e-02 -8.90458167e-01
3.86435270e-01 -1.31705010e+00 1.66660830e-01 -1.60645068e+00
3.90517026e-01 -7.24956989e-02 -5.03672898e-01 9.03591752e-01
-2.17621297e-01 5.03625691e-01 3.53455961e-01 1.51785463e-01
-9.74856675e-01 4.81416762e-01 1.46454906e+00 -4.04221117e-01
7.38177896e-02 -2.99811035e-01 -7.25480080e-01 5.92679322e-01
8.72042000e-01 2.84022875e-02 -7.00070024e-01 -8.88850689e-01
-5.96773513e-02 -2.18132615e-01 6.33180439e-01 -7.45453000e-01
1.40295640e-01 -1.98688224e-01 2.73101300e-01 -4.68210310e-01
3.62176806e-01 -6.33686125e-01 -1.87619254e-01 1.63532317e-01
-5.53556979e-01 9.14099813e-02 7.45544910e-01 3.96117657e-01
-3.96260411e-01 -7.45441159e-03 6.95082307e-01 -1.70550793e-01
-1.37517262e+00 4.30145085e-01 -2.04558879e-01 2.31600076e-01
9.34392512e-01 -1.85270868e-02 -8.05464387e-01 -2.94751108e-01
-8.32673848e-01 3.67208093e-01 4.05998379e-01 7.90710866e-01
6.39541626e-01 -1.36098158e+00 -8.59032512e-01 4.57794219e-02
7.44259775e-01 -3.05973172e-01 7.08158076e-01 7.85376310e-01
-1.10057816e-01 6.75606906e-01 -5.00741661e-01 -8.78183067e-01
-1.12887704e+00 8.32637906e-01 1.61593363e-01 -9.04444307e-02
-1.71405166e-01 9.33492243e-01 8.19995940e-01 -5.27259588e-01
2.30495229e-01 9.94574428e-02 -1.03483349e-01 9.54183936e-02
5.27528822e-01 -1.73697263e-01 -1.02258362e-01 -7.78361499e-01
-3.83213013e-01 4.39657629e-01 -1.65858835e-01 -3.17037940e-01
9.18212533e-01 -2.29285777e-01 9.79488119e-02 4.54254448e-01
1.13944054e+00 -3.58217716e-01 -1.30030107e+00 -2.65088886e-01
-2.48443544e-01 9.30915214e-03 2.04324462e-02 -1.28486586e+00
-9.17578816e-01 1.26393378e+00 5.19388199e-01 -1.45939127e-01
1.24204671e+00 3.03112447e-01 5.68747163e-01 4.14083809e-01
3.45320441e-02 -9.45627809e-01 3.69969994e-01 6.48430586e-01
1.27474988e+00 -1.68921232e+00 -4.23995495e-01 -8.52104351e-02
-1.00724089e+00 8.29073787e-01 1.05816913e+00 3.74819070e-01
3.03137869e-01 1.39471255e-02 2.86850393e-01 -4.19209041e-02
-9.03234541e-01 -6.10785127e-01 7.10049391e-01 9.10877943e-01
5.65919280e-01 -1.20035075e-01 1.16539299e-01 5.87978005e-01
1.31673008e-01 -2.54400540e-02 8.49142820e-02 8.10449839e-01
-1.30384281e-01 -1.06899750e+00 -1.97178617e-01 -3.08397207e-02
-2.21940950e-01 -5.09001493e-01 -2.41260737e-01 1.04071629e+00
4.95241545e-02 6.67095006e-01 1.88720599e-02 -2.53505141e-01
3.05920392e-01 4.27745491e-01 7.10155070e-01 -6.27358913e-01
-4.83328044e-01 6.70737475e-02 4.54801396e-02 -6.36595368e-01
-5.55072427e-01 -4.35893089e-01 -9.90215838e-01 -6.85287639e-02
2.11788192e-01 -3.73409748e-01 5.44562459e-01 7.99051285e-01
5.66638172e-01 5.47244191e-01 1.78974733e-01 -9.71576393e-01
-3.59793097e-01 -9.65812266e-01 -3.27158906e-02 7.97276020e-01
3.49682659e-01 -4.78213608e-01 6.15527444e-02 5.19507945e-01] | [10.847809791564941, 1.588831901550293] |
7c030892-61b0-49e1-867d-4700d8fe3c95 | the-chime-7-dasr-challenge-distant-meeting | 2306.13734 | null | https://arxiv.org/abs/2306.13734v1 | https://arxiv.org/pdf/2306.13734v1.pdf | The CHiME-7 DASR Challenge: Distant Meeting Transcription with Multiple Devices in Diverse Scenarios | The CHiME challenges have played a significant role in the development and evaluation of robust speech recognition (ASR) systems. We introduce the CHiME-7 distant ASR (DASR) task, within the 7th CHiME challenge. This task comprises joint ASR and diarization in far-field settings with multiple, and possibly heterogeneous, recording devices. Different from previous challenges, we evaluate systems on 3 diverse scenarios: CHiME-6, DiPCo, and Mixer 6. The goal is for participants to devise a single system that can generalize across different array geometries and use cases with no a-priori information. Another departure from earlier CHiME iterations is that participants are allowed to use open-source pre-trained models and datasets. In this paper, we describe the challenge design, motivation, and fundamental research questions in detail. We also present the baseline system, which is fully array-topology agnostic and features multi-channel diarization, channel selection, guided source separation and a robust ASR model that leverages self-supervised speech representations (SSLR). | ['Sanjeev Khudanpur', 'Stefano Squartini', 'Zhong-Qiu Wang', 'Yoshiki Masuyama', 'Paola Garcia', 'Xuankai Chang', 'Desh Raj', 'Shinji Watanabe', 'Matthew Wiesner', 'Samuele Cornell'] | 2023-06-23 | null | null | null | null | ['robust-speech-recognition'] | ['speech'] | [ 0.20283559 -0.2901863 0.21535327 -0.45481366 -1.8407807 -0.88660604
0.55902404 -0.28754058 -0.00931375 0.28748435 0.67651534 -0.43465373
-0.1735164 0.2303351 -0.52890986 -0.35400018 -0.33774838 0.23698422
-0.05085959 -0.49953628 -0.10109179 0.72496283 -1.3642713 0.5444038
0.27442208 0.8960828 0.1054449 1.2923534 0.4904942 0.6478458
-1.3912065 0.11432554 0.29444352 -0.27659988 -0.4107668 -0.01236251
0.6286439 -0.01368905 -0.81376827 0.5157341 1.4286693 0.2687268
0.63751554 -0.8797845 -0.65938514 1.0852939 -0.2694108 0.49219114
0.34028327 -0.06089281 0.81098896 -1.1140195 0.08008159 1.2883227
0.77851117 0.6465611 -1.2454798 -0.8396383 0.22502261 -0.18524444
-1.8252698 -1.4882162 0.58489937 -0.1681663 1.3224945 0.47919253
-0.02756763 1.6892098 -0.7021833 0.866271 0.89909154 -0.4891576
0.49795124 -0.09621038 0.24916433 -0.07122624 -0.12876837 -0.08306358
-1.0496763 -0.12416408 0.52268153 -0.557793 -0.38114145 0.18894356
-1.4103163 0.37585577 0.10621389 0.5127844 -0.12053262 0.01783831
0.12278675 0.5702599 0.3775484 0.69797975 -0.35745114 -0.30485752
-1.1473614 0.08468946 0.80935305 1.2995123 -0.09565369 0.67755854
-0.47075018 1.5627127 0.38246354 0.92687017 0.4192452 -0.72202283
0.8707221 -0.27367392 0.02685397 -0.36257547 -0.28420955 -0.8353428
-0.5760665 -0.24381045 -0.04536041 -0.51607215 -1.1980262 1.4137888
-0.37430423 0.2511172 0.28136373 0.8568697 0.92822856 0.70422375
-0.4274095 0.01660345 0.89529985 -1.044339 -0.6497418 -0.4891005
0.2554695 -1.1684858 0.6752677 0.7556671 -1.1404376 -0.47871357
-1.4485579 0.22624947 -0.23439066 0.47379503 0.22495982 1.1733695
-1.6063768 -0.02393778 -0.612223 -0.4688505 0.09327 0.37809005
-0.1137085 -0.10447275 -0.9920861 0.5237246 -0.21964073 0.26495895
-1.1190057 -0.6289458 -0.67692965 -0.1616449 0.12554385 0.02037316
1.6131009 -0.43014243 -1.9785691 0.5106528 -0.12320317 -0.575932
-0.06894814 -0.34197855 -1.1930126 -0.01008546 -0.18271779 0.38243407
0.8087558 -1.1992115 -0.1770012 -0.38437355 -0.40206194 0.38413808
-0.19755177 0.50632554 -0.50854075 -1.1202754 0.25607595 -0.9561751
-0.1579166 -0.84328973 -0.71710426 0.20968837 0.50876164 -0.8907332
1.2862686 -2.4969273 -0.17005081 0.5641703 0.04039738 0.48636237
-0.41637477 0.47926727 -0.20336245 0.10876397 0.02781549 -0.6380242
0.17023787 -0.37071785 -0.45360014 0.37323672 -0.08233538 0.41922393
-0.3091765 0.04672877 0.18532014 0.5996212 -0.51227695 0.42396003
0.19179156 0.29501548 -0.24601978 0.8423129 0.51742136 0.04463356
-0.02872632 -0.13144639 -0.20258044 0.8588816 -1.4249078 1.8616866
-0.8465461 0.920308 0.79216754 -0.9598308 1.1849713 0.5476291
0.15521209 -0.7293143 -0.03312979 0.67484283 -0.11836785 -0.07906963
0.4852238 0.4304186 -0.1645415 0.2333331 0.26662195 -0.21933417
-0.20756337 0.3443687 1.5082294 -0.6785143 0.01032363 -0.24846667
0.13116062 -0.49976745 0.02798404 1.2454877 -0.02916662 1.3316089
-0.28941527 0.4830352 -0.7139367 -1.5531633 -0.12203227 1.1635389
-0.14857812 -0.40863988 -0.5628265 -0.2452576 -0.22702776 0.7228725
-0.06985258 0.20807481 -0.71010256 -0.6976319 1.1688502 0.66309357
0.3629453 -0.5961125 0.1969669 0.2419674 -0.22484538 -1.3499961
-0.7501811 0.50926507 -0.12785602 -0.20291422 -0.94652194 -0.7261227
-0.03738288 0.70412326 0.92049867 -0.43270576 -0.32019833 0.82250464
-0.6105789 -0.29343367 -0.4139314 0.2790172 0.3434016 0.02195843
0.1248691 -0.64013326 -0.48975816 0.61759794 -0.2919695 -0.71649945
0.7918819 0.45825896 0.29214802 -0.2670602 1.0818318 -0.35741812
0.74457633 -0.59652656 -0.18625136 0.41162732 -0.20486578 -0.09823614
0.5073361 -0.3951607 -1.102531 -0.279455 -0.88709146 -0.31330636
-0.3057855 0.05192256 -0.5760403 -0.01199848 1.1549587 0.15116435
-0.28283462 -0.7224511 0.328289 1.6996169 0.6739729 -0.6408493
0.7098008 0.07503775 -0.626948 -1.5779544 -0.36774215 -0.8474592
-0.19238704 0.2924498 0.5361871 -1.5402377 -0.21541603 0.5877516
-1.004688 -0.62194437 -0.09242789 0.65533155 -0.19276059 -0.02389883
-0.51906157 -1.033927 -0.3929682 -1.1847434 1.5291169 -0.11612765
-0.3011716 -0.62908876 0.02556017 0.69193095 0.87425673 -0.40265542
0.09025498 -1.2244856 -0.47549182 -0.12267586 0.1397318 0.56025255
-0.04969032 -0.2976127 -1.5472609 -0.44343865 -0.06965117 -0.40559027
0.7220681 0.4494294 0.9805097 -0.22629778 -0.31506935 0.721588
0.63961595 0.41165397 0.48082116 -0.03479324 0.5515057 0.34562698
-0.02262006 0.1749094 0.2156838 1.1302181 -0.35173 -0.09359316
-0.67932516 -0.18099986 0.8774704 1.2520432 0.61561596 -0.8699857
-0.79597074 0.61459184 -1.2566531 -0.85002136 0.12429603 2.1077507
0.65027905 -0.08481893 0.3766532 0.27079913 0.84235686 0.35356185
-0.31095406 -0.34988078 -0.6038581 0.38409242 0.52216524 0.61983347
-0.9121636 0.72338283 7.0704775 1.1306849 -1.1096452 0.20660244
0.50699246 -0.465339 -0.18067716 -0.6801621 -1.0732158 0.10936907
1.4466316 0.30377316 0.8119249 0.6269845 -0.03908637 0.34254393
-1.2447399 1.3530535 0.6146265 -1.0476514 -0.37959182 -0.14296955
0.5544676 0.653561 0.44438237 0.3875663 0.5545534 -1.2858063
0.7663454 -0.06431966 1.0679432 -0.39972666 0.14387794 -0.04882857
-1.2769532 -0.29246515 0.06093836 0.35226053 0.10952649 0.537414
-1.1115222 0.5392177 0.68629766 0.33658862 -0.5368321 0.84894234
-0.02313551 1.2254661 -0.6809407 0.10127068 -0.17712648 0.49798578
0.89187783 1.6955608 0.5581099 0.02068232 0.07798167 0.45964453
-0.2311719 -0.23011473 -0.53697115 0.06168304 1.1875961 0.87002593
-0.4307449 0.14126748 -0.30186883 0.9014589 -0.01710827 0.70027304
-0.5246271 -0.44287908 0.72705805 0.09663963 0.2931612 -0.5002344
-0.20550019 -1.0938777 0.04217619 -1.3797547 0.13039936 -0.6199963
-1.3869094 0.92817926 -0.04478655 -0.8586965 -0.3616144 -0.4542081
-0.5581673 0.9267362 -1.2478924 -1.1364356 0.10707609 0.6221382
0.9638378 -0.8420758 0.7380508 0.64415854 -0.64397895 1.285368
0.41830552 0.12280466 0.84753335 -1.1136955 0.9645261 1.1225395
0.4931108 0.8716953 0.5671248 -0.29064193 -1.4207941 -1.1769533
0.6207037 -0.82369304 0.5480639 -1.3037078 -0.48227116 0.7574631
0.24363397 0.14548992 0.74292004 0.5520188 -0.51391774 -0.43992358
-0.6467116 0.5352854 1.412386 -0.8210749 -0.3284128 0.20266624
0.881888 -0.50060415 -0.63801914 0.19929945 0.37591788 -0.6726951
1.3638465 0.12157091 -0.50195366 -0.24614494 -0.8282522 -1.6458706
-0.09477113 -1.3449597 0.04445092 1.7306224 0.87530863 -0.4832493
0.342726 0.2272817 -0.7191384 -0.11848333 -0.8894759 -1.22723
-0.07238854 -0.67454475 0.5467916 0.49287006 -0.09336799 0.6348478
-0.5390611 0.5785072 0.521036 -0.38181198 0.7338685 -0.74900514
-0.7325998 -0.34419423 -0.14727502 -1.6486791 -0.06686667 -0.78894097
0.44309026 -1.1876959 -0.5389668 -0.6794498 -0.16177925 0.03728405
0.20335872 0.0441444 0.11096712 0.15986241 -0.93374133 0.48541975
0.32782918 -0.23576972 -0.4412679 0.27240288 -0.9804552 0.20815639
0.82621753 -0.19872326 -0.67364794 -0.6759657 -0.15626967 0.02572725
0.19139758 -1.3922698 0.5914283 0.535452 0.2258728 -0.41023484
0.66990536 -0.46575713 -0.10271529 -0.50223196 -0.83239245 -0.05986277
0.30675492 0.3028401 -0.13954917 0.17122851 0.6241985 0.36193335
-0.5210698 -0.17562863 -0.58872485 0.40991125 0.5131068 0.02468738
-0.4272963 -0.5938467 -0.60781 0.08394426 -0.21366626 0.6911269
0.6447077 -1.0688958 -1.0030457 0.3371452 0.22105724 -0.08915931
0.38651678 0.37733722 0.02657595 0.55157626 0.48849827 -0.53928244
-1.0462552 0.14183305 0.5350704 0.20365873 -0.4116614 1.1818362
0.06261468 -0.6587004 0.71319026 -0.08118771 -0.04192459 -0.30542493
0.69533235 0.46031743 0.6919231 -0.53667563 -0.6359063 0.18022624
-0.0505246 -0.79214406 1.181555 -0.3179504 0.5798895 0.56233865
1.3319348 0.5220838 -0.869085 -0.30000663 -0.16598727 -0.149848
0.24159662 -1.246089 -0.6817179 0.8316527 0.79081696 0.1329672
0.88332736 0.28777102 0.7475 0.5757715 0.2294917 -1.0783001
0.3651006 0.5538617 1.2536191 -0.81598693 -0.5172413 -0.21647117
-0.6306657 0.60904187 0.58292687 0.30343166 0.7063049 1.0078331
0.314458 -0.01365523 -0.5997857 -0.2020801 0.15840095 1.0735888
0.6622412 0.0688361 0.57153064 0.9820985 -0.45829323 -0.63299704
0.2640567 0.75208783 -0.32973439 -1.0489237 -0.72706 0.27592823
-0.42456102 -0.33977917 -0.61951196 0.26891208 -0.40586203 1.7453805
-0.12433586 -0.6717148 0.59175605 -0.07733753 0.23790284 -0.8721258
-0.6728828 0.40562627 0.61568505 -0.25345436 0.07267479 -0.81639564
-0.8157692 0.04807554 -0.28725797 0.26969746 0.9239243 0.6489033
0.9522031 0.87852436 0.88514155 -0.82899576 -0.82780534 -1.0259576
-0.63721186 -0.03969685 0.720602 -0.3669826 -0.42162463 -0.25485143] | [14.814629554748535, 6.078979015350342] |
072c0a74-a971-459c-9b62-cfef5963f8fa | discovars-a-new-data-analysis-perspective | 2304.03983 | null | https://arxiv.org/abs/2304.03983v1 | https://arxiv.org/pdf/2304.03983v1.pdf | DiscoVars: A New Data Analysis Perspective -- Application in Variable Selection for Clustering | We present a new data analysis perspective to determine variable importance regardless of the underlying learning task. Traditionally, variable selection is considered an important step in supervised learning for both classification and regression problems. The variable selection also becomes critical when costs associated with the data collection and storage are considerably high for cases like remote sensing. Therefore, we propose a new methodology to select important variables from the data by first creating dependency networks among all variables and then ranking them (i.e. nodes) by graph centrality measures. Selecting Top-$n$ variables according to preferred centrality measure will yield a strong candidate subset of variables for further learning tasks e.g. clustering. We present our tool as a Shiny app which is a user-friendly interface development environment. We also extend the user interface for two well-known unsupervised variable selection methods from literature for comparison reasons. | ['Ayhan Demiriz'] | 2023-04-08 | null | null | null | null | ['variable-selection'] | ['methodology'] | [ 1.17228970e-01 -4.40692119e-02 -2.18968213e-01 -5.28410971e-01
-2.57106096e-01 -4.33469176e-01 2.53733844e-01 7.56796658e-01
-5.07197618e-01 1.11432648e+00 -2.43467361e-01 -5.48978031e-01
-6.46267295e-01 -1.36017954e+00 -5.03895730e-02 -7.77603090e-01
-4.85409379e-01 7.48323202e-01 -8.15735385e-02 -3.63668911e-02
5.77240467e-01 7.23599672e-01 -1.96777678e+00 -4.06719595e-01
1.04920590e+00 5.72810590e-01 4.94279146e-01 5.22028506e-01
-5.49255252e-01 4.26205188e-01 -4.58409369e-01 -1.57763377e-01
-7.93678686e-03 -4.97615427e-01 -9.34484720e-01 -2.93262362e-01
-2.10520113e-04 3.56720328e-01 5.39686561e-01 1.03481233e+00
4.02645081e-01 4.81854439e-01 7.45110989e-01 -1.30035949e+00
-5.90556860e-02 1.05253220e+00 -6.98417723e-01 2.08448291e-01
6.47178814e-02 -4.23524022e-01 1.43664038e+00 -9.02640641e-01
8.73067141e-01 1.16791308e+00 4.32362527e-01 4.10017222e-02
-1.36852038e+00 -8.32126915e-01 1.90814778e-01 5.23772240e-01
-1.55884469e+00 -2.09542945e-01 1.11629558e+00 -5.98233163e-01
6.58037901e-01 8.33559632e-01 6.03850126e-01 4.19950873e-01
-1.97196051e-01 2.92208433e-01 9.84008372e-01 -4.26312625e-01
4.98593420e-01 4.91254270e-01 7.00582266e-01 5.94150305e-01
4.46716398e-01 -3.06046069e-01 -5.00591218e-01 -3.38069856e-01
5.42213500e-01 4.84139509e-02 -3.00597847e-02 -2.51869082e-01
-8.77660811e-01 1.14589512e+00 4.09464866e-01 2.42209136e-01
-6.38969719e-01 -1.48149729e-01 1.77095920e-01 6.04837596e-01
5.91779709e-01 9.18574929e-01 -6.75050199e-01 -1.78620487e-01
-9.03395355e-01 2.99414918e-02 6.40349925e-01 8.05825412e-01
1.12823963e+00 -1.47062317e-01 2.71434560e-02 9.15658772e-01
3.35644126e-01 2.46753737e-01 -1.72956064e-02 -4.24370289e-01
2.63144910e-01 9.80476618e-01 -1.62686333e-01 -1.29288089e+00
-7.39891827e-01 -6.03485167e-01 -8.43431771e-01 2.93210447e-01
1.69895887e-01 -4.32566106e-01 -6.80827141e-01 1.49522233e+00
4.50624257e-01 -2.43035004e-01 -4.58777726e-01 7.20697582e-01
1.14672303e+00 3.49071026e-01 3.16910595e-01 -6.11029267e-01
1.16262734e+00 -3.72641355e-01 -5.69177747e-01 2.95987457e-01
3.55898708e-01 -4.93463963e-01 9.93814230e-01 3.86279762e-01
-6.05502725e-01 -4.48540747e-01 -6.20323598e-01 2.55977392e-01
-6.96909130e-01 -9.49182883e-02 9.88963068e-01 4.31243420e-01
-1.08674288e+00 6.64226234e-01 -5.13433397e-01 -5.47560453e-01
2.87395597e-01 6.76845372e-01 -3.66387397e-01 3.56859028e-01
-1.10982287e+00 7.78055310e-01 2.41279513e-01 -4.79142368e-02
-2.19641969e-01 -3.36554825e-01 -6.25117540e-01 1.47451073e-01
4.70147371e-01 -5.65426946e-01 4.06721860e-01 -1.16047311e+00
-1.03031194e+00 5.58726728e-01 -4.29342836e-01 3.53171155e-02
3.57618511e-01 8.03875029e-02 -1.53808802e-01 -3.57213505e-02
4.75229807e-02 4.58881408e-02 8.05707991e-01 -1.26597011e+00
-8.49121511e-01 -4.11664873e-01 2.13275086e-02 1.83956295e-01
-5.23536921e-01 1.02289960e-01 -2.91419327e-01 -6.10728741e-01
2.70930380e-01 -5.76783538e-01 -7.23960698e-01 -1.84056088e-01
-4.68006074e-01 -3.69415283e-01 7.20213354e-01 -4.65204656e-01
1.70037413e+00 -1.67824161e+00 4.05419320e-01 1.21935499e+00
6.24279439e-01 -2.61571437e-01 1.11549713e-01 4.24956650e-01
-3.22986186e-01 6.66916668e-01 -2.59482950e-01 3.39823402e-02
-3.81004959e-01 -1.88685030e-01 3.58797967e-01 4.93378669e-01
6.26668781e-02 1.99076265e-01 -1.10126591e+00 -7.74179757e-01
4.42977697e-01 3.02008718e-01 -3.53442490e-01 2.78426290e-01
1.84819046e-02 4.49583262e-01 -7.14649200e-01 8.67651701e-01
5.19370735e-01 -2.48016551e-01 3.60188484e-01 1.16018571e-01
-2.30019510e-01 -1.59434099e-02 -1.42831051e+00 8.96935344e-01
-3.41136962e-01 6.99723721e-01 1.96211427e-01 -1.09477627e+00
1.15781403e+00 7.59849846e-02 6.62110507e-01 -2.02224836e-01
1.76776946e-01 1.42049551e-01 1.05014602e-02 -5.85374475e-01
5.64536035e-01 -2.13686880e-02 1.68169424e-01 3.07652026e-01
-2.15811834e-01 3.29481289e-02 4.16492790e-01 1.00273937e-01
9.60041523e-01 -2.81198323e-01 7.59708524e-01 -6.33543253e-01
7.14188218e-01 1.38077646e-01 3.92846793e-01 7.06088841e-01
1.71029910e-01 2.36565888e-01 6.93071604e-01 -3.08732808e-01
-5.35116613e-01 -6.92544758e-01 -2.95401186e-01 1.30727506e+00
-1.60124093e-01 -5.66299558e-01 -1.69725806e-01 -7.17781305e-01
1.40870139e-01 6.11434340e-01 -6.83200777e-01 3.35671753e-01
-4.00538296e-02 -6.27719879e-01 -2.42541179e-01 2.91101132e-02
-5.78584112e-02 -1.25115204e+00 -4.07844514e-01 1.40279472e-01
9.28049162e-02 -1.46641582e-01 3.18510681e-01 6.75760508e-01
-9.99015868e-01 -1.10793650e+00 -2.87355930e-01 -7.51882792e-01
1.08498144e+00 4.26893175e-01 1.51520467e+00 7.03651011e-01
-5.95815219e-02 -1.24238245e-01 -8.46499264e-01 -3.90121371e-01
1.17413752e-01 5.05857706e-01 1.53791280e-02 -2.82724202e-01
5.50083160e-01 -7.27024615e-01 -4.69000667e-01 2.85367910e-02
-5.66911519e-01 -1.30765334e-01 5.27855098e-01 7.14169145e-01
7.47423828e-01 7.83912182e-01 4.15487677e-01 -1.39115560e+00
7.92317510e-01 -8.94305110e-01 -6.79273188e-01 1.29821956e-01
-9.99083340e-01 2.72873819e-01 7.07884610e-01 -6.20531216e-02
-5.57121873e-01 2.40626648e-01 -7.37718195e-02 7.64712244e-02
-1.00772828e-01 1.31983113e+00 -2.80805230e-01 -9.39655006e-02
4.89111036e-01 -2.60098100e-01 -2.11606413e-01 -6.02456391e-01
1.18815638e-01 6.92254245e-01 -3.69632304e-01 -3.85534853e-01
9.31067050e-01 1.08855799e-01 4.00003076e-01 -1.18996465e+00
-2.46245507e-02 -6.38593912e-01 -8.97408128e-01 -4.17392701e-01
5.06780446e-01 -3.46301377e-01 -9.04842257e-01 -1.19379230e-01
-7.47982800e-01 1.47707999e-01 4.29706089e-02 5.44984400e-01
1.15139186e-01 -8.02750438e-02 1.88315675e-01 -9.27925050e-01
-4.13056880e-01 -1.06059361e+00 4.79262412e-01 3.15519333e-01
-5.22591412e-01 -1.09835005e+00 5.39693460e-02 -1.17986448e-01
4.13105428e-01 5.18695056e-01 1.08290648e+00 -7.73198962e-01
-4.40993369e-01 -1.43366426e-01 -1.22602306e-01 -2.73827791e-01
3.18520427e-01 5.76800227e-01 -7.96902120e-01 -2.05329031e-01
-5.59809923e-01 2.06515595e-01 9.44015563e-01 5.55363417e-01
1.32823491e+00 -1.23577990e-01 -6.03337526e-01 5.66527843e-01
1.64689052e+00 1.18624099e-01 2.22422436e-01 4.76582795e-01
8.81239235e-01 9.38026667e-01 7.90993154e-01 7.23129034e-01
1.93616316e-01 3.76245499e-01 4.20543551e-01 -4.97105032e-01
4.83911395e-01 2.27034941e-01 -1.07076786e-01 6.52087808e-01
-4.29684967e-01 -4.85705510e-02 -1.28516519e+00 5.18488169e-01
-1.70819843e+00 -1.01950908e+00 -7.87792027e-01 2.29069829e+00
8.29188943e-01 7.67488182e-02 1.98908553e-01 2.94561565e-01
7.47155249e-01 1.55012637e-01 -4.02417958e-01 -2.93769747e-01
2.09866747e-01 7.08960176e-01 6.89736009e-01 9.31427717e-01
-1.10969985e+00 8.34568858e-01 5.65989923e+00 6.75782561e-01
-1.26359832e+00 -2.58883119e-01 5.30471563e-01 -1.08659409e-01
-6.68841541e-01 9.63330194e-02 -5.57250977e-01 2.60681480e-01
6.25378549e-01 -3.71948451e-01 2.54055768e-01 9.66983438e-01
7.22701132e-01 -2.93116957e-01 -7.42962778e-01 9.45096850e-01
-4.30493981e-01 -1.13645947e+00 2.82878652e-02 1.18881293e-01
3.02459002e-01 -6.78576156e-02 -3.68604213e-01 6.55558929e-02
5.24903595e-01 -1.30694914e+00 2.02291831e-01 4.23546404e-01
7.43340850e-01 -1.08821011e+00 7.83296108e-01 -1.94517709e-02
-1.60561192e+00 8.72636959e-02 -3.26811582e-01 -3.77822608e-01
-2.55282879e-01 1.05403948e+00 -8.58957112e-01 5.42650104e-01
7.36372769e-01 8.23852181e-01 -9.49872494e-01 1.11920679e+00
-1.63989022e-01 7.44182527e-01 -3.31770390e-01 -4.35390353e-01
-1.92243338e-01 -4.91691887e-01 4.91482735e-01 1.14737988e+00
3.12879920e-01 4.58031381e-03 1.30945414e-01 5.03850400e-01
9.84985530e-02 6.72160625e-01 -6.49236917e-01 1.69610202e-01
9.55191731e-01 1.53035605e+00 -1.31620705e+00 -8.71606544e-02
-1.25219837e-01 4.25027728e-01 3.60616446e-01 2.49264181e-01
-5.41489601e-01 -8.81027579e-01 6.44594133e-01 8.01108032e-02
6.19283915e-02 -2.53802717e-01 -7.34612107e-01 -8.64688635e-01
-1.37641579e-01 -7.09540009e-01 5.36981523e-01 -1.90808877e-01
-1.05652034e+00 8.81040692e-01 -4.00994159e-02 -1.25390363e+00
-7.76392296e-02 -2.94499904e-01 -4.81423348e-01 1.16488659e+00
-1.49810219e+00 -6.25277221e-01 -6.39132380e-01 7.47873247e-01
2.96310544e-01 -1.60517156e-01 9.41010237e-01 2.85427600e-01
-6.23437703e-01 2.97943771e-01 3.38419646e-01 -1.87355392e-02
4.59554851e-01 -1.52987552e+00 6.63330257e-02 8.55249166e-01
-3.13790627e-02 9.52849448e-01 9.21992064e-01 -7.87764728e-01
-1.21804118e+00 -8.27879071e-01 1.14594495e+00 -2.22071245e-01
4.98623937e-01 -4.01708364e-01 -7.14291155e-01 1.79290965e-01
-2.03689188e-02 -4.10759002e-01 1.07392097e+00 8.02128196e-01
2.50815898e-01 -2.77429402e-01 -1.10056555e+00 5.71022093e-01
8.05261552e-01 -2.14504227e-01 -8.23286623e-02 1.82984203e-01
2.95741737e-01 3.83309603e-01 -8.61326158e-01 5.88415824e-02
5.35772681e-01 -7.66014576e-01 8.00187647e-01 -5.18617868e-01
3.74162018e-01 -3.07966530e-01 -1.74404792e-02 -1.39968181e+00
-6.23577476e-01 -5.77210665e-01 2.98978478e-01 1.45154929e+00
9.18618739e-01 -6.26975179e-01 9.97143328e-01 6.28168344e-01
5.28644562e-01 -3.80146235e-01 -6.92607045e-01 -2.97919780e-01
-3.19690615e-01 -3.68277699e-01 6.97272241e-01 1.35337281e+00
1.39149040e-01 6.35921538e-01 -1.99414745e-01 1.18340932e-01
7.21691132e-01 2.35688150e-01 8.27475011e-01 -2.03259969e+00
4.11457419e-02 -6.29595637e-01 -2.80151486e-01 -3.12411070e-01
6.74003549e-03 -8.53107274e-01 -5.75883649e-02 -1.67096889e+00
1.95193827e-01 -9.80256915e-01 -4.34282184e-01 3.07254016e-01
-6.17136002e-01 -1.09578326e-01 -3.55135277e-02 4.14261609e-01
-2.04190120e-01 1.58630311e-01 8.25923979e-01 9.19049755e-02
-7.54827559e-01 2.93145508e-01 -9.15898919e-01 4.92428392e-01
9.63961601e-01 -7.51799226e-01 -5.28469265e-01 -1.42254144e-01
3.57297480e-01 -4.81947027e-02 -2.90007442e-01 -7.83782244e-01
-1.52542507e-02 -8.63699734e-01 3.01804692e-01 -5.81637919e-01
-1.52738675e-01 -1.45503247e+00 3.59925210e-01 3.04088831e-01
-4.11884964e-01 2.89471447e-01 -2.76470929e-01 1.70907244e-01
-1.13760285e-01 -1.98446348e-01 6.75872087e-01 -1.89176008e-01
-8.09333742e-01 3.39606553e-01 -2.66072780e-01 -3.74994189e-01
1.08307540e+00 -2.01080099e-01 -1.42934963e-01 -4.40762043e-01
-7.45483994e-01 3.35886091e-01 4.55336809e-01 1.50979713e-01
7.31483877e-01 -8.63898635e-01 -8.08007121e-01 2.78392553e-01
1.14764184e-01 -8.42066109e-02 -3.30195367e-01 5.94996810e-01
-6.73752308e-01 -2.24837847e-02 -3.14273089e-01 -3.48184556e-01
-1.93541503e+00 3.88969511e-01 -1.22885630e-01 -2.37668663e-01
-3.26334685e-01 1.01906824e+00 -4.74653006e-01 -2.28481054e-01
1.99101448e-01 -9.00125429e-02 -9.37750936e-01 8.99481177e-01
3.95206034e-01 5.52691698e-01 6.85287490e-02 -5.13611078e-01
-8.54267836e-01 4.69317883e-01 3.13206613e-01 1.72660090e-02
1.84729445e+00 -1.95071593e-01 -5.17847896e-01 4.63760018e-01
1.07455230e+00 3.23138595e-01 -6.23556018e-01 -9.96905193e-02
4.09119815e-01 -6.29633546e-01 2.47219622e-01 -4.12478894e-01
-1.37401557e+00 6.03459001e-01 4.67856795e-01 6.80509150e-01
1.31150246e+00 2.89076641e-02 -2.47030735e-01 4.31705773e-01
4.12922382e-01 -1.25015402e+00 -5.03188193e-01 4.43737537e-01
7.59019792e-01 -1.37968969e+00 6.32078230e-01 -4.15697277e-01
-5.86492002e-01 1.20145237e+00 3.50899637e-01 -2.61051604e-03
1.18050492e+00 1.00123778e-01 2.05660731e-01 -5.24360001e-01
-5.12009203e-01 -6.67687297e-01 1.47273257e-01 7.82408237e-01
9.35265005e-01 4.05072838e-01 -1.01323128e+00 4.07376170e-01
-4.79951471e-01 -2.68675983e-01 3.00593317e-01 7.45217562e-01
-5.54759443e-01 -1.19359243e+00 -4.87556338e-01 1.09835041e+00
-4.35806572e-01 -4.24947202e-01 -7.16223359e-01 7.56470799e-01
3.98011282e-02 1.24068058e+00 -1.25451043e-01 -7.52063394e-01
1.37289301e-01 -2.37745494e-01 -2.62016863e-01 -8.01512778e-01
-9.21120286e-01 -1.25468850e-01 3.00810635e-01 -1.43455788e-01
-5.49118757e-01 -6.48596346e-01 -1.24440980e+00 -6.58756971e-01
-5.08470058e-01 6.35813236e-01 6.97154760e-01 5.83939731e-01
1.68115869e-01 5.31767190e-01 9.42603588e-01 -7.52078831e-01
1.43831670e-01 -9.35407519e-01 -9.09422159e-01 3.65474463e-01
2.17335105e-01 -9.15405810e-01 -4.52546597e-01 -7.28617534e-02] | [7.813797473907471, 4.679835319519043] |
ce4e4982-ae93-4a2a-b434-0bc30d031e26 | solution-of-physics-based-bayesian-inverse | 2107.02926 | null | https://arxiv.org/abs/2107.02926v2 | https://arxiv.org/pdf/2107.02926v2.pdf | Solution of Physics-based Bayesian Inverse Problems with Deep Generative Priors | Inverse problems are ubiquitous in nature, arising in almost all areas of science and engineering ranging from geophysics and climate science to astrophysics and biomechanics. One of the central challenges in solving inverse problems is tackling their ill-posed nature. Bayesian inference provides a principled approach for overcoming this by formulating the inverse problem into a statistical framework. However, it is challenging to apply when inferring fields that have discrete representations of large dimensions (the so-called "curse of dimensionality") and/or when prior information is available only in the form of previously acquired solutions. In this work, we present a novel method for efficient and accurate Bayesian inversion using deep generative models. Specifically, we demonstrate how using the approximate distribution learned by a Generative Adversarial Network (GAN) as a prior in a Bayesian update and reformulating the resulting inference problem in the low-dimensional latent space of the GAN, enables the efficient solution of large-scale Bayesian inverse problems. Our statistical framework preserves the underlying physics and is demonstrated to yield accurate results with reliable uncertainty estimates, even in the absence of information about underlying noise model, which is a significant challenge with many existing methods. We demonstrate the effectiveness of proposed method on a variety of inverse problems which include both synthetic as well as experimentally observed data. | ['Assad A Oberai', 'Deep Ray', 'Dhruv V Patel'] | 2021-07-06 | null | null | null | null | ['geophysics'] | ['miscellaneous'] | [ 4.92115617e-01 -2.28205863e-02 4.68629628e-01 -1.84779823e-01
-6.25496864e-01 -5.30945778e-01 6.23934865e-01 -6.38116062e-01
-1.05462924e-01 1.18104148e+00 1.60945743e-01 -1.55356422e-01
-6.96398973e-01 -8.45772505e-01 -8.55923057e-01 -9.56105649e-01
3.35367978e-01 6.40832126e-01 -1.48913950e-01 -4.10184637e-02
3.68452191e-01 4.18406785e-01 -1.13248742e+00 -4.92337883e-01
1.08824086e+00 9.34331715e-01 2.79165506e-01 5.85479498e-01
3.92981954e-02 3.82652521e-01 -4.07416344e-01 -5.52186854e-02
1.85067803e-01 -5.19142747e-01 -4.18998957e-01 -1.31311491e-01
5.69456676e-03 -4.77454543e-01 -3.36864591e-01 1.07621014e+00
3.84937525e-01 3.51758897e-01 1.04054081e+00 -9.99635577e-01
-7.57182419e-01 1.00338928e-01 -7.62983382e-01 1.86994001e-02
8.56257752e-02 9.02197696e-03 6.19679272e-01 -6.56967163e-01
3.16246241e-01 1.04055309e+00 6.55904949e-01 2.84421593e-01
-1.46177769e+00 -5.80575049e-01 -1.82924882e-01 1.05458461e-01
-1.41740465e+00 -4.21008170e-01 9.38077331e-01 -5.41870952e-01
3.95254344e-01 1.19378567e-01 3.96240771e-01 1.17826521e+00
4.51549351e-01 3.11971366e-01 1.55281174e+00 -2.49010667e-01
5.50304294e-01 -6.44312426e-02 -1.41372114e-01 3.80558282e-01
3.40055615e-01 3.38460922e-01 -5.19035041e-01 -2.35706434e-01
8.60362709e-01 8.23638588e-02 -3.97023231e-01 -1.78863660e-01
-1.13588941e+00 8.37446034e-01 3.61515433e-01 -2.51259059e-02
-6.51881814e-01 3.63804221e-01 -2.75452435e-01 -2.56437421e-01
3.62634689e-01 2.94657379e-01 -1.47289544e-01 -1.21022023e-01
-8.67197096e-01 3.37237746e-01 8.34759712e-01 7.93414235e-01
8.56407762e-01 2.78228104e-01 2.01723501e-01 6.12989783e-01
7.49713719e-01 1.16012120e+00 2.11994261e-01 -1.26507354e+00
7.11432770e-02 -5.89898787e-02 3.37870717e-01 -1.19198489e+00
1.30168125e-01 -6.22261167e-01 -1.22757018e+00 2.97186017e-01
2.78882295e-01 -1.26342550e-01 -1.08261132e+00 2.07023954e+00
4.81746316e-01 6.55084074e-01 1.25177205e-01 9.64282990e-01
4.02868211e-01 7.97131598e-01 -1.61245674e-01 -2.02059612e-01
1.25373495e+00 -2.10873291e-01 -9.60316718e-01 -3.62111300e-01
-3.47277671e-01 -7.38626420e-01 7.05442965e-01 4.13957804e-01
-7.42315233e-01 -1.97842479e-01 -1.03905141e+00 3.76041909e-03
-5.96412644e-02 -1.58730179e-01 3.87738973e-01 4.74488169e-01
-7.07244635e-01 4.99620467e-01 -1.04246688e+00 -2.14863360e-01
1.82335556e-01 1.45600125e-01 -2.58360535e-01 -1.28271267e-01
-1.16183937e+00 7.72363782e-01 3.12622301e-02 4.85248059e-01
-9.81583595e-01 -6.47606730e-01 -7.60936320e-01 -7.18625588e-03
4.45237339e-01 -8.25983465e-01 8.63180459e-01 -3.04200232e-01
-1.76321423e+00 1.43310159e-01 -2.43949175e-01 -1.12159938e-01
4.65437949e-01 -4.43033427e-01 -1.98955357e-01 -1.35782108e-01
2.28107408e-01 1.32212430e-01 1.14213526e+00 -1.27522767e+00
6.32068366e-02 -4.74658310e-01 -5.95603622e-02 -3.36378515e-02
-5.56851365e-03 -6.37528896e-01 -9.61563513e-02 -5.63891828e-01
5.41175127e-01 -1.01741755e+00 -2.58057505e-01 -7.90306088e-03
-5.21280169e-01 3.40072721e-01 9.82157350e-01 -9.68908370e-01
7.09121168e-01 -1.82525849e+00 6.40090764e-01 3.61117929e-01
1.24305725e-01 -1.03919983e-01 1.81833223e-01 3.89222324e-01
2.75201678e-01 -2.17014533e-02 -7.95398891e-01 -1.22959584e-01
1.14298590e-01 6.07362986e-01 -6.66878521e-01 6.00348294e-01
1.54063612e-01 6.84237659e-01 -8.77809167e-01 -2.01758012e-01
4.15191650e-01 8.76204669e-01 -5.67316890e-01 4.47288871e-01
-2.82663524e-01 1.15576482e+00 -8.47348094e-01 2.19742358e-01
8.58428121e-01 -2.89986730e-01 3.84339094e-02 -3.79967630e-01
1.54674640e-02 9.19235349e-02 -1.51278877e+00 1.86554408e+00
-5.90365827e-01 2.45828748e-01 2.81718671e-01 -1.32103598e+00
7.38425791e-01 1.38327658e-01 2.97501147e-01 -3.48459005e-01
-4.62218337e-02 1.82757303e-01 -2.18444858e-02 -4.45650011e-01
3.32130790e-01 -6.78831100e-01 6.05061650e-02 6.57851100e-01
-5.11054136e-02 -6.68372512e-01 -1.30778268e-01 2.21603706e-01
1.04523468e+00 5.76540232e-01 2.10694626e-01 -2.75800675e-01
4.71242249e-01 -2.89211690e-01 6.75091028e-01 7.96225786e-01
3.28201711e-01 7.90052176e-01 2.51540542e-01 -6.87883422e-02
-9.94677067e-01 -1.34184325e+00 -3.76647532e-01 2.57021397e-01
-2.64657568e-02 1.57010317e-01 -6.32109523e-01 -1.62942186e-01
4.97393012e-02 7.87044585e-01 -5.21966398e-01 -2.17029691e-01
-4.07798469e-01 -1.04176342e+00 2.93719143e-01 3.29314172e-01
7.41251230e-01 -7.42083549e-01 -4.06258166e-01 1.83942720e-01
-3.59916687e-01 -1.22653317e+00 -1.81771040e-01 -3.11858300e-02
-8.77515554e-01 -7.33559072e-01 -7.00365543e-01 7.57252797e-02
6.44359767e-01 -1.42210871e-01 8.83353174e-01 -4.19567585e-01
-3.69353890e-01 4.53665137e-01 -3.77685158e-03 -3.24946970e-01
-3.53465289e-01 -2.89754868e-01 2.10143834e-01 2.17870578e-01
-5.17028309e-02 -1.08071077e+00 -5.76977491e-01 1.95696965e-01
-1.01526284e+00 7.39831617e-03 7.40581214e-01 9.85127270e-01
5.85686564e-01 1.80756077e-01 6.56843424e-01 -8.33513141e-01
4.57367986e-01 -6.33059144e-01 -7.27066517e-01 1.06477201e-01
-4.37698722e-01 5.38696051e-01 4.25564796e-01 -2.71209776e-01
-1.44285119e+00 -1.41709402e-01 -3.67282093e-01 -3.32957387e-01
-1.97331309e-01 7.51754880e-01 -1.42996386e-01 -1.04600284e-02
4.54355508e-01 4.08289909e-01 1.40234351e-01 -5.73328316e-01
3.18796366e-01 5.83810091e-01 7.05288768e-01 -1.01652491e+00
1.05979168e+00 6.98544264e-01 5.66135764e-01 -9.59989011e-01
-9.79172349e-01 -8.50250795e-02 -4.01541054e-01 -2.23668460e-02
8.10587227e-01 -7.51160502e-01 -6.19230151e-01 4.81803149e-01
-1.07948601e+00 3.06403618e-02 -1.83163792e-01 6.20184839e-01
-6.48630440e-01 4.50684875e-01 -2.26257250e-01 -9.52719688e-01
-3.91927734e-02 -1.18729651e+00 1.08139443e+00 2.30676517e-01
-2.80896425e-02 -1.11988342e+00 2.68318266e-01 3.03938776e-01
6.09604895e-01 5.33464909e-01 8.72292340e-01 7.95678422e-02
-8.80616546e-01 -8.28337222e-02 -1.01295128e-01 4.67890978e-01
3.19952846e-01 -2.63374180e-01 -9.86133993e-01 -5.95722124e-02
4.96677130e-01 -1.93889990e-01 7.43596077e-01 5.97878873e-01
1.03090131e+00 -9.24420059e-02 -2.04508394e-01 6.91107333e-01
1.51575243e+00 -8.21826383e-02 7.26052582e-01 -2.23842368e-01
5.40785313e-01 3.77973497e-01 1.47557914e-01 5.48632681e-01
1.16513304e-01 5.19877613e-01 4.14474487e-01 2.93233275e-01
6.90817162e-02 -3.10153514e-01 -4.98673916e-02 9.41460371e-01
-2.42829695e-01 -1.95877612e-01 -9.15077984e-01 3.46212178e-01
-1.80530775e+00 -8.62271011e-01 4.09819297e-02 2.28803635e+00
7.51951754e-01 -1.87211573e-01 -7.47677743e-01 -3.02772336e-02
6.00335121e-01 -7.40113929e-02 -8.28590691e-01 6.15676120e-02
9.12396386e-02 3.73738855e-01 2.84824282e-01 7.03604162e-01
-7.21160948e-01 4.45608318e-01 6.43203354e+00 6.31055534e-01
-1.00592589e+00 2.32227966e-01 4.03115481e-01 3.20001006e-01
-5.69581926e-01 2.66659170e-01 -4.88323718e-01 5.53238153e-01
7.60800838e-01 3.68399024e-02 6.88858330e-01 3.84343863e-01
2.86613852e-01 -4.94178861e-01 -9.70607102e-01 6.86679602e-01
-3.10012512e-02 -1.12128448e+00 -2.24943683e-02 2.49714226e-01
8.35876822e-01 -2.02489063e-01 1.28980145e-01 1.59593970e-01
3.77384871e-01 -1.07005167e+00 5.88067353e-01 9.29872751e-01
7.43814647e-01 -4.71728265e-01 5.07094562e-01 6.52143717e-01
-5.95901430e-01 2.79260367e-01 -2.91820168e-01 -2.67634749e-01
5.55174470e-01 1.08973467e+00 -4.76189137e-01 5.68433285e-01
7.12852895e-01 7.33773649e-01 1.69804752e-01 6.27130270e-01
-3.78702432e-01 6.26837909e-01 -7.68385649e-01 3.10156107e-01
-7.02364966e-02 -8.11673641e-01 7.99705148e-01 6.26698732e-01
8.32935095e-01 4.94709581e-01 -2.53959954e-01 1.55310750e+00
5.18775024e-02 -5.47829211e-01 -7.13691175e-01 -2.55074918e-01
2.75156587e-01 1.18380928e+00 -5.92965066e-01 -4.04221788e-02
-1.37675345e-01 5.10669470e-01 2.83109229e-02 8.00660908e-01
-8.79624248e-01 -1.55634478e-01 5.77972889e-01 -5.12210503e-02
2.70978093e-01 -4.71669286e-01 -2.87365943e-01 -1.16481626e+00
-2.69866586e-02 -5.65777361e-01 -1.10221151e-02 -9.74952877e-01
-1.36513698e+00 2.18655959e-01 2.38091946e-01 -9.78975058e-01
-5.27824461e-01 -6.67115331e-01 -5.09047270e-01 1.30265415e+00
-1.43405330e+00 -1.03482580e+00 -5.10207415e-01 3.74806434e-01
9.41696465e-02 5.99397458e-02 7.68634439e-01 9.81286541e-02
-2.63036042e-01 -1.13894872e-01 6.00911796e-01 -2.53206819e-01
5.35953164e-01 -1.18690681e+00 -1.96324680e-02 9.07747149e-01
-1.24745980e-01 6.54008627e-01 1.16887486e+00 -6.91150665e-01
-1.76862538e+00 -7.80342281e-01 1.14215188e-01 -2.20657229e-01
6.60868585e-01 -2.83125997e-01 -8.99092793e-01 5.89350045e-01
9.55684576e-03 2.52972722e-01 3.00253034e-01 -1.44867569e-01
-5.36634549e-02 -1.00404181e-01 -1.27024078e+00 2.08146647e-01
6.95936561e-01 -6.13725483e-01 -5.83717465e-01 2.20506757e-01
4.84028906e-01 -5.05994916e-01 -8.35682452e-01 6.13075435e-01
6.37335479e-01 -8.10787678e-01 1.11276507e+00 -3.54565769e-01
5.71853340e-01 -4.97781336e-01 -4.02004331e-01 -1.47080648e+00
-1.01055205e-02 -5.49109578e-01 -2.95843422e-01 1.18206739e+00
5.67556545e-03 -8.56050134e-01 5.66960096e-01 5.52279353e-01
5.08793518e-02 -4.37359661e-01 -9.97086525e-01 -5.67729652e-01
1.05198681e-01 -5.93047202e-01 4.57194179e-01 6.25170946e-01
-7.53809571e-01 7.69745186e-02 -6.33102953e-01 5.49962938e-01
1.28543341e+00 5.68813682e-02 6.92146122e-01 -1.13015711e+00
-6.62205160e-01 1.02227502e-01 -2.60768056e-01 -1.14198351e+00
2.19668746e-01 -4.63131994e-01 5.34390092e-01 -1.47271562e+00
1.90603033e-01 -5.52868068e-01 -1.54366553e-01 -6.81946427e-02
-2.04632822e-02 3.30076128e-01 -2.05666333e-01 1.63593546e-01
9.70464051e-02 9.60912168e-01 1.21014977e+00 -7.50647485e-02
2.80860066e-01 -6.55823722e-02 -5.87144315e-01 7.25139678e-01
4.32543993e-01 -7.24804163e-01 -5.74753821e-01 -6.19730234e-01
3.41012836e-01 2.56956339e-01 6.52169049e-01 -9.65728521e-01
2.74463952e-01 -4.56956178e-01 3.69152904e-01 -3.88275415e-01
6.64816380e-01 -8.41091156e-01 6.39224291e-01 2.12603137e-01
2.47882996e-02 -2.68564105e-01 8.63758996e-02 1.02345347e+00
-1.79046646e-01 -2.95164138e-01 7.90519893e-01 -7.82298297e-02
-4.63710546e-01 1.47976890e-01 -1.00538477e-01 1.89826548e-01
6.65206432e-01 1.75967455e-01 -1.78174779e-01 -6.50069654e-01
-7.27269650e-01 -7.82691613e-02 3.27163339e-01 8.93419702e-03
6.45208538e-01 -1.16044271e+00 -7.07436204e-01 4.42594826e-01
-3.11233282e-01 2.72675633e-01 4.28028435e-01 9.92475390e-01
-4.41777587e-01 2.08741009e-01 -1.26572579e-01 -8.99358332e-01
-3.95326614e-01 7.02967942e-02 3.40643823e-01 -1.17302522e-01
-5.49814761e-01 4.99509722e-01 4.75321442e-01 -6.84168935e-01
-1.85952827e-01 -1.62743613e-01 2.20416784e-01 -5.60215771e-01
2.80185550e-01 3.74187440e-01 -1.48793802e-01 -6.70164347e-01
-2.32595935e-01 8.62489462e-01 4.01767552e-01 -4.42434549e-01
1.51269805e+00 -2.95672864e-01 -5.05422711e-01 4.98002172e-01
9.55097377e-01 -3.31187136e-02 -1.22544634e+00 -2.80930400e-01
-3.88865024e-01 -4.92831141e-01 4.21850592e-01 -7.36606359e-01
-8.97276342e-01 1.09013784e+00 4.82212782e-01 -6.15946539e-02
8.81572366e-01 -1.24972999e-01 6.17959380e-01 4.82913643e-01
5.19577086e-01 -8.94876659e-01 -3.36508499e-03 5.60429871e-01
1.01897490e+00 -1.20886290e+00 3.26906800e-01 -2.68025726e-01
-1.39645606e-01 8.28139126e-01 1.98797807e-01 -3.23893279e-01
1.14275050e+00 4.03271526e-01 -2.18715101e-01 -6.16025254e-02
-3.13649654e-01 1.76762134e-01 2.98224002e-01 4.96067047e-01
1.60610110e-01 -1.05536856e-01 1.26830358e-02 5.06047547e-01
-2.73086689e-02 1.57616749e-01 5.11205077e-01 9.15198147e-01
-2.36482397e-01 -9.02301788e-01 -7.20171809e-01 4.08314854e-01
-2.60331720e-01 4.73890081e-02 1.94981828e-01 5.55528402e-01
-8.30372870e-02 6.71838939e-01 -4.37110700e-02 -4.89685871e-02
-1.51426390e-01 -7.59830792e-03 4.09266919e-01 -3.30961585e-01
4.83937532e-01 1.74557284e-01 -3.42272997e-01 -4.01861757e-01
-5.47367036e-01 -6.79345191e-01 -1.10275376e+00 -2.63466686e-01
-3.06809068e-01 3.15593123e-01 1.00353169e+00 1.37170029e+00
3.57651353e-01 6.48559690e-01 2.20729187e-01 -9.86636996e-01
-9.39056635e-01 -1.07339191e+00 -8.09622347e-01 3.27959061e-01
4.60337132e-01 -1.13724315e+00 -6.28129661e-01 3.65863703e-02] | [6.763023376464844, 3.5579512119293213] |
3ebd4d79-f559-4494-bc79-c2620c1ae68c | low-resource-named-entity-recognition-based | 2109.07118 | null | https://arxiv.org/abs/2109.07118v3 | https://arxiv.org/pdf/2109.07118v3.pdf | Low-Resource Named Entity Recognition Based on Multi-hop Dependency Trigger | This paper presents a simple and effective approach in low-resource named entity recognition (NER) based on multi-hop dependency trigger. Dependency trigger refer to salient nodes relative to a entity in the dependency graph of a context sentence. Our main observation is that there often exists trigger which play an important role to recognize the location and type of entity in sentence. Previous research has used manual labelling of trigger. Our main contribution is to propose use a syntactic parser to automatically annotate trigger. Experiments on two English datasets (CONLL 2003 and BC5CDR) show that the proposed method is comparable to the previous trigger-based NER model. | ['Jiangxu Wu'] | 2021-09-15 | null | https://aclanthology.org/2022.ccl-1.85 | https://aclanthology.org/2022.ccl-1.85.pdf | ccl-2022-10 | ['low-resource-named-entity-recognition'] | ['natural-language-processing'] | [-1.13023840e-01 3.35396707e-01 1.32421613e-01 -6.14819527e-01
-6.00875437e-01 -9.13812876e-01 5.76558888e-01 5.61863184e-01
-9.16005433e-01 9.83680248e-01 8.03825200e-01 -1.40357107e-01
-1.30718844e-02 -6.52014911e-01 -1.68268710e-01 -1.23188533e-01
-1.74513429e-01 3.98284018e-01 6.11014247e-01 -5.40195584e-01
4.60798383e-01 6.86649203e-01 -9.19913828e-01 3.68701994e-01
1.93079680e-01 2.45236129e-01 5.32338917e-01 8.55527520e-01
-6.10049725e-01 1.07620275e+00 -1.14743745e+00 -2.59468973e-01
-7.15456530e-02 -4.03559744e-01 -1.55458188e+00 -4.33284640e-01
-1.58648431e-01 3.57841611e-01 1.23254374e-01 9.09442902e-01
7.28245378e-01 2.83547550e-01 4.10049617e-01 -5.86716473e-01
-1.33033127e-01 1.29553223e+00 -3.66472840e-01 1.13102555e+00
8.04919243e-01 -7.76541889e-01 1.01440597e+00 -8.37703109e-01
1.28479934e+00 1.26411438e+00 5.88925958e-01 4.31120813e-01
-5.62048078e-01 -3.36007833e-01 6.93288296e-02 -1.61294844e-02
-1.46610975e+00 -1.58148512e-01 6.31155193e-01 -6.08969145e-02
1.40363193e+00 1.15848541e-01 -1.45640731e-01 8.16194654e-01
2.66521156e-01 3.81357789e-01 1.29220164e+00 -1.01781964e+00
2.51655020e-02 -8.11144710e-02 7.03081012e-01 3.84787560e-01
8.01997036e-02 -3.23146701e-01 -4.44398165e-01 -1.79322362e-01
5.72518110e-01 -7.00029254e-01 5.79472519e-02 8.42387378e-01
-1.01503539e+00 7.25166857e-01 4.56627756e-02 1.37147331e+00
-6.51433170e-01 -1.73848599e-01 6.95887864e-01 8.63005668e-02
1.60507634e-01 4.85400915e-01 -7.49548912e-01 -2.82698184e-01
-6.30088031e-01 -1.13611490e-01 1.28656995e+00 9.64969993e-01
2.84038991e-01 -9.87601355e-02 -2.63913929e-01 4.23325121e-01
2.97818601e-01 3.45488876e-01 5.25180042e-01 -2.71079659e-01
6.00055814e-01 7.21508920e-01 3.68813634e-01 -1.12613201e+00
-7.51262724e-01 -1.70499951e-01 -3.44945431e-01 -2.54240453e-01
1.84750646e-01 -5.78595877e-01 -9.78641391e-01 1.43166518e+00
5.86176813e-01 2.73182187e-02 4.79399264e-01 6.73757315e-01
1.51555741e+00 7.92321920e-01 7.77828097e-01 -3.34424853e-01
1.67350054e+00 -4.30558771e-01 -1.09034014e+00 1.80454493e-01
9.23010707e-01 -1.13855410e+00 4.15302634e-01 -1.11686029e-01
-7.10947454e-01 -4.08859104e-01 -7.77474761e-01 -6.09659068e-02
-9.56607103e-01 2.86934912e-01 5.80460072e-01 5.90449810e-01
-8.21155012e-01 3.70110631e-01 -4.04812336e-01 -8.33397508e-01
-5.42264223e-01 5.76494455e-01 -5.97181439e-01 4.03209746e-01
-1.65110803e+00 1.17327273e+00 1.19656587e+00 2.33808503e-01
-3.82958710e-01 -3.86296660e-02 -8.99055541e-01 -2.79648125e-01
4.21109617e-01 9.21146385e-03 1.26466477e+00 -7.41590023e-01
-1.11195445e+00 1.08545208e+00 -3.78122926e-01 -4.96070057e-01
-2.75832504e-01 -3.36197883e-01 -1.05141938e+00 2.31502891e-01
5.16244650e-01 3.65007818e-01 1.31144822e-01 -9.59462345e-01
-9.91666615e-01 -1.01886138e-01 2.32707292e-01 1.63120121e-01
7.92629644e-02 1.34011364e+00 -2.28631735e-01 -6.34684205e-01
-1.49926469e-01 -4.86910224e-01 -3.32115650e-01 -1.62592292e+00
-4.65578020e-01 -9.18376088e-01 7.28439510e-01 -8.25485170e-01
1.79821002e+00 -1.91257584e+00 -2.17827380e-01 4.60531339e-02
-1.09351061e-01 3.11840922e-01 1.20893553e-01 9.94711161e-01
-5.59750795e-01 6.71447039e-01 2.58948989e-02 2.42794484e-01
-1.90044299e-01 6.36341095e-01 -3.88280272e-01 1.31083310e-01
2.84183860e-01 7.00739264e-01 -8.84643734e-01 -1.05134392e+00
-1.50505364e-01 3.60376358e-01 3.42421532e-01 5.84996343e-02
2.56899267e-01 2.74509281e-01 -8.79715323e-01 4.29827392e-01
2.95528173e-01 3.67586732e-01 3.52919757e-01 -2.51972377e-01
-8.08438897e-01 6.09611392e-01 -1.55132747e+00 1.30542898e+00
-1.50051087e-01 4.20895219e-01 -3.05009540e-02 -7.55691528e-01
9.63870049e-01 9.20356512e-01 -5.72032109e-03 -2.59371161e-01
3.12287897e-01 1.13286272e-01 -4.79875095e-02 -7.04855502e-01
7.78277278e-01 -6.09277748e-02 -6.57337070e-01 6.14829734e-02
2.07634121e-01 6.21820629e-01 6.90680623e-01 2.56465375e-01
1.42680192e+00 2.18651086e-01 9.00240779e-01 -5.14460921e-01
8.05485725e-01 2.60943025e-01 9.23744082e-01 5.94675064e-01
-2.04949170e-01 4.50929105e-02 5.21318793e-01 -3.55827153e-01
-7.76408851e-01 -4.93462563e-01 -1.74280331e-01 1.22251487e+00
-2.41644308e-01 -5.06005287e-01 -7.73490846e-01 -1.10205710e+00
-7.72345781e-01 7.64727712e-01 -7.10295141e-01 8.33156645e-01
-1.22218800e+00 -6.02306008e-01 1.04443634e+00 6.98305905e-01
5.20837665e-01 -1.51187015e+00 -8.75383317e-01 6.18448019e-01
-2.13681966e-01 -1.37584257e+00 -1.76562220e-01 5.85013986e-01
-5.80022812e-01 -1.04332232e+00 -1.98645264e-01 -1.14863276e+00
5.98591745e-01 -2.76173234e-01 1.17820132e+00 -7.54171312e-02
-3.55970144e-01 2.88603485e-01 -9.47726727e-01 -6.31609321e-01
-4.17251110e-01 2.52614319e-01 -1.53262153e-01 -5.22404790e-01
8.10739338e-01 -1.30040988e-01 -1.68925226e-02 1.19347818e-01
-9.01081979e-01 -5.65175653e-01 5.38976014e-01 3.55178952e-01
4.29663479e-01 3.10268402e-01 5.97595930e-01 -1.39363337e+00
7.48903513e-01 -5.71550846e-01 -2.48774290e-01 3.67183149e-01
-1.71362042e-01 2.16959834e-01 5.22164166e-01 4.15149964e-02
-1.50640988e+00 2.27685302e-01 -4.38144147e-01 4.64118958e-01
-8.84490430e-01 7.51610518e-01 -2.51198858e-01 2.80176193e-01
6.80643559e-01 -3.52725565e-01 -1.30392933e+00 -7.04901218e-01
2.18460381e-01 6.16963387e-01 5.72637200e-01 -6.84560061e-01
6.39410853e-01 9.58666205e-02 2.24210471e-01 -8.82385015e-01
-1.07261014e+00 -1.10195827e+00 -1.19605386e+00 -1.59063175e-01
1.20770335e+00 -6.91188455e-01 -2.36994848e-01 -1.58593670e-01
-1.52793503e+00 1.92762271e-01 -1.46791562e-01 6.38669550e-01
1.96384951e-01 4.41511035e-01 -6.91737235e-01 -9.18478489e-01
-4.76011157e-01 -1.65717706e-01 8.16450477e-01 5.33343911e-01
-2.92100132e-01 -1.28430283e+00 2.82976031e-01 -1.39399379e-01
-1.10711064e-03 6.57253385e-01 4.52377588e-01 -1.25748038e+00
3.86521995e-01 -3.90830517e-01 2.80637555e-02 -2.58040726e-01
1.12860277e-01 7.53605440e-02 -6.55241728e-01 3.92656803e-01
5.44030033e-02 5.40211983e-02 4.12778676e-01 -4.66092154e-02
6.29099309e-02 -8.32698420e-02 -5.33484578e-01 -3.32127512e-01
1.73267758e+00 4.97564733e-01 6.41400099e-01 4.04288292e-01
5.48821449e-01 6.23086393e-01 9.41699326e-01 1.07335210e-01
3.35409164e-01 3.09103638e-01 -2.02513382e-01 -9.53605771e-02
5.01301549e-02 -1.87883452e-01 3.83657664e-01 1.04960215e+00
-2.46498063e-01 -3.02057922e-01 -1.17239702e+00 6.30752146e-01
-1.70047736e+00 -1.07773352e+00 -8.48866642e-01 1.59581518e+00
1.00567508e+00 3.67496669e-01 5.85997924e-02 8.22061822e-02
1.21683049e+00 -3.35704014e-02 4.20541406e-01 -8.09557676e-01
-3.49630773e-01 5.85141003e-01 7.06082940e-01 5.16755283e-01
-1.50255704e+00 1.40564799e+00 6.80661917e+00 6.02298260e-01
-9.11075115e-01 4.33700562e-01 3.89230065e-02 6.86874151e-01
2.52407134e-01 2.12265372e-01 -1.58026314e+00 5.37445173e-02
1.48050845e+00 -2.11348698e-01 -3.42953086e-01 6.48058355e-01
3.45398426e-01 -3.35985571e-01 -7.50895083e-01 4.04043853e-01
9.66414660e-02 -9.87233877e-01 -3.33758414e-01 -2.23650292e-01
3.70529920e-01 1.24407403e-01 -9.44890618e-01 2.64227509e-01
5.14680326e-01 -6.68928027e-01 3.47432733e-01 3.80268395e-01
3.58792037e-01 -8.66151094e-01 1.24412322e+00 1.82151750e-01
-1.80826378e+00 3.31959158e-01 -5.03097773e-01 -1.09765507e-01
3.81301612e-01 3.63310635e-01 -9.64870572e-01 1.03618288e+00
6.49685025e-01 9.47057307e-02 -6.55500770e-01 9.17403221e-01
-7.57267475e-01 1.03119421e+00 -5.52186608e-01 -3.34053665e-01
4.18707252e-01 3.26536208e-01 7.28153884e-01 2.11099100e+00
-8.38394612e-02 7.03914344e-01 2.93738008e-01 1.64253861e-01
-3.84281129e-02 8.46990824e-01 -7.23211765e-01 -5.92640601e-02
5.07653832e-01 1.53196168e+00 -1.54503298e+00 -4.75212127e-01
-2.68698335e-01 9.64377403e-01 2.56048203e-01 -5.55904955e-02
-6.13383412e-01 -7.86788702e-01 -4.98561084e-01 -2.34863609e-01
6.51991725e-01 -3.53414178e-01 -1.36801610e-02 -7.94655204e-01
-8.24220031e-02 -4.74472225e-01 8.85104001e-01 -5.23653746e-01
-1.05962133e+00 1.02148795e+00 3.06063414e-01 -6.55082583e-01
-4.88753170e-02 -5.84332585e-01 -8.89191270e-01 8.19064498e-01
-1.12960887e+00 -1.12367606e+00 3.37741673e-01 4.88933265e-01
5.13259888e-01 1.92108363e-01 1.23873651e+00 3.56794119e-01
-6.51070714e-01 1.34988353e-01 -5.90326607e-01 7.65071213e-01
6.70053720e-01 -1.56866753e+00 2.01079220e-01 1.40381169e+00
3.74427080e-01 8.98709178e-01 8.60898495e-01 -1.10988092e+00
-9.00166154e-01 -8.32743108e-01 2.02005982e+00 -6.37067616e-01
6.24771357e-01 -6.19066171e-02 -5.12604535e-01 6.32904112e-01
9.06187296e-01 -2.75765359e-01 8.89661789e-01 2.42001861e-01
-1.12466641e-01 2.78698415e-01 -1.25453186e+00 2.72871733e-01
8.15182030e-01 -2.32810572e-01 -1.48792171e+00 1.32564679e-01
5.48989415e-01 -2.87893862e-01 -1.05257702e+00 3.46011162e-01
-2.32912689e-01 -3.57162833e-01 3.05741370e-01 -8.35525334e-01
-3.98802478e-03 -6.34143054e-01 -4.14420366e-02 -7.78266907e-01
-3.39453161e-01 -6.74424112e-01 4.08700198e-01 2.10872984e+00
7.40796208e-01 -1.62857413e-01 1.53732389e-01 6.70522273e-01
-6.22773245e-02 1.02564543e-02 -1.05371523e+00 -5.70699573e-01
-3.00454348e-01 -3.06206375e-01 2.96251476e-01 1.33144474e+00
1.68539003e-01 9.21454191e-01 2.59973276e-02 4.35143232e-01
2.40454525e-01 -3.02536070e-01 8.20813999e-02 -1.42294526e+00
1.05627738e-01 8.99875313e-02 -4.38671529e-01 -4.02162045e-01
3.27672273e-01 -5.92282653e-01 1.41668439e-01 -1.87165010e+00
1.40159754e-02 -2.62455046e-01 -4.40991044e-01 7.43070066e-01
-1.88513502e-01 -8.74692649e-02 -3.92028168e-02 1.04713552e-02
-8.58311594e-01 -2.44018406e-01 5.64327836e-01 5.56680441e-01
-2.45664164e-01 -3.90153587e-01 -3.95822048e-01 9.61792588e-01
8.20793033e-01 -1.00723267e+00 1.12249114e-01 -1.19052283e-01
5.16641855e-01 -1.04601070e-01 -2.25725710e-01 -6.79347515e-01
4.01387513e-01 -1.45048916e-01 4.14810151e-01 -8.19223702e-01
-4.51789856e-01 -8.67229700e-01 -5.88545902e-03 2.53149867e-01
-2.00768039e-01 7.09125817e-01 1.98804483e-01 2.95327365e-01
-3.61002058e-01 -1.05704212e+00 3.63840878e-01 -4.63363856e-01
-1.32296646e+00 -3.70999098e-01 -7.09020674e-01 2.61410803e-01
1.07241035e+00 9.77156013e-02 -1.00854546e-01 2.19269708e-01
-8.29056323e-01 2.63141319e-02 -1.69234246e-01 4.61645961e-01
3.21175873e-01 -9.65287328e-01 -5.68599105e-01 -6.28668368e-01
-2.07022309e-01 -2.94049740e-01 -2.83917129e-01 4.42509472e-01
-5.46933532e-01 6.57661676e-01 -3.00987028e-02 1.44862190e-01
-1.49630165e+00 5.58381081e-01 6.69194683e-02 -7.52551675e-01
-6.87698603e-01 7.25999475e-01 -3.60544324e-01 -3.48151922e-02
-5.95429167e-03 -2.06014454e-01 -1.16736555e+00 2.48763442e-01
5.23919702e-01 2.20793352e-01 -1.76011667e-01 -1.32306314e+00
-7.00229764e-01 3.76445025e-01 -3.89862433e-02 -3.45976353e-01
1.20127702e+00 -1.16325445e-01 -5.65964818e-01 7.28209376e-01
7.43645132e-01 5.58164060e-01 -2.78145615e-02 -2.07020435e-03
1.12699723e+00 1.03409663e-01 -9.72069502e-02 -9.19638634e-01
-3.56139779e-01 2.19815060e-01 5.12028217e-01 5.49190521e-01
9.54970956e-01 1.58939332e-01 5.29919982e-01 6.88994229e-01
5.25302649e-01 -1.63434684e+00 -6.54551744e-01 9.37678337e-01
8.14153671e-01 -9.12024498e-01 -9.78619307e-02 -7.30089486e-01
-6.63715661e-01 1.28906608e+00 5.38039923e-01 -1.54113680e-01
7.63235033e-01 6.43168867e-01 2.40415260e-01 -5.83648562e-01
-5.23811460e-01 -6.96245551e-01 6.57835305e-02 3.66453290e-01
1.07175493e+00 2.90258918e-02 -1.41966379e+00 7.94290423e-01
4.17219028e-02 -1.47942275e-01 5.96379995e-01 1.49058497e+00
-6.48491383e-01 -1.48225749e+00 -4.95245665e-01 -1.56394631e-01
-1.64562869e+00 -3.17455858e-01 -9.09848034e-01 1.01183248e+00
3.27859879e-01 1.39416170e+00 -2.58343250e-01 -6.88767582e-02
6.72576129e-01 4.21510875e-01 3.29957306e-01 -1.13112044e+00
-1.28401721e+00 2.12176010e-01 8.54209781e-01 -2.50663459e-01
-1.07879496e+00 -6.43485308e-01 -2.00473237e+00 4.27368969e-01
-4.52967405e-01 8.55576634e-01 6.97093487e-01 1.23216438e+00
1.70183003e-01 2.79929906e-01 5.38983524e-01 -1.25028566e-01
1.30183816e-01 -1.24061680e+00 -6.24631405e-01 2.89024800e-01
-3.94481778e-01 -4.18467343e-01 -1.83547541e-01 4.53804255e-01] | [9.725305557250977, 9.576695442199707] |
87e4bf19-ae1b-4a5f-aecb-24338ec9dec8 | qpic-query-based-pairwise-human-object | 2103.05399 | null | https://arxiv.org/abs/2103.05399v1 | https://arxiv.org/pdf/2103.05399v1.pdf | QPIC: Query-Based Pairwise Human-Object Interaction Detection with Image-Wide Contextual Information | We propose a simple, intuitive yet powerful method for human-object interaction (HOI) detection. HOIs are so diverse in spatial distribution in an image that existing CNN-based methods face the following three major drawbacks; they cannot leverage image-wide features due to CNN's locality, they rely on a manually defined location-of-interest for the feature aggregation, which sometimes does not cover contextually important regions, and they cannot help but mix up the features for multiple HOI instances if they are located closely. To overcome these drawbacks, we propose a transformer-based feature extractor, in which an attention mechanism and query-based detection play key roles. The attention mechanism is effective in aggregating contextually important information image-wide, while the queries, which we design in such a way that each query captures at most one human-object pair, can avoid mixing up the features from multiple instances. This transformer-based feature extractor produces so effective embeddings that the subsequent detection heads may be fairly simple and intuitive. The extensive analysis reveals that the proposed method successfully extracts contextually important features, and thus outperforms existing methods by large margins (5.37 mAP on HICO-DET, and 5.7 mAP on V-COCO). The source codes are available at $\href{https://github.com/hitachi-rd-cv/qpic}{\text{this https URL}}$. | ['Tomoaki Yoshinaga', 'Hiroki Ohashi', 'Masato Tamura'] | 2021-03-09 | null | http://openaccess.thecvf.com//content/CVPR2021/html/Tamura_QPIC_Query-Based_Pairwise_Human-Object_Interaction_Detection_With_Image-Wide_Contextual_Information_CVPR_2021_paper.html | http://openaccess.thecvf.com//content/CVPR2021/papers/Tamura_QPIC_Query-Based_Pairwise_Human-Object_Interaction_Detection_With_Image-Wide_Contextual_Information_CVPR_2021_paper.pdf | cvpr-2021-1 | ['human-object-interaction-concept-discovery'] | ['computer-vision'] | [-2.92732954e-01 -2.21197903e-01 -1.12912543e-01 -1.17010951e-01
-9.52901125e-01 -3.17905486e-01 5.31119525e-01 1.40824825e-01
-5.78389764e-01 4.72042471e-01 3.46248478e-01 -9.82496366e-02
-2.21674234e-01 -7.97176063e-01 -5.61256230e-01 -7.98309803e-01
-2.27643088e-01 2.18648046e-01 4.50150579e-01 -1.56249568e-01
8.01927876e-03 5.52568316e-01 -1.67371094e+00 9.88417566e-02
6.31137431e-01 1.31810009e+00 2.95329154e-01 5.81401408e-01
2.74460912e-01 5.14270961e-01 -6.81590557e-01 -2.08409652e-01
4.09188688e-01 -1.18536897e-01 -4.49062735e-01 -5.88549003e-02
4.29790527e-01 -5.64620018e-01 -6.72914147e-01 9.06982839e-01
6.45278573e-01 1.31385371e-01 6.57446086e-01 -1.41687381e+00
-6.56332016e-01 2.05243994e-02 -8.33461046e-01 4.95029151e-01
2.27970660e-01 4.78591919e-01 1.13101721e+00 -1.31893790e+00
4.95442271e-01 1.01580298e+00 5.18130004e-01 2.30459332e-01
-6.77204013e-01 -8.17522347e-01 -1.00647308e-01 2.21980184e-01
-1.87684512e+00 -1.45959780e-01 5.74967384e-01 -4.68503684e-01
7.58146107e-01 5.19542456e-01 6.74701333e-01 8.37529361e-01
5.92550635e-02 9.54926670e-01 5.77590823e-01 -1.39382526e-01
-1.95591807e-01 3.13050807e-01 4.39176634e-02 6.55782282e-01
4.52307135e-01 3.17809880e-02 -5.31408250e-01 -1.61505267e-01
8.67022872e-01 5.81415832e-01 -2.34821752e-01 -2.19280332e-01
-1.41017127e+00 1.01876140e+00 9.80375409e-01 3.33457708e-01
-3.48636836e-01 8.32379907e-02 2.17509687e-01 -1.01472005e-01
1.09609678e-01 4.06948239e-01 -6.49481714e-02 1.41859185e-02
-6.59980595e-01 4.54878509e-01 2.10379750e-01 1.21797872e+00
1.00072134e+00 -4.11401302e-01 -3.68066043e-01 7.42688477e-01
8.62845704e-02 6.31921947e-01 3.59501988e-02 -7.82770932e-01
6.31052852e-01 8.89181256e-01 1.97935045e-01 -1.33117688e+00
-3.40001881e-01 -4.36183870e-01 -1.02700305e+00 -2.01458205e-02
2.94170111e-01 -2.78939575e-01 -7.54118443e-01 1.51713586e+00
3.36862803e-01 -3.22570890e-01 -4.52144146e-01 1.13302600e+00
9.65155363e-01 6.22156203e-01 8.02260544e-03 2.70840198e-01
1.66634357e+00 -9.12863314e-01 -5.51624477e-01 -2.59020060e-01
5.60071588e-01 -6.34984851e-01 1.10794961e+00 7.24363653e-03
-9.26160693e-01 -5.33072054e-01 -1.07823384e+00 -1.36746824e-01
-7.64225125e-01 3.81038576e-01 5.40748596e-01 1.86577991e-01
-8.19082499e-01 -1.21095441e-01 -1.49271384e-01 -5.29148757e-01
7.02264309e-01 3.16604704e-01 -4.73085880e-01 -6.88939840e-02
-1.24303925e+00 6.22476459e-01 1.14143446e-01 1.76193565e-01
-7.03699172e-01 -2.85114497e-01 -6.86920404e-01 2.07393840e-01
7.44175375e-01 -5.05509317e-01 8.99686933e-01 -5.20273924e-01
-6.06909037e-01 7.25072205e-01 -3.92122418e-01 -2.91031033e-01
5.34168303e-01 -3.90384883e-01 -2.46356636e-01 4.64800239e-01
4.00662631e-01 8.27991843e-01 6.39962614e-01 -1.23863220e+00
-1.16184843e+00 -3.92624795e-01 2.05945313e-01 2.63460577e-01
-6.36261582e-01 1.59732074e-01 -8.58152151e-01 -6.23405218e-01
-1.29652590e-01 -8.26546371e-01 -1.18079416e-01 3.32864821e-02
-6.43702269e-01 -2.25841761e-01 8.86301875e-01 -3.86468351e-01
1.60273921e+00 -2.20760965e+00 -2.42482066e-01 2.78399318e-01
7.48124361e-01 3.60370725e-01 4.30947952e-02 6.10335112e-01
2.05480471e-01 2.23038495e-01 -6.85506165e-02 -1.35347560e-01
-6.76330552e-02 -1.23460218e-01 -7.56481439e-02 4.13419396e-01
2.85545319e-01 1.28306377e+00 -8.55139077e-01 -6.06657147e-01
4.65986699e-01 4.68352139e-01 -3.11065197e-01 2.93335915e-01
2.99847752e-01 2.25848392e-01 -5.70948899e-01 8.67186427e-01
6.10667109e-01 -3.93801749e-01 -1.88099489e-01 -2.70439267e-01
-1.42743811e-01 -2.13078365e-01 -1.20213902e+00 1.05197728e+00
-6.32377118e-02 6.86056256e-01 -6.84637651e-02 -7.87275791e-01
6.14298344e-01 2.36377522e-01 5.11803687e-01 -7.47371376e-01
1.72510937e-01 7.78381899e-02 -7.03607574e-02 -4.17895824e-01
5.51878035e-01 4.26663041e-01 -2.82446861e-01 2.07168415e-01
-9.57296565e-02 3.46062273e-01 1.22879948e-02 4.64408338e-01
1.15540528e+00 -5.78175247e-01 3.79698247e-01 -1.62174683e-02
3.06383848e-01 -1.64757296e-02 5.49442649e-01 1.00178468e+00
-4.16369826e-01 9.00504291e-01 4.28822815e-01 -4.36103642e-01
-9.92372036e-01 -9.86421347e-01 -7.30585456e-02 1.15574527e+00
4.46435541e-01 -5.24166584e-01 -5.95705628e-01 -8.37539196e-01
1.44358948e-01 8.40779319e-02 -7.76911736e-01 -4.68533449e-02
-3.66286129e-01 -4.38432753e-01 3.31079543e-01 6.94415987e-01
6.97360814e-01 -1.10628998e+00 -7.39217341e-01 -1.25380173e-01
-3.50558341e-01 -9.19918180e-01 -6.25294507e-01 1.18042029e-01
-2.41878241e-01 -1.15042186e+00 -1.03263700e+00 -5.27676702e-01
6.79880559e-01 9.60367620e-01 8.66126895e-01 3.03785115e-01
-7.42686391e-01 2.89982438e-01 -3.27493757e-01 -5.49789608e-01
5.15209258e-01 2.22552851e-01 -5.07472716e-02 1.89662412e-01
7.76167750e-01 -1.13939315e-01 -1.01775646e+00 5.93904376e-01
-7.69078016e-01 -1.52037427e-01 6.87140584e-01 8.93331468e-01
3.90517682e-01 6.47789165e-02 3.40570182e-01 -6.07456744e-01
4.48609680e-01 -7.19320655e-01 -3.70524168e-01 1.58601090e-01
-1.84629530e-01 -4.87507254e-01 1.59459725e-01 -3.34084630e-01
-6.91064417e-01 1.75875276e-02 5.61826602e-02 -4.99950409e-01
-3.54603946e-01 2.08962917e-01 -2.66128033e-01 -8.70248023e-03
5.92165530e-01 4.68737893e-02 -1.78433105e-01 -2.61946827e-01
2.33310059e-01 8.62472475e-01 4.20848370e-01 -1.90430924e-01
1.05567932e+00 7.15019643e-01 -3.49800229e-01 -9.02618527e-01
-7.72834003e-01 -1.00693071e+00 -5.08506775e-01 -2.46633649e-01
1.06806028e+00 -1.08984733e+00 -7.83023119e-01 3.91787499e-01
-1.10847414e+00 -6.72443435e-02 -5.05366102e-02 4.10626680e-01
-1.50838673e-01 1.56090543e-01 -5.24609685e-01 -9.65642571e-01
-2.37666041e-01 -1.09134936e+00 1.21374631e+00 4.64562774e-01
-3.09149474e-01 -5.55504501e-01 -4.02863026e-01 3.85237098e-01
4.49667603e-01 2.73249239e-01 4.71411645e-01 -4.71552849e-01
-8.86111200e-01 -6.12353802e-01 -9.24400747e-01 1.26872674e-01
9.03791413e-02 -2.76179332e-02 -1.03400421e+00 -3.64446938e-01
-3.42729986e-01 -2.19974324e-01 7.95657754e-01 3.71586800e-01
1.11920130e+00 -2.47692198e-01 -6.29992902e-01 4.20987934e-01
1.34176540e+00 1.65641934e-01 7.49333262e-01 3.88510793e-01
9.65217412e-01 6.28857255e-01 8.50480735e-01 6.12701595e-01
4.21463877e-01 8.07216823e-01 5.79042256e-01 -6.23345554e-01
-3.46867852e-02 -1.48632497e-01 -8.16822127e-02 2.63752609e-01
-2.95963287e-01 -2.23019764e-01 -9.52408075e-01 8.25258493e-01
-2.11578059e+00 -1.02264071e+00 -2.24040046e-01 2.29444265e+00
2.26028517e-01 9.12247226e-02 5.10690033e-01 -8.56771469e-02
6.55333638e-01 1.94887787e-01 -2.27135763e-01 2.31087431e-01
-8.67637843e-02 -1.49741277e-01 7.67532110e-01 3.20303679e-01
-1.23757839e+00 7.37620234e-01 5.62078047e+00 9.46826816e-01
-9.87568080e-01 3.56053770e-01 5.27288377e-01 -3.38957846e-01
2.47403421e-02 -2.24780813e-01 -1.05429661e+00 6.35687888e-01
3.96184772e-01 1.03426859e-01 1.09930448e-01 9.62570012e-01
9.26195979e-02 -4.25908923e-01 -6.89467072e-01 1.31471729e+00
4.59046960e-02 -1.43802774e+00 -9.26571861e-02 5.81259906e-01
3.38140160e-01 6.12027459e-02 3.07990283e-01 2.25096673e-01
3.25146392e-02 -1.09290063e+00 7.56270945e-01 3.26871753e-01
8.35575104e-01 -8.52913678e-01 9.98683810e-01 2.89714158e-01
-1.56582761e+00 -4.12733465e-01 -3.98479849e-01 1.30900173e-02
1.29454033e-02 5.65196276e-01 -5.85399568e-01 4.81045127e-01
1.16953206e+00 3.52885395e-01 -8.72830033e-01 1.15424895e+00
-5.14741242e-02 2.98412144e-01 -3.60429138e-01 -1.13641188e-01
3.55900228e-01 1.09267130e-01 5.33821523e-01 1.37866473e+00
4.17213678e-01 1.02329277e-01 1.85704544e-01 7.97561884e-01
-1.36467768e-02 8.32102373e-02 -1.00963295e+00 2.96325266e-01
6.84431911e-01 1.42941117e+00 -7.19434500e-01 -4.26682144e-01
-6.59565687e-01 9.53985929e-01 4.30590361e-01 5.47310114e-01
-1.09858525e+00 -9.38021719e-01 6.11978590e-01 4.72682327e-01
4.65660065e-01 -2.43328705e-01 -2.07670301e-01 -1.09453249e+00
4.10931855e-01 -5.55649161e-01 4.82708424e-01 -6.93686247e-01
-1.14364767e+00 6.64668858e-01 9.50079933e-02 -1.51412284e+00
3.59998941e-02 -5.77373207e-01 -7.03201234e-01 7.79129505e-01
-1.17721415e+00 -1.33038843e+00 -7.34280288e-01 7.47299790e-01
3.18574667e-01 -1.34677052e-01 4.82934564e-01 6.08197153e-01
-6.58394933e-01 7.84689486e-01 -2.94894334e-02 4.23491299e-01
6.83600187e-01 -1.12508523e+00 1.98375762e-01 8.36516380e-01
8.43489170e-02 8.63840938e-01 3.73519450e-01 -4.51172650e-01
-1.25562000e+00 -9.75934565e-01 8.62173080e-01 -7.99849093e-01
4.19229060e-01 -7.06325710e-01 -7.68956602e-01 4.79188561e-01
8.19285288e-02 2.31665760e-01 5.06205738e-01 1.15341552e-01
-2.96421945e-01 -2.10326999e-01 -9.76900280e-01 6.08896554e-01
1.07399428e+00 -5.93292117e-01 -3.48802000e-01 2.89925545e-01
4.08207268e-01 -1.06299706e-01 -6.56626344e-01 3.49105418e-01
6.14034474e-01 -1.22410583e+00 1.07724464e+00 -8.09535235e-02
2.36417010e-01 -4.27020997e-01 -1.76244050e-01 -8.25281024e-01
-5.83545744e-01 -4.55169320e-01 -9.11710933e-02 1.17245662e+00
3.52713466e-01 -7.94333160e-01 6.57418787e-01 5.31225801e-01
1.66896805e-01 -1.04876006e+00 -7.93619037e-01 -7.58075655e-01
-3.41971666e-01 -3.66730481e-01 7.48779356e-01 7.66118169e-01
-1.16968624e-01 3.12988430e-01 -4.44239199e-01 2.88087636e-01
6.26860559e-01 -2.22881973e-01 1.18443501e+00 -1.08337069e+00
-5.22499457e-02 -3.95385653e-01 -4.25570577e-01 -1.07457006e+00
-6.39392376e-01 -3.23643804e-01 -4.12366055e-02 -1.50052881e+00
5.11448026e-01 -6.38040841e-01 -3.75391245e-01 5.41712344e-01
-3.71966958e-01 5.45991659e-01 3.13505113e-01 5.13493359e-01
-8.54038000e-01 3.61774504e-01 1.00065768e+00 -6.26438782e-02
-3.42569612e-02 -1.78606525e-01 -8.63493502e-01 4.71193165e-01
6.86771691e-01 -2.34010786e-01 -2.83508778e-01 -2.36683264e-01
8.99026617e-02 -2.80952305e-01 7.45827496e-01 -1.13909054e+00
2.79165059e-01 -1.07455611e-01 6.74576640e-01 -8.14962506e-01
4.23406899e-01 -8.08516145e-01 3.77451926e-02 2.47714669e-01
-1.60485674e-02 3.75659950e-02 -4.43123141e-03 5.57792306e-01
-3.77167016e-01 6.15220293e-02 6.64130867e-01 -2.31099397e-01
-8.25699508e-01 6.19343162e-01 -4.13240761e-01 -1.19763147e-03
1.22265828e+00 -3.65937114e-01 -3.94290149e-01 -6.10140264e-01
-2.26602942e-01 4.36973870e-01 4.49271679e-01 4.71714705e-01
6.79339707e-01 -1.59363151e+00 -5.14141977e-01 1.98867381e-01
7.00069785e-01 1.18003778e-01 4.16400552e-01 1.02632213e+00
-2.61377752e-01 4.92902130e-01 -1.58462673e-01 -5.75467885e-01
-1.32133400e+00 6.55027270e-01 1.13479510e-01 -8.02671015e-02
-5.93363464e-01 8.55831981e-01 7.76625633e-01 -2.35663310e-01
4.23443735e-01 -5.02411388e-02 -7.88133293e-02 1.66196793e-01
7.18617260e-01 2.72588283e-01 -1.99795023e-01 -9.29410994e-01
-3.88800651e-01 5.71276009e-01 -6.83538914e-02 8.88327733e-02
1.04160941e+00 -3.47296864e-01 1.85236424e-01 6.31870702e-02
1.37838447e+00 8.76235068e-02 -1.16373110e+00 -3.81482750e-01
-2.33635023e-01 -7.95903921e-01 -1.96411029e-01 -4.37956423e-01
-9.65410054e-01 1.12187588e+00 6.98143065e-01 3.92462581e-01
9.32246089e-01 3.50124180e-01 1.01025760e+00 1.90519914e-01
3.80690813e-01 -1.08439291e+00 3.37231278e-01 4.42678720e-01
9.00702655e-01 -1.49644125e+00 -5.72495088e-02 -4.95452791e-01
-8.48782182e-01 6.68835163e-01 8.67996991e-01 5.91422059e-02
5.34033000e-01 -7.00306371e-02 -2.01546848e-02 -7.15665102e-01
-3.39336991e-01 -6.36828840e-01 3.97626191e-01 6.61670089e-01
1.10308252e-01 2.16054976e-01 5.21744937e-02 4.59178239e-01
8.09847265e-02 -3.56355757e-01 1.46655049e-02 9.02143240e-01
-6.06328726e-01 -5.94801605e-01 -5.16853750e-01 6.76877379e-01
-2.96549171e-01 1.38168558e-02 -3.93597662e-01 1.02561879e+00
3.53250235e-01 8.39324772e-01 1.92996010e-01 -5.82345545e-01
4.04748946e-01 -2.88848579e-01 -4.57017794e-02 -5.55763185e-01
-4.27934110e-01 8.54750797e-02 -1.85853336e-02 -7.93433249e-01
5.19437380e-02 -2.55012184e-01 -1.04587102e+00 -3.98491859e-01
-3.73100370e-01 -4.41616401e-02 2.48165533e-01 7.12015986e-01
6.45374298e-01 3.00193250e-01 6.56558812e-01 -1.18766522e+00
-1.38172910e-01 -8.80763590e-01 -6.12678051e-01 5.84073484e-01
3.97236049e-01 -1.02482200e+00 -4.16310966e-01 -3.37497652e-01] | [9.512704849243164, 1.2649269104003906] |
2bd2e6be-8fa7-4952-b87e-1baa271b5d48 | automated-hypothesis-generation-via | null | null | https://openreview.net/forum?id=PnraKzlFvp | https://openreview.net/pdf?id=PnraKzlFvp | Automated hypothesis generation via Evolutionary Abduction | Abduction is a powerful form of causal inference employed in many artificial intelligence tasks, such as medical diagnosis, criminology, root cause analysis, intent recognition. Given an effect, the abductive reasoning allows advancing a plausible set of explanatory hypotheses for its causes. This paper presents a new evolutionary strategy - called Evolutionary Abduction (EVA) - for automated abductive inference, aiming at effectively generating sets of hypotheses for explaining an occurred effect and/or predicting an effect that could occur in the future. EVA defines a set of abductive operators to repeatedly construct hypothetical cause-effect instances, and then automatically assesses their plausibility as well as their novelty with respect to already known instances - a mechanism mimicking the human reasoning employed whenever we need to select the best candidates from a set of hypotheses. Experiments with four datasets confirm that, given a background knowledge, EVA can construct new and realistic multiple-cause hypotheses for a given effect. EVA outperforms alternative strategies based on causal structure discovery, generating closer-to-real instances in most settings and datasets. | ['Roberto Pietrantuono'] | 2021-09-29 | null | null | null | null | ['intent-recognition'] | ['natural-language-processing'] | [ 5.00317276e-01 7.55903184e-01 -4.87668604e-01 -4.35179353e-01
-2.55773753e-01 -2.73222089e-01 8.35241258e-01 4.17758524e-01
2.55511720e-02 1.11596000e+00 4.13860947e-01 -7.37396538e-01
-9.82521772e-01 -1.01926017e+00 -7.42438078e-01 -3.28246713e-01
-4.71439928e-01 9.45732832e-01 4.59211320e-02 -2.03464627e-01
4.55853581e-01 5.12780905e-01 -1.76401317e+00 4.02623296e-01
1.02201104e+00 4.09560800e-01 -1.76136456e-02 4.28134590e-01
1.85046270e-01 9.90639806e-01 -6.40105665e-01 -8.74279678e-01
-8.20964873e-02 -7.97736287e-01 -1.04372501e+00 -2.41126060e-01
-1.83349356e-01 -1.57191351e-01 7.81597942e-02 7.92133570e-01
-4.18082997e-02 -7.52935931e-02 9.08798456e-01 -1.36606395e+00
-4.00457233e-01 1.12672877e+00 -3.06275338e-01 2.82349676e-01
9.21151638e-01 2.48812541e-01 1.08473003e+00 -5.13102591e-01
8.81389499e-01 1.71959853e+00 6.13541365e-01 7.15743005e-01
-1.23976791e+00 -5.75377405e-01 1.73005372e-01 5.57347834e-01
-1.02961147e+00 -4.94013615e-02 6.55042589e-01 -1.98846847e-01
9.90040600e-01 7.52518237e-01 8.57842207e-01 1.14658821e+00
4.82101709e-01 6.92925870e-01 1.00550020e+00 -6.00343466e-01
6.15102768e-01 -1.38842076e-01 -1.00760229e-01 6.50247633e-01
6.31584167e-01 4.12711591e-01 -6.15736783e-01 -9.48435485e-01
3.94379109e-01 -8.80806744e-02 -2.74241090e-01 2.23508060e-01
-1.11275756e+00 1.01679134e+00 3.93867344e-01 1.41514003e-01
-1.00194442e+00 2.17529669e-01 9.29540619e-02 -3.88245322e-02
1.03683293e-01 1.21507084e+00 -7.71289945e-01 1.28599197e-01
-6.17368758e-01 7.67335832e-01 7.10705698e-01 4.18058008e-01
3.93723637e-01 -3.17541152e-01 -2.09101260e-01 2.39931211e-01
3.99636596e-01 2.60140628e-01 4.85460550e-01 -6.75489128e-01
1.57982215e-01 1.23815989e+00 1.53896566e-02 -9.38851595e-01
-2.80875832e-01 -1.89003602e-01 -5.07521033e-01 1.05765730e-01
9.31330249e-02 -1.28920242e-01 -7.57547617e-01 1.72022474e+00
6.58063829e-01 5.73600292e-01 1.39433695e-02 6.50429010e-01
5.34467876e-01 5.66854954e-01 3.44656408e-01 -6.44823968e-01
1.28763068e+00 -5.81867173e-02 -7.38472700e-01 -3.44092339e-01
4.67518687e-01 -1.51729360e-01 6.24400377e-01 5.45506895e-01
-9.16400254e-01 -3.14753354e-02 -8.94807875e-01 4.76602048e-01
-1.85188651e-01 -2.96001077e-01 1.08147585e+00 5.19016564e-01
-4.56877649e-01 6.95388734e-01 -4.85937387e-01 -6.99925749e-03
3.62154722e-01 3.94537002e-01 -2.48539999e-01 -9.10811941e-04
-1.54714453e+00 9.77351844e-01 8.79561365e-01 -8.03832635e-02
-1.07414043e+00 -9.65818465e-01 -6.32192492e-01 1.84731901e-01
8.87563348e-01 -1.09267819e+00 9.87219214e-01 -6.65510952e-01
-9.31609392e-01 5.64640582e-01 -3.20527047e-01 -1.02505326e+00
5.17472744e-01 -2.40836754e-01 -5.68243027e-01 -1.85103729e-01
1.04492947e-01 6.32483006e-01 8.52264464e-01 -1.17009640e+00
-6.39783621e-01 -5.51118791e-01 1.51603088e-01 -8.37613121e-02
2.65759259e-01 2.67814789e-02 2.74872959e-01 -5.34985006e-01
2.06422910e-01 -8.42665732e-01 -6.17927074e-01 -3.21213752e-01
-8.09108734e-01 -5.11336923e-01 5.12910962e-01 -3.32505941e-01
1.51924467e+00 -1.47472918e+00 4.25507307e-01 5.00021875e-01
2.74150819e-01 4.31267880e-02 4.50904042e-01 3.84150535e-01
-4.61589545e-01 4.57933843e-01 -3.59742433e-01 3.64156932e-01
-2.76332825e-01 5.95665276e-01 -7.56632626e-01 -5.69296815e-02
4.59937125e-01 8.57597232e-01 -1.17510879e+00 -4.57435876e-01
8.70763659e-02 -1.27028629e-01 -7.27436185e-01 1.82671756e-01
-7.34889567e-01 1.03214473e-01 -6.41099274e-01 4.02211428e-01
2.29221597e-01 4.31946218e-02 5.00784338e-01 1.66339487e-01
2.16393605e-01 3.21934164e-01 -1.19350338e+00 8.43446493e-01
-2.88001269e-01 2.74087459e-01 -7.69262969e-01 -8.19292009e-01
1.01025152e+00 4.15673256e-01 4.13344763e-02 -2.62843728e-01
2.50113487e-01 3.86668444e-01 3.09738755e-01 -8.86097491e-01
2.12311909e-01 -6.21820331e-01 -5.77136688e-02 4.31307524e-01
-1.49445221e-01 -1.48927867e-01 1.38655841e-01 2.60712147e-01
1.28558373e+00 -1.50804043e-01 1.05719197e+00 -1.00901969e-01
4.95052099e-01 2.87923604e-01 8.16873670e-01 8.49237919e-01
3.54143143e-01 2.69944459e-01 8.42326045e-01 -8.64712894e-01
-7.55141675e-01 -8.78105879e-01 -1.80071935e-01 4.55957472e-01
-1.02005797e-02 -4.52023447e-01 -3.77605796e-01 -9.75252867e-01
2.03811750e-01 1.64673293e+00 -1.06199670e+00 -5.44703782e-01
-5.32975733e-01 -8.57257187e-01 6.16580963e-01 3.43082070e-01
2.61865348e-01 -1.24705935e+00 -1.20318627e+00 1.61523029e-01
-1.69402003e-01 -2.77941972e-01 4.16728407e-01 2.11198434e-01
-9.50876892e-01 -1.46936429e+00 1.58636034e-01 9.07751992e-02
5.84974229e-01 -5.20715415e-01 1.08637917e+00 4.66818690e-01
-4.27555323e-01 -7.34863952e-02 -1.75361246e-01 -8.72478604e-01
-9.71765578e-01 -5.32462060e-01 1.45622894e-01 -9.99492183e-02
2.70600170e-01 -3.18461955e-01 -1.50220171e-01 7.75473192e-02
-8.16926897e-01 3.63939884e-03 6.95451200e-01 9.36774373e-01
4.72535700e-01 6.81555688e-01 7.73085117e-01 -1.13115799e+00
8.93456161e-01 -7.79394925e-01 -2.67748713e-01 5.34085512e-01
-9.46786523e-01 4.58948284e-01 5.22785962e-01 -6.25621080e-01
-1.53320146e+00 -1.90535679e-01 -1.16318621e-01 -2.93340951e-01
-3.29994351e-01 8.46816957e-01 -2.74122775e-01 7.32080400e-01
1.16851699e+00 -1.18171588e-01 -3.15407217e-01 -2.80092865e-01
3.87679994e-01 3.25416148e-01 6.22528613e-01 -7.73935199e-01
4.48394567e-01 3.01239431e-01 4.63582367e-01 -1.98298633e-01
-8.54629517e-01 1.51104433e-02 -5.38635612e-01 -2.99360156e-01
6.11512125e-01 -2.23353699e-01 -8.25502038e-01 -2.96794891e-01
-1.30994976e+00 1.32296711e-01 -4.80770111e-01 4.21686739e-01
-3.16827357e-01 -2.47190759e-01 5.20960093e-02 -1.10453510e+00
-1.68977872e-01 -1.08468306e+00 8.46842885e-01 1.00151964e-01
-1.08777761e+00 -1.02558041e+00 7.01803416e-02 2.01775342e-01
-3.07255626e-01 5.65694749e-01 1.40901887e+00 -9.47524607e-01
-1.26196787e-01 -3.05116296e-01 4.39210832e-01 -4.15138364e-01
3.02202790e-03 3.71062428e-01 -6.63683951e-01 3.48909348e-01
-1.64193973e-01 -2.47771107e-02 6.74773693e-01 5.57437003e-01
1.10557139e+00 -1.03790450e+00 -9.79780614e-01 -1.27314389e-01
1.00641835e+00 6.44959688e-01 8.28374088e-01 2.97163963e-01
6.25054091e-02 7.97804654e-01 8.73634577e-01 6.12098575e-01
-1.27991304e-01 5.14334738e-01 7.16683090e-01 2.48713955e-01
2.32776746e-01 -5.88473439e-01 -2.14662388e-01 -5.96587777e-01
-3.29245329e-01 -2.89265573e-01 -9.92313087e-01 6.40661597e-01
-1.89010346e+00 -1.47829950e+00 -5.15084386e-01 2.12856865e+00
9.92554009e-01 3.24513406e-01 1.88165799e-01 4.88493383e-01
6.89064562e-01 -5.03382027e-01 -5.55940926e-01 -6.25716269e-01
-7.22414330e-02 2.48184279e-01 5.91666885e-02 5.12926579e-01
-5.48096538e-01 5.01996517e-01 6.67143583e+00 6.04175925e-01
-7.32457876e-01 -2.79682308e-01 6.86925888e-01 -1.75814144e-02
-8.89603674e-01 3.13793153e-01 -5.45808017e-01 1.76620454e-01
8.97079468e-01 -4.57940906e-01 1.26317188e-01 7.89073646e-01
2.31419310e-01 -1.09356754e-01 -1.36040294e+00 2.13167936e-01
-2.45030612e-01 -1.74775791e+00 5.69338679e-01 9.97861996e-02
7.34924376e-01 -8.28578651e-01 -1.59645185e-01 4.09938507e-02
4.69339311e-01 -1.14022088e+00 8.56561303e-01 5.44533551e-01
3.11010361e-01 -9.11900997e-01 7.54914582e-01 5.57655454e-01
-5.09419978e-01 -6.85012460e-01 -1.15075104e-01 -3.72553855e-01
2.39140719e-01 7.06710815e-01 -1.55691826e+00 7.65742600e-01
5.38285375e-01 3.48866105e-01 -4.88203198e-01 8.07013094e-01
-8.24668348e-01 9.82932746e-01 -1.74270406e-01 -3.76728773e-01
2.24665385e-02 2.99580753e-01 8.61407220e-01 9.63657498e-01
2.14747772e-01 5.37752092e-01 -3.94117594e-01 1.19446313e+00
2.16683924e-01 -2.17210978e-01 -6.48515046e-01 2.21755624e-01
8.44051301e-01 7.23587811e-01 -6.75395966e-01 -4.82188106e-01
4.90014970e-01 4.68252808e-01 1.03648238e-01 -6.99528977e-02
-9.57975626e-01 1.09325871e-01 2.40740553e-01 1.55680671e-01
-3.10377359e-01 6.58824921e-01 -5.37058592e-01 -5.62806249e-01
-2.85074085e-01 -1.09690845e+00 1.05967021e+00 -1.09558582e+00
-9.03888702e-01 5.25492728e-01 8.39286387e-01 -8.59158814e-01
-8.43916178e-01 -3.56134534e-01 -1.00606334e+00 6.21118248e-01
-6.70321167e-01 -8.57124388e-01 2.02810228e-01 2.53418714e-01
7.23846018e-01 -3.38097475e-03 7.07602739e-01 -4.56654161e-01
-4.35591877e-01 2.62697712e-02 -8.81520450e-01 -5.87587595e-01
1.82709023e-01 -1.21145165e+00 1.03786223e-01 9.27957475e-01
6.39016777e-02 9.26668406e-01 1.43884110e+00 -1.23334074e+00
-1.19698048e+00 -9.48661566e-01 1.20563817e+00 -4.50581998e-01
5.78225017e-01 2.25212410e-01 -1.09374607e+00 7.97782004e-01
1.95056777e-02 -5.43404162e-01 4.44586307e-01 3.44042152e-01
-2.21416414e-01 3.09143275e-01 -1.51047158e+00 1.09130597e+00
1.23531032e+00 2.41650105e-01 -1.30706000e+00 2.72027642e-01
7.18621790e-01 -1.81009352e-01 -5.78875065e-01 5.65998554e-01
4.90789920e-01 -9.67822373e-01 1.17678845e+00 -1.18733704e+00
9.55248952e-01 -3.10391158e-01 2.13843748e-01 -1.33186400e+00
-4.66346145e-01 -6.63641095e-01 -2.75143445e-01 9.75788414e-01
7.01016724e-01 -6.16819322e-01 3.63834739e-01 9.39743459e-01
8.62822905e-02 -1.01584291e+00 -9.93631840e-01 -5.90858757e-01
-4.05038327e-01 -6.37380958e-01 1.23822379e+00 1.17926073e+00
2.74574071e-01 5.62793195e-01 -1.86199546e-01 5.03700674e-01
5.43102324e-01 3.08152586e-01 4.88984346e-01 -1.46956241e+00
-4.72661555e-01 -4.42092657e-01 -4.17152673e-01 -1.08575307e-01
2.22157300e-01 -7.78296590e-01 -9.27496105e-02 -1.46348631e+00
2.31860891e-01 -4.17915285e-01 2.78402895e-01 8.90999615e-01
-6.74871087e-01 -3.43895227e-01 -1.94836050e-01 2.79014576e-02
-4.61185165e-02 4.57800448e-01 1.02764070e+00 -1.22729793e-01
-3.83595705e-01 1.31911248e-01 -8.00790370e-01 1.07409585e+00
6.22745216e-01 -8.53612065e-01 -6.50047541e-01 2.65584290e-01
7.22872138e-01 5.77590883e-01 7.42641330e-01 -3.94840181e-01
-5.23853116e-02 -8.17849815e-01 2.93603152e-01 -4.12144214e-01
-5.48340380e-02 -6.46141350e-01 8.78132701e-01 1.11872828e+00
-7.54548490e-01 2.94951592e-02 1.94826335e-01 5.78417659e-01
4.83453907e-02 -7.14418650e-01 3.20144862e-01 4.05951180e-02
-8.05099070e-01 -3.26832771e-01 -4.33024853e-01 -2.63285995e-01
1.19951332e+00 -1.05287872e-01 -3.05685878e-01 -1.79064482e-01
-6.68237448e-01 5.06479740e-02 -1.16328396e-01 4.41568434e-01
9.40999329e-01 -1.03797865e+00 -1.03786325e+00 -1.66308090e-01
1.50295362e-01 -1.79411888e-01 1.07491314e-01 6.22915208e-01
-1.69273943e-01 2.73681283e-01 1.13520198e-01 -2.73072362e-01
-1.33877516e+00 8.72083604e-01 3.51205885e-01 -3.82724762e-01
-4.81174469e-01 9.11888659e-01 1.35059521e-01 -2.08146840e-01
-4.11462784e-01 -3.00508499e-01 -2.17022732e-01 -2.70861447e-01
5.81819654e-01 6.39163673e-01 -1.11380100e-01 -6.85354844e-02
-5.12042642e-01 -1.68424010e-01 2.05028951e-01 -1.06252737e-01
1.30903172e+00 3.71634513e-01 -2.77271956e-01 2.43538693e-01
2.74973601e-01 -8.88301898e-03 -7.05065846e-01 1.96350873e-01
2.65858680e-01 -6.06240571e-01 -1.28223911e-01 -1.35336089e+00
-4.49918032e-01 2.99181104e-01 -1.46460637e-01 3.70356530e-01
1.31778467e+00 3.80017549e-01 -4.37214598e-02 4.92425919e-01
2.94386178e-01 -5.95579207e-01 -6.93305880e-02 -1.09833382e-01
1.56665957e+00 -8.45592141e-01 3.66071343e-01 -5.87059259e-01
-7.01265395e-01 1.05049682e+00 5.28202295e-01 1.53475866e-01
2.31496796e-01 2.74125189e-01 -5.65283120e-01 -6.23775542e-01
-1.35608232e+00 1.30010530e-01 5.54776311e-01 4.65030521e-01
2.78677404e-01 3.30180407e-01 -6.62558436e-01 5.68494081e-01
-4.32907462e-01 -1.59546986e-01 5.29288590e-01 6.81951046e-01
-1.78111568e-01 -8.13568652e-01 -8.50108743e-01 6.64937437e-01
-2.04542473e-01 1.88407060e-02 -9.55456614e-01 1.03456092e+00
5.07995605e-01 1.05804515e+00 -7.43683726e-02 -3.56506497e-01
2.07421318e-01 4.07280074e-03 4.96587545e-01 -4.49006826e-01
-5.83178282e-01 -3.11618656e-01 4.74168718e-01 -3.69776398e-01
-2.50719845e-01 -9.35719252e-01 -1.63198781e+00 -1.44517124e-01
-6.01454139e-01 2.38430440e-01 3.21261734e-01 1.35425627e+00
2.71829575e-01 6.45455599e-01 3.81034046e-01 -3.09419453e-01
-4.59396005e-01 -7.03638375e-01 -8.91707316e-02 5.07108569e-01
5.29741533e-02 -9.46721733e-01 -2.02247515e-01 6.01942018e-02] | [8.684447288513184, 5.654546737670898] |
749a94b6-8a17-46a7-a67c-724070dffead | explore-image-deblurring-via-encoded-blur | null | null | http://openaccess.thecvf.com//content/CVPR2021/html/Tran_Explore_Image_Deblurring_via_Encoded_Blur_Kernel_Space_CVPR_2021_paper.html | http://openaccess.thecvf.com//content/CVPR2021/papers/Tran_Explore_Image_Deblurring_via_Encoded_Blur_Kernel_Space_CVPR_2021_paper.pdf | Explore Image Deblurring via Encoded Blur Kernel Space | This paper introduces a method to encode the blur operators of an arbitrary dataset of sharp-blur image pairs into a blur kernel space. Assuming the encoded kernel space is close enough to in-the-wild blur operators, we propose an alternating optimization algorithm for blind image deblurring. It approximates an unseen blur operator by a kernel in the encoded space and searches for the corresponding sharp image. Unlike recent deep-learning-based methods, our system can handle unseen blur kernel, while avoiding using complicated handcrafted priors on the blur operator often found in classical methods. Due to the method's design, the encoded kernel space is fully differentiable, thus can be easily adopted in deep neural network models. Moreover, our method can be used for blur synthesis by transferring existing blur operators from a given dataset into a new domain. Finally, we provide experimental results to confirm the effectiveness of the proposed method. | ['Minh Hoai', 'Quynh Phung', 'Anh Tuan Tran', 'Phong Tran'] | 2021-06-19 | null | null | null | cvpr-2021-1 | ['blind-image-deblurring'] | ['computer-vision'] | [ 7.39692226e-02 -2.86075622e-01 3.71195972e-01 -2.99697608e-01
-2.99966305e-01 -5.41982830e-01 2.98093081e-01 -7.02444971e-01
-2.78403997e-01 7.86980569e-01 3.92666847e-01 -1.17447570e-01
-2.60021597e-01 -3.27342808e-01 -7.92081594e-01 -8.22061598e-01
3.30785573e-01 -1.98799506e-01 -5.80122285e-02 2.15430647e-01
2.90859759e-01 3.27136427e-01 -9.71491098e-01 -2.37119235e-02
1.26999032e+00 9.47882056e-01 6.87421143e-01 6.78631127e-01
2.06827864e-01 1.12151659e+00 -4.15794253e-01 -3.90094697e-01
3.56528312e-01 -6.12675548e-01 -9.53377128e-01 3.43834400e-01
5.57733774e-01 -7.92983532e-01 -9.69915509e-01 1.51332033e+00
2.40098178e-01 2.25760937e-01 5.44713318e-01 -7.41788447e-01
-1.43472540e+00 4.47816491e-01 -4.76703405e-01 2.45117560e-01
3.23535427e-02 1.90194800e-01 5.69780409e-01 -1.05333507e+00
2.59945452e-01 1.08917451e+00 7.22306073e-01 3.56361508e-01
-1.31030619e+00 -3.11414927e-01 -2.39119485e-01 4.31290746e-01
-1.35194743e+00 -5.52901864e-01 9.26143110e-01 -3.35740298e-01
2.75641888e-01 3.85046750e-01 4.75511640e-01 9.04529989e-01
1.05422921e-02 7.11845517e-01 1.19802928e+00 -3.04881960e-01
3.56777102e-01 -2.55657025e-02 9.18910131e-02 6.70763552e-01
3.08263779e-01 1.45093635e-01 -2.78813004e-01 -1.01959500e-02
1.35790586e+00 1.14906728e-01 -1.15247905e+00 -5.10825992e-01
-1.37535954e+00 5.73243499e-01 1.01352668e+00 1.96171030e-01
-4.35658187e-01 3.98944139e-01 -4.49707247e-02 2.34862536e-01
4.99726564e-01 7.05358386e-01 -2.13843346e-01 -1.99004754e-01
-1.12730122e+00 1.98284704e-02 6.49646640e-01 9.30311382e-01
8.69020641e-01 -2.41599139e-02 -3.54566723e-01 9.01204288e-01
1.19108088e-01 1.81751832e-01 5.59696317e-01 -1.13833129e+00
6.10085279e-02 2.40018647e-02 6.95339918e-01 -6.89043999e-01
1.52703017e-01 -2.90271401e-01 -1.12064302e+00 9.56633538e-02
5.89260340e-01 -3.29926014e-01 -8.75958383e-01 1.40732074e+00
-2.71009598e-02 8.61790240e-01 5.80152199e-02 1.49972248e+00
3.38103205e-01 4.86419886e-01 -5.77921033e-01 -1.50377020e-01
1.33260059e+00 -1.27970791e+00 -9.28055704e-01 -4.10876662e-01
-5.36131300e-02 -1.01095295e+00 8.91411960e-01 2.59325445e-01
-1.13608837e+00 -7.05978990e-01 -9.07183886e-01 -3.11888427e-01
1.58335611e-01 3.58964592e-01 4.87694025e-01 3.78718644e-01
-1.16879404e+00 7.43066847e-01 -6.50115252e-01 -1.96423873e-01
5.23991048e-01 -3.12956348e-02 -1.18834250e-01 -2.61443496e-01
-1.16059530e+00 1.11322010e+00 4.65646893e-01 5.86375654e-01
-8.65797639e-01 -6.68566823e-01 -8.77075851e-01 2.49778703e-01
2.00237501e-02 -9.61533606e-01 1.44340444e+00 -1.09329152e+00
-1.56060040e+00 4.30936664e-01 -3.93473059e-01 -4.76507396e-01
7.14717448e-01 -8.46581697e-01 -1.96964830e-01 1.86451986e-01
-1.25220641e-01 3.69438350e-01 1.56483006e+00 -1.26439095e+00
-2.20046356e-01 1.47137329e-01 1.32446364e-01 1.87857956e-01
-2.49295726e-01 -1.69244617e-01 -4.60747510e-01 -1.03518414e+00
3.14300247e-02 -6.33569777e-01 -2.02002972e-01 2.66913563e-01
-2.68131971e-01 3.51828873e-01 8.70653510e-01 -8.39239895e-01
1.13921905e+00 -2.42049813e+00 3.75497550e-01 -2.72695571e-01
2.94103414e-01 3.07171553e-01 -3.75290699e-02 1.54971823e-01
-2.21599102e-01 -3.79348934e-01 -5.68095922e-01 -3.61822903e-01
-1.20144591e-01 1.47341758e-01 -5.35389304e-01 7.26954222e-01
2.02531591e-01 9.80166852e-01 -1.19403362e+00 -1.14842899e-01
4.46271896e-01 5.68564355e-01 -3.64510089e-01 7.67108798e-01
1.21129543e-01 5.17621398e-01 -1.59889102e-01 1.66110232e-01
1.20196271e+00 -2.86441147e-01 -2.09689870e-01 -6.77825332e-01
-5.86321354e-02 -2.24098772e-01 -1.00869823e+00 1.87340069e+00
-6.67145729e-01 9.39819574e-01 2.41251171e-01 -7.72319794e-01
7.57554948e-01 2.55795985e-01 -4.45386022e-02 5.85820340e-02
1.43701077e-01 3.66527170e-01 -1.74779937e-01 -7.68478870e-01
4.95136350e-01 -1.23650543e-01 4.56155211e-01 2.57934391e-01
5.34431785e-02 -3.53021830e-01 -3.08286637e-01 -1.53757811e-01
8.77138972e-01 1.80747397e-02 1.26836777e-01 -4.45690751e-01
6.55469656e-01 -5.42711198e-01 2.33303159e-01 8.99944723e-01
-3.46511245e-01 1.03521192e+00 -1.25512462e-02 -5.63303232e-01
-1.01018608e+00 -1.15317392e+00 -2.79095411e-01 5.64700544e-01
5.46080649e-01 1.64108634e-01 -1.21198666e+00 -5.28957903e-01
-3.66303958e-02 6.13731086e-01 -5.33078372e-01 -2.75819302e-01
-4.65394348e-01 -5.19606531e-01 4.05608833e-01 3.73685062e-01
1.14861393e+00 -7.89822102e-01 -4.62675154e-01 3.76978144e-02
-2.87109554e-01 -1.14617014e+00 -1.27069831e+00 -1.39868751e-01
-8.24212551e-01 -1.00829005e+00 -1.42147565e+00 -1.25935721e+00
8.60834420e-01 7.71744132e-01 6.74025893e-01 -1.16989844e-01
-4.90379855e-02 9.15668905e-02 -4.91122268e-02 2.52788156e-01
-4.10745502e-01 -4.02599394e-01 6.40025213e-02 5.88082731e-01
-1.30416272e-04 -5.02822757e-01 -1.01031768e+00 1.60002977e-01
-9.89108086e-01 2.15156749e-01 6.37824059e-01 1.16084671e+00
-7.47832283e-02 2.41988137e-01 2.65113324e-01 -4.43598360e-01
9.28602219e-01 -2.55768538e-01 -6.50680780e-01 1.87670767e-01
-5.00077784e-01 4.20954555e-01 5.34742117e-01 -5.90974212e-01
-1.48219311e+00 -2.37886030e-02 2.85803467e-01 -9.96519506e-01
-2.38159433e-01 3.83456707e-01 3.06352321e-02 -3.29044670e-01
7.04798698e-01 3.66380453e-01 7.90320411e-02 -8.80490482e-01
8.33754718e-01 9.42045271e-01 1.10596991e+00 -3.91331822e-01
9.59412515e-01 4.11304861e-01 -3.10970277e-01 -6.83920622e-01
-8.58659029e-01 -3.43699336e-01 -6.82754993e-01 3.40428539e-02
8.02363992e-01 -9.76056874e-01 -5.32214820e-01 1.14020765e+00
-1.52981222e+00 -2.70229697e-01 -1.18749671e-01 6.18068576e-01
-5.60221672e-01 8.73197019e-01 -1.06258547e+00 -6.74369395e-01
-2.09458292e-01 -1.21144176e+00 8.86084914e-01 4.70596910e-01
-4.27393541e-02 -1.22683716e+00 -9.36019644e-02 1.70236915e-01
6.92858934e-01 -3.48386377e-01 5.20867884e-01 6.70399442e-02
-8.17360103e-01 6.02188632e-02 -7.85183370e-01 6.35251343e-01
7.36711264e-01 -3.58504951e-01 -1.20872128e+00 -3.53862405e-01
6.56630874e-01 -1.07026376e-01 1.05111766e+00 5.46913505e-01
1.41932368e+00 -5.93067110e-01 -4.96309102e-02 1.00219131e+00
1.38645279e+00 -1.02891736e-01 6.96642876e-01 3.16689789e-01
9.26978707e-01 7.45071024e-02 2.31644839e-01 1.76826105e-01
1.19998120e-02 6.33053780e-01 3.31540436e-01 -2.35058025e-01
-3.21112514e-01 -1.05628975e-01 2.85852462e-01 4.93668318e-01
-2.06200443e-02 1.23302847e-01 -3.37909907e-01 6.85089469e-01
-1.94097900e+00 -8.99224937e-01 -7.61846779e-03 2.06060982e+00
1.26830828e+00 -1.84089348e-01 -4.49326217e-01 -2.70606250e-01
9.91268039e-01 2.66680151e-01 -5.70289075e-01 -2.58447498e-01
5.74309602e-02 2.05374673e-01 5.11382997e-01 8.44804466e-01
-1.17806160e+00 8.18763256e-01 6.19015408e+00 6.89517260e-01
-1.13966942e+00 -1.81928910e-02 4.02009517e-01 2.95545459e-01
-9.79186073e-02 1.63168702e-02 2.14027837e-02 7.24589527e-01
5.16460061e-01 -2.57860154e-01 1.18189943e+00 7.65220821e-01
3.34319413e-01 -5.42857274e-02 -1.25592649e+00 1.44896007e+00
-1.56086534e-01 -1.39224672e+00 -2.39151165e-01 -2.69566506e-01
7.75712788e-01 -1.76805347e-01 2.12639645e-01 -1.30160034e-01
1.74843788e-01 -8.92954171e-01 7.58248270e-01 8.48805606e-01
8.03053081e-01 -3.51062924e-01 7.24806488e-01 2.77639449e-01
-6.05582297e-01 4.09818776e-02 -5.72526634e-01 -2.29860872e-01
4.19380926e-02 8.65713298e-01 -6.65011644e-01 3.88373673e-01
7.21876442e-01 9.10337865e-01 -4.82509583e-01 1.50595260e+00
-2.63684511e-01 4.62444991e-01 2.46357903e-01 4.09519285e-01
1.38654441e-01 -5.83294690e-01 5.89688420e-01 1.32378447e+00
6.35770440e-01 -1.17369227e-01 -3.58867884e-01 1.27495253e+00
-3.20766181e-01 -6.17558539e-01 -4.70702112e-01 2.18008652e-01
4.39812571e-01 1.19665873e+00 -1.75118938e-01 -3.26762229e-01
-3.60273361e-01 1.92437446e+00 2.20058069e-01 9.94799674e-01
-9.81094539e-01 -6.81860030e-01 9.65763628e-01 -4.80476171e-01
6.76595867e-01 -4.14029658e-02 -3.51009876e-01 -1.63577807e+00
-5.97368143e-02 -6.74125075e-01 -1.77482560e-01 -1.36152554e+00
-1.65237439e+00 7.01299429e-01 -1.72109962e-01 -1.26684856e+00
1.23914488e-01 -6.28992617e-01 -8.43345582e-01 1.35103810e+00
-1.48014891e+00 -8.68905008e-01 -6.75158381e-01 5.06566107e-01
6.67688966e-01 9.98432785e-02 4.60769624e-01 3.97730879e-02
-4.17151898e-01 2.37927258e-01 5.32332599e-01 2.74870276e-01
9.65196967e-01 -1.43734396e+00 6.26825690e-01 1.25899196e+00
-2.33443268e-02 7.91686356e-01 8.98227036e-01 -3.10489625e-01
-1.12881637e+00 -1.17232072e+00 4.43266988e-01 -2.84310728e-01
8.87866735e-01 -4.33004126e-02 -1.20004284e+00 5.98298252e-01
8.88587654e-01 3.03487927e-01 -3.90563346e-02 -3.60387743e-01
-2.98596203e-01 -2.49294505e-01 -1.10042417e+00 6.44477367e-01
7.68220484e-01 -8.46004605e-01 -8.90825450e-01 2.45153233e-01
8.01017165e-01 -8.17384303e-01 -6.36396050e-01 1.21334232e-01
3.22717667e-01 -9.50434089e-01 1.05700135e+00 -4.69501436e-01
4.84731793e-01 -6.08187318e-01 1.93288416e-01 -1.93655777e+00
-8.64473283e-01 -9.61207986e-01 -5.16953290e-01 9.88599420e-01
-2.68173933e-01 -5.84302843e-01 3.58504385e-01 6.01052046e-01
-2.31379271e-01 -4.48900312e-01 -7.45943904e-01 -6.37215614e-01
-1.37887016e-01 -3.21654342e-02 7.96585917e-01 9.10818279e-01
-2.02638239e-01 1.57535881e-01 -7.70670712e-01 5.33609629e-01
9.08050478e-01 2.50872403e-01 4.13254440e-01 -6.56763613e-01
-5.00549138e-01 -4.99994069e-01 -1.76761404e-01 -1.61391759e+00
6.55296445e-02 -4.60925907e-01 3.63542229e-01 -1.35171127e+00
2.16581792e-01 -1.75258324e-01 -3.38567406e-01 3.46484740e-04
-5.99654377e-01 2.13682160e-01 7.99494311e-02 4.30431873e-01
-8.51813480e-02 6.86412275e-01 1.58835721e+00 -2.20145971e-01
-3.36482525e-02 2.76704878e-02 -7.17062593e-01 7.32477069e-01
4.70493019e-01 3.56206903e-03 -4.78583485e-01 -8.14609051e-01
-4.94917482e-01 -1.16262965e-01 7.64123797e-01 -8.37441385e-01
3.04565012e-01 -2.79111058e-01 3.64633322e-01 1.00011177e-01
1.57090768e-01 -8.17052484e-01 2.25257367e-01 1.99441880e-01
-3.04078430e-01 -3.88084531e-01 1.16918460e-01 6.15941107e-01
-2.61168450e-01 -5.96289039e-01 9.71829951e-01 2.71504074e-02
-7.11343408e-01 4.57019433e-02 -4.96201292e-02 -6.45764247e-02
6.68079913e-01 -1.39544457e-01 -4.63995337e-01 -4.79529291e-01
-5.73378325e-01 -1.71957418e-01 7.81124115e-01 2.65300095e-01
7.75916040e-01 -1.32360613e+00 -4.19285744e-01 2.52951711e-01
-1.94939405e-01 1.10096104e-01 2.69358516e-01 7.90419877e-01
-8.87321770e-01 1.35093957e-01 -1.38199225e-01 -4.12922084e-01
-7.27050424e-01 9.56750393e-01 7.06319511e-01 3.82151008e-01
-5.68543971e-01 1.00468683e+00 5.79259396e-01 2.33206287e-01
9.24121067e-02 -6.79229140e-01 3.18554819e-01 -5.76741219e-01
6.36571229e-01 3.30787778e-01 -1.71352118e-01 -4.05258209e-01
-1.07845098e-01 3.82337213e-01 -7.70999044e-02 2.60160007e-02
1.19604194e+00 -4.01368946e-01 -5.32044649e-01 1.20623007e-01
1.32238364e+00 9.43081453e-02 -1.73394907e+00 -3.01341563e-01
-1.39786154e-01 -9.77867067e-01 3.81092429e-01 -5.18016100e-01
-1.10750496e+00 7.22306728e-01 6.41379833e-01 1.69060186e-01
1.35955238e+00 -4.54917252e-02 8.47311378e-01 2.10421517e-01
1.15937039e-01 -6.13709331e-01 1.00717224e-01 2.51707077e-01
1.07486868e+00 -1.09045088e+00 -2.71692127e-01 -3.16709757e-01
-4.94593859e-01 1.31796801e+00 3.34358811e-01 -3.17168415e-01
5.85838020e-01 -4.56435606e-02 1.33364797e-01 1.78657442e-01
-2.33016595e-01 2.50000111e-03 4.57309484e-01 5.77880561e-01
1.99008301e-01 -2.46587291e-01 1.10807218e-01 1.58793867e-01
9.13367793e-02 5.27499855e-01 8.63343358e-01 4.44289744e-01
-3.66044313e-01 -7.63552368e-01 -6.23025239e-01 7.23643154e-02
-7.53725022e-02 -3.61012787e-01 5.63510507e-02 7.56849051e-02
1.08447559e-01 8.21714461e-01 1.79554080e-03 -7.82366619e-02
1.16943300e-01 -1.01755299e-01 6.55883431e-01 -1.72940850e-01
3.85605469e-02 5.23862056e-02 -3.90592068e-01 -3.37172449e-01
-2.88826883e-01 -4.40687001e-01 -7.39779949e-01 -2.75805831e-01
-3.93695772e-01 1.74758077e-01 6.23494126e-02 8.58278871e-01
3.50884348e-01 2.56526709e-01 7.18688786e-01 -9.04019892e-01
-7.56767869e-01 -1.09726131e+00 -6.73018456e-01 6.17892087e-01
1.00743461e+00 -4.98257399e-01 -5.13927877e-01 5.73735952e-01] | [11.584650993347168, -2.7253878116607666] |
1bc1162c-592c-4b2a-b9ca-b50358e4baef | large-language-models-are-not-abstract | 2305.19555 | null | https://arxiv.org/abs/2305.19555v1 | https://arxiv.org/pdf/2305.19555v1.pdf | Large Language Models Are Not Abstract Reasoners | Large Language Models have shown tremendous performance on a large variety of natural language processing tasks, ranging from text comprehension to common sense reasoning. However, the mechanisms responsible for this success remain unknown, and it is unclear whether LLMs can achieve human-like cognitive capabilities or whether these models are still fundamentally limited. Abstract reasoning is a fundamental task for cognition, consisting of finding and applying a general pattern from few data. Evaluating deep neural architectures on this task could give insight into their potential limitations regarding reasoning and their broad generalisation abilities, yet this is currently an under-explored area. In this paper, we perform extensive evaluations of state-of-the-art LLMs on abstract reasoning tasks, showing that they achieve very limited performance in contrast with other natural language tasks, and we investigate the reasons for this difference. We apply techniques that have been shown to improve performance on other NLP tasks and show that in most cases their impact on abstract reasoning performance is limited. In the course of this work, we have generated a new benchmark for evaluating language models on abstract reasoning tasks. | ['Gillian Dobbie', 'Michael Witbrock', 'Qiming Bao', 'Gaël Gendron'] | 2023-05-31 | null | null | null | null | ['reading-comprehension', 'common-sense-reasoning'] | ['natural-language-processing', 'reasoning'] | [ 1.56320244e-01 5.00153482e-01 9.04324204e-02 -3.17334265e-01
-4.51291114e-01 -5.73571980e-01 1.05318892e+00 4.60424304e-01
-6.57755136e-01 5.93501389e-01 3.81811321e-01 -7.45930076e-01
-4.60217565e-01 -7.01394439e-01 -4.92893815e-01 -2.53442526e-01
8.12137797e-02 8.80057454e-01 4.52714801e-01 -3.75797570e-01
5.97230434e-01 5.89445770e-01 -1.56822419e+00 6.02027297e-01
6.56977475e-01 5.58644891e-01 9.45878476e-02 7.66603470e-01
-7.00625122e-01 1.40049255e+00 -7.01049685e-01 -5.30860007e-01
-2.43634582e-01 -4.62361544e-01 -1.40584612e+00 -4.75950420e-01
4.63250518e-01 -1.64075270e-01 4.42307331e-02 7.57755160e-01
3.03404480e-01 2.74668247e-01 6.81752563e-01 -8.25304627e-01
-6.23906910e-01 7.36334920e-01 -1.74575925e-01 6.45526707e-01
8.20748627e-01 1.25085011e-01 1.15598941e+00 -6.96443200e-01
4.72334176e-01 1.63638425e+00 6.07635379e-01 5.52392483e-01
-1.23288035e+00 -3.14999521e-01 3.47772509e-01 6.72926679e-02
-1.12494075e+00 -5.32122254e-01 2.68841773e-01 -2.78469950e-01
1.61707568e+00 8.71222392e-02 4.85547334e-01 8.08597803e-01
4.73506242e-01 9.60689306e-01 1.44947851e+00 -6.40449464e-01
3.29357982e-01 6.72445968e-02 5.99074066e-01 8.52533758e-01
4.36985821e-01 -3.76763433e-01 -5.95584810e-01 -1.15432918e-01
5.50968230e-01 -1.91907257e-01 -2.92780161e-01 -5.54419123e-02
-1.18746233e+00 8.12894523e-01 4.57472861e-01 5.88069499e-01
-3.42386931e-01 1.80641368e-01 4.40900475e-01 5.69779098e-01
5.66495396e-02 9.11585808e-01 -4.83914316e-01 -1.52950838e-01
-9.06723797e-01 6.15472853e-01 1.29679763e+00 4.95648861e-01
4.64633256e-01 -2.31702134e-01 -2.70645559e-01 7.34374166e-01
1.47391647e-01 2.05507964e-01 6.43015862e-01 -1.05648160e+00
5.10074377e-01 8.07013571e-01 -1.95785481e-02 -8.36563766e-01
-8.84157419e-01 -3.73447299e-01 -4.84869003e-01 4.45942551e-01
7.21771002e-01 1.03206135e-01 -5.40320396e-01 1.97645676e+00
-3.78281415e-01 -4.94337201e-01 3.23136508e-01 7.53516734e-01
8.81106555e-01 4.25673068e-01 4.27980512e-01 3.05552948e-02
1.56400907e+00 -8.31834316e-01 -5.60386777e-01 -7.67686009e-01
7.17952311e-01 -4.63696420e-01 1.39973032e+00 5.90923369e-01
-1.54120862e+00 -4.57159489e-01 -9.94075775e-01 -4.18367177e-01
-6.85526073e-01 -2.92509049e-01 1.11114728e+00 5.72366357e-01
-1.24007678e+00 4.94557083e-01 -6.46008313e-01 -7.44511783e-01
2.70961195e-01 2.63080508e-01 -1.25449732e-01 -2.11350903e-01
-1.37491310e+00 1.34969890e+00 4.84918594e-01 -9.13621783e-02
-2.56934077e-01 -4.17856485e-01 -7.59449065e-01 4.86772060e-01
5.13465285e-01 -9.14645076e-01 1.46192801e+00 -5.10442436e-01
-1.18971384e+00 1.27052462e+00 -1.86835423e-01 -7.60371745e-01
3.87913823e-01 -1.78882703e-01 -3.28865737e-01 2.61217561e-02
4.32258472e-02 7.14270175e-01 2.18570501e-01 -9.39322054e-01
-2.67461717e-01 -2.15539128e-01 5.68744421e-01 2.39053786e-01
-1.85755603e-02 4.54025082e-02 -4.15749475e-02 -4.30227458e-01
1.87996820e-01 -9.22014296e-01 2.63624694e-02 1.92020333e-03
5.57124056e-03 -5.79537272e-01 1.80048600e-01 -4.74992633e-01
9.61568236e-01 -1.68908715e+00 9.19139236e-02 -1.17921010e-01
2.16468364e-01 4.21135396e-01 -9.70740393e-02 7.08164036e-01
4.21327092e-02 3.85828018e-01 -3.52059007e-01 -8.57289284e-02
4.36752766e-01 2.91945338e-01 -7.58420110e-01 -1.16551749e-01
1.92263514e-01 1.25256193e+00 -7.69345760e-01 -3.38164955e-01
1.02204075e-02 5.92678972e-03 -4.56208706e-01 8.03440884e-02
-7.95268297e-01 -1.14471242e-01 -3.47236305e-01 1.20382957e-01
2.40181074e-01 -3.82574618e-01 2.45244727e-01 3.41434330e-01
1.02760727e-02 8.05199206e-01 -9.36197758e-01 1.76039207e+00
-5.32644570e-01 8.92338574e-01 -2.21171603e-01 -9.28376377e-01
7.10154116e-01 2.44536221e-01 -3.42116177e-01 -9.67391133e-01
-8.32364634e-02 1.27454966e-01 6.21198833e-01 -5.33213913e-01
3.73519897e-01 -4.40043449e-01 -6.76122457e-02 9.77483094e-01
-1.22549169e-01 -5.21394372e-01 4.48812455e-01 2.94239998e-01
1.22986293e+00 9.80004594e-02 5.03297269e-01 -5.13046384e-01
6.30070746e-01 2.89940447e-01 -1.78550899e-01 1.35218537e+00
-9.40666050e-02 4.24982101e-01 7.49117970e-01 -6.23920441e-01
-5.21738648e-01 -1.11164534e+00 -1.63872186e-02 1.13963699e+00
-8.24413896e-02 -5.51848471e-01 -6.35509431e-01 -2.68643826e-01
-3.07343513e-01 1.26156795e+00 -4.15361494e-01 -2.55414754e-01
-3.67550015e-01 -7.41351008e-01 8.04124713e-01 6.41583562e-01
7.83385575e-01 -1.59718871e+00 -9.83152747e-01 1.60477370e-01
-8.37516412e-03 -1.26675594e+00 3.99042994e-01 2.69508719e-01
-9.02103424e-01 -9.29048181e-01 -3.30759078e-01 -6.18145883e-01
1.87865213e-01 2.54816025e-01 1.80046892e+00 5.32280624e-01
9.01746750e-03 4.66089278e-01 -1.79347128e-01 -7.01201379e-01
-5.47060847e-01 1.81699321e-01 -9.50938389e-02 -8.36271048e-01
6.82415545e-01 -4.97050166e-01 -3.78628045e-01 -3.10868677e-02
-1.07530010e+00 9.77464393e-02 8.04115772e-01 5.31427741e-01
-1.11089200e-01 1.34602189e-01 5.89169502e-01 -9.05209303e-01
1.38417435e+00 -2.58994371e-01 -1.85686901e-01 2.78085202e-01
-5.03145099e-01 5.03292501e-01 5.85824847e-01 -1.20986409e-01
-1.18315113e+00 -6.55213833e-01 -2.62827367e-01 4.01489109e-01
-4.35906202e-01 8.89126182e-01 3.10631901e-01 2.02281147e-01
8.22332203e-01 2.99608558e-01 -7.47469366e-02 -2.28508383e-01
1.43594474e-01 2.02761874e-01 3.69789928e-01 -1.10742211e+00
4.14040536e-01 2.22672820e-01 2.65222248e-02 -9.08862472e-01
-1.37270653e+00 -8.28088522e-02 -3.56214166e-01 4.31887031e-01
9.05443907e-01 -8.43519628e-01 -7.39093363e-01 1.83275770e-02
-1.30162239e+00 -6.56403840e-01 -8.50008577e-02 2.70347536e-01
-6.66257560e-01 2.17887238e-01 -7.97763288e-01 -7.19663382e-01
-2.20328629e-01 -1.11828947e+00 8.55343044e-01 1.86401844e-01
-9.48312640e-01 -1.28223085e+00 8.82817879e-02 5.46249807e-01
5.61773181e-01 -1.89677507e-01 1.44911802e+00 -9.02037084e-01
-6.13713324e-01 1.00112204e-02 -3.87730062e-01 8.99257287e-02
-3.58512133e-01 -4.19690222e-01 -9.65233922e-01 -1.54039145e-01
9.31515023e-02 -6.99760258e-01 1.15964961e+00 1.29774407e-01
1.09958351e+00 1.27403736e-01 -2.45256096e-01 1.62678048e-01
1.19409168e+00 -9.00925249e-02 7.51226068e-01 4.55303818e-01
8.01600069e-02 7.73238897e-01 2.18317330e-01 -1.20346852e-01
4.94455367e-01 5.17500699e-01 -6.56786188e-02 1.27945051e-01
-2.10987017e-01 9.34623461e-03 3.24685752e-01 2.60302901e-01
-1.03139177e-01 -3.52964789e-01 -1.37622094e+00 4.50920641e-01
-1.80744600e+00 -1.18831944e+00 1.17921811e-02 1.92637229e+00
7.53087878e-01 7.45410323e-01 -1.35023281e-01 8.17653760e-02
9.02906731e-02 2.16765925e-01 -4.33087736e-01 -1.08530414e+00
-8.73000845e-02 4.38887030e-01 -2.10371792e-01 5.07859111e-01
-4.59918290e-01 9.86260653e-01 6.82041645e+00 3.94057244e-01
-8.30022752e-01 -1.53704166e-01 2.89246500e-01 2.02768907e-01
-3.47110838e-01 -9.71281379e-02 -6.02361321e-01 -1.76109090e-01
1.09574544e+00 3.36934961e-02 4.85390902e-01 3.93806130e-01
-3.29993367e-02 -3.81649077e-01 -1.45404732e+00 6.60218835e-01
2.78970838e-01 -1.24267066e+00 3.01503301e-01 -2.34276593e-01
3.74749094e-01 1.02029622e-01 -1.83860749e-01 8.37136209e-01
3.67228746e-01 -1.40579605e+00 6.07407212e-01 5.53014755e-01
3.27635817e-02 -4.44224864e-01 6.75563991e-01 7.38498569e-01
-6.13917589e-01 7.62048960e-02 -4.94537711e-01 -8.36624086e-01
-7.89722148e-03 3.24302137e-01 -6.75033987e-01 1.55108020e-01
5.11117101e-01 8.19856767e-03 -9.89615202e-01 6.92143619e-01
-7.37737775e-01 5.21836519e-01 -2.32682094e-01 -3.08572888e-01
3.62032950e-01 1.72150150e-01 2.56962150e-01 1.38829994e+00
-9.93214771e-02 2.03061327e-01 7.12166503e-02 1.27449059e+00
6.58307299e-02 -1.72813624e-01 -5.80942988e-01 -2.84011692e-01
2.02297553e-01 8.65626574e-01 -9.94035661e-01 -4.18869019e-01
-6.70552611e-01 7.88562775e-01 8.06389809e-01 3.62011850e-01
-5.85357249e-01 -2.50723094e-01 5.75078428e-01 -3.65475789e-02
-9.48115215e-02 -3.72815430e-01 -4.69695479e-01 -1.20219958e+00
-5.95209561e-02 -8.86547685e-01 6.25275433e-01 -1.32128143e+00
-1.02972126e+00 4.35986161e-01 1.57672971e-01 -2.75092781e-01
-5.74110150e-01 -1.02631068e+00 -6.81237280e-01 1.08956826e+00
-1.46114159e+00 -7.15549886e-01 -3.01546484e-01 2.37371296e-01
6.26991928e-01 -1.17786191e-01 1.16004169e+00 -5.13580203e-01
-1.87360123e-01 2.33425051e-01 -4.14535105e-01 5.89275993e-02
4.91089851e-01 -1.34724903e+00 7.38521934e-01 6.03150845e-01
3.61025214e-01 1.18395126e+00 9.13367808e-01 -2.96378136e-01
-1.24280894e+00 -4.47036594e-01 9.65702593e-01 -8.76370907e-01
6.05548918e-01 -4.53155160e-01 -1.23445833e+00 9.78029072e-01
4.71167713e-01 -4.55659866e-01 5.85185468e-01 5.30501544e-01
-5.82475066e-01 4.22058165e-01 -9.45236444e-01 1.03921783e+00
1.07959783e+00 -6.22276962e-01 -1.48420143e+00 3.49783927e-01
7.09334075e-01 -2.41501227e-01 -5.71563244e-01 7.69344792e-02
5.84136665e-01 -1.36401737e+00 1.17656255e+00 -7.88580775e-01
7.23465621e-01 -9.40185189e-02 -3.34055647e-02 -1.35996795e+00
-3.81640851e-01 -2.30412304e-01 6.69728592e-02 1.02762628e+00
5.02613485e-01 -9.37706947e-01 4.87996727e-01 7.72176206e-01
7.58973137e-02 -7.64164150e-01 -4.02557701e-01 -6.39658630e-01
6.18124902e-01 -5.51409423e-01 4.43557024e-01 8.04129541e-01
3.18072021e-01 8.75948310e-01 3.11784744e-01 3.58728394e-02
3.44657928e-01 3.75245094e-01 6.73938692e-01 -1.40389729e+00
-2.84035534e-01 -9.23125923e-01 -2.73690879e-01 -8.66691828e-01
7.31524169e-01 -1.07150853e+00 -2.20182449e-01 -2.09402537e+00
2.04540625e-01 5.18850870e-02 -6.04810491e-02 4.66243297e-01
-2.25951552e-01 -1.70395151e-01 2.56271899e-01 5.07268719e-02
-7.74215758e-01 8.17996338e-02 1.03548801e+00 -8.56802762e-02
1.21086337e-01 -1.59561634e-01 -1.04381442e+00 9.54845130e-01
1.04147279e+00 -9.96404886e-02 -7.72149503e-01 -6.37983263e-01
6.42046809e-01 -6.05654791e-02 4.47164208e-01 -1.35218334e+00
2.80407965e-01 -1.89994246e-01 4.22713518e-01 -2.51662284e-01
3.13799411e-01 -5.17631114e-01 -4.15198952e-01 7.38933325e-01
-7.81975806e-01 4.19259638e-01 6.62810206e-01 1.43858477e-01
-1.74164414e-01 -4.76203322e-01 5.82442641e-01 -6.51995420e-01
-9.79848206e-01 -3.34897071e-01 -6.73264742e-01 4.44957852e-01
3.88767719e-01 -1.18859053e-01 -5.20975590e-01 -4.96733636e-01
-5.41850507e-01 2.34058022e-01 4.24147785e-01 5.21714866e-01
4.06286776e-01 -8.07589412e-01 -6.71951234e-01 -1.38980776e-01
1.64821237e-01 -5.11171743e-02 3.70488828e-03 6.14112973e-01
-6.68778062e-01 9.57813799e-01 -2.03813896e-01 -3.90043378e-01
-9.12322998e-01 7.65886307e-01 4.68285114e-01 -4.97100532e-01
-7.31044769e-01 6.92125201e-01 4.22852844e-01 -6.55149519e-01
1.32834196e-01 -6.83250189e-01 -2.01331139e-01 -1.82681307e-01
7.38689899e-01 -2.08417177e-02 2.28098959e-01 -6.57861009e-02
-4.12895054e-01 4.49590772e-01 -2.40144968e-01 -1.58043459e-01
1.19976807e+00 3.20489369e-02 -3.25542927e-01 6.60437286e-01
6.68993771e-01 -1.50989056e-01 -5.73123157e-01 -2.19891563e-01
4.11711007e-01 9.12613422e-02 -2.35193655e-01 -1.12214601e+00
-3.44485462e-01 1.21532297e+00 -5.19849965e-03 4.52858210e-01
8.59784663e-01 1.49300218e-01 3.64369512e-01 1.05002332e+00
3.22676420e-01 -7.52022326e-01 5.48639148e-02 9.72546101e-01
1.13304019e+00 -1.24368536e+00 2.74063140e-01 -1.04124337e-01
-4.21909273e-01 1.22444570e+00 6.24895573e-01 -4.51291651e-02
4.28005382e-02 3.21272165e-01 -4.67959754e-02 -6.42389774e-01
-1.17478490e+00 -3.06193352e-01 4.64247689e-02 3.93682390e-01
9.67454493e-01 -5.59667535e-02 -3.76628488e-01 3.56646627e-01
-6.33969605e-01 1.40570492e-01 6.14186287e-01 9.83822107e-01
-6.79953098e-01 -8.20007026e-01 -4.69149560e-01 4.02999282e-01
-5.11985600e-01 -2.19216570e-01 -7.17114806e-01 9.88450587e-01
-3.45747620e-01 9.22842503e-01 -3.83978263e-02 2.78984576e-01
4.13721949e-01 6.30480647e-01 8.02344739e-01 -8.23288441e-01
-5.31531036e-01 -8.16115856e-01 3.55493098e-01 -4.23377931e-01
-3.23554784e-01 -5.91217756e-01 -1.57490528e+00 -4.22140986e-01
2.38942042e-01 5.76604307e-02 3.40195894e-01 1.44847441e+00
1.33032277e-02 7.14159846e-01 -7.52646863e-01 -5.07681251e-01
-6.55935049e-01 -8.07433367e-01 -2.23280117e-01 4.98765349e-01
3.18845868e-01 -5.27615190e-01 -2.32831284e-01 -2.85225689e-01] | [9.588494300842285, 7.3509840965271] |
93c7d141-1b56-45be-9050-763d59f5f364 | pose-recognition-in-the-wild-animal-pose | 2111.08259 | null | https://arxiv.org/abs/2111.08259v1 | https://arxiv.org/pdf/2111.08259v1.pdf | Pose Recognition in the Wild: Animal pose estimation using Agglomerative Clustering and Contrastive Learning | Animal pose estimation has recently come into the limelight due to its application in biology, zoology, and aquaculture. Deep learning methods have effectively been applied to human pose estimation. However, the major bottleneck to the application of these methods to animal pose estimation is the unavailability of sufficient quantities of labeled data. Though there are ample quantities of unlabelled data publicly available, it is economically impractical to label large quantities of data for each animal. In addition, due to the wide variety of body shapes in the animal kingdom, the transfer of knowledge across domains is ineffective. Given the fact that the human brain is able to recognize animal pose without requiring large amounts of labeled data, it is only reasonable that we exploit unsupervised learning to tackle the problem of animal pose recognition from the available, unlabelled data. In this paper, we introduce a novel architecture that is able to recognize the pose of multiple animals fromunlabelled data. We do this by (1) removing background information from each image and employing an edge detection algorithm on the body of the animal, (2) Tracking motion of the edge pixels and performing agglomerative clustering to segment body parts, (3) employing contrastive learning to discourage grouping of distant body parts together. Hence we are able to distinguish between body parts of the animal, based on their visual behavior, instead of the underlying anatomy. Thus, we are able to achieve a more effective classification of the data than their human-labeled counterparts. We test our model on the TigDog and WLD (WildLife Documentary) datasets, where we outperform state-of-the-art approaches by a significant margin. We also study the performance of our model on other public data to demonstrate the generalization ability of our model. | ['Sk Shahnawaz', 'Samayan Bhattacharya'] | 2021-11-16 | null | null | null | null | ['animal-pose-estimation'] | ['computer-vision'] | [ 1.52749091e-01 -8.87111854e-03 1.42034382e-01 -2.54249305e-01
-2.94789106e-01 -5.70088208e-01 4.68892545e-01 2.96684861e-01
-9.72735584e-01 5.95043659e-01 -2.67513245e-01 -2.14993209e-02
-6.08018413e-02 -6.47710621e-01 -7.84764171e-01 -7.58918524e-01
-2.94188678e-01 6.79760635e-01 3.77918392e-01 -1.17655033e-02
2.56482307e-02 7.68010020e-01 -1.71119976e+00 -1.25587672e-01
5.35624862e-01 8.65684211e-01 4.00412083e-01 6.23798251e-01
1.29992574e-01 3.34202468e-01 -4.25270945e-01 -3.52088600e-01
2.45643869e-01 -5.98149121e-01 -7.20900536e-01 4.25316423e-01
5.53008974e-01 -3.58568221e-01 1.31994262e-01 6.87252522e-01
4.47336823e-01 -1.43663168e-01 8.80610406e-01 -1.01104283e+00
1.30790666e-01 1.34784222e-01 -8.63797128e-01 1.50261059e-01
9.42934975e-02 1.57505319e-01 8.05761158e-01 -4.10284787e-01
7.06656575e-01 9.25418317e-01 6.87048078e-01 5.07578373e-01
-1.37407219e+00 -5.10253727e-01 -6.13633618e-02 9.44502205e-02
-1.17845953e+00 -3.33588332e-01 5.52539110e-01 -6.10952318e-01
5.52900076e-01 6.27132133e-02 9.16264653e-01 8.11636865e-01
3.48143503e-02 6.25837505e-01 1.00679648e+00 -3.63861680e-01
2.41509125e-01 -2.34833837e-01 -2.93772280e-01 7.12750614e-01
3.38095576e-01 1.72970608e-01 -3.30001533e-01 -5.61003126e-02
6.92324996e-01 1.40297219e-01 -7.82530233e-02 -8.96596789e-01
-1.24288952e+00 7.27563560e-01 5.67727745e-01 3.48356143e-02
-3.34266186e-01 1.74600035e-01 3.56823593e-01 9.65854973e-02
4.69689786e-01 4.86814976e-01 -5.73833048e-01 1.71020940e-01
-1.03722835e+00 2.77878582e-01 7.59835184e-01 4.91540074e-01
6.08239293e-01 -2.13755071e-01 3.97368699e-01 7.52563655e-01
4.69675064e-01 6.49937749e-01 3.46265644e-01 -7.59775460e-01
6.54281229e-02 6.22900605e-01 2.84117945e-02 -1.01455212e+00
-6.63976669e-01 -3.39907408e-01 -7.11471558e-01 4.15327251e-01
8.39089632e-01 -2.48148173e-01 -1.00751460e+00 1.71535242e+00
5.55867851e-01 -1.29730880e-01 -8.32046494e-02 9.01118398e-01
6.85029268e-01 4.93958205e-01 2.26839662e-01 6.04573376e-02
1.37151742e+00 -5.58914900e-01 -1.27390057e-01 -5.37309527e-01
6.97461307e-01 -6.21703148e-01 3.98767680e-01 3.15591574e-01
-7.72176266e-01 -2.49746844e-01 -1.02599359e+00 2.35442385e-01
-4.57525045e-01 2.56817341e-01 6.55474424e-01 4.37807590e-01
-6.78281724e-01 7.08495319e-01 -1.10189819e+00 -8.03329647e-01
5.29246807e-01 5.81723213e-01 -7.90786684e-01 1.56583622e-01
-6.30700290e-01 1.10250425e+00 4.12025154e-01 4.44729835e-01
-8.51106465e-01 -1.99866563e-01 -9.85966206e-01 -3.17784399e-01
3.50670993e-01 -4.42786217e-01 1.00888407e+00 -9.62141991e-01
-1.08814919e+00 1.32223451e+00 3.70962679e-01 -6.51581585e-01
8.49330902e-01 -7.31036216e-02 5.47621772e-02 2.98441321e-01
-1.07256725e-01 1.06484437e+00 7.46239603e-01 -1.11764848e+00
-8.46540093e-01 -6.53857708e-01 -8.43668655e-02 2.52278656e-01
-4.55310903e-02 -9.62131768e-02 -4.54782873e-01 -5.46116829e-01
3.61688137e-01 -1.21668637e+00 -3.29082489e-01 4.08015400e-01
-1.14464406e-02 5.07181771e-02 6.69089198e-01 -7.49103546e-01
6.94630444e-01 -2.28310895e+00 3.55190247e-01 -3.36521715e-02
2.07054570e-01 1.82735115e-01 -1.80638865e-01 2.69243956e-01
1.49913505e-01 -2.20349550e-01 -4.13452148e-01 -7.54981413e-02
-2.57151932e-01 4.20823783e-01 2.62715697e-01 1.07669675e+00
2.12440699e-01 7.40957201e-01 -8.68466973e-01 -8.07021856e-01
3.03285122e-01 2.82431751e-01 -6.92862570e-01 1.74641252e-01
-6.14238232e-02 5.81680894e-01 -2.56656706e-01 5.01059234e-01
5.53556442e-01 -2.49454775e-03 3.39925528e-01 -2.28546467e-02
-8.88819098e-02 4.19450924e-02 -1.13047647e+00 1.44428790e+00
-2.92714804e-01 4.69379872e-01 3.03293705e-01 -1.11804116e+00
5.71084917e-01 1.10300764e-01 6.95217550e-01 -2.90969074e-01
2.51146704e-01 2.84450769e-01 2.85446107e-01 -4.25386518e-01
8.30087140e-02 -2.39338264e-01 -1.50111899e-01 3.07289869e-01
1.34371072e-01 -2.81533837e-01 4.42113489e-01 -1.46896273e-01
9.34691072e-01 4.02101278e-01 3.27459633e-01 -3.80223513e-01
2.77318776e-01 1.37519434e-01 2.99878359e-01 3.80940497e-01
-2.22703293e-01 5.74850202e-01 2.64497519e-01 -5.75732887e-01
-9.96936679e-01 -9.61592138e-01 -2.64687777e-01 1.23181629e+00
1.27116784e-01 4.86405380e-02 -8.34116876e-01 -6.96081102e-01
1.82609379e-01 5.30453250e-02 -7.02739418e-01 -1.19483761e-01
-5.82484365e-01 -9.73562062e-01 5.32917261e-01 5.09367049e-01
4.29478049e-01 -1.21822858e+00 -1.30771041e+00 1.13669463e-01
-1.76845379e-02 -9.62946117e-01 -7.90078193e-02 6.66539490e-01
-8.13619077e-01 -9.60133791e-01 -7.74368525e-01 -9.49001670e-01
7.73101747e-01 -4.53143660e-03 9.05003428e-01 2.52497703e-01
-5.95569909e-01 -2.14407649e-02 -2.33788311e-01 -3.71740520e-01
-2.98448801e-01 7.07664564e-02 1.61683202e-01 -3.31052393e-01
2.19649985e-01 -3.67672503e-01 -6.47976518e-01 5.31565666e-01
-8.40937018e-01 -5.09948470e-02 7.89679050e-01 7.42375255e-01
4.41023737e-01 -1.72074977e-02 3.15871328e-01 -7.53423631e-01
-1.60615161e-01 -2.63811260e-01 -7.12015867e-01 7.31259957e-02
-9.41629261e-02 8.02902579e-02 3.37029248e-01 -5.58185637e-01
-5.86711049e-01 5.23409486e-01 -1.04906105e-01 2.11779904e-02
-4.91818905e-01 3.98320347e-01 -1.61249459e-01 -1.86149031e-01
5.09592116e-01 -1.20286264e-01 2.20457628e-01 -5.26164234e-01
2.45189130e-01 6.40232623e-01 6.86047018e-01 -2.90961921e-01
6.09617352e-01 5.61767340e-01 2.12050915e-01 -1.17634535e+00
-5.04825056e-01 -5.62198818e-01 -8.24325323e-01 -3.72069120e-01
1.06493127e+00 -7.95810580e-01 -7.27376103e-01 5.84299266e-01
-8.01059604e-01 -2.64523834e-01 -1.25008464e-01 7.20115423e-01
-5.66591918e-01 5.52965581e-01 -5.13762057e-01 -7.44986832e-01
-1.27397075e-01 -9.71515536e-01 1.03198850e+00 -3.41800340e-02
-3.63429755e-01 -7.07270443e-01 -4.33155894e-02 4.62230325e-01
2.84914840e-02 4.17956412e-01 7.72830367e-01 -8.11098456e-01
-2.81100482e-01 -5.41102350e-01 4.81003486e-02 1.96402937e-01
4.70539145e-02 9.03952308e-03 -6.78113282e-01 -3.28690827e-01
-1.10632159e-01 -5.49136937e-01 7.25144446e-01 4.35580671e-01
7.14089453e-01 2.18556188e-02 -4.69742566e-01 4.64319438e-01
1.06903625e+00 -1.06153391e-01 2.93526292e-01 3.83869141e-01
4.87006396e-01 1.10096884e+00 5.61698854e-01 3.10327768e-01
2.40698040e-01 6.93446159e-01 6.28169715e-01 -2.90227383e-01
-1.07817706e-02 -1.51747078e-01 1.47381142e-01 4.31021750e-01
-3.15862149e-01 -1.47828147e-01 -1.04340219e+00 5.80691040e-01
-1.63504767e+00 -8.33216608e-01 -2.63396502e-01 2.36673069e+00
5.73193908e-01 1.39461383e-01 4.31736648e-01 1.97803956e-02
6.81797266e-01 -2.32311308e-01 -4.29184139e-01 -1.49568155e-01
9.35090482e-02 -2.26199590e-02 8.50883245e-01 -8.28400925e-02
-1.38156676e+00 6.57164991e-01 6.24181986e+00 5.58336318e-01
-1.10225785e+00 -4.24880892e-01 5.55259645e-01 -1.32241705e-02
4.53834802e-01 -2.54830450e-01 -7.39318013e-01 4.30496246e-01
5.79822242e-01 3.14021796e-01 2.56940722e-01 7.74935782e-01
-3.00651100e-02 -4.69817758e-01 -1.22182572e+00 8.65041256e-01
7.63072968e-02 -5.28695762e-01 -3.35761279e-01 1.59334868e-01
4.48724687e-01 3.55441757e-02 -2.20165789e-01 4.25928794e-02
1.57105327e-01 -9.80390549e-01 7.13174582e-01 1.35276109e-01
3.94647628e-01 -7.42987692e-01 9.41711068e-01 7.32364058e-01
-1.01649392e+00 6.65731803e-02 -3.71339262e-01 -1.45400658e-01
-1.24185875e-01 1.01206183e-01 -7.76845098e-01 3.41297895e-01
8.52672935e-01 4.46207583e-01 -7.75817871e-01 1.36363864e+00
-1.76767588e-01 2.41152793e-01 -8.00853908e-01 -1.92249715e-02
1.74413994e-01 -2.98646152e-01 1.76654294e-01 9.38075364e-01
2.32855245e-01 -7.92732090e-02 3.41374934e-01 3.81862760e-01
-7.70387286e-03 2.18480259e-01 -3.92708123e-01 -1.79939762e-01
2.26498898e-02 1.32745290e+00 -1.20303416e+00 -8.94748978e-03
-4.83684182e-01 9.73340392e-01 2.44272605e-01 -1.20819263e-01
-7.70231545e-01 -6.20460175e-02 3.36808681e-01 2.78559804e-01
5.38559318e-01 -1.67537868e-01 -2.53424998e-02 -7.37744153e-01
-1.18898034e-01 -7.98233092e-01 5.06700277e-01 -3.63720477e-01
-1.18226612e+00 4.25419062e-01 2.34555736e-01 -1.32752013e+00
-4.15097415e-01 -6.33660257e-01 2.82478388e-02 6.01271391e-01
-1.19012034e+00 -1.22871971e+00 -1.16331674e-01 1.28810983e-02
2.72952735e-01 2.16751784e-01 7.10927308e-01 3.33752602e-01
-1.99115261e-01 1.85729980e-01 1.87540814e-01 2.68272281e-01
6.65616751e-01 -1.03123069e+00 1.61572218e-01 6.15991056e-01
2.70925224e-01 3.60796392e-01 9.69008625e-01 -6.90479279e-01
-1.13165271e+00 -8.22868049e-01 7.11456537e-01 -4.42699879e-01
6.70882225e-01 -4.52610046e-01 -6.68927133e-01 5.33782244e-01
-2.04892129e-01 1.56333432e-01 6.02257788e-01 -1.21262267e-01
7.55142719e-02 6.99127316e-02 -1.13341868e+00 4.29284662e-01
8.69249582e-01 -5.09024896e-02 -7.44068563e-01 1.47864431e-01
-7.39561617e-02 -2.96050936e-01 -5.62177479e-01 6.91466570e-01
8.07233036e-01 -7.63482988e-01 9.50828850e-01 -4.90311176e-01
5.94052494e-01 -4.47208822e-01 1.41257092e-01 -1.15174294e+00
-4.32659313e-02 2.39436582e-01 2.57914364e-01 1.06514680e+00
2.98679799e-01 -2.76863426e-01 9.43126619e-01 3.69797170e-01
1.85013026e-01 -6.35076404e-01 -8.48153830e-01 -8.56169939e-01
1.13168187e-01 1.71507876e-02 1.20618433e-01 5.76445401e-01
-2.43518099e-01 3.30544293e-01 -4.82438684e-01 1.94319174e-01
6.12535715e-01 4.61541384e-01 9.49026704e-01 -1.49639690e+00
-4.28653747e-01 -2.44746327e-01 -9.15654421e-01 -9.02197003e-01
7.18265586e-03 -7.16952801e-01 4.36766416e-01 -1.45302260e+00
2.83231348e-01 -1.19501211e-01 6.58648089e-02 6.13744199e-01
2.50435527e-02 7.94531882e-01 6.43582270e-03 1.23498656e-01
-5.47230422e-01 1.30355686e-01 1.09762812e+00 -2.10545529e-02
4.02096622e-02 -8.07753280e-02 -7.04914182e-02 1.06842041e+00
5.74949145e-01 -5.29114783e-01 -4.24751565e-02 -3.36002380e-01
3.83952968e-02 -8.19965750e-02 5.55990160e-01 -1.13984442e+00
-1.99204590e-03 8.31753239e-02 6.94189966e-01 -6.03719175e-01
2.27935165e-01 -1.10725665e+00 3.85788143e-01 8.86540473e-01
-9.65826809e-02 -2.65698075e-01 1.18342429e-01 5.38164020e-01
5.69255240e-02 -3.33172858e-01 1.09961486e+00 -3.05408388e-01
-6.84575319e-01 1.05426505e-01 -5.88715374e-01 6.05580211e-02
1.17575037e+00 -2.01338425e-01 1.36075288e-01 -5.27215078e-02
-7.04569340e-01 2.50772357e-01 8.81830752e-01 1.75922692e-01
1.58080876e-01 -8.41400743e-01 -7.25872517e-01 3.13641101e-01
3.62777442e-01 5.50772883e-02 2.00342551e-01 9.69732404e-01
-1.08668804e+00 9.16572064e-02 -4.41695482e-01 -8.13264668e-01
-1.63223922e+00 6.16623342e-01 1.95023611e-01 -3.55144180e-02
-5.50815761e-01 8.09965074e-01 3.62939566e-01 -4.85665113e-01
2.09941462e-01 -1.95136160e-01 -3.76585245e-01 3.09345365e-01
2.07341835e-01 2.94287771e-01 1.31777421e-01 -9.46438193e-01
-5.22577941e-01 6.60190642e-01 2.85209157e-02 1.92089096e-01
1.57785773e+00 -7.98574090e-02 7.90622607e-02 2.47543529e-01
9.54522610e-01 -6.38002008e-02 -1.33903861e+00 2.20826447e-01
1.82286143e-01 -3.77880245e-01 -1.65409535e-01 -4.86113727e-01
-9.56919670e-01 8.48319173e-01 7.66301930e-01 1.65476695e-01
9.68981922e-01 2.17641115e-01 4.81104583e-01 3.60087216e-01
5.52427113e-01 -1.04409182e+00 -1.97828263e-01 2.18585297e-01
4.75561023e-01 -1.44544804e+00 3.32318634e-01 -2.54712760e-01
-4.93520439e-01 8.64243329e-01 4.13606942e-01 -1.37431040e-01
3.05814832e-01 3.54587615e-01 2.58662879e-01 -2.26547450e-01
-3.87029737e-01 -5.18749535e-01 2.10144386e-01 5.42655528e-01
4.32420015e-01 -7.38618895e-02 -5.35677910e-01 1.41205177e-01
-2.03540087e-01 -2.00137064e-01 1.46574989e-01 1.12596273e+00
-4.90708441e-01 -9.87160385e-01 -4.41628307e-01 4.44070816e-01
-6.45559013e-01 2.20313847e-01 -5.45305490e-01 9.10456836e-01
3.16565812e-01 6.77454948e-01 -1.51067805e-02 3.32297273e-02
2.80692548e-01 -1.25357769e-02 6.52839422e-01 -7.27678955e-01
-4.07283425e-01 3.38618189e-01 -3.12936269e-02 -2.44822681e-01
-5.72737515e-01 -7.03358650e-01 -1.19296122e+00 -2.00717121e-01
-4.06495005e-01 8.41937661e-02 6.54001176e-01 1.01771140e+00
-2.04818472e-01 1.85048252e-01 3.20959717e-01 -1.17015862e+00
-4.09482718e-01 -8.52728188e-01 -7.80036390e-01 6.47196889e-01
9.77789164e-02 -8.05127859e-01 -4.62930381e-01 3.45629603e-01] | [7.6866888999938965, -0.9599111080169678] |
fd5c0912-e678-4986-992f-46af9def4bcf | mt-cgcnn-integrating-crystal-graph | 1811.05660 | null | http://arxiv.org/abs/1811.05660v1 | http://arxiv.org/pdf/1811.05660v1.pdf | MT-CGCNN: Integrating Crystal Graph Convolutional Neural Network with Multitask Learning for Material Property Prediction | Developing accurate, transferable and computationally inexpensive machine
learning models can rapidly accelerate the discovery and development of new
materials. Some of the major challenges involved in developing such models are,
(i) limited availability of materials data as compared to other fields, (ii)
lack of universal descriptor of materials to predict its various properties.
The limited availability of materials data can be addressed through transfer
learning, while the generic representation was recently addressed by Xie and
Grossman [1], where they developed a crystal graph convolutional neural network
(CGCNN) that provides a unified representation of crystals. In this work, we
develop a new model (MT-CGCNN) by integrating CGCNN with transfer learning
based on multi-task (MT) learning. We demonstrate the effectiveness of MT-CGCNN
by simultaneous prediction of various material properties such as Formation
Energy, Band Gap and Fermi Energy for a wide range of inorganic crystals (46774
materials). MT-CGCNN is able to reduce the test error when employed on
correlated properties by upto 8%. The model prediction has lower test error
compared to CGCNN, even when the training data is reduced by 10%. We also
demonstrate our model's better performance through prediction of end user
scenario related to metal/non-metal classification. These results encourage
further development of machine learning approaches which leverage multi-task
learning to address the aforementioned challenges in the discovery of new
materials. We make MT-CGCNN's source code available to encourage reproducible
research. | ['Abhishek Kumar', 'Naganand Yadati', 'Soumya Sanyal', 'Partha Talukdar', 'Janakiraman Balachandran', 'Suchismita Sanyal', 'Padmini Rajagopalan'] | 2018-11-14 | null | null | null | null | ['formation-energy'] | ['miscellaneous'] | [ 1.85524598e-01 -3.50562394e-01 -2.29653135e-01 -3.96396220e-02
-9.97858465e-01 -2.22702846e-01 4.32485878e-01 2.61675835e-01
-1.07266620e-01 1.02134132e+00 -1.58794448e-01 -3.01443368e-01
-1.95022240e-01 -1.07084203e+00 -9.32153821e-01 -1.00216079e+00
1.15113623e-01 5.45814335e-01 3.05718064e-01 -2.64887899e-01
4.08555359e-01 4.74694759e-01 -1.56906784e+00 4.20373946e-01
1.00766420e+00 1.28387487e+00 6.77705467e-01 4.41401422e-01
1.43407196e-01 4.71080244e-01 -2.32961670e-01 -2.78932694e-02
7.62216300e-02 -1.26277685e-01 -7.65752137e-01 -4.67676312e-01
2.80891359e-01 1.44932464e-01 1.04186401e-01 7.31589317e-01
7.81469524e-01 -6.27465174e-02 1.21272564e+00 -1.00835919e+00
-1.02598047e+00 3.33060712e-01 -3.25368583e-01 -2.36676306e-01
2.50638396e-01 1.97693214e-01 1.14934206e+00 -8.01428556e-01
6.43687308e-01 7.82701552e-01 7.50791609e-01 5.81856310e-01
-1.21161473e+00 -1.01904213e+00 -4.37798679e-01 3.64145428e-01
-1.17899752e+00 -3.41590401e-03 7.57816672e-01 -4.84510422e-01
1.32657754e+00 9.01381936e-05 4.19533134e-01 1.16623998e+00
5.79282701e-01 3.53227615e-01 1.21098697e+00 -5.56528807e-01
1.59875378e-01 -1.03544667e-02 1.53035959e-02 7.19173431e-01
4.23076719e-01 1.07095271e-01 -6.41038656e-01 -8.50002170e-02
5.74195325e-01 -1.19205013e-01 2.55775988e-01 -2.55080968e-01
-8.34504068e-01 6.77923977e-01 6.34467483e-01 4.02708262e-01
-1.77025020e-01 1.76008925e-01 4.12015170e-01 3.59320283e-01
4.59741294e-01 6.13469124e-01 -6.39365792e-01 1.42289087e-01
-5.43689370e-01 3.94762963e-01 5.90584099e-01 1.04253483e+00
9.18799520e-01 1.85471922e-02 -8.86440685e-04 8.65802467e-01
2.20340401e-01 6.63599551e-01 3.68866622e-01 -3.06555659e-01
4.04818326e-01 5.70297062e-01 -1.02291085e-01 -4.87176597e-01
-7.02925622e-01 -4.01327908e-01 -7.59686708e-01 2.98321187e-01
1.46785066e-01 -9.40585416e-03 -8.43352258e-01 1.42592382e+00
6.07758388e-02 -3.32734108e-01 1.44196749e-01 5.48655689e-01
1.20714295e+00 7.97066391e-01 2.68208206e-01 1.73920199e-01
1.02310765e+00 -1.04138148e+00 -1.85990423e-01 1.48052335e-01
8.51441205e-01 -7.41266012e-01 9.37302113e-01 4.27141488e-01
-8.50635111e-01 -7.16691434e-01 -1.25154018e+00 7.11854128e-03
-7.59344697e-01 1.63492367e-01 1.10256076e+00 7.42784679e-01
-7.75271833e-01 9.93550241e-01 -7.13415742e-01 -3.65018904e-01
5.18718302e-01 1.01789665e+00 -3.51725101e-01 -1.55836061e-01
-9.32388008e-01 8.37589443e-01 4.72033918e-01 -3.36767852e-01
-8.87208819e-01 -9.52667654e-01 -4.19496983e-01 -2.05870628e-01
9.68385264e-02 -5.34089744e-01 9.07627285e-01 -6.46192253e-01
-1.66953051e+00 6.64598882e-01 2.46617213e-01 1.38367002e-03
1.41047239e-01 -6.67987317e-02 -6.56700611e-01 -1.84859380e-01
2.62239486e-01 4.46270823e-01 4.25497025e-01 -1.05792415e+00
-1.77601874e-01 -1.03438705e-01 -3.79016310e-01 -2.58237064e-01
-5.14453948e-01 -5.96698448e-02 -8.06933045e-02 -4.65631425e-01
-3.88119332e-02 -9.73115623e-01 -9.66006890e-03 -2.62268513e-01
-4.83238518e-01 -4.25490737e-01 7.61486053e-01 -3.12689781e-01
6.99384689e-01 -1.70011568e+00 1.62585482e-01 1.96221173e-01
2.36816153e-01 3.60636383e-01 -3.06822091e-01 7.29709446e-01
-1.67683899e-01 3.02048549e-02 -4.16610055e-02 1.02477215e-01
-2.95279711e-01 -2.53338963e-01 3.23040038e-01 3.83902818e-01
3.49625081e-01 1.17431533e+00 -7.19943762e-01 -1.10101715e-01
2.87209600e-01 3.78400356e-01 -5.84865570e-01 7.46415332e-02
-5.60256779e-01 6.62658513e-01 -6.65693402e-01 1.09743333e+00
8.69083285e-01 -6.24910772e-01 -4.80669923e-02 -4.60140467e-01
-4.04790878e-01 3.01690429e-01 -7.34722733e-01 1.49472761e+00
-5.10527492e-01 2.07845882e-01 -2.99377739e-01 -1.06938779e+00
1.22291458e+00 2.50396520e-01 7.83827960e-01 -1.07776177e+00
5.39368764e-02 5.69608212e-01 3.56736988e-01 -5.17852962e-01
7.25506693e-02 -3.50010335e-01 3.75041887e-02 5.43486953e-01
3.98097962e-01 -2.28095710e-01 5.50631434e-02 -2.59003192e-01
1.10883284e+00 3.79833639e-01 2.41654441e-02 -5.45633256e-01
5.46427429e-01 -1.60612538e-01 2.21111700e-01 7.04894483e-01
1.20793134e-01 3.76288176e-01 8.01386591e-03 -5.88047504e-01
-1.33790374e+00 -1.03777969e+00 -5.13120592e-01 1.19647193e+00
-1.06807880e-01 -5.68140686e-01 -3.49335551e-01 -2.78660357e-01
1.85934335e-01 2.63783723e-01 -4.85610962e-01 -3.71472090e-01
-4.09937769e-01 -1.34930503e+00 2.23152831e-01 5.55347502e-01
5.49225628e-01 -1.17231655e+00 -3.28784771e-02 3.86127383e-01
2.69873381e-01 -1.02697551e+00 8.75267312e-02 6.29618704e-01
-7.43967593e-01 -1.03130448e+00 -5.74145675e-01 -9.88541961e-01
4.17011827e-01 -9.22626536e-03 8.34769428e-01 2.04590131e-02
-5.63391984e-01 1.04952671e-01 -3.91348034e-01 -5.07769227e-01
-6.72914803e-01 5.53388834e-01 9.75815877e-02 -3.63568574e-01
2.96986371e-01 -7.76978910e-01 -5.68112016e-01 2.47413978e-01
-5.44386625e-01 2.95733273e-01 9.60228741e-01 7.06327856e-01
6.39564037e-01 -6.89434111e-02 1.02825606e+00 -1.19760215e+00
4.90751266e-01 -5.46493173e-01 -5.13134897e-01 4.18216318e-01
-1.07911742e+00 4.50780630e-01 8.49712372e-01 -3.27842325e-01
-8.77152085e-01 6.63902760e-02 -3.40416163e-01 1.21449292e-01
8.64518359e-02 4.20601994e-01 -9.37318206e-02 -6.59234345e-01
6.36134744e-01 9.01905745e-02 -3.53808939e-01 -4.99687105e-01
5.39482199e-02 8.40078235e-01 1.33224681e-01 -1.08025980e+00
7.51260936e-01 6.24251494e-04 6.30499244e-01 -7.73295581e-01
-7.21113861e-01 -3.39970648e-01 -7.73114800e-01 -2.10093379e-01
9.45114434e-01 -9.97098446e-01 -1.02154326e+00 5.88097394e-01
-8.61857116e-01 -3.00405473e-01 3.81153345e-01 4.20627266e-01
-5.56557834e-01 1.01174407e-01 -7.02996731e-01 -5.33381939e-01
-8.24104488e-01 -1.15882540e+00 1.02895844e+00 6.37810752e-02
-8.22423771e-02 -1.17886508e+00 1.78732048e-03 6.21750236e-01
7.11849213e-01 4.66983110e-01 1.57207119e+00 -5.89701474e-01
-7.52334774e-01 -1.94337428e-01 -2.72815585e-01 1.89189345e-01
6.26287758e-01 -9.95157436e-02 -9.76660609e-01 -4.11274493e-01
-3.14674705e-01 -7.12569594e-01 9.50010896e-01 4.74500477e-01
1.31636572e+00 3.03700894e-01 -4.23098147e-01 5.85136771e-01
1.74590683e+00 3.34262550e-01 5.70078254e-01 5.46534717e-01
1.01686656e+00 1.35967717e-01 1.32317483e-01 2.79467881e-01
1.23594113e-01 6.39820755e-01 3.70209783e-01 -8.50600824e-02
-2.59580463e-01 -9.32691693e-02 2.39378214e-01 1.04773390e+00
-7.91492939e-01 -1.79640576e-01 -1.01782382e+00 -2.92477570e-02
-1.54270470e+00 -5.62052310e-01 -3.16054165e-01 2.10228133e+00
6.11744463e-01 2.73657143e-01 1.05527692e-01 -6.27121925e-02
5.35119176e-01 -1.43449843e-01 -9.13499832e-01 -2.67718494e-01
-3.08451325e-01 8.11998546e-01 6.24564350e-01 8.80364254e-02
-1.01235116e+00 9.83986259e-01 6.62811327e+00 8.74040961e-01
-1.42393005e+00 1.16240904e-01 5.66118002e-01 2.07602922e-02
-2.30877593e-01 -1.45944059e-01 -9.71056581e-01 4.19999003e-01
1.14869297e+00 -2.00403631e-02 2.74251044e-01 7.39791691e-01
-1.34854168e-01 9.08734575e-02 -1.22842991e+00 9.77441132e-01
4.14563790e-02 -1.70036709e+00 -1.46823227e-02 6.57971352e-02
9.35827017e-01 3.05273175e-01 2.18943551e-01 3.20331603e-01
3.83376889e-02 -1.17694890e+00 4.91152704e-01 3.45065206e-01
9.78733540e-01 -7.29669333e-01 4.62370723e-01 -8.39966163e-02
-1.23245358e+00 3.25209796e-02 -5.15852809e-01 -3.89169864e-02
-3.90447229e-01 4.99918967e-01 -8.31457436e-01 8.93096745e-01
6.29561186e-01 9.00561571e-01 -6.04044795e-01 8.41945350e-01
1.85718879e-01 5.39622903e-01 -9.47223827e-02 -5.00135601e-01
3.21871676e-02 -5.30054867e-01 5.00183692e-03 9.70239162e-01
5.13482690e-01 -2.57166475e-01 2.05535010e-01 1.10468626e+00
-3.98899615e-01 3.13784212e-01 -3.50410819e-01 -6.27573058e-02
2.19589114e-01 1.22006345e+00 -7.31969774e-01 7.84763619e-02
-7.43581533e-01 6.80149555e-01 4.11308587e-01 4.61876057e-02
-6.21934652e-01 -8.47263932e-02 1.61427647e-01 3.08099836e-01
3.48422378e-01 -8.59090909e-02 -2.34717324e-01 -6.61397099e-01
-4.54176776e-02 -4.69427615e-01 4.74617742e-02 -7.58737326e-01
-1.59392810e+00 4.64024454e-01 -2.34403774e-01 -1.02249563e+00
1.16920270e-01 -1.30810487e+00 -6.11274362e-01 7.98597097e-01
-1.54484165e+00 -1.52093053e+00 -1.12762205e-01 2.87285775e-01
1.69323981e-01 -5.78865230e-01 1.01576281e+00 4.28406596e-01
-5.93068719e-01 4.37379837e-01 6.59038663e-01 -2.60928839e-01
8.70930016e-01 -1.19464767e+00 2.13332474e-01 9.22889486e-02
-3.30282718e-01 1.42602459e-01 5.68374693e-01 -6.96607590e-01
-1.61534142e+00 -1.25405502e+00 4.15786803e-01 -4.53798234e-01
6.83017135e-01 -7.46339977e-01 -7.56275713e-01 2.89618999e-01
-5.24686612e-02 5.25368638e-02 1.05101442e+00 3.67513150e-01
-3.16606820e-01 -2.67006874e-01 -9.60795462e-01 2.61462361e-01
9.00887311e-01 -6.11898005e-01 6.02246374e-02 9.96488512e-01
6.77353621e-01 -1.51085719e-01 -1.28393269e+00 5.04256308e-01
5.96129000e-01 -8.54069054e-01 9.06195045e-01 -4.51982737e-01
7.06101656e-01 1.22685254e-01 -2.60602564e-01 -1.03674257e+00
-4.61781859e-01 -2.94333279e-01 4.63981032e-02 1.08698308e+00
8.20833564e-01 -4.62992698e-01 1.02955675e+00 5.56555331e-01
-3.32989663e-01 -1.04170561e+00 -9.49278712e-01 -1.03193355e+00
5.98142803e-01 -3.40940326e-01 5.93068421e-01 5.08159161e-01
-2.24468887e-01 4.11117941e-01 -5.49761117e-01 -1.34363428e-01
3.74086320e-01 4.15680140e-01 3.94766271e-01 -1.67956507e+00
-3.09056491e-01 -9.64857265e-02 -3.52713734e-01 -4.18558359e-01
3.15915346e-01 -1.40608263e+00 -3.85662913e-01 -1.42005169e+00
5.52320957e-01 -8.31630826e-01 -6.76988304e-01 4.68351483e-01
2.38358647e-01 6.14267290e-02 -1.18782304e-01 2.23825723e-01
-6.10888362e-01 6.61620498e-01 1.51202667e+00 -2.88862526e-01
-1.48900419e-01 1.17590137e-01 -7.18706310e-01 1.15233764e-01
7.91239798e-01 -4.97413099e-01 -2.75859404e-02 -1.44846234e-02
3.34297419e-01 -1.27670556e-01 1.34339124e-01 -1.57161772e+00
9.31474492e-02 4.21931669e-02 6.52376950e-01 -4.85659927e-01
3.27510715e-01 -6.24914408e-01 3.22872549e-01 6.93895102e-01
4.68192399e-02 -1.63176060e-01 2.13770553e-01 3.98976505e-01
1.19150847e-01 -3.21453780e-01 6.36118054e-01 -2.43739083e-01
-4.28175628e-01 5.12039781e-01 -2.45697096e-01 -5.45001388e-01
1.00180936e+00 -9.42537114e-02 -3.15853655e-01 9.88530070e-02
-4.62120205e-01 -1.74740881e-01 3.71763885e-01 2.44375274e-01
3.86450738e-01 -1.34707785e+00 -5.68894744e-01 3.56320411e-01
3.60627383e-01 -1.26179606e-01 1.72547683e-01 3.99509817e-01
-6.71562672e-01 7.75011539e-01 -5.77769935e-01 -4.61278021e-01
-1.10201240e+00 6.14030600e-01 2.49348015e-01 -3.30330878e-01
-3.78119916e-01 5.44512510e-01 -9.53939976e-04 -5.63630641e-01
-3.07772011e-01 -3.20452422e-01 1.11467624e-02 -2.95754939e-01
-2.91946139e-02 3.43521029e-01 5.96752763e-01 -4.43847001e-01
-3.01319033e-01 8.28562856e-01 -5.18597841e-01 6.04135394e-01
2.12724018e+00 4.75753844e-01 -2.27873564e-01 4.13587987e-01
1.33180130e+00 -2.27146074e-01 -1.05300963e+00 -1.77584976e-01
1.07464463e-01 1.75936997e-01 1.21422838e-02 -8.77404034e-01
-9.75933969e-01 6.81172788e-01 1.08159339e+00 -2.48406202e-01
9.53422189e-01 2.43886814e-01 8.37016761e-01 6.93856835e-01
4.95409697e-01 -1.21669769e+00 4.06691998e-01 3.96243125e-01
7.13459909e-01 -1.60551894e+00 2.75678486e-01 -6.56153619e-01
-1.28371090e-01 1.41452837e+00 8.39975297e-01 5.37703633e-02
8.01216424e-01 2.10483104e-01 -3.62558246e-01 -5.83672583e-01
-6.33893311e-01 3.10136601e-02 3.19611251e-01 7.81854033e-01
7.91569173e-01 1.91198885e-01 -3.86189371e-02 6.20334029e-01
-1.51161164e-01 -6.53257146e-02 1.41996041e-01 9.59593952e-01
-4.39209968e-01 -1.75829995e+00 -1.47136807e-01 7.65694797e-01
-2.82504648e-01 -2.64930546e-01 -2.29144424e-01 5.22039652e-01
3.58253598e-01 8.04208279e-01 -2.77405024e-01 -4.98486936e-01
7.38132149e-02 1.24083878e-02 8.02270830e-01 -6.58194363e-01
-5.27650297e-01 -1.24288969e-01 -1.07689977e-01 -1.04780920e-01
-6.17411673e-01 -5.18425465e-01 -1.14508152e+00 -2.58954406e-01
-5.64535677e-01 -3.05277016e-02 7.25793004e-01 1.09914744e+00
3.24727237e-01 6.70281410e-01 5.63064396e-01 -9.76884127e-01
-7.33069927e-02 -9.85986531e-01 -7.24280775e-01 3.97423029e-01
3.00224964e-02 -8.20996940e-01 1.22643761e-01 -2.65546918e-01] | [5.242095470428467, 5.449158191680908] |
fe5d0f5b-4be2-4815-9fab-9225f5b21dc6 | moss-monocular-shape-sensing-for-continuum | 2303.00891 | null | https://arxiv.org/abs/2303.00891v2 | https://arxiv.org/pdf/2303.00891v2.pdf | MoSS: Monocular Shape Sensing for Continuum Robots | Continuum robots are promising candidates for interactive tasks in medical and industrial applications due to their unique shape, compliance, and miniaturization capability. Accurate and real-time shape sensing is essential for such tasks yet remains a challenge. Embedded shape sensing has high hardware complexity and cost, while vision-based methods require stereo setup and struggle to achieve real-time performance. This paper proposes the first eye-to-hand monocular approach to continuum robot shape sensing. Utilizing a deep encoder-decoder network, our method, MoSSNet, eliminates the computation cost of stereo matching and reduces requirements on sensing hardware. In particular, MoSSNet comprises an encoder and three parallel decoders to uncover spatial, length, and contour information from a single RGB image, and then obtains the 3D shape through curve fitting. A two-segment tendon-driven continuum robot is used for data collection and testing, demonstrating accurate (mean shape error of 0.91 mm, or 0.36% of robot length) and real-time (70 fps) shape sensing on real-world data. Additionally, the method is optimized end-to-end and does not require fiducial markers, manual segmentation, or camera calibration. Code and datasets will be made available at https://github.com/ContinuumRoboticsLab/MoSSNet. | ['Jessica Burgner-Kahrs', 'David B. Lindell', 'Puspita Triana Dewi', 'Chaojun Chen', 'Enxu Li', 'Chengnan Shentu'] | 2023-03-02 | null | null | null | null | ['camera-calibration', 'stereo-matching-1'] | ['computer-vision', 'computer-vision'] | [ 1.27889559e-01 -2.28148606e-02 2.08170973e-02 8.32161680e-03
-7.91569591e-01 -6.50519490e-01 -2.11549029e-01 7.36574605e-02
-3.76847476e-01 3.31096470e-01 -6.07865095e-01 -3.05164278e-01
1.72129467e-01 -4.10395503e-01 -8.35884333e-01 -4.71517146e-01
1.60468608e-01 6.28661513e-01 3.36079568e-01 -9.20908973e-02
2.23470122e-01 5.72834074e-01 -9.15258586e-01 -3.55946332e-01
7.07706273e-01 1.30619991e+00 6.91877961e-01 5.51229119e-01
5.24291933e-01 1.47363394e-01 6.01906981e-03 -3.36781889e-02
4.65333432e-01 7.81047642e-02 -2.95595020e-01 8.37770626e-02
1.78985879e-01 -6.80755913e-01 -3.48249555e-01 8.76431108e-01
7.84118056e-01 -4.77663994e-01 3.61327320e-01 -9.14797068e-01
-5.31705678e-01 3.67454216e-02 -6.96096539e-01 -6.68163836e-01
4.54637945e-01 2.40352288e-01 4.57260758e-01 -9.90658522e-01
7.93795586e-01 7.10995734e-01 1.06437230e+00 6.27592921e-01
-1.30518639e+00 -5.28521240e-01 -7.03122556e-01 -4.43911791e-01
-1.42151892e+00 -4.12491709e-01 9.41807687e-01 -5.34738183e-01
6.23466194e-01 -8.18013772e-02 8.58387828e-01 7.32868314e-01
7.32512593e-01 3.56767356e-01 7.70628512e-01 -2.12901682e-01
2.00043559e-01 -7.83779174e-02 -5.55358648e-01 8.96924317e-01
4.64086264e-01 3.31915349e-01 -4.93114814e-02 1.18937880e-01
1.83414900e+00 1.57051504e-01 -2.97695994e-01 -1.01751161e+00
-1.35350776e+00 3.79336029e-01 4.89209443e-01 -2.22968068e-02
-3.30222338e-01 6.72833860e-01 4.77767326e-02 -5.20058498e-02
-1.44813105e-01 4.73449528e-01 -1.65262356e-01 -6.10688627e-01
-4.28470612e-01 -6.34794831e-02 6.85248375e-01 1.42918789e+00
4.14761752e-01 7.34380111e-02 6.48080707e-01 7.64470756e-01
4.68421161e-01 1.00133812e+00 3.57314944e-01 -1.39326191e+00
2.44256914e-01 4.95871603e-01 1.92176819e-01 -1.12729549e+00
-8.74765754e-01 -1.45882025e-01 -5.87108195e-01 5.12909770e-01
2.16620311e-01 -3.34349006e-01 -5.70171773e-01 1.18939650e+00
2.93319404e-01 -4.89985734e-01 -1.21050395e-01 1.39001560e+00
6.23725295e-01 1.43338308e-01 -5.45517802e-01 2.25728471e-02
1.14082456e+00 -6.32945299e-01 -3.14291805e-01 -3.41191888e-01
3.17781150e-01 -8.85527968e-01 1.02061880e+00 4.00037795e-01
-1.11987722e+00 -2.28297427e-01 -1.20368314e+00 -1.81976691e-01
3.13676745e-01 5.27056634e-01 4.78386760e-01 1.36193454e-01
-7.48141944e-01 5.19740641e-01 -1.20578921e+00 -1.54614896e-01
1.95276454e-01 6.71233535e-01 -2.11545080e-01 1.24780737e-01
-5.13003349e-01 7.02329457e-01 -1.21823050e-01 7.09737614e-02
-3.31630260e-01 -5.37551045e-01 -7.36607313e-01 -4.49275643e-01
3.44252735e-01 -6.53384507e-01 1.54505920e+00 -3.09985816e-01
-2.04768300e+00 8.53669047e-01 2.81827927e-01 -2.82210886e-01
5.21070957e-01 -3.22176069e-01 2.64580458e-01 5.53685606e-01
-7.09709302e-02 7.99728990e-01 6.00654125e-01 -1.19172251e+00
2.38772016e-02 -4.87130225e-01 -2.71574646e-01 1.70237377e-01
-3.05237826e-02 -4.17037547e-01 -3.88216078e-01 -5.22358418e-01
7.62263119e-01 -1.27410007e+00 -2.96721369e-01 9.16122258e-01
-1.83919519e-01 4.14293498e-01 8.49999905e-01 -4.84688073e-01
4.59936917e-01 -2.16471815e+00 -2.23347306e-01 1.22580327e-01
1.78708270e-01 -1.19800411e-01 1.25482872e-01 2.93092370e-01
5.42526901e-01 -4.18963462e-01 -3.16514164e-01 -1.36634717e-02
-2.11022094e-01 -1.68388151e-02 2.18034163e-01 7.91190803e-01
-2.22737849e-01 9.48570669e-01 -6.72725260e-01 -4.51862127e-01
2.90977597e-01 3.59576881e-01 -4.12959635e-01 -4.09845077e-02
-8.75794068e-02 6.48091793e-01 -2.94367790e-01 1.10058224e+00
4.99128014e-01 -2.63711303e-01 1.87763020e-01 -3.54314446e-01
-2.85665154e-01 2.94826250e-03 -1.14820385e+00 2.13713694e+00
-5.15811265e-01 6.23666704e-01 6.85432613e-01 -5.14780939e-01
1.42160273e+00 2.70954460e-01 1.01198494e+00 -9.78446066e-01
5.32356322e-01 7.75646329e-01 -1.78499877e-01 -4.61564213e-01
4.61257100e-01 -1.05812050e-01 -3.42996329e-01 2.54260480e-01
-5.35769939e-01 -8.98423791e-01 -2.56884187e-01 -3.86329740e-01
8.92196476e-01 5.43823421e-01 2.24597573e-01 -8.75093117e-02
-5.95608056e-02 4.68041986e-01 5.86653829e-01 -6.60778163e-03
-1.45491764e-01 7.84197867e-01 -1.11303225e-01 -1.03161462e-01
-1.51106143e+00 -1.22692347e+00 -2.51590282e-01 3.79170269e-01
9.11042988e-01 9.64251086e-02 -6.82368517e-01 3.21823001e-01
4.90941793e-01 -1.61342788e-02 4.04031947e-02 -3.77631299e-02
-8.08691442e-01 2.17172995e-01 2.86242455e-01 8.29304099e-01
3.27312469e-01 -7.37376451e-01 -1.46525621e+00 5.77864945e-01
6.28094450e-02 -1.34650517e+00 -6.26757145e-01 -4.42835651e-02
-1.15561545e+00 -1.15503705e+00 -5.90472400e-01 -9.35480654e-01
6.58907294e-01 3.32691133e-01 5.38773477e-01 -3.54779720e-01
-8.35389435e-01 5.89878559e-01 -8.43807012e-02 -3.48303050e-01
-1.62005469e-01 -1.55584723e-01 1.38105914e-01 -6.93003416e-01
-2.64009744e-01 -5.99946201e-01 -8.65684390e-01 6.96543217e-01
-2.82765985e-01 3.16643298e-01 8.56834412e-01 6.60413086e-01
1.07025564e+00 -6.51286662e-01 3.61712903e-01 -1.50861293e-01
4.60514903e-01 6.92634657e-02 -1.04700518e+00 -4.28431898e-01
-5.63130498e-01 -4.77309227e-01 5.79239726e-01 -4.65139508e-01
-4.97821689e-01 7.84256160e-01 7.86175281e-02 -7.17897475e-01
1.99736997e-01 4.37497646e-01 3.73049468e-01 -3.23575914e-01
7.44107008e-01 1.83888048e-01 8.16051781e-01 -2.31110603e-01
1.10433973e-01 9.39264357e-01 1.05232811e+00 -4.97125000e-01
4.51842695e-01 6.95999980e-01 3.88815477e-02 -9.76586580e-01
1.25050440e-01 -3.95303577e-01 -7.74419844e-01 -4.27010715e-01
5.73571622e-01 -9.18112814e-01 -1.17459071e+00 5.52051306e-01
-1.05137932e+00 -5.54631412e-01 -2.24594519e-01 7.01165020e-01
-1.17944407e+00 2.82144427e-01 -8.41998637e-01 -6.67958140e-01
-7.60149658e-01 -1.21758890e+00 1.18494415e+00 2.70261824e-01
-4.84830230e-01 -5.09413481e-01 -4.11568880e-01 2.79606551e-01
4.01868641e-01 5.94116151e-01 3.62406790e-01 4.57443118e-01
-7.38834023e-01 -4.83126760e-01 -1.52140826e-01 -2.09461376e-01
3.22081238e-01 -1.31771758e-01 -5.94362795e-01 -4.86199886e-01
-6.53891265e-02 -5.41764140e-01 9.06330869e-02 6.12701118e-01
9.62948322e-01 -2.76336260e-02 -4.61893320e-01 5.27060449e-01
1.48799610e+00 4.53127861e-01 2.60921180e-01 1.07961453e-01
6.95980132e-01 1.69437453e-01 8.89887154e-01 4.62002605e-01
3.51629883e-01 9.39873874e-01 5.66540420e-01 -1.11160226e-01
-1.25949532e-01 -1.56260625e-01 3.01846653e-01 8.82297933e-01
-1.15298142e-03 4.47468489e-01 -9.60053325e-01 3.50840449e-01
-1.51695776e+00 -4.79465991e-01 -2.42640227e-01 2.31160045e+00
9.73764777e-01 4.04132195e-02 2.12889791e-01 -2.62987968e-02
5.25619745e-01 -7.73326695e-01 -1.03286827e+00 -4.11734462e-01
2.96052516e-01 -1.63375363e-01 9.19557035e-01 3.40361923e-01
-9.14466381e-01 7.37209976e-01 5.11984348e+00 3.65839213e-01
-1.63017082e+00 -9.63048413e-02 -6.10696385e-03 -9.68774483e-02
-1.56769887e-01 -3.46799046e-01 -3.14384967e-01 2.50391006e-01
3.40202570e-01 2.94980733e-03 4.31088954e-01 9.72252846e-01
2.25131288e-01 -3.25978577e-01 -1.05561244e+00 1.28800893e+00
-2.10216865e-01 -1.23324072e+00 -7.77426600e-01 1.73654437e-01
3.19396377e-01 1.91704690e-01 2.98406892e-02 -5.02011836e-01
-5.21492865e-03 -7.09100366e-01 1.17320764e+00 4.19159055e-01
1.51009285e+00 -3.91537309e-01 3.01025271e-01 5.30854523e-01
-1.30399466e+00 -6.66935742e-02 -5.68719804e-01 -9.48205777e-03
2.85722941e-01 6.29682481e-01 -1.11096466e+00 1.66388258e-01
4.82653052e-01 5.18129885e-01 8.21526647e-02 9.63961065e-01
1.41951993e-01 4.39773761e-02 -6.16962731e-01 -4.05072540e-01
-2.18347773e-01 -2.70326644e-01 7.02700198e-01 8.16255033e-01
4.44969296e-01 1.71764359e-01 3.62581760e-01 1.19239938e+00
8.88553560e-02 -1.47323444e-01 -6.73307240e-01 -4.04651240e-02
8.08291316e-01 1.26017082e+00 -6.99194610e-01 3.61775845e-01
-2.16702744e-01 1.04577649e+00 -3.10918707e-02 -7.92345479e-02
-8.35243881e-01 -7.94295311e-01 4.91637737e-01 3.91558737e-01
1.13243766e-01 -9.70392764e-01 -9.76496875e-01 -6.32328033e-01
3.45429212e-01 -2.01766938e-01 -3.51774693e-01 -8.42724621e-01
-7.37738967e-01 1.81143224e-01 -5.54662049e-01 -1.75329876e+00
-1.49070948e-01 -7.81949282e-01 2.09338535e-02 4.90601033e-01
-1.08727074e+00 -1.08649766e+00 -5.30626595e-01 2.75378436e-01
4.52613115e-01 1.80946723e-01 8.65357459e-01 8.87041092e-02
-1.79314867e-01 4.61811393e-01 2.12779462e-01 2.71727830e-01
7.13050485e-01 -9.11751688e-01 2.23087788e-01 3.47079843e-01
-4.37352657e-01 5.16020536e-01 3.66195679e-01 -8.08841109e-01
-2.27752900e+00 -8.99825275e-01 1.66118190e-01 -4.21719924e-02
3.62444997e-01 -2.90822566e-01 -5.10622740e-01 3.62903357e-01
-3.83172661e-01 1.80485472e-01 3.07021111e-01 -6.18487895e-01
1.16796508e-01 -2.52357632e-01 -1.43905008e+00 5.75271785e-01
1.18659043e+00 -3.23289484e-01 -2.40298524e-01 3.06090806e-02
5.54440022e-01 -1.29132438e+00 -1.14684987e+00 3.01865369e-01
1.15924513e+00 -5.87092936e-01 8.55277717e-01 5.46927214e-01
5.22816300e-01 -2.92495102e-01 -7.99464956e-02 -9.89368618e-01
-3.08196783e-01 -7.24771082e-01 4.34335507e-02 5.98136067e-01
4.30170834e-01 -8.31313312e-01 1.01790369e+00 6.16811693e-01
-5.30422390e-01 -1.33113635e+00 -1.09027743e+00 -1.01789939e+00
-7.81679302e-02 -2.42184713e-01 2.33893529e-01 5.21187484e-01
5.77341974e-01 1.07195996e-01 8.72685686e-02 1.64869264e-01
6.23132467e-01 7.06149340e-01 6.82107508e-01 -1.16106975e+00
-6.41433820e-02 -3.03693503e-01 -5.79264402e-01 -1.27319705e+00
-4.89549160e-01 -7.38826811e-01 4.13810700e-01 -1.60435975e+00
-1.61684424e-01 -8.15980196e-01 6.20907664e-01 5.12633145e-01
6.00938797e-01 4.29056793e-01 1.44814759e-01 5.37202954e-01
-1.51920021e-01 3.91094446e-01 1.56903815e+00 1.15705751e-01
-2.61469722e-01 -1.17032833e-01 -2.46978506e-01 6.69632614e-01
9.95192409e-01 4.01659450e-03 -1.60515770e-01 -5.80527544e-01
-3.55784111e-02 6.79678261e-01 5.46446979e-01 -1.27260494e+00
5.76067090e-01 -1.66575208e-01 4.13048208e-01 -4.14860338e-01
6.16886318e-01 -1.15851772e+00 1.46347255e-01 1.02536571e+00
1.61751717e-01 2.46424470e-02 1.99288681e-01 2.67155498e-01
2.05413908e-01 4.14267853e-02 9.66878891e-01 -6.06653839e-02
-4.13907945e-01 2.71055192e-01 -3.76719236e-01 -1.92593813e-01
1.29116666e+00 -1.01645648e+00 -6.91467430e-03 -2.51058429e-01
-6.11457527e-01 1.42520010e-01 1.07898843e+00 1.44773692e-01
1.14605999e+00 -1.36703670e+00 -2.93540418e-01 2.86827862e-01
-7.25884661e-02 3.68850678e-01 4.87167723e-02 1.01653755e+00
-1.10583270e+00 4.56767768e-01 -2.64860868e-01 -1.13535273e+00
-1.05554342e+00 1.63227558e-01 5.64229846e-01 5.83585203e-01
-8.59172344e-01 5.86652398e-01 -3.47300738e-01 -6.47724330e-01
-9.71169993e-02 -5.16440451e-01 5.34244180e-01 -6.64951086e-01
-2.34881923e-01 6.69613302e-01 -2.85716299e-02 -2.43924692e-01
-3.26855332e-01 1.24569142e+00 5.40727019e-01 -1.97829589e-01
1.08850491e+00 -3.15685332e-01 1.49222091e-01 4.13033783e-01
1.03921854e+00 -8.75367597e-02 -1.68759549e+00 5.41978050e-03
-2.66644180e-01 -3.58651847e-01 -2.11679682e-01 -5.94335198e-01
-1.00413465e+00 8.62361193e-01 5.94337523e-01 -1.87882096e-01
8.45078945e-01 2.43250392e-02 1.31681085e+00 2.54814923e-01
8.81723166e-01 -1.33978701e+00 1.02537207e-01 3.54346812e-01
1.18627465e+00 -9.63265479e-01 7.55117387e-02 -9.18718696e-01
-3.61113518e-01 1.48484766e+00 6.93789244e-01 -2.56786704e-01
4.70802307e-01 8.75945807e-01 2.38290548e-01 -2.32266769e-01
-1.67362273e-01 3.24581683e-01 -1.28402829e-01 6.01077318e-01
3.50434750e-01 4.66977358e-01 -1.99475259e-01 4.61671025e-01
-3.93919826e-01 3.02256852e-01 5.26803732e-01 1.28525412e+00
-6.80345058e-01 -6.42314851e-01 -3.47842306e-01 3.66441667e-01
-1.09942198e-01 5.12867391e-01 -1.14105269e-01 5.43482542e-01
-2.23770425e-01 6.58258557e-01 2.22270548e-01 -5.08519948e-01
6.21352434e-01 -5.60325682e-01 8.08828354e-01 -4.59326208e-01
-2.28956640e-02 4.60738420e-01 -5.48867919e-02 -8.95646155e-01
7.56101534e-02 -7.91027844e-01 -2.04749823e+00 -5.26465178e-02
-2.13533849e-01 -4.19525832e-01 1.08461928e+00 3.14221531e-01
7.49262869e-01 1.97961822e-01 6.63882732e-01 -1.15511644e+00
-7.47100294e-01 -7.09160447e-01 -5.34675479e-01 1.44113284e-02
2.04137713e-01 -7.91681945e-01 3.69680196e-01 1.24322161e-01] | [6.217565536499023, -1.0098737478256226] |
36f76b5d-6056-467b-a3a8-035b582c3a03 | revealing-weaknesses-of-vietnamese-language | 2303.13355 | null | https://arxiv.org/abs/2303.13355v1 | https://arxiv.org/pdf/2303.13355v1.pdf | Revealing Weaknesses of Vietnamese Language Models Through Unanswerable Questions in Machine Reading Comprehension | Although the curse of multilinguality significantly restricts the language abilities of multilingual models in monolingual settings, researchers now still have to rely on multilingual models to develop state-of-the-art systems in Vietnamese Machine Reading Comprehension. This difficulty in researching is because of the limited number of high-quality works in developing Vietnamese language models. In order to encourage more work in this research field, we present a comprehensive analysis of language weaknesses and strengths of current Vietnamese monolingual models using the downstream task of Machine Reading Comprehension. From the analysis results, we suggest new directions for developing Vietnamese language models. Besides this main contribution, we also successfully reveal the existence of artifacts in Vietnamese Machine Reading Comprehension benchmarks and suggest an urgent need for new high-quality benchmarks to track the progress of Vietnamese Machine Reading Comprehension. Moreover, we also introduced a minor but valuable modification to the process of annotating unanswerable questions for Machine Reading Comprehension from previous work. Our proposed modification helps improve the quality of unanswerable questions to a higher level of difficulty for Machine Reading Comprehension systems to solve. | ['Ngan Luu-Thuy Nguyen', 'Kiet Van Nguyen', 'Phong Nguyen-Thuan Do', 'Son Quoc Tran'] | 2023-03-16 | null | null | null | null | ['reading-comprehension', 'machine-reading-comprehension'] | ['natural-language-processing', 'natural-language-processing'] | [ 1.59393579e-01 3.82983536e-01 -7.54342005e-02 -3.39856327e-01
-1.21522868e+00 -8.84301782e-01 4.46130127e-01 5.55087626e-01
-7.03394532e-01 7.78697908e-01 5.32322943e-01 -9.92875457e-01
8.19856599e-02 -7.67941236e-01 -7.95162082e-01 -5.35711162e-02
5.67799926e-01 5.92864990e-01 6.40082583e-02 -8.51615846e-01
1.98581234e-01 -3.20039913e-02 -1.29903817e+00 3.56725901e-01
1.66416001e+00 9.87009257e-02 5.19766629e-01 8.48690391e-01
-1.43892601e-01 1.02924156e+00 -7.26454854e-01 -6.31869435e-01
-9.47164148e-02 -6.94178104e-01 -1.40670800e+00 -4.14518595e-01
7.88385391e-01 -4.65744108e-01 5.32569513e-02 8.91851902e-01
3.09227765e-01 -7.38242455e-03 5.07941723e-01 -8.17000210e-01
-1.10775101e+00 9.35885429e-01 -8.97536576e-02 3.40067774e-01
1.04675376e+00 1.76961362e-01 1.20607579e+00 -7.75665164e-01
7.10080981e-01 1.31014693e+00 2.39632875e-01 5.48382163e-01
-7.52621949e-01 -3.75017554e-01 2.73888916e-01 6.95366740e-01
-8.90940845e-01 -3.14784318e-01 4.31096196e-01 -2.92142123e-01
1.40918446e+00 3.78167689e-01 4.50542986e-01 1.24857962e+00
4.08238322e-01 8.17115605e-01 1.47818935e+00 -9.74664092e-01
-4.74609852e-01 1.34056181e-01 5.21112323e-01 6.96853161e-01
1.03009917e-01 -3.31385642e-01 -4.56044674e-01 5.42749763e-01
3.34868580e-01 -7.39896894e-01 -3.74137223e-01 1.54615417e-01
-1.54335570e+00 8.53337407e-01 1.48000330e-01 4.81268615e-01
1.07246011e-01 -6.92284048e-01 4.55900192e-01 7.81768322e-01
2.82429159e-01 8.54721487e-01 -6.56431317e-01 -4.28077042e-01
-5.41389525e-01 2.67530769e-01 1.05047894e+00 1.09453034e+00
6.31534994e-01 -2.03648180e-01 -1.16651617e-01 1.00726950e+00
1.48141190e-01 8.61070395e-01 3.70805413e-01 -7.84702361e-01
1.17838216e+00 6.59057915e-01 -9.70700383e-02 -8.87642026e-01
-2.46171787e-01 -4.67253953e-01 -5.52397788e-01 -4.02706981e-01
8.06449175e-01 -5.03668115e-02 -4.15420532e-01 1.51219392e+00
-9.82357115e-02 -7.62245119e-01 3.51590335e-01 7.10176110e-01
8.28940988e-01 9.34962988e-01 2.70667255e-01 -2.43224278e-01
1.40534151e+00 -1.46263456e+00 -8.97221386e-01 -2.39699259e-01
9.29401994e-01 -1.12959945e+00 1.76440430e+00 4.16192472e-01
-1.39083040e+00 -8.66554379e-01 -1.01880407e+00 -7.76787400e-01
-5.73984385e-01 -3.77105288e-02 3.12718362e-01 7.09378541e-01
-1.01914406e+00 -1.22019649e-01 -4.97911185e-01 -5.97141683e-01
-3.57599795e-01 -3.04985251e-02 -2.62933135e-01 -4.87803489e-01
-1.50941873e+00 1.48433447e+00 6.25518680e-01 1.89263240e-01
-9.27709699e-01 -4.38407928e-01 -1.13859820e+00 2.69294493e-02
3.98590773e-01 -4.16697174e-01 1.63257849e+00 -8.54617417e-01
-1.32756305e+00 8.22678447e-01 -4.45909888e-01 -1.44463837e-01
4.08278614e-01 -5.88852108e-01 -4.03478384e-01 -1.84442196e-02
2.84204155e-01 6.24005258e-01 1.81000575e-01 -1.00629556e+00
-9.40978348e-01 -2.74745554e-01 2.33366176e-01 5.47933459e-01
-3.06363940e-01 2.07392365e-01 -3.43466014e-01 -4.72368091e-01
-6.04302771e-02 -6.69883490e-01 2.45991215e-01 -6.98672712e-01
-9.79767181e-03 -7.84733355e-01 2.65154660e-01 -1.33838081e+00
1.46895552e+00 -1.52926898e+00 4.25793231e-01 -2.44215563e-01
2.86086828e-01 2.68755883e-01 -6.46777630e-01 8.04627359e-01
2.31883511e-01 3.22827101e-01 2.39886856e-03 -1.03742555e-01
6.75365478e-02 3.23135138e-01 -3.07916403e-01 9.08502340e-02
2.36138299e-01 1.19395161e+00 -1.17622459e+00 -4.51592416e-01
1.26820341e-01 3.25069241e-02 -3.47174942e-01 4.74918246e-01
-3.85308474e-01 4.52142209e-01 -2.41279110e-01 6.70760751e-01
6.18580461e-01 1.26921698e-01 3.09772976e-02 3.82527828e-01
-4.01502699e-01 7.10046768e-01 -5.30338824e-01 1.77707517e+00
-7.38477349e-01 7.05791175e-01 -2.64371298e-02 -6.68254554e-01
7.25440145e-01 2.60140389e-01 -3.43819708e-01 -1.29650223e+00
1.66100293e-01 3.50694567e-01 4.74593520e-01 -9.55244362e-01
8.83738458e-01 2.14449078e-01 -1.62908196e-01 3.74266833e-01
4.55772020e-02 -2.15256795e-01 5.32810748e-01 2.51795471e-01
6.14665389e-01 5.79833053e-02 3.64770919e-01 -4.34510499e-01
9.73179519e-01 3.66452545e-01 7.29480460e-02 8.00351441e-01
-3.02197456e-01 1.82091177e-01 4.11841959e-01 -5.04151136e-02
-1.02922392e+00 -1.02333009e+00 1.60104446e-02 1.59666824e+00
-2.59655327e-01 -5.50751805e-01 -1.16446888e+00 -6.77357495e-01
-5.42482257e-01 1.05483162e+00 -3.13838989e-01 2.61347175e-01
-1.05385506e+00 -3.15193355e-01 9.38402057e-01 5.11558473e-01
6.49428189e-01 -1.11155808e+00 -4.46135312e-01 3.27842802e-01
-8.35443735e-01 -1.30614471e+00 -1.19547702e-01 -7.07578734e-02
-6.87163532e-01 -1.07277620e+00 -6.93124712e-01 -1.44121289e+00
3.83861631e-01 2.01759264e-01 1.43094087e+00 3.01747918e-01
4.10937876e-01 3.93750310e-01 -8.55700195e-01 -6.96054697e-01
-9.02901769e-01 8.30027461e-01 -3.44212174e-01 -7.84311116e-01
6.81025147e-01 3.07432860e-01 -1.97252501e-02 1.71372935e-01
-7.87644029e-01 2.67974317e-01 5.46345651e-01 8.15524757e-01
1.27187818e-01 -2.88634658e-01 7.46665657e-01 -8.61840248e-01
1.24128687e+00 -5.66083252e-01 -5.74637473e-01 7.83230901e-01
-6.37494624e-01 4.88872342e-02 8.98041129e-01 -1.20795369e-01
-1.32988429e+00 -5.72143674e-01 -6.20616019e-01 6.00930572e-01
-3.15493912e-01 8.93191636e-01 -1.91375479e-01 2.94379950e-01
6.01040423e-01 3.11308920e-01 -8.62948075e-02 -5.53766668e-01
4.39044654e-01 7.30939925e-01 6.04598522e-01 -8.62485349e-01
6.04261637e-01 -4.06826615e-01 -5.17840207e-01 -1.16841984e+00
-8.91710818e-01 -4.49141413e-01 -8.62638533e-01 -1.11245319e-01
1.09711325e+00 -9.98876393e-01 -5.60887873e-01 6.84943736e-01
-1.37050843e+00 -4.81321812e-01 2.00440198e-01 1.04629837e-01
-2.10935190e-01 6.20289266e-01 -9.34195697e-01 -5.01988113e-01
-2.76399255e-01 -1.45296693e+00 9.02554750e-01 1.29483059e-01
-4.29174811e-01 -1.45800567e+00 1.33207634e-01 1.11086071e+00
4.51081067e-01 -2.31404051e-01 1.66455626e+00 -4.39091861e-01
-4.52071190e-01 1.77804157e-01 -2.13168859e-01 6.09285891e-01
-2.37112463e-01 -1.55104578e-01 -6.17099881e-01 -2.73855090e-01
1.35495409e-01 -7.64900982e-01 4.17665124e-01 -1.37826771e-01
7.03244925e-01 -2.01872408e-01 3.53136361e-01 1.79327533e-01
1.19504559e+00 6.50881305e-02 7.65025079e-01 6.77874207e-01
7.92392969e-01 1.04009724e+00 8.06096196e-01 -2.19502985e-01
1.26296771e+00 3.77910346e-01 8.58788863e-02 -2.10487649e-01
-2.36048579e-01 -6.79974556e-01 6.12083316e-01 1.90693927e+00
-8.37156698e-02 -5.41992009e-01 -1.36501515e+00 7.22240269e-01
-1.52231658e+00 -4.48031694e-01 -7.83414304e-01 1.89996910e+00
1.08507228e+00 -1.40732318e-01 -2.07386747e-01 -1.24290965e-01
1.77943319e-01 5.41885803e-03 3.14316526e-03 -9.20030713e-01
-4.10196275e-01 4.25967962e-01 1.94904655e-01 1.03379023e+00
-6.78931952e-01 1.57552135e+00 6.35126257e+00 7.40090072e-01
-8.87385905e-01 2.36533269e-01 2.54202038e-01 6.96342349e-01
-5.19083738e-01 7.42055252e-02 -9.26115096e-01 -8.29374194e-02
1.22876632e+00 -1.85441747e-01 4.35237497e-01 3.57828051e-01
2.47239351e-01 -4.13263470e-01 -1.17533243e+00 6.05566859e-01
3.89935791e-01 -7.52213597e-01 3.38681519e-01 -3.64552677e-01
7.26934910e-01 -1.33957844e-02 -1.41246110e-01 6.97144628e-01
-4.73267362e-02 -1.30500650e+00 6.31716430e-01 2.45274857e-01
5.59590876e-01 -5.92894018e-01 9.25394177e-01 6.85849309e-01
-9.03055429e-01 8.81589055e-02 -4.62777585e-01 -6.90107405e-01
1.87112227e-01 -2.58670241e-01 -8.59507918e-01 5.32701612e-01
3.24952930e-01 5.64149737e-01 -1.21624291e+00 4.88294542e-01
-7.67016351e-01 9.10083830e-01 2.13374764e-01 -3.89446288e-01
4.64960903e-01 -2.41860256e-01 2.07737982e-01 1.32808030e+00
6.56200945e-02 1.13161519e-01 2.40199909e-01 5.01263440e-01
5.09746885e-03 7.58671403e-01 -5.35186946e-01 -1.18476376e-02
2.36384273e-01 7.66892016e-01 -2.07580194e-01 -3.90893757e-01
-8.32496345e-01 9.82163131e-01 7.81063735e-01 3.59779030e-01
-5.75520992e-01 -3.63940179e-01 -3.45924720e-02 6.28784001e-02
-4.68298674e-01 -8.66673827e-01 -5.12748122e-01 -1.46530223e+00
1.60061523e-01 -1.83237708e+00 3.59813899e-01 -7.56924093e-01
-1.12747657e+00 6.38344526e-01 1.32554010e-01 -6.15431726e-01
-3.04593533e-01 -7.30284691e-01 -3.28338474e-01 1.26930368e+00
-1.68501866e+00 -1.34393311e+00 -4.86417413e-01 6.28554702e-01
1.28861713e+00 -9.34728980e-02 1.00426745e+00 1.29816502e-01
-5.56701064e-01 8.04596066e-01 -1.23918578e-01 1.26755223e-01
9.93389487e-01 -1.30888844e+00 5.78139186e-01 1.09439850e+00
-7.07880557e-02 8.20927441e-01 5.44360936e-01 -7.88190663e-01
-1.38816404e+00 -5.85343301e-01 1.87865782e+00 -8.39019656e-01
1.00100648e+00 -3.69155914e-01 -1.18618441e+00 9.14630949e-01
1.12089682e+00 -1.35252428e+00 5.84114790e-01 4.11393583e-01
-1.42530382e-01 1.94039181e-01 -5.65663040e-01 9.13587630e-01
8.07317376e-01 -7.28287101e-01 -1.30198920e+00 4.13868159e-01
7.62719333e-01 -4.00842279e-01 -1.04546750e+00 2.15472609e-01
4.46903497e-01 -6.81437731e-01 4.52747792e-01 -5.89278042e-01
6.79837167e-01 -3.07435058e-02 5.20841628e-02 -1.44557106e+00
-1.19485836e-02 -3.85209203e-01 5.56758761e-01 1.13539112e+00
8.33840311e-01 -6.83218718e-01 1.68319479e-01 5.46960652e-01
-3.19577962e-01 -3.63724291e-01 -7.22528160e-01 -4.44779515e-01
1.02480710e+00 -3.99972945e-01 3.00906986e-01 9.23037231e-01
2.54820228e-01 7.48515785e-01 -7.66636729e-02 8.19401890e-02
2.83279717e-01 -3.56305182e-01 8.73318374e-01 -8.96887779e-01
-3.91535833e-02 -2.72586167e-01 2.64660567e-01 -1.50472820e+00
5.89557707e-01 -9.90410388e-01 1.36795431e-01 -1.73208916e+00
1.78922236e-01 -1.67884529e-01 1.44085705e-01 3.02010894e-01
-4.74205226e-01 -2.13594168e-01 3.56369346e-01 1.72600094e-02
-6.40067816e-01 4.35842931e-01 1.76678634e+00 -1.25968739e-01
1.90191735e-02 -2.99231440e-01 -7.31934071e-01 4.62235242e-01
8.86076391e-01 -1.09505177e-01 -4.98622835e-01 -1.39702201e+00
3.39798212e-01 1.87881082e-01 -6.21922463e-02 -8.14847827e-01
3.03207576e-01 -4.97746989e-02 1.44219264e-01 -4.93018538e-01
-2.46292248e-01 -4.61131036e-01 -5.47298431e-01 4.14294422e-01
-5.88490427e-01 9.91356254e-01 2.02504396e-01 -2.41464078e-01
-6.13284886e-01 -5.29209316e-01 3.71678770e-01 -2.30460718e-01
-8.97360146e-01 -4.29139733e-01 -8.53375971e-01 5.16436815e-01
7.82538772e-01 -1.92418415e-02 -7.71439970e-01 -5.85245728e-01
-3.64380717e-01 6.45751834e-01 3.91152263e-01 9.02688503e-01
4.22682703e-01 -6.04560137e-01 -1.04384363e+00 3.42717022e-01
3.08549970e-01 -1.49941951e-01 1.53867453e-01 5.71196318e-01
-1.14463627e+00 1.13612568e+00 -5.58823228e-01 -3.29226166e-01
-1.19576693e+00 3.50527555e-01 1.00284919e-01 -5.33603549e-01
-1.76264152e-01 5.04688084e-01 -1.39387906e-01 -6.96574152e-01
3.86441916e-01 -6.23270690e-01 -5.80722570e-01 -4.24614958e-02
4.87796009e-01 6.90799832e-01 3.26611340e-01 -8.01153243e-01
7.05224872e-02 5.36199868e-01 -3.46005559e-01 -2.08585426e-01
6.37368083e-01 -4.54700649e-01 -4.18520778e-01 5.34775615e-01
8.94560575e-01 3.42403710e-01 -4.34886962e-01 1.88472867e-01
1.08142100e-01 -2.67221499e-02 -3.30521703e-01 -1.32882929e+00
-1.76416576e-01 1.18920720e+00 2.40413785e-01 7.70597607e-02
1.04560041e+00 -8.51978362e-02 9.59352374e-01 8.29189003e-01
4.21417087e-01 -1.33725333e+00 -2.68780887e-01 1.44158769e+00
1.09492993e+00 -1.50774491e+00 -4.00769204e-01 -5.38689852e-01
-6.54198527e-01 1.15255654e+00 1.03213680e+00 3.70057166e-01
3.08765601e-02 -4.84211408e-02 7.46220171e-01 1.16293123e-02
-6.20728016e-01 -1.40191153e-01 4.18853164e-01 5.83878040e-01
1.01403379e+00 3.13469917e-01 -9.75355148e-01 4.05826002e-01
-8.61564040e-01 -4.71255451e-01 7.56565750e-01 7.77585030e-01
-3.85005772e-01 -1.54757571e+00 -3.33281279e-01 4.48700897e-02
-3.95619631e-01 -5.85215092e-01 -5.29709101e-01 1.22259390e+00
-1.10135481e-01 1.48378527e+00 -2.25316495e-01 -7.17717484e-02
6.60748959e-01 1.57612607e-01 5.63535035e-01 -7.50302672e-01
-9.67914224e-01 -1.46308139e-01 2.83973396e-01 -2.78931707e-01
-1.37159005e-01 -4.18275893e-01 -7.88766026e-01 -4.18863893e-01
-8.78591165e-02 2.47724399e-01 4.83054399e-01 1.16046953e+00
-1.94715455e-01 3.33089113e-01 -1.10645741e-01 -7.96032175e-02
-6.47290826e-01 -1.27789295e+00 -6.07886910e-02 2.79071420e-01
2.00145155e-01 -1.65127274e-02 -2.05121800e-01 5.62444739e-02] | [11.309456825256348, 8.394969940185547] |
1957db81-6171-4b07-99a0-e18692f788e5 | context-guided-triple-matching-for-multiple-1 | null | null | https://openreview.net/forum?id=3fBNtKp72iV | https://openreview.net/pdf?id=3fBNtKp72iV | Context-guided Triple Matching for Multiple Choice Question Answering | The task of multiple choice question answering (MCQA) refers to identifying a suitable answer from multiple candidates, by estimating the matching score among the \emph{triple} of the passage, question and answer. Despite the general research interest in this regard, existing methods decouple the process into several pair-wise or \emph{dual} matching steps, that limited the ability of assessing cases with multiple evidence sentences. To alleviate this issue, this paper introduces a novel \textbf{C}ontext-guided \textbf{T}riple \textbf{M}atching algorithm, which is achieved by integrating a Triple Matching (TM) module and a Contrastive Regularization (CR). The former is designed to enumerate one component from the triple as the background context, and estimate its semantic matching with the other two. Additionally, the contrastive term is further proposed to capture the dissimilarity between the correct answer and distractive ones. We validate the proposed algorithm on several benchmarking MCQA datasets, which exhibits competitive performances against state-of-the-arts. | ['Anonymous'] | 2022-01-16 | null | null | null | acl-arr-january-2022-1 | ['multiple-choice-qa'] | ['natural-language-processing'] | [ 1.2370108e-01 -8.4471449e-02 1.4359394e-01 -4.4738337e-01
-1.4792269e+00 -6.0248142e-01 4.5229784e-01 3.7264374e-01
-6.3789165e-01 4.9942115e-01 2.5037694e-01 -1.8953741e-01
-3.2708156e-01 -4.5352721e-01 -4.9439967e-01 -6.8612784e-01
7.1021557e-01 4.0329641e-01 5.5966985e-01 -2.9485735e-01
6.3773990e-01 -7.1168341e-02 -1.7044922e+00 6.4322180e-01
1.3031216e+00 1.2728670e+00 3.3837658e-01 4.5979491e-01
-6.2365711e-01 1.1514601e+00 -5.6603652e-01 -6.7270207e-01
-9.3211189e-02 -7.6143837e-01 -8.1216550e-01 -1.4812687e-01
6.9371527e-01 1.0440968e-01 -3.0917453e-02 1.1883465e+00
6.6835934e-01 4.7635567e-01 5.1821327e-01 -9.9718225e-01
-5.3371596e-01 3.5592750e-01 -4.4933254e-01 7.2441399e-01
5.8688235e-01 1.4019793e-02 1.4282176e+00 -1.3298857e+00
3.8752943e-01 1.3427846e+00 2.3018841e-01 4.3002030e-01
-8.6770946e-01 -5.1573324e-01 3.6798716e-01 6.5193683e-01
-1.0989304e+00 -3.5184067e-01 1.0586436e+00 -3.6822736e-01
4.9522963e-01 6.0142076e-01 3.5427697e-02 8.5276169e-01
-2.2847174e-01 1.0926237e+00 1.2918035e+00 -4.3475249e-01
1.7029394e-01 2.4547870e-01 5.5226129e-01 4.0022588e-01
-5.2529448e-01 -3.3468550e-01 -4.9442294e-01 -2.6349708e-01
-4.4362232e-02 -2.9951921e-01 -4.6488830e-01 5.5006310e-02
-9.0616244e-01 6.9726574e-01 3.4674144e-01 4.3835142e-01
-3.1988370e-01 -3.0361310e-01 1.3184825e-01 3.4825492e-01
2.4444021e-01 2.1382816e-01 -3.4675419e-01 1.4137960e-02
-9.7855616e-01 4.2237604e-01 7.5671089e-01 6.6685617e-01
7.5007433e-01 -4.4614729e-01 -8.2350731e-01 1.1114197e+00
4.0168038e-01 4.2265382e-01 5.2172673e-01 -9.7126484e-01
9.7840643e-01 7.6279473e-01 2.6021475e-01 -1.1184819e+00
-2.3764277e-01 -4.7666067e-01 -4.3667653e-01 -2.8455007e-01
6.6026986e-01 5.8000937e-02 -2.1836969e-01 1.7284324e+00
5.9118056e-01 4.3295734e-02 -2.8211117e-01 1.0980136e+00
1.1147635e+00 5.8125031e-01 1.3648732e-01 -2.2986534e-01
1.3954878e+00 -1.2339902e+00 -7.9786009e-01 -2.4143486e-01
3.7501854e-01 -1.0672026e+00 1.4633092e+00 2.9959893e-01
-1.1245049e+00 -5.6235129e-01 -1.0080173e+00 -2.9459003e-01
-4.1688543e-02 2.7816865e-01 -1.9990247e-01 2.9343492e-01
-5.6730980e-01 2.1698019e-01 4.7615860e-02 1.4072657e-01
4.1697428e-02 -7.0842318e-02 6.1037220e-02 -2.7450788e-01
-1.4401243e+00 9.5905650e-01 1.1548089e-01 2.6372764e-01
-4.0850717e-01 -4.4977844e-01 -5.8065343e-01 1.1431600e-01
5.9850568e-01 -6.8627506e-01 1.1185164e+00 -1.0917935e+00
-1.2595960e+00 9.9755257e-01 -4.0261394e-01 1.4555485e-02
5.4352766e-01 -1.6296221e-01 -6.3731384e-01 3.3058554e-01
3.6079052e-01 3.0480671e-01 1.0261406e+00 -1.3311943e+00
-6.9633257e-01 -5.4974079e-01 1.8259422e-01 4.3902665e-01
-1.4281073e-01 3.0507225e-01 -5.6309623e-01 -6.2153655e-01
4.3408087e-01 -5.7979614e-01 2.7281731e-01 -1.4609915e-01
-2.8633440e-01 -5.5712146e-01 4.9415749e-01 -8.6442953e-01
1.6866869e+00 -2.1281464e+00 3.0234024e-01 9.1350980e-02
2.1336181e-01 1.8005526e-01 -2.7905735e-01 2.4545673e-01
1.6959339e-01 -2.5808096e-01 -2.1168916e-01 -3.4124130e-01
9.3082882e-02 -2.2029098e-02 -1.7627786e-01 3.1325096e-01
2.6654264e-01 6.5328503e-01 -9.3393016e-01 -6.9525665e-01
-9.0207942e-02 7.2288886e-03 -4.3348590e-01 4.2174444e-01
-4.3577698e-01 5.9666091e-01 -7.0898664e-01 5.2127475e-01
7.2169966e-01 -3.0954984e-01 6.8139300e-02 -1.3817945e-01
-1.1279462e-01 5.3613853e-01 -1.4157590e+00 1.3321013e+00
-2.9088771e-01 1.0187951e-01 4.0900207e-01 -9.4725913e-01
1.0917041e+00 1.0037198e-01 8.6936027e-02 -1.1488671e+00
2.7300876e-01 5.3216708e-01 7.3911756e-02 -1.0266387e+00
4.7661689e-01 -8.7864123e-02 3.7605792e-02 3.1674278e-01
-4.6259567e-02 -3.4791313e-02 2.0366588e-01 1.6567160e-01
9.6940875e-01 -9.0654343e-03 1.9824458e-02 -3.9997819e-01
1.3096069e+00 -1.3310006e-01 4.8654687e-01 6.1123532e-01
-4.3666020e-01 6.9887853e-01 5.2150273e-01 -1.1483321e-02
-5.7354474e-01 -1.0095382e+00 2.0472421e-01 1.2951369e+00
5.5886734e-01 -1.3039111e-01 -8.5556084e-01 -8.4023422e-01
-2.9046956e-01 9.5081079e-01 -4.8814803e-01 -6.5629512e-02
-9.8035258e-01 -6.3812876e-01 3.7605292e-01 2.2082499e-01
6.0372263e-01 -1.0109198e+00 -3.1910574e-01 2.1360606e-01
-9.2425352e-01 -8.9349431e-01 -7.6041126e-01 -1.3641483e-01
-5.6186163e-01 -1.2048638e+00 -4.0975010e-01 -8.8899833e-01
3.1589887e-01 3.1395233e-01 1.3255787e+00 3.4850356e-01
1.8990910e-01 2.6239678e-01 -5.2906573e-01 -9.9693865e-02
-1.9266447e-01 -1.8003809e-01 -4.2312756e-01 4.4440126e-01
5.5693084e-01 -3.4210324e-01 -7.7110463e-01 4.3604633e-01
-8.6940151e-01 -2.0117344e-01 4.6898609e-01 7.7872330e-01
8.3281231e-01 -3.0743060e-01 8.6129642e-01 -6.6910756e-01
9.1199785e-01 -6.3110852e-01 -1.9019875e-01 5.3036112e-01
-3.9874887e-01 -1.7754944e-02 6.9765484e-01 -3.5775146e-01
-1.2210293e+00 -4.3961766e-01 -5.0492752e-01 -1.3356614e-01
8.2020506e-02 4.7511432e-01 -4.2415446e-01 2.5683209e-01
5.2782452e-01 1.9214222e-01 -3.2870135e-01 -5.3029406e-01
4.3670154e-01 5.8034945e-01 4.4548401e-01 -6.0392064e-01
5.2753478e-01 1.2394527e-01 -3.2346657e-01 -4.4172812e-01
-1.4525578e+00 -7.8297341e-01 -3.6822796e-01 -5.5101240e-01
8.1607592e-01 -4.9777457e-01 -6.8733543e-01 1.7092882e-01
-1.2353566e+00 2.7852196e-01 -1.0055604e-01 3.8186941e-01
-2.9449877e-01 5.5709946e-01 -4.8612356e-01 -8.9314938e-01
-2.6109269e-01 -1.0844257e+00 8.4587604e-01 3.0855858e-01
-2.5488809e-01 -6.4192265e-01 6.9549762e-02 1.1767412e+00
2.4669577e-01 -3.4361936e-02 1.2420129e+00 -1.1425517e+00
-6.0337710e-01 -2.5519335e-01 -2.6326758e-01 4.8223421e-01
-3.0635232e-01 -3.7801355e-01 -9.9840558e-01 -3.5687178e-02
6.4451492e-01 -4.4865471e-01 8.8852447e-01 4.0474698e-02
1.0291197e+00 -2.9969335e-01 1.0580193e-01 -1.0101246e-01
1.1576258e+00 4.5066599e-02 5.3815216e-01 2.9638731e-01
4.7072855e-01 7.5885463e-01 9.4653487e-01 2.3625417e-01
7.1332490e-01 6.7910790e-01 4.4834083e-01 2.9547551e-01
-3.4647968e-01 -2.1909747e-01 2.4910483e-01 1.2797656e+00
3.0902353e-01 -3.5869187e-01 -6.6066027e-01 5.3097254e-01
-1.8030666e+00 -1.0136379e+00 -4.9097854e-01 2.1544962e+00
7.8995371e-01 1.8354316e-01 -1.1346137e-01 2.7908272e-01
1.0010103e+00 2.6308888e-01 -5.3145856e-01 -5.9418809e-02
-4.2095837e-01 2.7031949e-01 -4.4141510e-01 6.3643074e-01
-9.3646294e-01 4.2251202e-01 4.8138804e+00 1.4041175e+00
-6.1436117e-01 3.2315531e-01 6.2033343e-01 6.6487186e-02
-7.5849181e-01 4.1355804e-02 -6.8982315e-01 5.9508842e-01
5.8285332e-01 1.0070569e-01 1.5173575e-01 4.7125667e-01
2.9499245e-01 -3.5224256e-01 -8.5751313e-01 7.4335170e-01
4.3827939e-01 -9.4224864e-01 -3.0525243e-02 -6.5241253e-01
4.2879096e-01 -4.1120091e-01 7.5553641e-02 3.9796841e-01
-3.2838270e-01 -6.8874472e-01 9.2458755e-01 7.2200608e-01
2.1906480e-01 -5.2416742e-01 8.6537254e-01 7.9914862e-01
-1.2632072e+00 -2.6138550e-01 2.3453247e-02 1.2139459e-02
2.2818113e-02 5.5874807e-01 -8.8607803e-02 7.8044754e-01
7.0475882e-01 2.4468881e-01 -7.8409410e-01 9.8166484e-01
-4.3121243e-01 6.3438576e-01 4.0075246e-02 -3.3102828e-01
2.0664193e-01 -3.9759883e-01 7.6970011e-01 9.4313192e-01
1.7686394e-01 2.5781840e-01 1.2864490e-01 9.5128173e-01
-2.1053468e-01 5.6732464e-01 2.9146689e-01 4.3717852e-01
7.2748131e-01 1.2502357e+00 -4.3119290e-01 -1.1442191e-01
-5.1919407e-01 9.3619400e-01 6.9886661e-01 1.7691469e-01
-8.7904060e-01 -2.6721987e-01 -4.5453802e-02 -7.0688963e-02
2.6339984e-01 2.7497077e-01 -3.2764888e-01 -1.2154485e+00
6.0067892e-01 -1.1024574e+00 7.6844597e-01 -7.1882159e-01
-1.7211711e+00 6.5753347e-01 -3.6783391e-01 -1.2563030e+00
2.4638826e-01 -2.7216971e-01 -5.9646094e-01 1.1023760e+00
-1.7670147e+00 -8.4623891e-01 -2.9742137e-01 5.4343516e-01
7.4235892e-01 8.3778821e-02 4.8860043e-01 7.2095799e-01
-7.2261786e-01 6.0712051e-01 8.7187979e-03 -2.2022025e-01
7.9170978e-01 -1.1591991e+00 -2.9222050e-01 8.9683139e-01
3.8190586e-03 4.1094318e-01 7.7528733e-01 -3.3077902e-01
-9.8168075e-01 -7.0986784e-01 1.3502551e+00 -5.1450557e-01
5.4210788e-01 8.4590785e-02 -1.1919540e+00 8.2289197e-02
1.7590125e-01 -1.3510999e-01 5.8436179e-01 -2.5915122e-01
-5.4999638e-01 -1.2554860e-01 -1.1937953e+00 4.1385356e-01
5.5652606e-01 -6.3930428e-01 -9.9528259e-01 6.4989686e-02
7.9112595e-01 -3.3994183e-01 -5.5049795e-01 4.5017639e-01
1.7925984e-01 -1.2533063e+00 8.5787928e-01 -3.8584915e-01
5.7227808e-01 -4.6083593e-01 -3.4210289e-01 -7.8302854e-01
-1.0133592e-01 -2.0652269e-01 -3.3211917e-01 1.5106083e+00
5.9154105e-01 -1.6574946e-01 5.0182062e-01 5.3846908e-01
-2.4552213e-01 -8.6745512e-01 -1.1635404e+00 -3.3010393e-01
1.6029318e-01 -1.5819193e-01 3.4172171e-01 8.8205332e-01
-1.7871282e-01 6.8020236e-01 -1.9522308e-01 3.9508741e-02
2.7467474e-01 3.8854185e-01 1.7946701e-01 -1.0380414e+00
-4.1107532e-01 -6.1242789e-01 2.5726917e-01 -1.5377290e+00
1.7284864e-01 -9.2336416e-01 2.6503342e-01 -1.3763101e+00
3.1054977e-01 -3.5574993e-01 -4.6263301e-01 -2.5658441e-01
-8.9987427e-01 -8.3125331e-02 1.4681095e-01 2.8413334e-01
-9.6159363e-01 8.3127266e-01 1.3306425e+00 -1.9565134e-01
6.7738608e-02 2.2115007e-01 -6.2394834e-01 7.8433257e-01
6.9415534e-01 -6.6216743e-01 -4.1450170e-01 -5.2687681e-01
4.1615430e-01 3.1371149e-01 3.6175892e-01 -8.3411700e-01
4.5055884e-01 -6.4514177e-03 -2.7981378e-02 -7.6294470e-01
4.4527629e-01 -5.9725106e-01 -4.7358453e-01 1.4279965e-01
-7.5765014e-01 2.4579641e-01 -6.0527187e-02 7.1572554e-01
-4.7880679e-01 -7.1833199e-01 7.8018272e-01 -2.0868319e-01
-5.6984836e-01 6.5966077e-02 -1.6712505e-01 7.5400370e-01
6.1706883e-01 1.1107103e-01 -4.5029643e-01 -3.1788474e-01
-6.1446297e-01 5.2164167e-01 -8.8069320e-02 4.3597507e-01
7.3414057e-01 -1.2979978e+00 -8.8124478e-01 -3.4387332e-01
2.9591882e-01 -3.5866743e-01 6.8345529e-01 1.0938071e+00
-1.8673380e-01 1.4445551e-01 3.6025721e-01 -2.9132643e-01
-1.2853236e+00 5.3539699e-01 4.5882770e-01 -4.9223018e-01
-1.4905675e-01 1.0463825e+00 -7.7307448e-02 -6.2209350e-01
3.9307541e-01 1.6084929e-01 -7.2775787e-01 3.3923262e-01
5.5761456e-01 6.6875207e-01 2.0260736e-01 -8.3895761e-01
-4.0108630e-01 5.6246293e-01 5.9528641e-02 -2.4295524e-01
7.7236509e-01 -6.0274482e-01 -4.3819028e-01 5.8365929e-01
1.0241792e+00 2.5753221e-01 -8.6579341e-01 -4.7360960e-01
3.5445464e-01 -4.7861943e-01 -1.7355685e-01 -9.6615732e-01
-6.4667213e-01 8.8191175e-01 4.5029962e-01 1.9981669e-01
1.0938843e+00 1.3954146e-01 9.9720407e-01 3.5005397e-01
-8.4272832e-02 -1.3453017e+00 4.8407933e-01 5.2036220e-01
1.0272003e+00 -1.2783643e+00 -3.2477397e-01 -4.7945070e-01
-7.1376926e-01 9.2238170e-01 8.4375596e-01 1.2494550e-02
5.0060332e-01 -3.8088250e-01 1.2327365e-01 -3.3428317e-01
-7.1240979e-01 -2.6232272e-01 6.3203895e-01 -2.8025113e-02
4.6872288e-01 -1.6771846e-01 -8.8024509e-01 1.0068120e+00
1.9193232e-02 -4.6839800e-01 8.0648698e-02 7.4637043e-01
-7.2511989e-01 -9.7110301e-01 -2.8627291e-01 4.7993252e-01
-5.6671906e-01 -1.6376734e-01 -3.4403142e-01 3.8465163e-01
4.2582154e-01 1.6336085e+00 -2.8347659e-01 -2.3288903e-01
5.9168434e-01 1.8029173e-01 3.3739391e-01 -3.1582782e-01
-1.0813898e+00 -4.7788019e-03 2.0923604e-01 -3.5500577e-01
-6.7671913e-01 -7.1913606e-01 -1.0906317e+00 -4.2426836e-02
-5.8688873e-01 3.6837181e-01 1.8006812e-01 1.2830936e+00
9.3790367e-02 5.2140808e-01 7.5312638e-01 -1.1598974e-01
-8.5555494e-01 -9.4801015e-01 -3.1679967e-01 8.0994940e-01
3.4122545e-01 -4.4752666e-01 -5.6810164e-01 -3.9978462e-01] | [11.395598411560059, 8.036478042602539] |
74b5b1b1-6070-42b4-9ded-52455784d20d | utterance-weighted-multi-dilation-temporal | 2205.08455 | null | https://arxiv.org/abs/2205.08455v3 | https://arxiv.org/pdf/2205.08455v3.pdf | Utterance Weighted Multi-Dilation Temporal Convolutional Networks for Monaural Speech Dereverberation | Speech dereverberation is an important stage in many speech technology applications. Recent work in this area has been dominated by deep neural network models. Temporal convolutional networks (TCNs) are deep learning models that have been proposed for sequence modelling in the task of dereverberating speech. In this work a weighted multi-dilation depthwise-separable convolution is proposed to replace standard depthwise-separable convolutions in TCN models. This proposed convolution enables the TCN to dynamically focus on more or less local information in its receptive field at each convolutional block in the network. It is shown that this weighted multi-dilation temporal convolutional network (WD-TCN) consistently outperforms the TCN across various model configurations and using the WD-TCN model is a more parameter efficient method to improve the performance of the model than increasing the number of convolutional blocks. The best performance improvement over the baseline TCN is 0.55 dB scale-invariant signal-to-distortion ratio (SISDR) and the best performing WD-TCN model attains 12.26 dB SISDR on the WHAMR dataset. | ['Thomas Hain', 'Stefan Goetze', 'William Ravenscroft'] | 2022-05-17 | null | null | null | null | ['speech-dereverberation'] | ['speech'] | [ 1.97946638e-01 -7.89498836e-02 1.65372714e-01 -2.84439981e-01
-2.92255819e-01 -2.70447075e-01 6.23737037e-01 -5.04965484e-01
-6.09423041e-01 1.59949720e-01 6.23350799e-01 -8.29307258e-01
-4.71131951e-02 -2.44180456e-01 -4.76457000e-01 -9.41173613e-01
-1.18040852e-01 -2.39093989e-01 5.39148808e-01 -2.46454760e-01
2.27087061e-03 8.13807547e-01 -1.25435913e+00 7.12411642e-01
3.74752820e-01 1.00622165e+00 6.55628145e-01 1.21430349e+00
2.38357782e-01 7.65301764e-01 -8.94602060e-01 -4.87224683e-02
2.18156517e-01 -2.96905935e-01 -7.06329226e-01 -1.54215217e-01
3.46853286e-01 -3.06745172e-01 -9.77023482e-01 8.59925330e-01
1.02654922e+00 5.65114021e-01 2.43286178e-01 -6.83881164e-01
-5.64750433e-01 8.55055511e-01 -2.37172410e-01 9.49960887e-01
-3.53324443e-01 9.28780362e-02 7.50234127e-01 -9.48718369e-01
2.72128463e-01 1.54617870e+00 6.19082332e-01 5.76236546e-01
-9.33340847e-01 -5.38726866e-01 -8.28882381e-02 5.51593006e-01
-1.04947996e+00 -7.21943378e-01 6.04843616e-01 -8.04257914e-02
1.72107387e+00 4.46755707e-01 5.29453456e-01 9.53249514e-01
7.08171204e-02 7.58959234e-01 7.35371888e-01 -5.36999881e-01
1.16397753e-01 -4.58476126e-01 -8.23700726e-02 1.04629308e-01
-3.92344117e-01 6.50703132e-01 -7.54593372e-01 3.77820373e-01
8.06222022e-01 -4.58190471e-01 -2.97670245e-01 3.71024162e-01
-1.08504593e+00 5.26186466e-01 6.96167111e-01 7.66220689e-01
-3.04390550e-01 5.28555989e-01 6.65225744e-01 5.07392168e-01
7.09773362e-01 1.80716276e-01 -7.00514734e-01 -5.33961117e-01
-1.23212051e+00 5.66320680e-02 3.17031026e-01 7.16513872e-01
-4.54577729e-02 7.48337388e-01 -5.04551470e-01 1.26044512e+00
4.99779657e-02 1.05192587e-01 7.95592844e-01 -8.50710332e-01
5.75837135e-01 -4.01673168e-02 -2.77858227e-01 -3.38618606e-01
-2.71583021e-01 -8.80274475e-01 -7.81817079e-01 1.30432203e-01
2.32067600e-01 -2.31286898e-01 -1.48664963e+00 1.38844693e+00
-1.04578540e-01 3.16463798e-01 1.98542029e-01 8.58001173e-01
7.85228193e-01 1.08694124e+00 -6.13073558e-02 -6.60227612e-02
1.09450245e+00 -1.28690934e+00 -1.02163994e+00 -2.01949760e-01
5.24653256e-01 -1.02385128e+00 6.07944787e-01 2.95170605e-01
-1.20594215e+00 -7.69594252e-01 -1.27178860e+00 -2.65412480e-01
-4.09162045e-01 3.19787085e-01 -8.17699265e-03 9.69844937e-01
-1.56705272e+00 6.78855777e-01 -8.42172503e-01 -2.38106161e-01
2.89967775e-01 5.01208425e-01 4.85840924e-02 2.44416781e-02
-1.30208707e+00 9.76570547e-01 4.11668509e-01 3.68538976e-01
-1.10454392e+00 -8.65851104e-01 -5.83478332e-01 3.34248394e-01
3.16445008e-02 -2.52281219e-01 1.78595030e+00 -8.82244706e-01
-1.71337962e+00 3.59659404e-01 -2.86131889e-01 -1.18752623e+00
4.80816752e-01 -6.48272410e-02 -7.04032063e-01 1.48659945e-01
-5.71105897e-01 8.45671654e-01 9.62175369e-01 -7.75724530e-01
-5.48060298e-01 2.34846361e-02 -2.98758745e-01 1.32400632e-01
-3.44748795e-01 4.42486376e-01 -3.13695222e-01 -1.26304018e+00
2.32548654e-01 -7.81981587e-01 -1.26759946e-01 -1.48237154e-01
-1.86138377e-01 -2.26726681e-01 1.57762659e+00 -1.01939905e+00
1.34430945e+00 -2.28318167e+00 4.22260873e-02 -3.41855228e-01
2.66320948e-02 1.17668235e+00 -2.75861323e-01 3.00473392e-01
-4.13032144e-01 1.07052706e-01 2.44112741e-02 -5.79571426e-01
-9.87015963e-02 2.70311505e-01 -2.21095413e-01 3.57191145e-01
1.23881213e-01 8.22264791e-01 -5.28836727e-01 9.96597335e-02
5.51454544e-01 9.60313857e-01 -3.25705707e-01 -1.83837600e-02
-4.35014889e-02 1.82102457e-01 3.23627323e-01 1.44244641e-01
7.31094003e-01 4.06619698e-01 -9.35439020e-02 -1.06881417e-01
-4.48124260e-01 7.00306654e-01 -5.61461449e-01 1.57716441e+00
-7.12646484e-01 1.61269021e+00 -6.36937544e-02 -1.02250481e+00
9.53153014e-01 9.36351955e-01 8.30586329e-02 -1.04169106e+00
1.77003562e-01 4.32739526e-01 7.35150933e-01 -3.90542954e-01
4.91806567e-01 -1.33835107e-01 5.98131061e-01 1.43943876e-01
1.98913649e-01 1.15999341e-01 -1.19653888e-01 -1.65255487e-01
1.16392589e+00 -3.98190558e-01 -3.15940708e-01 -3.67825657e-01
6.08686388e-01 -7.35530019e-01 4.66617525e-01 4.81160074e-01
-3.60500038e-01 7.39003003e-01 2.48436794e-01 -6.17921829e-01
-1.64137602e+00 -7.59260654e-01 -9.21284556e-02 9.45127964e-01
-4.06133205e-01 2.24624444e-02 -9.55079317e-01 3.32257748e-02
-4.87511903e-01 8.17375004e-01 -3.88807803e-01 -2.08196908e-01
-9.48255539e-01 -5.04517496e-01 9.80265737e-01 6.52418315e-01
9.12587285e-01 -1.13559699e+00 -5.43119192e-01 5.35004437e-01
-2.29577184e-01 -1.36538267e+00 -7.61970878e-01 4.59267318e-01
-8.37326884e-01 -2.57874161e-01 -1.04957151e+00 -9.73705411e-01
2.19892457e-01 3.56218070e-01 5.40978670e-01 -1.85142353e-01
-1.88677698e-01 -2.52394408e-01 -6.25090718e-01 -3.80483299e-01
-6.11794829e-01 1.99006423e-01 -7.18620792e-02 7.59444982e-02
3.46865237e-01 -5.79913378e-01 -6.78388536e-01 3.04412603e-01
-8.89321387e-01 -2.12919246e-02 3.10913026e-01 9.32501435e-01
4.02155854e-02 2.50469655e-01 6.47608876e-01 -2.05142349e-01
7.29626298e-01 1.34534702e-01 -5.09548783e-01 -8.45253095e-02
-1.97096139e-01 1.60804391e-01 5.75860739e-01 -7.79402196e-01
-1.04942751e+00 -2.74735928e-01 -5.27000844e-01 -8.25422645e-01
8.36935490e-02 3.93511534e-01 9.78691205e-02 3.79925291e-03
4.74363297e-01 3.25150371e-01 -1.40645191e-01 -6.02033377e-01
1.56788453e-01 8.99368405e-01 3.85522932e-01 -4.14154567e-02
6.04093134e-01 1.87658846e-01 -2.95218825e-01 -1.31790257e+00
-2.35487878e-01 -5.08141398e-01 -5.65922618e-01 -2.50429302e-01
9.84214604e-01 -8.19028318e-01 -4.41702932e-01 9.12801504e-01
-1.34906471e+00 -7.53554702e-01 -2.70400196e-01 6.30831301e-01
-2.68063486e-01 1.09314583e-01 -6.39252126e-01 -1.00735319e+00
-4.42515343e-01 -1.25472987e+00 6.33104920e-01 -5.70921786e-02
1.05508395e-01 -9.76361275e-01 -3.21661264e-01 2.37058043e-01
9.46405947e-01 -3.35393310e-01 9.42027390e-01 -5.54799139e-01
-3.29839855e-01 -9.56628285e-03 -1.78926170e-01 1.01197290e+00
5.84264509e-02 -7.01203346e-02 -1.42028892e+00 -4.69907075e-01
1.77603602e-01 1.83589622e-01 1.09917569e+00 8.91046524e-01
1.23347020e+00 -2.71797270e-01 9.18967426e-02 5.72038591e-01
1.02670729e+00 8.68989944e-01 1.04911172e+00 3.02172124e-01
4.38384116e-01 3.10657799e-01 1.47812009e-01 1.39643669e-01
-2.25504011e-01 8.04010749e-01 2.85372317e-01 -4.20789599e-01
-9.09857392e-01 4.30657752e-02 5.34783483e-01 1.00636792e+00
4.87585040e-03 -5.78987956e-01 -9.00132716e-01 8.26042354e-01
-1.59005010e+00 -1.08655167e+00 -4.05371375e-02 1.91572893e+00
5.07325649e-01 2.19011664e-01 -3.59385870e-02 6.13204539e-01
8.35840523e-01 4.05930340e-01 -3.42920512e-01 -1.16712248e+00
-2.71386564e-01 6.19568884e-01 6.00180686e-01 6.80323005e-01
-8.76053751e-01 9.57407713e-01 5.90624571e+00 1.11895585e+00
-1.41471577e+00 3.08972508e-01 7.23816693e-01 -2.91568041e-01
1.29236788e-01 -3.90869617e-01 -5.01977146e-01 2.93123603e-01
1.51501572e+00 -5.76526225e-02 5.45475900e-01 4.50713515e-01
9.91845608e-01 1.97559550e-01 -7.85748124e-01 8.64728212e-01
-1.36914968e-01 -1.24593449e+00 -2.09777012e-01 1.91437677e-01
7.52456605e-01 2.99019247e-01 5.82292616e-01 1.94617912e-01
4.60381918e-02 -1.15310788e+00 9.02357042e-01 -1.94279123e-02
8.77974451e-01 -9.44923282e-01 7.27885008e-01 2.03210786e-01
-1.21905553e+00 -3.93437684e-01 -4.75799829e-01 1.20556187e-02
4.53136787e-02 3.51251155e-01 -1.20548487e+00 2.63707727e-01
9.47274685e-01 3.83903354e-01 -1.90522507e-01 1.28794503e+00
-1.14688883e-02 9.42537606e-01 -1.42762929e-01 3.23748514e-02
7.26316810e-01 4.09410626e-01 6.34312332e-01 1.56337178e+00
6.03271186e-01 -3.70716378e-02 -5.19515932e-01 4.86324638e-01
-1.94715694e-01 -4.06950146e-01 -1.50478512e-01 -2.27052215e-02
5.11500001e-01 6.71250880e-01 -3.45922709e-01 -2.43336082e-01
-3.04269016e-01 8.60619247e-01 -3.79094183e-02 6.00508630e-01
-7.41462350e-01 -5.65362275e-01 8.99143100e-01 8.27937275e-02
1.02279592e+00 -4.95644510e-01 -2.61517793e-01 -5.05943537e-01
-8.96928534e-02 -7.72220373e-01 -1.63397983e-01 -9.10459936e-01
-5.42640865e-01 1.02638960e+00 -2.11128235e-01 -1.09397447e+00
-1.36862740e-01 -8.28645051e-01 -5.32228947e-01 1.33468294e+00
-1.65325737e+00 -1.10963011e+00 1.38506830e-01 4.82300103e-01
1.23071551e+00 -2.91105568e-01 5.69212556e-01 5.94197214e-01
-5.51430643e-01 7.69061744e-01 3.29705745e-01 1.58282697e-01
2.30548978e-01 -1.00000179e+00 8.89772415e-01 1.43530643e+00
-3.25383879e-02 3.91613603e-01 8.08786988e-01 -2.35703990e-01
-8.09925556e-01 -1.21237981e+00 1.15108550e+00 3.26326936e-01
5.13824284e-01 -5.49048781e-01 -8.98317456e-01 4.09267157e-01
6.63230300e-01 1.41762629e-01 1.55685455e-01 -3.65519047e-01
-4.54904974e-01 -2.74373263e-01 -9.58015084e-01 6.08158529e-01
7.85950541e-01 -7.26895988e-01 -3.36593032e-01 -2.60290056e-02
1.18292570e+00 -5.56739628e-01 -4.66694504e-01 2.56863683e-01
3.77429575e-01 -9.13855314e-01 9.57524955e-01 -2.34350592e-01
1.65654734e-01 -4.01140481e-01 -2.32569337e-01 -1.41529119e+00
-1.20113648e-01 -9.04150426e-01 -1.36912271e-01 8.85801017e-01
5.62980473e-01 -3.24036390e-01 6.71481669e-01 8.56960937e-02
-8.01336229e-01 -5.58259308e-01 -1.53713536e+00 -1.10404491e+00
2.75155097e-01 -7.93117762e-01 5.39935708e-01 3.91310424e-01
-4.08235043e-01 3.46795805e-02 -4.28585380e-01 2.86777169e-01
4.94360551e-02 -7.79056728e-01 6.76706582e-02 -5.84320188e-01
-3.51203084e-02 -5.04043221e-01 -5.21062553e-01 -1.44085872e+00
-1.03293560e-01 -6.36730731e-01 6.54535964e-02 -1.53214061e+00
-4.67544287e-01 -1.43917710e-01 -3.32843333e-01 3.36363435e-01
3.12272072e-01 1.12875635e-02 1.64355159e-01 -2.74609238e-01
2.57024318e-01 5.08079171e-01 1.33073282e+00 -2.77810544e-01
-2.48130873e-01 2.28855759e-01 2.18481831e-02 2.16342092e-01
1.00374866e+00 -2.46113330e-01 -7.19558299e-01 -8.27814400e-01
-4.45756227e-01 1.83039367e-01 3.09275478e-01 -1.25797093e+00
5.84262192e-01 4.26133037e-01 2.64752418e-01 -8.80121231e-01
6.96826875e-01 -7.14422822e-01 4.57088836e-02 7.01046228e-01
-5.44461966e-01 9.26457718e-02 5.89580715e-01 2.23053530e-01
-4.57668662e-01 -1.19181223e-01 1.15379179e+00 6.57187551e-02
-6.75775528e-01 1.67259187e-01 -8.65264237e-01 -4.74303514e-01
6.16278410e-01 -4.26334947e-01 -2.10607246e-01 -4.41632122e-01
-8.83516312e-01 -3.40253502e-01 -4.38392192e-01 7.02767313e-01
8.68352354e-01 -1.23976827e+00 -7.13923514e-01 1.82054013e-01
-3.86645555e-01 -2.35871062e-01 4.87781465e-01 6.73221827e-01
-6.77389324e-01 8.83942306e-01 -1.36681991e-02 -4.40329850e-01
-1.60008967e+00 2.80568957e-01 1.00255501e+00 1.31425988e-02
-6.15974188e-01 1.31526506e+00 3.84056196e-03 -9.47811678e-02
6.69157147e-01 -7.32652068e-01 -3.87553833e-02 -2.50573963e-01
5.65122366e-01 5.50914288e-01 6.18057728e-01 -7.24852622e-01
-1.65575817e-01 5.92138283e-02 -3.34446341e-01 -5.55747390e-01
1.37198508e+00 -2.41298676e-02 1.37960851e-01 2.50097394e-01
1.43936443e+00 -6.34018302e-01 -1.35650551e+00 -4.01002109e-01
-8.29781070e-02 -3.91171873e-01 7.61994720e-01 -9.91951108e-01
-1.32361710e+00 1.46085823e+00 9.30944204e-01 2.74416864e-01
1.41065502e+00 -3.90713394e-01 8.36081028e-01 3.06949615e-02
-2.44163543e-01 -1.18333352e+00 2.76798517e-01 8.58859122e-01
1.13790417e+00 -6.93057954e-01 -4.20335501e-01 -9.00507718e-02
-3.77402335e-01 1.43418336e+00 3.68065596e-01 8.54891464e-02
7.30283320e-01 3.66804928e-01 1.64659306e-01 1.33074299e-01
-9.95795369e-01 -1.59743965e-01 1.96838036e-01 6.82921529e-01
5.26157558e-01 4.87791598e-02 -9.28276032e-02 -1.29653901e-01
-4.11280036e-01 -3.82937938e-01 3.68125260e-01 5.59970260e-01
-4.80893612e-01 -1.16836882e+00 -2.44909853e-01 1.09609529e-01
-5.90190351e-01 -4.38392788e-01 -9.90391597e-02 4.75423098e-01
1.23926729e-01 1.31006896e+00 2.64275402e-01 -6.29416406e-01
3.62921745e-01 -2.00985417e-01 2.64540464e-01 -3.62317502e-01
-1.02534735e+00 5.35586715e-01 1.37892365e-01 -6.15788698e-02
-3.50552738e-01 -6.17577672e-01 -9.86261964e-01 -5.26034892e-01
-4.53512013e-01 -2.14101508e-01 1.00342989e+00 9.24208939e-01
1.03361860e-01 1.10162473e+00 6.45995200e-01 -6.39597416e-01
-2.73845166e-01 -1.34522951e+00 -2.69149244e-01 -2.09381342e-01
1.02404165e+00 -1.56685933e-01 -3.20219249e-01 8.62073004e-02] | [14.79050350189209, 5.968505382537842] |
da13f8a0-1ff0-4698-a188-d2eb9020409e | fastaudio-a-learnable-audio-front-end-for | 2109.02774 | null | https://arxiv.org/abs/2109.02774v1 | https://arxiv.org/pdf/2109.02774v1.pdf | FastAudio: A Learnable Audio Front-End for Spoof Speech Detection | Voice assistants, such as smart speakers, have exploded in popularity. It is currently estimated that the smart speaker adoption rate has exceeded 35% in the US adult population. Manufacturers have integrated speaker identification technology, which attempts to determine the identity of the person speaking, to provide personalized services to different members of the same family. Speaker identification can also play an important role in controlling how the smart speaker is used. For example, it is not critical to correctly identify the user when playing music. However, when reading the user's email out loud, it is critical to correctly verify the speaker that making the request is the authorized user. Speaker verification systems, which authenticate the speaker identity, are therefore needed as a gatekeeper to protect against various spoofing attacks that aim to impersonate the enrolled user. This paper compares popular learnable front-ends which learn the representations of audio by joint training with downstream tasks (End-to-End). We categorize the front-ends by defining two generic architectures and then analyze the filtering stages of both types in terms of learning constraints. We propose replacing fixed filterbanks with a learnable layer that can better adapt to anti-spoofing tasks. The proposed FastAudio front-end is then tested with two popular back-ends to measure the performance on the LA track of the ASVspoof 2019 dataset. The FastAudio front-end achieves a relative improvement of 27% when compared with fixed front-ends, outperforming all other learnable front-ends on this task. | ['Douglas C. Schmidt', 'Maria Powell', 'Jules White', 'Zhongwei Teng', 'Quchen Fu'] | 2021-09-06 | null | null | null | null | ['voice-anti-spoofing', 'speaker-identification'] | ['audio', 'speech'] | [ 2.62213141e-01 1.16899811e-01 -4.20826860e-02 -5.56725442e-01
-8.75509381e-01 -7.30688393e-01 2.01226249e-01 -3.26284170e-02
-4.44379777e-01 3.86143565e-01 2.98880816e-01 -4.78821874e-01
4.02233079e-02 -4.88700271e-01 -4.83664811e-01 -6.88564360e-01
-3.10472981e-03 4.62026238e-01 4.93441485e-02 -6.62726834e-02
-2.34616265e-01 4.83422309e-01 -1.51023400e+00 4.75015253e-01
7.23178864e-01 1.12765062e+00 -1.43481478e-01 8.75122666e-01
9.40127075e-02 4.48794663e-01 -9.03518260e-01 -4.76881921e-01
-1.67366728e-01 -1.60079181e-01 -6.90398991e-01 -3.89476091e-01
4.72287714e-01 -5.06499052e-01 -1.69616550e-01 1.02672851e+00
9.78246510e-01 -2.94747859e-01 3.21250319e-01 -1.13401031e+00
2.84560001e-03 1.21651268e+00 -7.17047527e-02 2.74180263e-01
5.96808910e-01 -1.65444747e-01 1.02115226e+00 -3.21100175e-01
-5.60322702e-02 1.47040999e+00 5.94936073e-01 8.53320479e-01
-1.11108160e+00 -1.21055710e+00 7.01550543e-02 2.57268548e-01
-1.33374739e+00 -1.08482873e+00 7.34562874e-01 -3.07559192e-01
5.84722996e-01 4.85117793e-01 8.28876272e-02 1.47615266e+00
-2.46962622e-01 6.70333207e-01 5.53511918e-01 -4.27059591e-01
2.34768525e-01 4.98153627e-01 3.56314898e-01 3.63037169e-01
7.51238167e-02 -1.71441063e-02 -7.85457253e-01 -3.62061143e-01
1.52117521e-01 -3.97092253e-01 -5.05160451e-01 2.27886483e-01
-8.37797105e-01 6.72629118e-01 9.12433565e-02 5.72699010e-01
-2.71087527e-01 -1.10452600e-01 5.14087200e-01 4.18337286e-01
7.50847235e-02 6.24130294e-02 -3.73554736e-01 -2.96563059e-01
-1.04350901e+00 5.52108772e-02 1.11844695e+00 6.29928827e-01
1.72364622e-01 1.27440661e-01 -6.29512966e-02 8.46967161e-01
5.53287864e-01 6.87745869e-01 6.75986648e-01 -4.67053473e-01
6.17987156e-01 1.86842605e-01 6.59572482e-02 -6.73621893e-01
-4.25169408e-01 -7.17095077e-01 -8.73774230e-01 -1.11695550e-01
5.24963379e-01 -5.62269092e-01 -4.57113415e-01 1.75758696e+00
3.73419553e-01 3.92108768e-01 -6.80147558e-02 5.74507773e-01
8.34095001e-01 5.61475217e-01 -1.18933916e-01 -6.74612522e-02
1.68840063e+00 -4.94466245e-01 -8.20402384e-01 -1.77370772e-01
2.57254899e-01 -7.06203878e-01 6.95331097e-01 5.89825034e-01
-7.53826141e-01 -6.72514081e-01 -1.21488786e+00 3.47381979e-01
-2.59281665e-01 1.37139380e-01 9.01088268e-02 1.74940264e+00
-9.32917297e-01 3.86413932e-01 -7.99668849e-01 -1.56537041e-01
2.74044037e-01 7.82470882e-01 -2.78216362e-01 3.94591898e-01
-1.49004734e+00 3.62483114e-01 2.48174462e-02 2.79933531e-02
-8.60914230e-01 -8.57375741e-01 -6.16246939e-01 3.68362516e-01
-1.60823792e-01 -3.96015167e-01 1.37872446e+00 -8.97157788e-01
-1.83752680e+00 8.37143838e-01 -3.12839866e-01 -7.52448559e-01
5.82628310e-01 -3.42017621e-01 -1.05341041e+00 1.28471345e-01
-8.63886252e-02 1.43704757e-01 1.39408100e+00 -7.61562169e-01
-9.32553411e-01 -4.11090285e-01 -1.38632849e-01 -3.55376869e-01
-6.77500725e-01 3.33359003e-01 -9.54018086e-02 -4.46208179e-01
-1.41365658e-02 -9.73039329e-01 5.64808071e-01 -4.43024069e-01
-7.29010880e-01 -1.33135334e-01 1.04363143e+00 -9.80890393e-01
1.28504908e+00 -2.48294377e+00 -2.15087891e-01 3.20548743e-01
-2.57989645e-01 8.00557911e-01 1.41539574e-01 2.77739316e-01
-2.35524416e-01 3.20520326e-02 3.04265041e-02 -6.61440134e-01
7.22690150e-02 -3.12773257e-01 -4.99185175e-01 5.29054582e-01
-5.76878637e-02 1.93924293e-01 -5.13568282e-01 -3.50757837e-02
6.32906258e-02 9.17506456e-01 -5.41299164e-01 3.65514398e-01
3.34833533e-01 5.02629697e-01 -3.85504961e-01 3.18241715e-01
5.81056654e-01 8.24926272e-02 2.52607524e-01 -7.38882124e-02
1.21633209e-01 8.98255944e-01 -1.51676083e+00 1.07216763e+00
-6.15474522e-01 6.29811108e-01 8.38324368e-01 -8.20679128e-01
7.82110870e-01 9.29917097e-01 1.64806440e-01 -1.95696652e-01
3.44131917e-01 3.14016134e-01 2.23545790e-01 -4.01554763e-01
1.20633736e-01 9.97468829e-03 6.30743355e-02 3.58065784e-01
-1.12425983e-01 3.62253726e-01 -4.31778193e-01 4.14583497e-02
1.06999207e+00 -3.89395326e-01 3.74488300e-03 -2.19369695e-01
1.29917538e+00 -8.73731434e-01 3.64588648e-01 6.87103450e-01
-2.97164053e-01 4.32739049e-01 1.81680053e-01 8.06925073e-03
-2.56054550e-01 -1.01502740e+00 -2.20253304e-01 1.29543626e+00
-3.33351195e-01 -2.78671175e-01 -1.05069602e+00 -5.69465756e-01
7.22130761e-02 7.51499832e-01 -2.05988184e-01 -1.92474365e-01
-5.69474459e-01 -2.68541604e-01 1.14442134e+00 1.99280128e-01
6.49392724e-01 -7.91183054e-01 -3.03841889e-01 3.63716841e-01
-4.41305399e-01 -1.24930978e+00 -6.52933836e-01 1.23842187e-01
-4.27997500e-01 -8.07628393e-01 -6.03394687e-01 -7.71502197e-01
1.45930454e-01 6.47097006e-02 5.81184328e-01 -1.35685831e-01
1.20487973e-01 5.11277139e-01 -2.51022190e-01 -5.39365172e-01
-8.86948347e-01 5.55854440e-01 5.44544816e-01 7.95378208e-01
3.84539247e-01 -6.61276758e-01 -3.99556309e-01 4.74217296e-01
-5.32008827e-01 -5.82405806e-01 -2.50125397e-02 4.49019939e-01
-1.24575227e-01 2.05807641e-01 8.30756605e-01 -6.57873511e-01
5.31209946e-01 -2.37885788e-01 -3.83303463e-01 6.02233894e-02
-1.69994697e-01 4.48295996e-02 4.83121455e-01 -5.44186711e-01
-9.03683782e-01 1.24191105e-01 -7.60463178e-01 -3.22740451e-02
-2.47969419e-01 -1.13670424e-01 -6.42204821e-01 8.67031962e-02
3.93075019e-01 2.33861387e-01 -8.80919024e-02 -7.53927588e-01
-9.43050236e-02 1.57331240e+00 5.81463695e-01 -2.14098051e-01
8.11626434e-01 3.65254551e-01 -6.16430879e-01 -1.35345674e+00
-3.30018699e-01 -6.75039649e-01 -2.72494406e-01 -2.48837993e-02
7.54955411e-01 -1.02355313e+00 -1.41821909e+00 9.41761792e-01
-1.23168075e+00 1.18405245e-01 3.58736008e-01 4.77334350e-01
-2.44274855e-01 2.45766103e-01 -5.90194643e-01 -1.01346874e+00
-7.05593169e-01 -1.16665316e+00 1.00044382e+00 9.55537707e-02
-5.35107017e-01 -7.14082003e-01 -2.67411023e-01 6.98347986e-01
5.68690300e-01 -3.53891909e-01 6.60842240e-01 -1.11572647e+00
-6.93005547e-02 -3.80903363e-01 2.48333648e-01 6.53621733e-01
5.04899740e-01 -2.17581421e-01 -1.75191510e+00 -5.97056389e-01
1.29898474e-01 1.25322908e-01 6.13224208e-01 4.77629244e-01
7.70154893e-01 -5.04806757e-01 -3.90453279e-01 5.35638750e-01
6.93340778e-01 2.22487137e-01 5.22980392e-01 -5.27372351e-03
4.62945938e-01 7.38406837e-01 -2.69588325e-02 1.94474131e-01
2.19018415e-01 8.22543859e-01 2.02896968e-01 4.32104796e-01
-7.51053989e-02 -2.61992157e-01 9.03952181e-01 7.37387538e-01
2.15848312e-01 -2.96633571e-01 -5.95407009e-01 1.90707862e-01
-1.21556783e+00 -1.05804133e+00 2.92486101e-01 2.45967078e+00
8.28389466e-01 1.87380522e-01 4.47713375e-01 7.81449199e-01
1.05319917e+00 1.12654202e-01 -4.79052067e-01 -3.56915474e-01
2.20985711e-01 2.91829437e-01 3.15929294e-01 7.37400115e-01
-1.23751545e+00 6.72142565e-01 5.16033602e+00 5.77516973e-01
-1.55535579e+00 2.27051795e-01 4.78130132e-01 -2.51917280e-02
8.51251930e-02 -5.97289085e-01 -1.33736062e+00 6.15898371e-01
1.39547420e+00 1.49666637e-01 2.88600713e-01 7.09714651e-01
3.79545569e-01 3.91607225e-01 -1.36020434e+00 1.29453659e+00
5.27180824e-03 -8.84149253e-01 -4.22397912e-01 -3.99291925e-02
-4.05396447e-02 -6.44530207e-02 1.15146659e-01 2.43054882e-01
-2.20244732e-02 -8.76557767e-01 8.94912362e-01 3.50620858e-02
6.45724833e-01 -8.01355064e-01 5.76879203e-01 3.27637643e-01
-1.16414952e+00 -2.55158961e-01 2.06644550e-01 1.56732574e-01
2.05821276e-01 4.52448100e-01 -1.16860938e+00 1.41768962e-01
8.68704677e-01 1.98928013e-01 -2.32412159e-01 8.67784560e-01
-3.44260514e-01 1.12075162e+00 -4.58519429e-01 2.57903934e-02
-2.95000583e-01 3.68069112e-01 7.91558206e-01 1.21813369e+00
3.34397137e-01 -1.86957702e-01 -3.88041228e-01 4.85151321e-01
-1.69355005e-01 -6.22937158e-02 -1.71527535e-01 -1.77155554e-01
8.48444104e-01 1.04671443e+00 -2.03925386e-01 -6.02269843e-02
-1.43266469e-01 9.91749883e-01 -1.13367788e-01 3.19832176e-01
-6.48158312e-01 -6.51158094e-01 1.19790518e+00 4.59003896e-01
4.10279036e-01 4.33242433e-02 6.59948587e-02 -8.13083827e-01
-4.47145328e-02 -1.35219991e+00 2.68607348e-01 -2.83506423e-01
-8.71565580e-01 6.97330296e-01 -5.05143881e-01 -9.16772485e-01
-5.67047417e-01 -4.57767069e-01 -4.92429882e-01 1.05485320e+00
-1.28408766e+00 -9.20379162e-01 8.37529153e-02 6.06680274e-01
3.56307000e-01 -2.87236303e-01 9.22492743e-01 7.33871102e-01
-5.84764183e-01 1.28534734e+00 -4.40276302e-02 5.20628214e-01
8.15226257e-01 -8.33044231e-01 4.19861376e-01 7.43631542e-01
1.97585449e-01 8.90717983e-01 7.02429235e-01 -3.34321767e-01
-1.29323733e+00 -7.82826543e-01 1.20851088e+00 -1.37125552e-01
3.61075282e-01 -7.40562081e-01 -8.99079621e-01 6.02480829e-01
6.01795502e-02 -3.87633294e-01 7.86296427e-01 1.35850653e-01
-5.16405404e-01 -6.77442431e-01 -1.24059117e+00 2.03741595e-01
6.71637774e-01 -9.77510691e-01 -3.29333425e-01 9.44205746e-03
7.15663433e-01 -1.16400803e-02 -5.94790041e-01 7.56547451e-02
7.99637735e-01 -1.12812114e+00 1.08696604e+00 -4.06475186e-01
-3.28005463e-01 -6.88620880e-02 -6.57454282e-02 -1.13458693e+00
-8.66708755e-02 -1.22459352e+00 -4.98571433e-02 1.91530883e+00
5.42995989e-01 -1.06882882e+00 8.57273161e-01 4.99431878e-01
2.11792424e-01 -1.11515217e-01 -1.20163417e+00 -6.66906714e-01
-2.32348174e-01 -5.68810880e-01 9.85326469e-01 6.88358784e-01
2.06914395e-02 4.24664021e-01 -4.66884941e-01 7.94079065e-01
4.80015308e-01 -6.17318094e-01 7.24251807e-01 -1.32040977e+00
-4.89114612e-01 -3.03096980e-01 -4.38432485e-01 -1.13426399e+00
2.42820963e-01 -7.08766580e-01 -1.71062499e-01 -8.88875604e-01
-5.70454419e-01 -4.42849785e-01 -2.89301366e-01 3.10608417e-01
5.11253849e-02 -1.54049143e-01 5.94123341e-02 -2.35459924e-01
1.54190529e-02 8.21583495e-02 4.64216739e-01 -4.80268002e-01
-5.68112016e-01 8.58981013e-01 -6.64686680e-01 6.07500732e-01
7.86867261e-01 -3.86193752e-01 -2.88984448e-01 -2.74937898e-01
9.61019695e-02 7.58648664e-02 9.92914364e-02 -1.33811903e+00
3.10694724e-01 6.23107255e-01 2.70777140e-02 -2.32024878e-01
4.40149277e-01 -1.02405441e+00 3.27757075e-02 6.64969921e-01
-5.27906120e-01 -3.71864378e-01 4.21779081e-02 3.05874676e-01
-8.41414109e-02 -1.83892459e-01 9.70262289e-01 3.97236049e-01
-2.59012412e-02 9.06247366e-03 -6.36238039e-01 -1.99020460e-01
4.72252101e-01 7.60266907e-04 -4.26671207e-02 -7.64388919e-01
-7.51539826e-01 -3.60477194e-02 -2.04753265e-01 7.52410769e-01
3.01358849e-01 -9.05714452e-01 -8.03518713e-01 6.64616287e-01
-2.83378541e-01 -4.14484710e-01 1.47584394e-01 5.50644934e-01
-1.63318127e-01 5.82382083e-01 2.45578021e-01 -5.15337884e-01
-1.93728340e+00 2.70499796e-01 5.04996181e-01 1.50723636e-01
-2.27049574e-01 1.12279499e+00 -1.97917178e-01 -4.18970197e-01
8.72822106e-01 -4.20011401e-01 -3.75018895e-01 2.48536855e-01
1.20190907e+00 6.28052413e-01 4.18091118e-01 -8.41192126e-01
-7.54689813e-01 2.76093632e-01 -1.45107687e-01 -1.45974964e-01
1.09125304e+00 -1.74379095e-01 9.15177241e-02 3.03611636e-01
1.36877990e+00 3.73719156e-01 -5.26170015e-01 -3.51211458e-01
-6.48261905e-02 -6.71589421e-03 8.88815224e-02 -5.91185629e-01
-1.11454248e+00 1.05059230e+00 1.01357234e+00 4.47088540e-01
9.11971509e-01 -1.23946346e-01 8.68549049e-01 2.41603136e-01
4.43702340e-01 -8.07559609e-01 -2.96101689e-01 2.46819094e-01
7.63364196e-01 -1.12361538e+00 -3.78702313e-01 -5.40043712e-01
-3.37718815e-01 9.03384864e-01 1.59759466e-02 3.43981296e-01
1.02551997e+00 2.06649080e-01 3.03886056e-01 3.30897242e-01
-1.63872451e-01 2.25145996e-01 1.78985909e-01 7.98338175e-01
4.18266088e-01 2.76468754e-01 1.60301760e-01 8.88937354e-01
-6.32322073e-01 -2.57810175e-01 1.53510883e-01 5.37775099e-01
-3.50403130e-01 -1.26070130e+00 -7.91287959e-01 1.95374891e-01
-8.70475173e-01 2.27068923e-02 -3.24249387e-01 5.38079776e-02
1.01410456e-01 1.70012057e+00 -3.32186520e-02 -5.03648400e-01
3.07346463e-01 4.68673170e-01 8.02086070e-02 -4.46123153e-01
-8.91887963e-01 -8.05394128e-02 3.94465029e-01 -2.61408418e-01
-2.21043423e-01 -8.30615401e-01 -1.03280222e+00 -2.82442421e-01
-2.83523440e-01 3.72277141e-01 8.29653382e-01 8.86013448e-01
2.90028095e-01 5.17186046e-01 7.95715451e-01 -6.07673705e-01
-9.48823392e-01 -9.63354230e-01 -7.39609778e-01 7.57618919e-02
9.56599355e-01 -2.39006937e-01 -5.11387527e-01 -1.31125087e-02] | [14.168641090393066, 5.954559803009033] |
81b3fcb2-2368-4d8a-a05a-2bcd0993c79f | a-simple-episodic-linear-probe-improves | null | null | http://openaccess.thecvf.com//content/CVPR2022/html/Liang_A_Simple_Episodic_Linear_Probe_Improves_Visual_Recognition_in_the_CVPR_2022_paper.html | http://openaccess.thecvf.com//content/CVPR2022/papers/Liang_A_Simple_Episodic_Linear_Probe_Improves_Visual_Recognition_in_the_CVPR_2022_paper.pdf | A Simple Episodic Linear Probe Improves Visual Recognition in the Wild | Understanding network generalization and feature discrimination is an open research problem in visual recognition. Many studies have been conducted to assess the quality of feature representations. One of the simple strategies is to utilize a linear probing classifier to quantitatively evaluate the class accuracy under the obtained features. The typical linear probe is only applied as a proxy at the inference time, but its efficacy in measuring features' suitability for linear classification is largely neglected in training. In this paper, we propose an episodic linear probing (ELP) classifier to reflect the generalization of visual representations in an online manner. ELP is trained with detached features from the network and re-initialized episodically. It demonstrates the discriminability of the visual representations in training. Then, an ELP-suitable Regularization term (ELP-SR) is introduced to reflect the distances of probability distributions between ELP classifier and the main classifier. ELP-SR leverages a re-scaling factor to regularize each sample in training, which modulates the loss function adaptively and encourages the features to be discriminative and generalized. We observe significant improvements in three real-world visual recognition tasks, including fine-grained visual classification, long-tailed visual recognition, and generic object recognition. The performance gains show the effectiveness of our method in improving network generalization and feature discrimination. | ['Yi Yang', 'Xiaohan Wang', 'Linchao Zhu', 'Yuanzhi Liang'] | 2022-01-01 | null | null | null | cvpr-2022-1 | ['fine-grained-image-classification'] | ['computer-vision'] | [ 1.69890746e-01 -3.93617004e-01 -5.87636709e-01 -6.54090881e-01
-4.33470577e-01 -5.11183679e-01 5.69864154e-01 -1.07025154e-01
-4.11741227e-01 5.07805228e-01 -1.48895472e-01 -1.68367639e-01
-3.97543758e-01 -6.90040827e-01 -5.98838508e-01 -9.79226589e-01
-2.43818119e-01 4.83949147e-02 2.68791348e-01 2.56545544e-01
3.03186536e-01 9.53948796e-01 -1.59054899e+00 4.86051112e-01
6.66303873e-01 1.70506024e+00 -1.75216086e-02 5.04179060e-01
2.12339330e-02 3.68958771e-01 -4.38451171e-01 -5.29783182e-02
2.69331068e-01 -2.75062919e-01 -7.12038636e-01 1.12520410e-02
4.28632528e-01 8.05677474e-02 -6.18858933e-01 1.03945255e+00
4.36522543e-01 3.74074042e-01 8.76002371e-01 -1.55703211e+00
-8.99353325e-01 3.08818400e-01 -5.31333327e-01 8.01011145e-01
1.54698521e-01 3.22387755e-01 9.79899168e-01 -1.07734358e+00
2.17183024e-01 1.29520988e+00 7.54234493e-01 5.36682844e-01
-1.22524846e+00 -8.43098402e-01 5.26663065e-01 4.24699157e-01
-1.47263539e+00 -3.92010748e-01 6.68332696e-01 -4.32777286e-01
7.64425397e-01 2.80476809e-01 5.88287830e-01 1.23986101e+00
1.80780917e-01 8.47030759e-01 1.19680119e+00 -2.64256001e-01
2.96012193e-01 2.87946582e-01 5.87127626e-01 7.18334198e-01
-6.09764010e-02 4.98831958e-01 -8.04665923e-01 -2.41606370e-01
7.52109408e-01 2.36325070e-01 -3.84229869e-01 -3.27939421e-01
-5.89177430e-01 8.37635040e-01 8.24963629e-01 1.94136336e-01
-5.38463116e-01 -2.98918989e-02 3.56235504e-01 4.70505834e-01
4.50839490e-01 1.32863820e-01 -2.78387725e-01 -1.44086197e-01
-8.15932393e-01 -6.90570399e-02 5.62517703e-01 6.44996405e-01
6.98099494e-01 1.23092897e-01 -7.48077989e-01 1.07127953e+00
3.15936983e-01 2.24074841e-01 7.85112858e-01 -6.21189654e-01
7.89262578e-02 7.75641739e-01 -3.40570211e-01 -1.07843578e+00
-1.87637389e-01 -8.95075679e-01 -8.08860481e-01 2.49060676e-01
2.44749352e-01 1.79164410e-01 -8.89836907e-01 1.78325689e+00
2.06805050e-01 2.98807144e-01 -1.47129253e-01 7.24121571e-01
6.82978094e-01 6.47792041e-01 2.04836473e-01 -1.10846162e-01
1.15978837e+00 -9.21851873e-01 -1.41045004e-01 -2.49656200e-01
1.40218496e-01 -2.20104739e-01 1.26601660e+00 2.85632312e-01
-8.29750538e-01 -7.04786360e-01 -9.94506359e-01 3.68725777e-01
-3.76618594e-01 3.72277498e-02 5.29564261e-01 6.63756013e-01
-8.18296850e-01 8.11230123e-01 -7.66735673e-01 -3.00891846e-01
1.02580214e+00 2.17678025e-01 -3.73031557e-01 -2.03277409e-01
-9.57205415e-01 5.12869895e-01 3.67007196e-01 3.19129288e-01
-1.07651627e+00 -7.42529094e-01 -5.78283608e-01 3.85988742e-01
6.32456020e-02 -4.17217076e-01 7.90807486e-01 -1.27647066e+00
-1.44096005e+00 6.97819769e-01 -5.33082075e-02 -4.56860006e-01
4.47640061e-01 1.58338517e-01 -4.17954892e-01 4.73902561e-02
1.06494874e-02 6.72493577e-01 1.23087347e+00 -1.10306311e+00
-5.61390638e-01 -3.73165518e-01 -1.08974293e-01 2.52722412e-01
-6.37293398e-01 -2.61523932e-01 -3.81124854e-01 -6.98151410e-01
2.13017792e-01 -7.51978755e-01 1.44747198e-01 3.45148861e-01
-3.55699390e-01 -5.26298165e-01 7.82310843e-01 -2.47559190e-01
1.21555281e+00 -2.48641658e+00 -9.01634023e-02 5.61465204e-01
1.21602856e-01 3.17542195e-01 -3.36950541e-01 7.88565576e-02
-5.28669618e-02 -8.99707992e-03 -5.56891188e-02 -9.35552418e-02
-4.09480855e-02 2.95697838e-01 -6.17756844e-01 5.14031351e-01
3.97257566e-01 1.00557220e+00 -6.42641246e-01 -1.57718033e-01
-1.14063300e-01 5.62599778e-01 -5.25316298e-01 2.35854685e-01
1.20154239e-01 2.54978031e-01 -5.88739932e-01 8.59519064e-01
5.58378816e-01 -5.55947423e-01 -1.29431179e-02 -3.38465095e-01
1.13669112e-01 -5.17950952e-02 -9.89351928e-01 1.15311360e+00
-2.39500821e-01 5.93995154e-01 -4.16145921e-01 -1.29149795e+00
1.05828524e+00 -2.56856441e-01 -2.73514315e-02 -7.94681966e-01
1.46687075e-01 -1.02524526e-01 -7.06666112e-02 -4.00759518e-01
1.57099932e-01 2.79669780e-02 1.35670483e-01 2.62422711e-01
7.67838657e-02 6.45471752e-01 -1.65620804e-01 -1.30924746e-01
9.36426163e-01 3.18554193e-02 2.86884695e-01 -2.59889573e-01
4.47740048e-01 -2.95545638e-01 5.26948452e-01 1.19898117e+00
-5.09870946e-01 3.70348454e-01 4.53092933e-01 -3.50060880e-01
-4.74688828e-01 -1.11244762e+00 -6.05525911e-01 1.57771313e+00
8.02842826e-02 5.91221154e-02 -3.55818063e-01 -1.00850725e+00
3.49812776e-01 4.60130721e-01 -1.01098025e+00 -8.14570725e-01
-1.95095107e-01 -8.66502941e-01 6.09633505e-01 9.10558045e-01
6.06983781e-01 -1.18236637e+00 -5.01564622e-01 -1.03220433e-01
2.74706185e-01 -6.39344394e-01 -4.67952371e-01 4.15849209e-01
-8.87638509e-01 -9.87080812e-01 -5.13616621e-01 -7.65004218e-01
7.98840821e-01 2.59933889e-01 7.13477075e-01 1.34948269e-01
-4.15659398e-01 4.70744610e-01 -2.19337940e-01 3.08212545e-02
-8.34616553e-03 3.18729021e-02 1.63532093e-01 4.74717438e-01
4.32128072e-01 -5.82842588e-01 -7.39632308e-01 5.00146568e-01
-6.14572823e-01 -6.29129887e-01 6.99059784e-01 1.27684498e+00
6.93970323e-01 -8.65315199e-02 6.97263598e-01 -7.37424374e-01
7.34219551e-01 -7.06042349e-01 -2.96623796e-01 3.76042753e-01
-7.99321711e-01 1.16562314e-01 7.60154426e-01 -9.91704285e-01
-7.93228567e-01 -3.05495083e-01 3.44800800e-02 -6.65760279e-01
1.63472723e-02 6.37221634e-01 5.74548580e-02 -4.68455017e-01
8.06792796e-01 5.42986214e-01 -1.44161843e-02 -2.22449169e-01
2.85440683e-01 5.52597284e-01 4.03517216e-01 -6.09305978e-01
6.25090003e-01 2.98556298e-01 -1.61583945e-01 -8.35018873e-01
-1.09820056e+00 -2.65594065e-01 -3.15680802e-01 -2.74048209e-01
2.63344914e-01 -6.05648458e-01 -9.87947166e-01 3.98709983e-01
-6.23752832e-01 -3.19884390e-01 -5.17871499e-01 3.35468799e-01
-1.92799270e-01 2.56440401e-01 -3.18916351e-01 -7.19275534e-01
-2.48706535e-01 -9.27862525e-01 5.43790638e-01 6.03170633e-01
6.49843216e-02 -9.17670727e-01 -6.36395067e-02 -1.05773114e-01
5.97389698e-01 -7.69733116e-02 9.37505186e-01 -7.86360681e-01
-2.77924478e-01 -3.07695538e-01 -6.42886817e-01 7.10681379e-01
-2.51426008e-02 -5.51757924e-02 -1.31255364e+00 -7.26426423e-01
-1.50239512e-01 -7.33661771e-01 1.14914358e+00 4.35365111e-01
1.65602589e+00 -3.56248528e-01 -5.59642434e-01 8.53867888e-01
1.19530618e+00 2.67927527e-01 5.06262481e-01 2.32216522e-01
3.43930542e-01 3.07277828e-01 3.91731858e-01 4.58971143e-01
-2.75682434e-02 4.17123049e-01 2.49788851e-01 2.85651773e-01
-8.07142258e-02 -3.70314598e-01 2.90986061e-01 4.20194387e-01
1.48795888e-01 -1.72621623e-01 -6.96978748e-01 2.24870488e-01
-1.60456383e+00 -9.21011150e-01 6.17918730e-01 2.06757140e+00
6.59972191e-01 3.37501556e-01 1.96980461e-02 -4.95644733e-02
5.63399613e-01 2.33051434e-01 -1.12523901e+00 -6.02319650e-02
-1.86377481e-01 1.29473343e-01 1.44260496e-01 3.26059490e-01
-1.03948927e+00 6.25147998e-01 6.59581041e+00 9.68862653e-01
-1.47927916e+00 -3.98483500e-02 7.59509325e-01 -3.45479846e-02
6.50469512e-02 -1.01592928e-01 -9.82732654e-01 5.27339876e-01
8.27352345e-01 -1.00766815e-01 4.62878913e-01 1.02491403e+00
-1.19853005e-01 1.55842632e-01 -1.24639046e+00 1.04283845e+00
6.10562339e-02 -1.27831829e+00 3.55452985e-01 -7.39625618e-02
4.46585208e-01 2.10334528e-02 6.46189094e-01 6.99980795e-01
-5.46211824e-02 -1.15001094e+00 6.74742043e-01 9.06266451e-01
7.49444306e-01 -5.73243618e-01 3.58966559e-01 2.81923920e-01
-1.28570223e+00 -5.74268043e-01 -6.17053270e-01 1.00508057e-01
-4.23641711e-01 3.30637485e-01 -6.63884103e-01 1.14450604e-01
8.44104230e-01 8.10982764e-01 -8.03617120e-01 1.16136563e+00
6.01904951e-02 8.69984508e-01 -2.74286926e-01 -8.95599276e-02
3.54225188e-01 -2.02639345e-02 3.79039943e-01 1.20478547e+00
8.11545774e-02 -6.14789836e-02 4.65690225e-01 8.72510195e-01
-3.18569988e-01 -6.26260340e-02 -2.96328753e-01 -1.20253041e-01
8.20142686e-01 1.24786639e+00 -6.10015750e-01 -1.59232900e-01
-1.80664808e-01 8.47721636e-01 8.46636713e-01 4.46721494e-01
-7.64647603e-01 -4.31761384e-01 5.13762951e-01 1.20334122e-02
4.93850738e-01 1.40668094e-01 -9.96335000e-02 -9.12143052e-01
8.61826017e-02 -7.26151168e-01 6.13194406e-01 -4.76996452e-01
-1.73379672e+00 8.51503611e-01 -8.71905759e-02 -1.23895955e+00
3.23390663e-02 -6.66908979e-01 -7.87207723e-01 8.58411729e-01
-1.56007087e+00 -8.39921474e-01 -5.57058036e-01 7.45428205e-01
3.90719891e-01 -4.25242841e-01 6.22477949e-01 2.58406848e-01
-9.82981145e-01 1.19312739e+00 2.39784122e-01 6.12489916e-02
4.16076303e-01 -8.62900853e-01 -1.11801341e-01 5.93756735e-01
1.89254090e-01 6.39240623e-01 3.51272821e-01 -2.77176917e-01
-1.13976622e+00 -1.19213593e+00 2.57433176e-01 -1.72679737e-01
5.78086317e-01 -1.40795469e-01 -1.34281588e+00 6.40722752e-01
-2.10933343e-01 6.76958740e-01 7.20332801e-01 2.47047339e-02
-8.32509160e-01 -4.74699199e-01 -1.18119550e+00 2.74646491e-01
1.13645601e+00 -7.20978916e-01 -4.91579086e-01 1.59381405e-01
2.96887785e-01 -5.09672984e-02 -8.52731705e-01 3.14911008e-01
7.34483421e-01 -8.07673037e-01 1.00400972e+00 -1.02471423e+00
-8.86210892e-03 -8.33850130e-02 -2.92125911e-01 -1.24442804e+00
-7.76737928e-01 -1.25866368e-01 -2.16954559e-01 1.16603982e+00
2.80315369e-01 -1.00287497e+00 8.94144773e-01 2.19042450e-01
1.07727036e-01 -1.16943741e+00 -1.07639575e+00 -9.32878435e-01
1.80765651e-02 -3.27206492e-01 1.23779856e-01 8.46945345e-01
-2.14016542e-01 2.10927531e-01 -2.89993677e-02 2.88829714e-01
5.85324049e-01 2.66644031e-01 2.56280750e-01 -1.36860275e+00
-4.28348720e-01 -6.99664712e-01 -6.41212642e-01 -1.33245146e+00
4.10053253e-01 -1.31836236e+00 -6.91576153e-02 -9.30404544e-01
3.83433342e-01 -6.41778409e-01 -7.85624802e-01 6.75011456e-01
-8.18143561e-02 2.75005341e-01 -3.15709077e-02 4.92414117e-01
-6.05835319e-01 8.18414211e-01 9.11165118e-01 -2.24045858e-01
-1.60818115e-01 2.42922693e-01 -7.36527503e-01 5.68112731e-01
6.25456095e-01 -6.04081571e-01 -4.08661991e-01 -1.85197845e-04
-1.99161932e-01 -2.58019596e-01 7.27582633e-01 -9.31948364e-01
4.29274440e-01 -4.37378176e-02 8.81124735e-01 -3.13151896e-01
2.59887218e-01 -5.24550140e-01 -2.31606588e-01 5.06695867e-01
-6.22202277e-01 -5.20498306e-03 2.36035317e-01 9.35192108e-01
-1.93994984e-01 -2.56336182e-01 1.16136169e+00 1.91859365e-01
-7.84592927e-01 7.56385148e-01 -6.23807348e-02 2.97528714e-01
9.73562777e-01 -3.41487169e-01 -3.66862595e-01 6.41167089e-02
-7.93713331e-01 2.19271764e-01 3.05693597e-02 3.98208320e-01
9.94556308e-01 -1.55726326e+00 -5.05233526e-01 5.51193893e-01
3.31007630e-01 -4.51170743e-01 3.23712170e-01 7.71304965e-01
-1.92038342e-03 1.26638755e-01 -3.54250044e-01 -9.52225983e-01
-1.01815283e+00 5.34147620e-01 6.40164554e-01 -6.52327761e-02
-6.43948019e-01 1.08482313e+00 1.49491817e-01 -1.91081017e-01
6.68706536e-01 -2.19135389e-01 -2.14854732e-01 1.68695733e-01
6.90049827e-01 3.80377501e-01 -1.89485654e-01 -3.34123045e-01
-4.31983382e-01 4.76342171e-01 -4.63311464e-01 3.28024656e-01
1.29834974e+00 -6.58595264e-02 1.19105130e-01 6.28679216e-01
1.40097046e+00 -4.97265488e-01 -1.75352561e+00 -6.17042780e-01
9.35974251e-03 -5.38211703e-01 7.59863406e-02 -8.01817477e-01
-1.16363215e+00 7.75286734e-01 1.03445232e+00 3.85830730e-01
1.01335251e+00 3.45424823e-02 2.14419872e-01 5.43745518e-01
1.28315851e-01 -8.35704684e-01 3.70623916e-01 5.28507233e-01
1.03258693e+00 -1.18135333e+00 -2.04727948e-01 -9.57417712e-02
-6.26639366e-01 1.17947102e+00 6.48022175e-01 -3.34609509e-01
9.12261367e-01 6.19199462e-02 -2.13019356e-01 -1.34071991e-01
-8.03011298e-01 1.72775030e-01 5.62465191e-01 5.29639184e-01
-1.26360646e-02 -5.15480898e-02 1.95883870e-01 7.19850540e-01
1.60402507e-01 -8.79305005e-02 -3.62451039e-02 8.39283705e-01
-5.79240441e-01 -5.44435918e-01 -4.30854075e-02 7.97924519e-01
-2.74216291e-02 -2.51432993e-02 -1.16223626e-01 5.01240194e-01
-1.57891646e-01 5.50014973e-01 1.91314727e-01 -3.94169867e-01
4.70854133e-01 2.53002524e-01 4.54755872e-01 -3.13746631e-01
-5.33888817e-01 -4.76946354e-01 -4.31630790e-01 -7.11331189e-01
-1.80935904e-01 -4.84855920e-01 -1.07710528e+00 -1.16891108e-01
-3.57791871e-01 4.68295105e-02 2.17914820e-01 8.14412832e-01
5.31673729e-01 5.90350688e-01 1.07414198e+00 -8.39414358e-01
-1.23708534e+00 -7.05767751e-01 -6.75248563e-01 4.28416997e-01
3.62065136e-01 -1.07459855e+00 -6.36612594e-01 -4.59789157e-01] | [9.616520881652832, 2.661231756210327] |
4f98e676-c905-474b-95b5-9889faa839a1 | analyzing-the-generalizability-of-deep | 2303.12936 | null | https://arxiv.org/abs/2303.12936v1 | https://arxiv.org/pdf/2303.12936v1.pdf | Analyzing the Generalizability of Deep Contextualized Language Representations For Text Classification | This study evaluates the robustness of two state-of-the-art deep contextual language representations, ELMo and DistilBERT, on supervised learning of binary protest news classification and sentiment analysis of product reviews. A "cross-context" setting is enabled using test sets that are distinct from the training data. Specifically, in the news classification task, the models are developed on local news from India and tested on the local news from China. In the sentiment analysis task, the models are trained on movie reviews and tested on customer reviews. This comparison is aimed at exploring the limits of the representative power of today's Natural Language Processing systems on the path to the systems that are generalizable to real-life scenarios. The models are fine-tuned and fed into a Feed-Forward Neural Network and a Bidirectional Long Short Term Memory network. Multinomial Naive Bayes and Linear Support Vector Machine are used as traditional baselines. The results show that, in binary text classification, DistilBERT is significantly better than ELMo on generalizing to the cross-context setting. ELMo is observed to be significantly more robust to the cross-context test data than both baselines. On the other hand, the baselines performed comparably well to ELMo when the training and test data are subsets of the same corpus (no cross-context). DistilBERT is also found to be 30% smaller and 83% faster than ELMo. The results suggest that DistilBERT can transfer generic semantic knowledge to other domains better than ELMo. DistilBERT is also favorable in incorporating into real-life systems for it requires a smaller computational training budget. When generalization is not the utmost preference and test domain is similar to the training domain, the traditional ML algorithms can still be considered as more economic alternatives to deep language representations. | ['Berfu Buyukoz'] | 2023-03-22 | null | null | null | null | ['news-classification'] | ['natural-language-processing'] | [-6.06617443e-02 -7.13081509e-02 -5.95586240e-01 -7.48238862e-01
-7.21101165e-01 -6.73192918e-01 8.08131158e-01 3.32600683e-01
-6.21464312e-01 7.51331806e-01 1.81384027e-01 -7.14909613e-01
4.77047339e-02 -8.12707543e-01 -6.78946733e-01 -3.81133288e-01
1.06122367e-01 5.74636340e-01 -1.50819704e-01 -6.48058236e-01
1.52031377e-01 2.02876255e-01 -1.40403831e+00 7.89606869e-01
4.65977252e-01 1.06342542e+00 4.56789210e-02 5.20410061e-01
-1.74623773e-01 9.64798927e-01 -5.62591791e-01 -6.53332710e-01
1.50343791e-01 -2.45315984e-01 -8.38403165e-01 -7.18755573e-02
4.07246172e-01 8.18093941e-02 2.71431208e-01 9.05539155e-01
2.54265636e-01 1.67493060e-01 6.95676625e-01 -7.71378636e-01
-1.12079394e+00 7.69446611e-01 -3.88979316e-01 4.07990664e-01
3.09136629e-01 -3.47381562e-01 1.20310271e+00 -8.10903192e-01
7.50743985e-01 1.36718869e+00 8.97930086e-01 3.36886048e-01
-1.06831360e+00 -5.36848366e-01 4.95241851e-01 -3.08326520e-02
-8.42640162e-01 -1.39342889e-01 5.28634846e-01 -4.35917109e-01
1.32412422e+00 4.47662659e-02 2.18471333e-01 1.35111499e+00
7.21777320e-01 6.73616230e-01 1.27341080e+00 -6.39712393e-01
3.46646488e-01 8.90010357e-01 4.83580738e-01 4.10594523e-01
3.64675105e-01 6.84873248e-03 -3.28561157e-01 -2.24354371e-01
1.29098296e-01 -1.76451504e-01 8.26439559e-02 -1.31308436e-02
-7.09687531e-01 1.46924257e+00 5.35190463e-01 5.75675309e-01
-3.35898370e-01 -1.54629931e-01 8.46105099e-01 6.26193166e-01
9.95744288e-01 7.64034569e-01 -9.49411392e-01 1.17077321e-01
-8.94149303e-01 1.47024184e-01 1.13199210e+00 5.65040231e-01
4.08157736e-01 8.49746466e-02 3.03806096e-01 9.60576415e-01
3.63846153e-01 4.56664413e-01 1.06276321e+00 -1.77824840e-01
5.45448601e-01 6.54269099e-01 -1.35413125e-01 -1.16175199e+00
-5.81427395e-01 -8.12305927e-01 -2.57903934e-01 6.66296631e-02
2.04616219e-01 -3.65452021e-01 -8.62405002e-01 1.40900981e+00
6.04656078e-02 -2.33481288e-01 7.14178979e-01 6.53868914e-01
8.34636807e-01 9.61827874e-01 2.40616530e-01 5.07586859e-02
1.56165588e+00 -1.04857409e+00 -4.51356977e-01 -8.67735803e-01
1.06092083e+00 -1.02588642e+00 1.23734677e+00 4.37523276e-01
-5.91667116e-01 -5.98171949e-01 -1.43294775e+00 -1.19103014e-01
-1.05784011e+00 2.60653883e-01 7.65187204e-01 8.14901114e-01
-7.42942333e-01 3.72831672e-01 -6.09256387e-01 -5.44512689e-01
5.57703637e-02 1.69344440e-01 -3.17914933e-01 -1.73612252e-01
-1.52832496e+00 1.26962829e+00 3.82433593e-01 9.37149078e-02
-5.69592357e-01 -3.29161257e-01 -1.05326760e+00 -6.67155981e-02
3.92043851e-02 -2.11605191e-01 1.35254800e+00 -1.67796397e+00
-1.47973871e+00 9.37040329e-01 8.41928348e-02 -6.83810830e-01
2.78399736e-01 -3.61796409e-01 -9.23800886e-01 -1.57867178e-01
3.70248795e-01 2.37765968e-01 4.04762745e-01 -1.14056802e+00
-6.51418507e-01 -1.88914925e-01 1.91301093e-01 2.22425923e-01
-2.92925775e-01 2.77703017e-01 -2.61721686e-02 -5.45633376e-01
-1.76128104e-01 -1.06192136e+00 -1.61740452e-01 -6.63993537e-01
-2.38713264e-01 -1.39452502e-01 8.45020056e-01 -5.45462370e-01
9.83477294e-01 -2.16405225e+00 -3.72600317e-01 4.50005708e-03
-5.37367582e-01 2.02751741e-01 -1.30662009e-01 3.80535960e-01
-2.87339658e-01 7.10318238e-02 -2.83254340e-05 -3.03127527e-01
-7.14364350e-02 3.15875858e-01 -5.72354734e-01 4.40310746e-01
2.34035626e-01 8.12446654e-01 -6.58485830e-01 -1.14561252e-01
-3.52285281e-02 4.03023452e-01 -3.76558900e-01 -3.23122084e-01
-2.76111603e-01 -6.94736987e-02 -2.32402518e-01 5.14056921e-01
2.78881878e-01 -3.77257228e-01 4.68359470e-01 1.89668044e-01
7.40181357e-02 6.92789376e-01 -8.40935528e-01 1.21778095e+00
-8.82916868e-01 8.02043259e-01 -2.49059349e-01 -1.21616542e+00
1.12074769e+00 1.65834665e-01 -2.51352847e-01 -9.21255291e-01
3.34496498e-01 3.33887428e-01 4.20862734e-02 -3.76470983e-01
6.38395488e-01 -5.30824900e-01 -4.79904890e-01 3.60550255e-01
1.05690695e-01 -7.95285776e-02 2.49303892e-01 -4.69484776e-02
5.22155344e-01 2.03091994e-01 5.20642281e-01 -5.93583524e-01
4.36107308e-01 3.03925604e-01 4.50737476e-01 6.22701108e-01
1.38701394e-01 2.60804147e-01 5.71218312e-01 -5.44281185e-01
-7.67487884e-01 -7.43432105e-01 -5.74769080e-01 1.57646549e+00
3.61464471e-02 -2.87688345e-01 -4.97837096e-01 -1.06526101e+00
7.54723772e-02 1.15793002e+00 -8.32761765e-01 4.21099551e-03
-2.63390094e-01 -1.19908679e+00 3.42182517e-01 5.14941454e-01
3.73773873e-01 -1.06042325e+00 -4.31094438e-01 2.71076143e-01
2.19172210e-01 -1.15025556e+00 1.12287529e-01 5.42187691e-01
-8.75414550e-01 -8.15084875e-01 -3.88814688e-01 -9.80498612e-01
3.27188909e-01 -1.26117542e-01 1.26901841e+00 -4.33324486e-01
1.43011689e-01 1.57028764e-01 -6.36793733e-01 -6.40121818e-01
-6.85236573e-01 2.50029296e-01 1.44220442e-02 -1.19062409e-01
9.17395115e-01 -1.11965492e-01 -3.38189185e-01 3.34421396e-01
-9.19235170e-01 -2.98124880e-01 5.90189457e-01 1.04734790e+00
3.85937840e-01 1.69329837e-01 9.86783624e-01 -1.46605539e+00
7.55220592e-01 -9.94369805e-01 -3.53036046e-01 1.11787893e-01
-9.37782586e-01 -1.61228985e-01 8.78139734e-01 -4.82430190e-01
-1.17884409e+00 -4.35873002e-01 -1.86325908e-01 2.72343397e-01
-1.61661699e-01 1.14859676e+00 2.21477345e-01 3.69261354e-01
1.07916832e+00 -2.42436945e-01 -1.76051751e-01 -4.33334976e-01
2.01152578e-01 8.65631461e-01 -1.68556832e-02 -4.60946947e-01
-1.92345679e-02 3.76228303e-01 -5.45829713e-01 -8.19163203e-01
-1.11978245e+00 -3.00189823e-01 -3.65046322e-01 2.29299799e-01
8.11880291e-01 -1.16032267e+00 1.77192762e-02 1.72475561e-01
-6.67531371e-01 -4.22236353e-01 -1.82286933e-01 6.63579404e-01
-2.77610958e-01 -2.41708513e-02 -9.64946866e-01 -5.59011161e-01
-2.25981876e-01 -1.15454924e+00 9.17245686e-01 1.34656891e-01
-4.40954745e-01 -1.58282959e+00 2.04648115e-02 4.06282008e-01
3.05976212e-01 1.40209258e-01 9.82755899e-01 -1.48709798e+00
3.37334216e-01 -6.56901300e-01 -9.20076892e-02 8.48711252e-01
1.04125381e-01 -2.93290410e-02 -1.16443157e+00 -3.80364984e-01
3.35159570e-01 -5.33316970e-01 8.80313396e-01 3.07918727e-01
4.10278499e-01 -2.34099030e-01 -2.11659133e-01 1.46805704e-01
1.45012975e+00 3.71183306e-01 2.71921813e-01 7.87127435e-01
2.53754884e-01 7.50357687e-01 7.32734919e-01 -5.63980676e-02
4.78844970e-01 3.53473395e-01 5.91908768e-02 -1.74477369e-01
7.03804120e-02 -1.20248914e-01 9.25686955e-01 9.06572223e-01
5.98049700e-01 -3.22260827e-01 -1.04162276e+00 6.08496070e-01
-1.63484466e+00 -7.85764515e-01 1.29084708e-02 1.78887141e+00
7.31965721e-01 6.46646559e-01 -1.74712256e-01 -4.51462679e-02
6.05540752e-01 1.76282287e-01 -3.95554334e-01 -1.10999560e+00
-3.28611225e-01 2.22722173e-01 4.13548857e-01 5.34423351e-01
-1.39834857e+00 9.69944954e-01 6.19666052e+00 6.82751894e-01
-1.48724902e+00 3.62726301e-01 1.07612550e+00 -1.26044497e-01
-3.89401168e-02 1.10475319e-02 -1.16284251e+00 4.76446271e-01
1.54575300e+00 2.00176254e-01 -9.17349905e-02 1.32452965e+00
-1.29471928e-01 -3.02904278e-01 -1.14593434e+00 4.80447084e-01
4.67785567e-01 -1.30283773e+00 -8.82353261e-02 -1.73872590e-01
1.00296700e+00 4.40406114e-01 2.38107473e-01 8.26448679e-01
5.70057988e-01 -1.10714948e+00 6.89198613e-01 5.02386056e-02
5.05477011e-01 -8.02278042e-01 1.34894240e+00 2.51405656e-01
-5.81296980e-01 -3.50264966e-01 -4.76775199e-01 -3.37960154e-01
7.45429695e-02 6.20648503e-01 -9.37213302e-01 5.28378069e-01
7.96549499e-01 8.85651946e-01 -6.72885776e-01 2.24112988e-01
-1.50837258e-01 8.01785648e-01 -1.46902934e-01 -4.13757324e-01
7.44666040e-01 8.54164436e-02 1.41653493e-01 1.68204117e+00
4.99541499e-02 -3.62882227e-01 3.07128519e-01 3.77851754e-01
1.37892380e-01 5.51588774e-01 -7.86255777e-01 -1.60172433e-01
-1.79151595e-02 1.12745607e+00 -8.83112431e-01 -4.47302759e-01
-7.34740376e-01 6.86885655e-01 2.21559495e-01 4.08129960e-01
-6.88162386e-01 -3.30861270e-01 4.26900625e-01 -3.64369266e-02
3.60138267e-01 1.32262543e-01 -5.00886738e-01 -1.17617571e+00
-1.64124444e-01 -9.60038126e-01 5.98396182e-01 -7.02221155e-01
-1.72597802e+00 1.08027649e+00 -5.05709834e-02 -9.07198727e-01
-3.55208844e-01 -1.26542604e+00 -4.49878275e-01 8.27809632e-01
-1.65029931e+00 -1.11155558e+00 4.01005238e-01 3.30445498e-01
7.30337560e-01 -5.20601153e-01 9.96372998e-01 2.08815739e-01
-4.55130577e-01 5.64778805e-01 4.17325407e-01 2.60821044e-01
7.58919656e-01 -9.68369246e-01 2.13486254e-01 7.51796126e-01
2.50928938e-01 8.00464869e-01 6.47200465e-01 -7.23631978e-01
-7.97620475e-01 -1.08861983e+00 1.05942202e+00 -5.32520771e-01
1.02654123e+00 -5.98869920e-01 -7.40356088e-01 1.01857793e+00
3.60272378e-01 -2.02000573e-01 1.10616636e+00 5.05217731e-01
-4.79149550e-01 -8.81436989e-02 -1.00694549e+00 3.10997367e-01
2.04693109e-01 -8.13578248e-01 -1.01036716e+00 5.95889568e-01
6.52720809e-01 -1.53307632e-01 -7.06956863e-01 3.36965680e-01
4.70853448e-01 -6.87011540e-01 5.27675331e-01 -9.02356863e-01
8.12730670e-01 -1.41275814e-03 -6.70513451e-01 -1.29240274e+00
-1.35975897e-01 1.51074762e-02 4.45211321e-01 1.20051432e+00
1.16298115e+00 -1.05303800e+00 5.19970894e-01 2.08841071e-01
1.76361036e-02 -9.35174644e-01 -8.51466417e-01 -7.40351319e-01
5.88735461e-01 -6.58115268e-01 2.68664479e-01 1.31769431e+00
3.89810093e-02 7.96489060e-01 -6.37149811e-03 6.66299909e-02
-1.44379094e-01 3.34399909e-01 4.42549080e-01 -1.09492493e+00
-5.24500608e-01 -3.99035424e-01 -4.19096351e-01 -7.05848515e-01
4.91497546e-01 -1.15770376e+00 -6.65912554e-02 -1.34899127e+00
-7.69836828e-02 -4.75405753e-01 -5.06927550e-01 4.75002140e-01
6.94172159e-02 3.40923786e-01 1.03136403e-02 -7.44820088e-02
-1.23009585e-01 2.36276418e-01 8.37870419e-01 -1.49669915e-01
-2.18454689e-01 7.49060363e-02 -1.10451651e+00 9.08259392e-01
9.01989520e-01 -4.65084285e-01 -5.38455188e-01 -2.85755813e-01
6.53972745e-01 -3.17997605e-01 5.34483194e-02 -6.23922706e-01
-7.82347694e-02 1.31357625e-01 4.86044675e-01 -1.74773380e-01
2.57993996e-01 -6.85197413e-01 -2.61018068e-01 2.49861687e-01
-5.93245149e-01 3.28240663e-01 5.37054956e-01 4.65409607e-01
-4.12346452e-01 -5.27103066e-01 7.38324702e-01 -1.93053275e-01
-7.89813817e-01 -4.42395657e-01 -5.60703218e-01 2.68156808e-02
1.03210163e+00 -1.69069543e-02 -5.74924946e-01 -3.13057303e-01
-8.84487629e-01 -2.25021075e-02 2.30029359e-01 7.51774311e-01
2.10195303e-01 -9.54624474e-01 -6.79530799e-01 2.94772834e-01
1.70961916e-01 -4.20342386e-01 2.91761216e-02 7.18001008e-01
-6.32849216e-01 7.04058349e-01 1.76458180e-01 -5.00000715e-01
-9.48833644e-01 6.60174966e-01 3.08614731e-01 -4.13259596e-01
-3.15373212e-01 8.99892211e-01 2.90238738e-01 -7.48309910e-01
-1.43358767e-01 -5.15285492e-01 -3.40853602e-01 4.54451293e-01
4.18943882e-01 -9.00578424e-02 3.38560939e-01 -7.13138282e-01
-3.61812592e-01 2.75210351e-01 -5.11947513e-01 -3.49717215e-02
1.32970941e+00 1.37321865e-02 -2.59078983e-02 1.04114354e+00
1.24818420e+00 1.89120710e-01 -7.35667944e-01 -2.14860424e-01
3.00966054e-01 2.54599005e-02 2.04601660e-01 -1.25027156e+00
-8.54307234e-01 6.57578230e-01 4.70347673e-01 5.06292462e-01
8.44875574e-01 6.94273561e-02 3.91753614e-01 4.45133001e-01
5.82811646e-02 -1.31409478e+00 -2.63801336e-01 6.90652430e-01
8.46847594e-01 -1.35694146e+00 1.73357949e-01 -5.16555794e-02
-1.12954986e+00 1.03351450e+00 5.24785042e-01 -5.82187951e-01
9.71810102e-01 3.13803107e-01 4.95657355e-01 -2.12187633e-01
-8.70221674e-01 2.22779319e-01 2.26198062e-01 3.11506111e-02
7.70386517e-01 9.06324014e-02 -3.90624821e-01 8.60324979e-01
-5.29489160e-01 -3.22171062e-01 5.63164294e-01 9.16873991e-01
-1.90825582e-01 -8.97361934e-01 -2.13344693e-01 3.94441247e-01
-8.49801838e-01 -3.32761765e-01 -3.20446700e-01 1.10217524e+00
1.34087637e-01 1.18117857e+00 1.69099450e-01 -1.76396027e-01
2.74753362e-01 4.79537249e-01 -1.48582980e-01 -9.13635671e-01
-9.41544533e-01 -4.92931008e-02 5.70200741e-01 -2.37456948e-01
-5.08592844e-01 -7.06905007e-01 -1.06318831e+00 -3.90819758e-02
-4.88296211e-01 2.83923477e-01 1.06576800e+00 1.13883364e+00
4.43001807e-01 5.37914813e-01 4.74458545e-01 -5.52444518e-01
-6.44127250e-01 -1.09595871e+00 -5.75162053e-01 3.80305380e-01
2.92152345e-01 -6.32117033e-01 -4.35003340e-01 -9.29721966e-02] | [11.071942329406738, 7.205007553100586] |
d76dd62d-0ec5-46b3-8e74-188ffd715277 | self-supervised-robustifying-guidance-for | 2112.14382 | null | https://arxiv.org/abs/2112.14382v3 | https://arxiv.org/pdf/2112.14382v3.pdf | Self-Supervised Robustifying Guidance for Monocular 3D Face Reconstruction | Despite the recent developments in 3D Face Reconstruction from occluded and noisy face images, the performance is still unsatisfactory. Moreover, most existing methods rely on additional dependencies, posing numerous constraints over the training procedure. Therefore, we propose a Self-Supervised RObustifying GUidancE (ROGUE) framework to obtain robustness against occlusions and noise in the face images. The proposed network contains 1) the Guidance Pipeline to obtain the 3D face coefficients for the clean faces and 2) the Robustification Pipeline to acquire the consistency between the estimated coefficients for occluded or noisy images and the clean counterpart. The proposed image- and feature-level loss functions aid the ROGUE learning process without posing additional dependencies. To facilitate model evaluation, we propose two challenging occlusion face datasets, ReaChOcc and SynChOcc, containing real-world and synthetic occlusion-based face images for robustness evaluation. Also, a noisy variant of the test dataset of CelebA is produced for evaluation. Our method outperforms the current state-of-the-art method by large margins (e.g., for the perceptual errors, a reduction of 23.8% for real-world occlusions, 26.4% for synthetic occlusions, and 22.7% for noisy images), demonstrating the effectiveness of the proposed approach. The occlusion datasets and the corresponding evaluation code are released publicly at https://github.com/ArcTrinity9/Datasets-ReaChOcc-and-SynChOcc. | ['Yong-Sheng Chen', 'K. S. Venkatesh', 'Kevin Jou', 'Hung-Jen Chen', 'Hsien-Kai Kuo', 'Yi-Min Tsai', 'Min-Hung Chen', 'Hitika Tiwari'] | 2021-12-29 | null | null | null | null | ['3d-face-reconstruction', 'face-reconstruction'] | ['computer-vision', 'computer-vision'] | [-6.81896433e-02 8.69126245e-03 3.89405131e-01 -7.35297978e-01
-9.76449072e-01 -3.83212388e-01 5.11162460e-01 -5.29698014e-01
1.04729859e-02 4.58629519e-01 -2.94748023e-02 9.61004049e-02
-1.82293862e-01 -4.27613944e-01 -7.62907386e-01 -9.66257036e-01
-8.15729871e-02 1.97096676e-01 -4.80750322e-01 -3.27378437e-02
-2.18450697e-03 8.14663351e-01 -1.84060776e+00 8.94236192e-02
8.57987702e-01 1.27042520e+00 -6.85593858e-02 2.89359778e-01
3.29164207e-01 1.33449703e-01 -7.22747803e-01 -8.23354661e-01
4.90687817e-01 -1.80807680e-01 -2.44216561e-01 3.36476982e-01
8.32311451e-01 -3.41478258e-01 -2.81977355e-01 1.10200834e+00
9.31746006e-01 -1.61364805e-02 4.67000663e-01 -1.37589955e+00
-5.97182691e-01 5.47539927e-02 -6.57360196e-01 -2.06708312e-01
4.86062646e-01 4.30785656e-01 4.50251281e-01 -1.29381835e+00
7.12278008e-01 1.65122449e+00 6.96086824e-01 5.89217722e-01
-1.28461504e+00 -8.64684343e-01 -2.23065913e-02 1.80733055e-01
-1.52412319e+00 -1.29194772e+00 7.84666657e-01 -2.88222998e-01
4.62337911e-01 2.59586304e-01 8.54710788e-02 1.46964371e+00
3.46872211e-02 3.63619685e-01 1.24322474e+00 -1.67620897e-01
1.50624514e-01 2.07312837e-01 -9.80642065e-02 5.44709384e-01
2.09242985e-01 4.16113466e-01 -4.48061645e-01 -1.02071166e-01
4.43386227e-01 -4.47358727e-01 -3.03187132e-01 -2.63466567e-01
-4.10619676e-01 4.86773968e-01 2.73897022e-01 -7.31793866e-02
-2.00053170e-01 -2.31939778e-01 4.35872614e-01 4.13711071e-01
6.83721960e-01 1.38093261e-02 -4.25255239e-01 2.00282902e-01
-7.40115523e-01 2.81500220e-01 6.65117204e-01 9.84331369e-01
6.87192798e-01 3.63412976e-01 -1.61972478e-01 1.14499867e+00
5.99846005e-01 8.05574715e-01 -4.67583090e-02 -9.37254608e-01
5.98754585e-01 2.30522618e-01 5.37282117e-02 -1.28545189e+00
-3.23101342e-01 -5.55074394e-01 -9.00602996e-01 3.20629060e-01
4.49725896e-01 -2.57675108e-02 -9.96937811e-01 2.00137663e+00
6.80576444e-01 3.03213060e-01 1.76883433e-02 8.55051219e-01
1.11415303e+00 4.14223820e-01 -1.13561653e-01 -3.94308567e-01
1.29542041e+00 -5.82128584e-01 -8.99163306e-01 -7.94908702e-02
1.02026708e-01 -1.12302387e+00 1.13388908e+00 3.94658893e-01
-1.03013992e+00 -8.39646399e-01 -8.77998650e-01 1.67629644e-01
-7.10157603e-02 4.81805146e-01 2.92670578e-01 9.66815114e-01
-1.00402856e+00 4.59198713e-01 -5.87594092e-01 -6.49928376e-02
8.32106948e-01 4.10057127e-01 -8.78788412e-01 -4.01565909e-01
-8.31975281e-01 7.17242718e-01 -1.27357945e-01 6.48566186e-01
-1.22599792e+00 -7.60882556e-01 -1.10109150e+00 -4.98417020e-02
3.58624101e-01 -1.07027598e-01 7.17766166e-01 -6.59169197e-01
-1.55131614e+00 9.31042135e-01 -9.49566811e-02 1.69818569e-02
7.33023465e-01 -4.94523942e-01 -5.96261799e-01 5.41546531e-02
3.15014943e-02 4.95009303e-01 1.41445780e+00 -1.55990243e+00
1.34736029e-02 -5.97802937e-01 -1.68538153e-01 -1.68136671e-01
-1.96718737e-01 1.99148625e-01 -8.43664169e-01 -7.82244682e-01
1.84266523e-01 -9.07265246e-01 3.29344660e-01 2.00262889e-01
-4.52816784e-01 -5.83035685e-02 9.51035559e-01 -9.24180329e-01
7.38855541e-01 -2.36765671e+00 -4.62971926e-02 2.65699834e-01
-1.99643373e-01 4.15636539e-01 -6.64698243e-01 5.51666580e-02
-4.78136837e-01 -1.77888609e-02 -2.47303680e-01 -8.93317759e-01
8.75537172e-02 2.16222201e-02 -2.01519445e-01 8.21400642e-01
5.62194288e-01 5.39291918e-01 -4.03239340e-01 -2.50793606e-01
2.01193601e-01 9.37926233e-01 -6.67935133e-01 4.74648446e-01
1.27306744e-01 8.79758060e-01 -6.64751008e-02 9.71468568e-01
1.34609258e+00 4.48991209e-01 5.05453162e-02 -4.45176810e-01
2.76626408e-01 -1.01053596e-01 -1.49969983e+00 1.55016840e+00
-5.10670662e-01 3.83865833e-01 5.59443593e-01 -7.00464427e-01
1.10837960e+00 5.07891834e-01 2.76796013e-01 -7.08334744e-01
2.14722738e-01 2.94060737e-01 -1.53699607e-01 -7.42415547e-01
-9.34336632e-02 2.51892954e-01 2.23497748e-01 1.49251685e-01
2.57729828e-01 -3.10204148e-01 2.92919934e-01 -7.19410926e-02
8.20817769e-01 4.29791838e-01 -3.34777981e-01 -2.47442350e-01
8.93289089e-01 -8.85867953e-01 9.32090938e-01 2.98112184e-01
-3.61211479e-01 8.44748616e-01 5.89414179e-01 -2.96562701e-01
-8.26565027e-01 -9.89366472e-01 -6.55795157e-01 6.65577769e-01
-1.11638509e-01 -4.12856519e-01 -1.11907625e+00 -7.10344672e-01
1.08479060e-01 3.92163724e-01 -6.10707760e-01 -3.57747860e-02
-6.17970586e-01 -8.50932956e-01 5.77171147e-01 1.22297443e-01
6.58829510e-01 -1.00535154e+00 2.14297727e-01 -2.52074152e-01
-1.99197769e-01 -1.40813017e+00 -3.80737811e-01 -3.70180219e-01
-4.88092482e-01 -1.21021068e+00 -2.50883102e-01 -5.34645677e-01
9.73327041e-01 -9.18360520e-03 1.02835655e+00 2.74442583e-01
-4.91386116e-01 1.53495401e-01 -5.74110635e-02 -1.67611107e-01
-4.53228503e-01 -4.30062771e-01 4.18303758e-01 3.46923918e-01
-1.24749057e-01 -6.48521423e-01 -4.81558144e-01 8.46560299e-01
-8.09125543e-01 -4.24583882e-01 2.88545310e-01 1.02799094e+00
5.64317524e-01 7.25864619e-02 8.10914457e-01 -6.49006009e-01
2.45156467e-01 -3.71296644e-01 -9.69105363e-01 1.18524656e-01
-3.15096617e-01 -2.46747449e-01 5.95914841e-01 -2.66704500e-01
-1.36666560e+00 1.05285525e-01 -4.96376812e-01 -5.38531065e-01
-4.60616350e-01 6.10256307e-02 -9.27259207e-01 -2.09669068e-01
6.58323765e-01 -2.73215383e-01 1.78463198e-02 -7.33865857e-01
3.09562594e-01 6.07745111e-01 7.65661716e-01 -7.22877860e-01
1.29681289e+00 4.80835587e-01 2.03619543e-02 -7.91276991e-01
-7.16223598e-01 5.05770072e-02 -5.83088934e-01 -3.15354913e-01
4.08370346e-01 -1.13168371e+00 -7.96248436e-01 8.15659523e-01
-1.06474757e+00 -9.92907584e-02 -5.61036430e-02 2.88708180e-01
-5.65175653e-01 3.67428809e-01 -4.79927272e-01 -8.83369863e-01
-2.87723899e-01 -1.58413732e+00 1.29842246e+00 1.90020666e-01
1.11192808e-01 -5.49894810e-01 -4.13905472e-01 7.01143384e-01
3.81526887e-01 4.36450332e-01 6.03761971e-01 -2.59191453e-01
-3.68587434e-01 -3.66065115e-01 -2.07773745e-01 6.61692679e-01
2.19893634e-01 2.35216498e-01 -1.54862559e+00 -7.52440512e-01
2.57223219e-01 -5.31971574e-01 4.57571536e-01 2.09991947e-01
1.30013084e+00 -2.42794245e-01 8.40222389e-02 9.28811550e-01
1.11902356e+00 -1.32607237e-01 9.26370680e-01 -1.24828130e-01
4.58481401e-01 1.00279093e+00 7.66281545e-01 3.55643541e-01
-2.99560670e-02 9.07908082e-01 5.89467585e-01 -2.03621723e-02
-4.88392383e-01 -1.72778979e-01 4.80725884e-01 6.45154715e-01
8.14841911e-02 -1.16331823e-01 -5.61375916e-01 2.59496987e-01
-1.37566483e+00 -7.27662981e-01 1.01408795e-01 2.26444793e+00
8.06833088e-01 -5.61093055e-02 -2.93073297e-01 4.61344332e-01
7.38084197e-01 1.62052646e-01 -5.13026893e-01 -3.09579559e-02
-2.29074329e-01 2.34811828e-01 -1.85950816e-01 6.87158525e-01
-1.24451280e+00 9.62868333e-01 5.22684479e+00 1.02845430e+00
-9.89581883e-01 1.15002826e-01 9.74624455e-01 -7.32815117e-02
3.27916741e-02 -2.07675382e-01 -9.58463490e-01 4.77869391e-01
7.21942544e-01 3.67163748e-01 4.69127148e-01 7.52008677e-01
4.28594917e-01 2.58219289e-03 -9.17281985e-01 1.16236305e+00
3.43592554e-01 -7.58816063e-01 -2.81124741e-01 -3.51034962e-02
8.71572733e-01 -3.44378710e-01 3.51311415e-01 6.92121387e-02
-1.92204818e-01 -1.13854873e+00 6.33561671e-01 4.87778425e-01
1.02345705e+00 -7.39857435e-01 8.03926826e-01 -3.25445160e-02
-9.54269588e-01 -7.63465390e-02 -4.60889041e-01 4.71771151e-01
4.22244705e-02 9.15855706e-01 -4.76473361e-01 7.58453965e-01
9.36145604e-01 5.93207777e-01 -7.68753171e-01 8.64346385e-01
-6.47272170e-01 5.32009721e-01 -4.37086165e-01 8.62596333e-01
-3.46075863e-01 -3.40636283e-01 7.15855658e-01 8.52182746e-01
2.90674716e-01 -1.51274130e-01 -1.50266290e-01 7.62394488e-01
-3.46416980e-01 3.43118459e-02 -4.39162135e-01 3.78717571e-01
5.50784349e-01 1.45903349e+00 -3.57340336e-01 -3.16590467e-03
-1.58343703e-01 6.69665575e-01 2.74523109e-01 5.12927115e-01
-9.76753950e-01 -2.49540567e-01 9.08456147e-01 -4.41791900e-02
2.20986858e-01 4.93862443e-02 -1.84158921e-01 -1.08627415e+00
4.37515050e-01 -1.33888018e+00 1.56889960e-01 -5.43604255e-01
-1.27825415e+00 9.67780650e-01 4.69672913e-03 -1.11984289e+00
-9.33345109e-02 -5.50478101e-01 -5.01667619e-01 9.03933346e-01
-1.58959544e+00 -1.10942042e+00 -5.97964227e-01 5.60927749e-01
4.98913288e-01 -3.90236050e-01 7.85928071e-01 7.12796569e-01
-1.08320355e+00 1.07112169e+00 3.42480950e-02 -1.02401957e-01
1.00616598e+00 -6.65526927e-01 4.77930367e-01 1.01198804e+00
-3.50626139e-03 6.18335426e-01 7.59029090e-01 -4.59848821e-01
-1.60848582e+00 -1.15559900e+00 5.90861678e-01 -4.12684619e-01
2.09857509e-01 -7.14366913e-01 -1.03102922e+00 3.32123399e-01
-1.18743487e-01 5.27192056e-01 3.97341132e-01 -1.23465486e-01
-6.81321323e-01 -4.92240280e-01 -1.63769948e+00 4.61604446e-01
1.28609228e+00 -3.89386952e-01 -1.13707155e-01 4.64071006e-01
2.93477356e-01 -4.46631253e-01 -9.62143540e-01 7.72269666e-01
6.48077667e-01 -1.09237158e+00 1.00674832e+00 -2.43218243e-01
1.44225791e-01 -4.21614677e-01 -2.92767704e-01 -1.23202908e+00
4.41307127e-02 -7.35473752e-01 -1.39529079e-01 1.70648921e+00
1.81027979e-01 -7.59717166e-01 5.53412437e-01 4.57135946e-01
4.08827240e-04 -6.22361124e-01 -1.19042218e+00 -8.38685393e-01
-2.46732756e-01 -5.35106540e-01 7.46199489e-01 6.64581299e-01
-6.35878861e-01 -4.91160620e-03 -4.97033447e-01 4.96087193e-01
9.27042365e-01 -1.04325339e-01 1.12405419e+00 -1.09769881e+00
-2.81358380e-02 2.08757482e-02 -4.52588886e-01 -6.07086480e-01
5.18690765e-01 -6.77918434e-01 7.11527616e-02 -8.54750872e-01
-1.18290231e-01 -5.71826398e-01 7.75147527e-02 5.76349020e-01
-2.68568546e-01 7.30383873e-01 1.72173962e-01 7.51759708e-02
-1.51350692e-01 8.51808906e-01 1.10777402e+00 -8.15368891e-02
-1.67815723e-02 7.31561854e-02 -4.92562950e-01 6.68021262e-01
8.14529598e-01 -4.31902915e-01 -3.53686124e-01 -5.18584132e-01
-2.65251428e-01 -2.33763278e-01 3.88826281e-01 -1.04524791e+00
-1.96150914e-01 2.53685474e-01 5.83224297e-01 -3.57284039e-01
6.86070621e-01 -7.89943874e-01 3.70447159e-01 2.37567857e-01
-5.97884916e-02 -2.36381143e-01 3.98366511e-01 3.23421001e-01
-1.49764627e-01 -1.01518892e-01 1.13550794e+00 4.45253313e-01
-8.03630427e-02 5.21035969e-01 2.49392301e-01 -1.16176359e-01
7.58149326e-01 -1.87352672e-02 -4.27828878e-01 -4.41475123e-01
-7.85400510e-01 6.23569228e-02 4.04731840e-01 6.42865956e-01
6.59231246e-01 -1.32849348e+00 -1.11561441e+00 9.12785530e-01
3.09412703e-02 -9.70589891e-02 4.34341848e-01 7.73023486e-01
-3.30312848e-01 5.42012677e-02 -2.61647522e-01 -5.79670787e-01
-1.56759512e+00 3.83543611e-01 3.32452148e-01 8.56042579e-02
-4.09434974e-01 9.04240370e-01 2.40374357e-01 -7.34383464e-01
5.01855910e-01 1.29966453e-01 -6.50739223e-02 1.17066288e-02
6.23027742e-01 4.04596627e-01 4.31580037e-01 -1.12804294e+00
-3.57781410e-01 8.12616527e-01 6.98171407e-02 9.58123356e-02
1.39522588e+00 -1.55067340e-01 -2.29925513e-01 -2.96942443e-01
1.33252919e+00 1.90452904e-01 -1.41166794e+00 -7.84249827e-02
-1.54721498e-01 -1.01749921e+00 -1.77318245e-01 -7.37694800e-01
-1.59441328e+00 7.25021124e-01 9.62721467e-01 -2.29393184e-01
1.31511688e+00 -2.90096819e-01 2.22907305e-01 1.12961590e-01
3.30514491e-01 -7.97462106e-01 7.31754899e-02 2.23416179e-01
1.56819332e+00 -1.35096014e+00 1.81236610e-01 -9.53454614e-01
-2.70279765e-01 7.18138158e-01 7.04993904e-01 1.11331247e-01
7.85576582e-01 2.89911181e-01 4.00436610e-01 -8.25654119e-02
-5.97530544e-01 1.26078039e-01 2.39571184e-01 8.14538598e-01
3.15962702e-01 -2.73359835e-01 -7.09194690e-02 4.19086576e-01
-3.20844442e-01 -4.01153147e-01 -9.24799684e-03 4.74431276e-01
1.44805759e-01 -1.12264752e+00 -8.21650386e-01 1.29462317e-01
-5.25093913e-01 2.33845010e-01 -5.83273843e-02 5.53734958e-01
2.55142152e-01 1.35437405e+00 -2.22283438e-01 -1.95680618e-01
5.95688760e-01 -3.76542397e-02 4.41766590e-01 -3.54949325e-01
-4.39351946e-01 3.09796214e-01 1.53547868e-01 -8.49976182e-01
-2.44379669e-01 -6.51668549e-01 -6.74124837e-01 -3.21861565e-01
-6.55479014e-01 1.84401553e-02 8.91789794e-01 5.62913537e-01
4.91713762e-01 3.13537031e-01 1.14759135e+00 -1.16443133e+00
-5.64638495e-01 -1.21833384e+00 -3.89857471e-01 6.80227816e-01
1.77132070e-01 -9.21669602e-01 -6.92232430e-01 3.26342359e-02] | [12.993574142456055, 0.27214401960372925] |
5508f1db-6d75-4c2a-89f8-17f983a3a66d | memory-based-dual-gaussian-processes-for | 2306.03566 | null | https://arxiv.org/abs/2306.03566v1 | https://arxiv.org/pdf/2306.03566v1.pdf | Memory-Based Dual Gaussian Processes for Sequential Learning | Sequential learning with Gaussian processes (GPs) is challenging when access to past data is limited, for example, in continual and active learning. In such cases, errors can accumulate over time due to inaccuracies in the posterior, hyperparameters, and inducing points, making accurate learning challenging. Here, we present a method to keep all such errors in check using the recently proposed dual sparse variational GP. Our method enables accurate inference for generic likelihoods and improves learning by actively building and updating a memory of past data. We demonstrate its effectiveness in several applications involving Bayesian optimization, active learning, and continual learning. | ['Mohammad Emtiyaz Khan', 'Arno Solin', 'S. T. John', 'Prakhar Verma', 'Paul E. Chang'] | 2023-06-06 | null | null | null | null | ['active-learning', 'gaussian-processes', 'bayesian-optimization', 'active-learning'] | ['methodology', 'methodology', 'methodology', 'natural-language-processing'] | [-6.11113459e-02 -6.89922720e-02 -2.13574007e-01 -2.83896029e-01
-1.20370829e+00 -4.28224891e-01 7.16360092e-01 5.81480801e-01
-5.93831956e-01 1.27122939e+00 1.95631515e-02 -3.55623662e-02
-3.38073581e-01 -7.84996569e-01 -1.03628147e+00 -8.61252487e-01
-7.64622912e-02 7.99432874e-01 2.54288584e-01 3.31988752e-01
3.48418266e-01 3.30215394e-01 -1.34027922e+00 -4.41048026e-01
1.12646139e+00 6.46675408e-01 2.15860292e-01 5.37082732e-01
-1.84663638e-01 7.22820699e-01 -6.50486171e-01 -3.89305234e-01
-1.01234064e-01 7.90992379e-02 -3.34159702e-01 2.81479567e-01
2.13030219e-01 -1.93049893e-01 -3.59012224e-02 1.08380353e+00
3.39134514e-01 7.83318281e-01 7.26368248e-01 -1.08183849e+00
-3.85913223e-01 3.18614632e-01 -6.32727504e-01 9.66954157e-02
1.92794487e-01 1.65918902e-01 7.85660386e-01 -1.01685560e+00
2.75716126e-01 1.34464419e+00 9.73093152e-01 1.87311500e-01
-1.43843126e+00 -3.54418933e-01 7.00571001e-01 -4.00787219e-02
-1.41913414e+00 -7.01948524e-01 6.04494810e-01 -4.27282512e-01
6.86610520e-01 -1.60141423e-01 8.56523573e-01 1.22091198e+00
2.64026850e-01 1.06577706e+00 6.28320217e-01 -3.19881111e-01
8.84549677e-01 9.63000804e-02 1.17880113e-01 6.89302325e-01
2.91285574e-01 2.82628704e-02 -9.44772124e-01 -7.23735213e-01
8.73105407e-01 3.05606544e-01 -1.34511188e-01 -3.85497212e-01
-6.86966062e-01 8.67381394e-01 -1.08144559e-01 -3.02610844e-01
-4.89258498e-01 2.36144215e-01 -1.38488680e-01 5.08460142e-02
9.50822771e-01 1.36589855e-01 -3.64854515e-01 -4.97366011e-01
-1.11548293e+00 3.48517001e-01 1.01268423e+00 7.74212599e-01
8.78151894e-01 1.43142156e-02 -1.21383041e-01 8.89835238e-01
7.33284056e-01 6.59421980e-01 1.48079336e-01 -1.18875074e+00
4.11485225e-01 1.65895503e-02 6.34466171e-01 -7.11823404e-01
-4.13155258e-02 -4.07626688e-01 -5.89528501e-01 4.18434925e-02
4.06952769e-01 -3.11709315e-01 -8.67660522e-01 1.56354976e+00
6.26571715e-01 6.07984722e-01 -4.73278649e-02 2.50553548e-01
2.20822126e-01 7.54050970e-01 2.62165219e-01 -7.54513085e-01
6.05012536e-01 -6.67622387e-01 -9.72098112e-01 -3.26969236e-01
1.31663412e-01 -5.30750751e-01 9.11005855e-01 5.27704298e-01
-1.21340954e+00 -3.93693745e-01 -7.61766136e-01 2.16646314e-01
-7.01415986e-02 -2.04218328e-01 9.13734913e-01 5.04719019e-01
-8.90937984e-01 9.73548234e-01 -1.62166488e+00 3.59130986e-02
7.00337589e-01 3.47485095e-01 1.11660473e-01 1.03776762e-02
-8.55381906e-01 6.31421149e-01 2.80286163e-01 2.41078645e-01
-1.04963160e+00 -9.16878402e-01 -7.18326628e-01 7.65424967e-02
5.36584377e-01 -5.74650049e-01 1.39673793e+00 -4.62400168e-01
-1.66468418e+00 6.67626783e-02 -4.30054694e-01 -6.03096426e-01
7.26861954e-01 -6.20615900e-01 -1.05152223e-02 -2.02155650e-01
-1.65409490e-01 3.99210691e-01 1.08924913e+00 -9.96835411e-01
-4.85354722e-01 -5.12708783e-01 -4.78246883e-02 4.75433648e-01
-1.17935121e-01 -5.61007798e-01 -5.60923159e-01 -4.52896982e-01
3.98165017e-01 -8.80931437e-01 -4.41541791e-01 1.29101381e-01
-1.82572156e-02 -3.93587202e-01 8.31925809e-01 -4.23614442e-01
8.49203765e-01 -2.18187332e+00 -3.71329039e-02 2.48831868e-01
-1.46330679e-02 -1.68115348e-01 3.33343625e-01 1.39854953e-01
6.59415722e-01 -8.93597975e-02 -3.90331745e-01 -1.03263068e+00
4.82330993e-02 7.56646514e-01 -4.40411210e-01 5.68886101e-01
2.34908871e-02 7.31896996e-01 -1.16986120e+00 -7.47232437e-01
5.54888882e-02 5.71352601e-01 -4.76048231e-01 1.91154510e-01
-4.49482769e-01 6.80959046e-01 -3.46337855e-01 6.24648333e-01
2.78443426e-01 -5.60701609e-01 8.03704932e-02 3.98202360e-01
-1.11620754e-01 1.39055267e-01 -1.60811424e+00 1.69056046e+00
-4.35452163e-01 4.77019399e-01 -4.30013500e-02 -7.90108562e-01
5.62200308e-01 3.25435340e-01 5.44526100e-01 -1.28251716e-01
-4.46989536e-01 7.21088201e-02 -6.82132125e-01 -1.32136017e-01
5.33566475e-01 -2.87098624e-02 3.73988360e-01 4.97812092e-01
2.54464060e-01 -4.08170670e-01 3.49017859e-01 1.78505689e-01
7.47996032e-01 2.91369438e-01 7.86193609e-02 1.52497059e-02
5.67671992e-02 -2.16837317e-01 8.89338195e-01 1.32563972e+00
-3.35246101e-02 4.62814957e-01 4.29294556e-01 3.90600264e-02
-7.44772911e-01 -1.29550099e+00 -3.05472225e-01 9.50881898e-01
-3.17407221e-01 -3.44488651e-01 -1.37329221e-01 -5.17311573e-01
1.47688583e-01 7.91228831e-01 -4.45365757e-01 -1.17272791e-02
-2.79750735e-01 -1.32022822e+00 -8.54998454e-02 6.57535195e-01
3.88225257e-01 -5.59225857e-01 -1.18063636e-01 5.31123161e-01
8.10914114e-02 -5.66369653e-01 -2.71242768e-01 2.08018780e-01
-1.46636140e+00 -7.30682492e-01 -7.07834542e-01 -2.03528509e-01
6.05247378e-01 -1.99019849e-01 1.00885391e+00 -3.73121679e-01
2.31081918e-01 8.94245923e-01 1.67663693e-01 -6.69023514e-01
-1.29745677e-01 -4.79941629e-03 1.06289171e-01 -1.45576879e-01
-8.71343687e-02 -6.90251172e-01 -2.42911234e-01 -1.46817774e-01
-6.05338335e-01 -4.21631604e-01 3.11486959e-01 8.91800046e-01
9.53421474e-01 1.33737698e-01 2.61033952e-01 -9.49955523e-01
6.43616736e-01 -6.00695729e-01 -1.09945309e+00 4.47161704e-01
-7.95272708e-01 6.75555924e-03 -1.40313894e-01 -7.39593267e-01
-1.51345706e+00 3.57998488e-03 -8.22529346e-02 -4.43409979e-01
2.13714212e-01 6.12724364e-01 1.14686243e-01 -5.68711050e-02
5.40209591e-01 5.94776571e-02 -2.94640437e-02 -5.59897721e-01
9.89622399e-02 2.27952972e-01 2.68300474e-01 -8.53854001e-01
3.82387102e-01 4.60068733e-01 9.11827981e-02 -8.65764737e-01
-1.25876689e+00 -2.82086104e-01 -5.66559315e-01 -1.51685387e-01
2.04608098e-01 -1.19457400e+00 -6.09977841e-01 6.38894975e-01
-9.35205638e-01 -5.93680799e-01 -6.67273045e-01 8.88054848e-01
-6.07943714e-01 4.93354857e-01 -6.69036746e-01 -1.27237189e+00
-1.42668277e-01 -9.63969171e-01 8.97458315e-01 4.10095781e-01
-1.46580115e-01 -1.65455008e+00 4.42926168e-01 2.44043022e-03
2.74775654e-01 -8.80204141e-02 4.68708098e-01 -4.46119606e-01
-7.31280982e-01 -2.13678777e-02 4.59637940e-01 2.06013709e-01
9.28869322e-02 2.97701269e-01 -8.20478976e-01 -4.58478987e-01
2.91062087e-01 -4.40778852e-01 8.70896637e-01 6.89439118e-01
1.08483994e+00 -3.60289097e-01 -4.44986999e-01 4.44286585e-01
1.20697463e+00 1.25536650e-01 9.96570289e-02 -9.80090424e-02
5.14792979e-01 2.67858624e-01 5.34723401e-01 7.50925541e-01
1.51895165e-01 2.29460955e-01 -4.89411652e-02 2.98893750e-01
4.44422364e-01 -2.79601246e-01 1.78093314e-01 6.97467566e-01
-8.38170201e-02 -1.47328258e-01 -9.97916341e-01 4.33349490e-01
-2.14569592e+00 -9.03540969e-01 8.12405497e-02 2.35535097e+00
1.26214254e+00 2.77993888e-01 -3.92873377e-01 -1.29891947e-01
4.79353517e-01 1.50268778e-01 -1.02073562e+00 3.92063558e-01
-5.22356145e-02 8.05005506e-02 5.05690515e-01 9.19339001e-01
-1.14063334e+00 6.10569656e-01 7.64898157e+00 8.00127983e-01
-7.13863969e-01 4.63968486e-01 5.53184628e-01 -4.50320691e-01
-3.56126726e-01 1.70715779e-01 -1.18465519e+00 4.71584082e-01
8.80422592e-01 -5.58097214e-02 2.19258964e-01 8.23924601e-01
2.58088350e-01 -6.18498385e-01 -1.04575431e+00 1.03046942e+00
-2.05656826e-01 -1.33964968e+00 -2.54461229e-01 -2.25345436e-02
1.15626788e+00 7.79815316e-02 1.80994362e-01 3.77934217e-01
9.01739240e-01 -6.12318218e-01 6.04690671e-01 1.12603486e+00
1.15507737e-01 -8.42681706e-01 4.46439683e-01 5.96333683e-01
-6.69915199e-01 -1.31351516e-01 -5.32179773e-01 1.12055726e-02
3.67991596e-01 1.15526676e+00 -5.12989461e-01 1.54552422e-02
4.55757082e-01 8.22550476e-01 -3.75657827e-01 1.24632299e+00
-4.04774249e-01 9.60598469e-01 -7.68206537e-01 3.95779535e-02
8.12313855e-02 -5.17861426e-01 6.91504478e-01 6.13904417e-01
3.29986572e-01 8.63283649e-02 3.10432076e-01 9.24737096e-01
8.88459906e-02 -1.79070458e-01 -3.21085006e-01 -8.75761583e-02
7.61966765e-01 7.78201938e-01 -5.71221471e-01 -5.12301862e-01
-3.63385439e-01 6.03676975e-01 5.05082309e-01 8.78154218e-01
-5.69019973e-01 2.35256106e-01 4.82091218e-01 -3.20527047e-01
2.06534356e-01 -7.16983914e-01 -1.08829826e-01 -1.26909661e+00
-1.48741424e-01 -3.76640260e-01 5.34556091e-01 -4.95302081e-01
-1.40356076e+00 -2.51859248e-01 1.66611135e-01 -5.63982010e-01
-4.48897302e-01 -1.10876627e-01 -5.38184702e-01 7.38897264e-01
-1.07696712e+00 -6.44564867e-01 2.42023870e-01 5.86723745e-01
5.66843212e-01 -5.95032722e-02 5.86186588e-01 4.75935265e-02
-6.77929699e-01 2.86290288e-01 7.69634962e-01 -3.53499919e-01
6.47835732e-01 -1.36440122e+00 2.81637430e-01 7.02097654e-01
4.75272268e-01 8.14347267e-01 8.92184675e-01 -9.60232913e-01
-1.36133909e+00 -7.79110193e-01 5.19363940e-01 -4.70405340e-01
8.21357548e-01 -3.39902341e-01 -1.21354663e+00 9.34268177e-01
-3.70201766e-01 -1.21689543e-01 8.16229641e-01 7.15279818e-01
1.72708839e-01 -9.81660187e-02 -1.07378423e+00 3.36602002e-01
6.79088295e-01 -4.49663430e-01 -4.42844987e-01 6.84685707e-01
5.81465304e-01 -6.89948976e-01 -8.97055268e-01 1.67973772e-01
3.57437819e-01 -5.43669581e-01 1.00324130e+00 -4.57207620e-01
-2.84326196e-01 -2.21333373e-02 2.76057292e-02 -1.30606461e+00
-8.59467685e-02 -8.35728347e-01 -9.36228275e-01 1.24470568e+00
3.60164672e-01 -8.14125478e-01 9.11924064e-01 9.94947433e-01
1.34133846e-01 -6.11153126e-01 -9.35854375e-01 -6.20134294e-01
-2.85358615e-02 -6.68464124e-01 4.28448558e-01 6.62208676e-01
-4.78706926e-01 4.93517965e-02 -2.86449164e-01 3.51982981e-01
1.09503913e+00 -1.08739957e-01 5.00537336e-01 -1.54582131e+00
-5.18157840e-01 2.98398808e-02 1.97048604e-01 -1.28307831e+00
1.60147831e-01 -2.49607757e-01 2.25606367e-01 -1.39996457e+00
1.26666561e-01 -7.16813087e-01 -7.47516528e-02 3.46492112e-01
-4.52305347e-01 -1.79329455e-01 -8.88167471e-02 5.49255311e-01
-8.91070962e-01 9.92237985e-01 9.61576402e-01 -6.13320991e-02
-4.61379319e-01 5.29523671e-01 -1.75132781e-01 9.22095299e-01
5.46050012e-01 -5.47739685e-01 -5.45523942e-01 -3.48208159e-01
5.50361812e-01 1.66804761e-01 3.84201080e-01 -1.01265836e+00
7.29664624e-01 -3.80684823e-01 6.59585357e-01 -9.55244482e-01
6.95506632e-01 -3.98707747e-01 2.88433909e-01 8.82610306e-02
-2.74727374e-01 -2.64292955e-01 9.99040306e-02 1.21028411e+00
-2.05069199e-01 -4.62838948e-01 6.45462751e-01 -3.35191786e-01
-4.66884285e-01 4.61087972e-01 -4.44941491e-01 2.16505796e-01
5.34295082e-01 4.63867560e-02 9.03820693e-02 -5.34115076e-01
-1.25274706e+00 5.23881614e-01 2.82323033e-01 -1.45943210e-01
5.04751921e-01 -1.15462685e+00 -3.37759972e-01 2.13146836e-01
-3.37508529e-01 5.34766614e-01 2.78172106e-01 8.97803664e-01
2.77738795e-02 6.98309690e-02 4.74240929e-01 -8.60905528e-01
-1.00170302e+00 1.62655532e-01 2.46922687e-01 -1.66948900e-01
-4.22808051e-01 9.65969741e-01 -2.67531306e-01 -2.21705154e-01
7.71542907e-01 -5.58285154e-02 -8.52733031e-02 4.33004826e-01
4.89745051e-01 7.51600444e-01 1.33431971e-03 7.30235502e-02
2.15635642e-01 1.90771803e-01 -9.57782865e-02 -5.18737614e-01
1.33936930e+00 -4.73463744e-01 3.83626595e-02 1.29411864e+00
6.61480725e-01 -8.82645771e-02 -2.11070585e+00 -6.35290086e-01
-1.46074034e-02 -4.31791365e-01 2.86754370e-01 -2.06982017e-01
-7.51406312e-01 7.61894941e-01 4.93859947e-01 -5.61140887e-02
3.72454435e-01 -2.19723642e-01 4.43302691e-01 8.93819332e-01
5.23679674e-01 -1.46309328e+00 2.00074762e-01 7.33557642e-01
4.35904175e-01 -1.33656597e+00 3.17588925e-01 6.50668293e-02
-4.19772834e-01 9.15313721e-01 2.17581481e-01 1.94030240e-01
1.05762088e+00 1.05729997e-01 -3.13033819e-01 -9.95815098e-02
-9.71846700e-01 1.07347116e-01 1.71616688e-01 4.80195701e-01
1.12512045e-01 -2.30142653e-01 1.47758305e-01 6.82174340e-02
1.31071731e-01 -2.36118156e-02 2.63518065e-01 1.34057844e+00
-6.10430479e-01 -8.74430656e-01 -5.73095500e-01 5.06533325e-01
-4.70890045e-01 -6.10516891e-02 2.67948329e-01 5.02060413e-01
-2.23436639e-01 9.23210442e-01 5.21560907e-01 6.68395281e-01
-6.30545989e-02 3.42452019e-01 5.97890615e-01 -6.61092937e-01
-1.02141043e-02 2.71729112e-01 -1.24312885e-01 -4.34681773e-01
-4.74430710e-01 -1.19999635e+00 -1.08507454e+00 -2.02863842e-01
-3.06693733e-01 3.46029699e-01 6.59274697e-01 1.15693653e+00
1.22068919e-01 1.58924997e-01 3.38447273e-01 -7.68593311e-01
-9.18646574e-01 -9.49298382e-01 -8.14325213e-01 -1.79621149e-02
3.55835676e-01 -8.95463049e-01 -5.66119313e-01 2.14331076e-02] | [6.958805561065674, 3.8535611629486084] |
90d8f48f-3fac-46b9-9873-45a3106d6de3 | extracting-clinical-concepts-from-user | 1912.06262 | null | https://arxiv.org/abs/1912.06262v2 | https://arxiv.org/pdf/1912.06262v2.pdf | Extracting clinical concepts from user queries | Clinical concept extraction often begins with clinical Named Entity Recognition (NER). Often trained on annotated clinical notes, clinical NER models tend to struggle with tagging clinical entities in user queries because of the structural differences between clinical notes and user queries. User queries, unlike clinical notes, are often ungrammatical and incoherent. In many cases, user queries are compounded of multiple clinical entities, without comma or conjunction words separating them. By using as dataset a mixture of annotated clinical notes and synthesized user queries, we adapt a clinical NER model based on the BiLSTM-CRF architecture for tagging clinical entities in user queries. Our contribution are the following: 1) We found that when trained on a mixture of synthesized user queries and clinical notes, the NER model performs better on both user queries and clinical notes. 2) We provide an end-to-end and easy-to-implement framework for clinical concept extraction from user queries. | ['Yue Zhao', 'John Handley'] | 2019-12-12 | null | null | null | null | ['clinical-concept-extraction'] | ['medical'] | [-3.12785767e-02 4.91573870e-01 -1.02651842e-01 -3.34989130e-01
-1.41031611e+00 -6.48983061e-01 6.49492443e-02 9.96942163e-01
-7.79969573e-01 8.41779709e-01 6.31632805e-01 -5.85554183e-01
-7.38773821e-03 -5.56822479e-01 6.30457848e-02 -8.51047859e-02
-1.68409362e-01 1.04513669e+00 -4.78366315e-02 2.23256983e-02
-3.60074610e-01 2.07423553e-01 -4.64284986e-01 5.67274332e-01
1.01381969e+00 4.71645802e-01 2.27046814e-02 1.12612808e+00
-1.24068908e-01 1.16729283e+00 -8.88560474e-01 -4.91690934e-01
-2.16451824e-01 -6.80284798e-01 -1.18475831e+00 -1.62837088e-01
1.96305830e-02 -1.98248297e-01 -6.84712604e-02 7.48428524e-01
8.24552000e-01 4.84618843e-02 5.77778101e-01 -4.84070569e-01
-8.08100700e-01 8.08698237e-01 2.95142561e-01 1.85908556e-01
5.78279614e-01 -2.91849319e-02 1.20939386e+00 -8.68060231e-01
1.14687645e+00 5.33231258e-01 9.63250935e-01 8.91126752e-01
-1.22279739e+00 -3.66896600e-01 -2.74631411e-01 -2.72385091e-01
-1.29887819e+00 -3.18283796e-01 -1.72026157e-01 -4.57792848e-01
1.55248141e+00 2.91942716e-01 5.61001182e-01 9.54199672e-01
3.50568295e-01 6.70015812e-01 5.46862125e-01 -3.62857401e-01
1.97612062e-01 1.60237327e-01 4.07942563e-01 5.88823557e-01
1.53427973e-01 -1.49242297e-01 1.66463926e-01 -8.81837666e-01
5.59212565e-01 7.24672228e-02 -2.49181852e-01 3.25168312e-01
-1.07396317e+00 8.41332376e-01 1.34519130e-01 6.63960695e-01
-5.38443387e-01 -2.39681229e-01 5.33328414e-01 -1.96941122e-02
1.63788423e-01 9.16709483e-01 -8.44033539e-01 -1.88033968e-01
-1.01191962e+00 -4.07243669e-02 1.27746785e+00 1.19812882e+00
9.18517821e-04 -3.91364306e-01 -3.67988586e-01 9.58644569e-01
8.46250635e-03 3.56365651e-01 8.23807657e-01 -4.23792183e-01
1.79475725e-01 3.51813406e-01 1.06938928e-01 -5.01202762e-01
-7.76306868e-01 -5.14785945e-01 -7.60012805e-01 -7.08823740e-01
1.72082707e-02 -5.60055137e-01 -1.24231994e+00 1.55929661e+00
-5.72904712e-03 -9.10765752e-02 7.02112973e-01 2.97198921e-01
1.49181032e+00 1.98035687e-01 1.05519378e+00 -4.16506857e-01
1.88997793e+00 -6.89181745e-01 -1.00570953e+00 -1.02065712e-01
1.33040202e+00 -1.08099794e+00 3.07687283e-01 -1.21157125e-01
-1.19793475e+00 -1.15941569e-01 -3.65027547e-01 -2.55561084e-01
-4.97095257e-01 2.67867953e-01 3.35866511e-01 3.01232696e-01
-1.15468073e+00 3.80044699e-01 -1.22780943e+00 -7.40370572e-01
2.45407358e-01 4.89345342e-01 -4.19249266e-01 2.00740919e-01
-1.23669231e+00 1.17948794e+00 5.37313640e-01 -6.62696362e-02
-3.41206223e-01 -9.08255935e-01 -1.17168355e+00 1.70005448e-02
2.52262354e-01 -1.13323331e+00 1.89706457e+00 -3.07176948e-01
-8.90054703e-01 1.06589317e+00 -2.52814651e-01 -3.37851077e-01
1.54510319e-01 -1.17860503e-01 -9.08345699e-01 2.83823907e-01
2.70121425e-01 5.17940581e-01 -1.67087704e-01 -8.99698794e-01
-6.48845911e-01 -1.97420016e-01 -3.12275529e-01 1.21046498e-01
2.95782924e-01 4.90948170e-01 -3.54622781e-01 -6.45673454e-01
-1.65788502e-01 -7.94773400e-01 -8.93350422e-01 -5.12547731e-01
-8.90987456e-01 -1.62193120e-01 1.85851172e-01 -8.78799856e-01
1.54993474e+00 -1.90516973e+00 -4.79464144e-01 -6.69254810e-02
6.99711800e-01 4.74384636e-01 2.66892929e-02 7.88520813e-01
-4.82854038e-01 6.03874207e-01 -2.14804292e-01 -2.74852395e-01
-3.46393943e-01 3.71920854e-01 -1.97266527e-02 -1.26752108e-01
6.44327343e-01 1.10413492e+00 -1.28414643e+00 -9.61065292e-01
-2.40411848e-01 5.55508077e-01 -5.47666669e-01 4.80762839e-01
-1.39652044e-01 4.46641833e-01 -7.51461387e-01 6.41155839e-01
2.13605568e-01 -7.78486192e-01 4.74110752e-01 6.83133006e-02
1.70307815e-01 6.41428053e-01 -6.19438410e-01 1.53929782e+00
-2.60363996e-01 7.50414506e-02 -1.67585537e-02 -3.39685291e-01
3.46798062e-01 1.26405418e+00 8.36480677e-01 -2.16439649e-01
-2.11211946e-02 4.66334641e-01 -8.66066590e-02 -9.33155000e-01
5.09111404e-01 -4.97616380e-01 -3.24731708e-01 3.48807544e-01
2.63475180e-01 5.29170595e-03 1.31470352e-01 4.60474759e-01
1.61091971e+00 -2.62755007e-01 1.28963220e+00 1.68348864e-01
3.49749535e-01 5.33253193e-01 7.05416024e-01 7.75927365e-01
-1.65421203e-01 4.63891149e-01 3.23679060e-01 -1.64457917e-01
-6.93152487e-01 -9.83207107e-01 -4.49561417e-01 7.53659010e-01
-5.67942560e-01 -8.32619309e-01 -4.43407208e-01 -1.02393889e+00
-3.42760831e-01 6.31988883e-01 -3.20195377e-01 2.12193698e-01
-5.91279626e-01 -8.02277744e-01 1.01747966e+00 7.93859184e-01
-2.62705594e-01 -1.35011435e+00 -5.76567292e-01 9.66497600e-01
-2.44792834e-01 -1.46431613e+00 -7.87881255e-01 5.02304077e-01
-1.01898336e+00 -1.33754671e+00 -7.01060474e-01 -1.16533256e+00
6.52958632e-01 -4.26015109e-01 1.58891213e+00 1.92411482e-01
-6.01744115e-01 5.23785651e-01 -4.91098881e-01 -5.47293901e-01
-8.77943397e-01 2.15035453e-01 -4.53843713e-01 -8.13864112e-01
8.44966292e-01 -2.19414696e-01 -5.17925739e-01 -1.43117294e-01
-1.17601156e+00 -1.92493394e-01 7.30285168e-01 1.07964313e+00
7.54254580e-01 -6.76743448e-01 6.66832566e-01 -1.65509820e+00
8.56224775e-01 -6.03192329e-01 1.70089215e-01 5.28850615e-01
-6.40669584e-01 8.62479806e-02 6.76274240e-01 -3.34982336e-01
-7.31378853e-01 2.59264797e-01 -8.37811053e-01 -2.14608070e-02
-6.42227590e-01 9.01835024e-01 3.58441740e-01 6.44820750e-01
8.97826135e-01 -4.06888202e-02 -3.98552030e-01 -5.71188390e-01
4.09519851e-01 8.64483237e-01 5.81153274e-01 -2.01203808e-01
1.56646118e-01 4.77500586e-03 -4.49841201e-01 -7.75052309e-01
-1.06729960e+00 -9.08084989e-01 -6.47075772e-01 5.14286041e-01
1.57219839e+00 -8.72564852e-01 -4.16525513e-01 -6.34889156e-02
-1.25244176e+00 -1.23680390e-01 -5.29165864e-01 6.21148765e-01
-3.28969330e-01 4.32176024e-01 -1.33488429e+00 -4.82027471e-01
-7.36326456e-01 -9.93370771e-01 1.24927187e+00 2.67472684e-01
-7.03359306e-01 -1.42613626e+00 5.00955224e-01 2.10090429e-02
1.42999858e-01 3.97007346e-01 1.32583213e+00 -1.74830365e+00
1.18044823e-01 -4.96725440e-01 -4.38371375e-02 -1.85764804e-01
4.44394678e-01 -1.22323476e-01 -6.74571395e-01 -1.02553308e-01
-1.35053456e-01 -3.41135353e-01 4.78439838e-01 1.54395133e-01
5.75112283e-01 -5.12040436e-01 -5.99462628e-01 2.64778227e-01
1.44764006e+00 4.50644106e-01 4.27322716e-01 -2.10565552e-01
4.63392287e-01 3.84939879e-01 2.11670578e-01 2.57013291e-01
5.84756136e-01 2.18375951e-01 -3.01110417e-01 -4.70571548e-01
1.98045835e-01 -1.81826890e-01 -8.35317299e-02 1.19945562e+00
1.87835857e-01 -1.73694119e-01 -1.29026198e+00 6.56698108e-01
-1.55282557e+00 -7.04138994e-01 -1.63248017e-01 1.94620740e+00
1.48943090e+00 -1.78440347e-01 -3.70491371e-02 -5.78232169e-01
4.06781673e-01 -5.71477056e-01 -3.03889900e-01 -2.61003792e-01
1.72181100e-01 8.34642947e-01 3.86089325e-01 5.52038491e-01
-1.03491509e+00 9.64463294e-01 6.75879288e+00 3.70266289e-01
-9.64256942e-01 2.84680128e-01 4.29016024e-01 2.46480197e-01
-5.35583720e-02 -1.02061920e-01 -9.61998582e-01 1.58408687e-01
1.33029568e+00 -9.92141441e-02 -5.56384861e-01 7.71029711e-01
1.63060009e-01 9.11167115e-02 -1.48315966e+00 6.39533043e-01
-1.81949273e-01 -1.29373658e+00 -2.74714604e-02 1.29452854e-01
4.07187909e-01 1.71392262e-01 -4.94013965e-01 4.91415948e-01
8.55326474e-01 -1.28144085e+00 -3.76639925e-02 4.63280469e-01
1.05751264e+00 -5.67382239e-02 1.18887138e+00 1.25625163e-01
-1.11770105e+00 3.16962272e-01 -1.23310007e-01 6.62406206e-01
4.62623090e-01 3.46461743e-01 -1.67659962e+00 8.72664988e-01
2.73031414e-01 5.32788396e-01 -5.42568147e-01 1.34781361e+00
-1.85360163e-01 5.22737145e-01 -1.31368786e-01 1.13598667e-01
2.54539669e-01 1.32638916e-01 3.11561584e-01 1.86462975e+00
1.24097668e-01 7.71476805e-01 7.18057454e-01 4.92522031e-01
-6.20282330e-02 5.40392935e-01 -6.34352028e-01 -6.37242913e-01
4.04184252e-01 1.29293334e+00 -7.84194231e-01 -8.44945252e-01
-3.96750718e-01 7.37876296e-01 1.34754092e-01 3.41759354e-01
-5.46715319e-01 -5.13761818e-01 3.21861237e-01 -9.46668014e-02
2.05187351e-01 1.56120241e-01 -3.52680571e-02 -1.27354920e+00
-3.28921676e-01 -9.32008564e-01 1.13180661e+00 -6.23845160e-01
-1.50285327e+00 1.15901220e+00 -3.94977361e-01 -1.31730521e+00
-7.72639751e-01 -5.37796915e-01 -2.68994600e-01 8.40863287e-01
-1.28094482e+00 -1.02719331e+00 -5.66957565e-03 4.54062134e-01
2.82427311e-01 3.35530370e-01 1.78660750e+00 5.96433222e-01
-4.26871836e-01 6.16762578e-01 -2.14961231e-01 9.30997729e-01
9.97980773e-01 -1.52834475e+00 2.41571277e-01 3.65694612e-01
6.42274842e-02 1.06003582e+00 4.66342032e-01 -9.33055103e-01
-5.26109815e-01 -1.14746737e+00 1.96345818e+00 -7.19907999e-01
6.20151162e-01 1.77724659e-01 -1.02242529e+00 1.03676724e+00
3.80945057e-01 -6.38639107e-02 1.76301539e+00 9.55128521e-02
-2.63969511e-01 7.53626525e-01 -1.19110727e+00 5.60307562e-01
6.27651155e-01 -8.53668690e-01 -1.18070054e+00 8.28397512e-01
8.71926248e-01 -6.73670709e-01 -1.60863197e+00 2.42761716e-01
9.51769426e-02 -1.73970014e-01 7.97389507e-01 -1.46165824e+00
2.78544933e-01 1.95481218e-02 1.29240036e-01 -1.12062109e+00
-1.44292295e-01 -6.24804318e-01 1.60083100e-01 7.92096198e-01
1.02382457e+00 -4.65138495e-01 5.59153676e-01 9.19780374e-01
-1.96983218e-01 -8.55526805e-01 -6.36694312e-01 -3.03055793e-01
1.66124865e-01 -1.72180519e-01 9.91171524e-02 1.27680743e+00
5.29781461e-01 7.57804692e-01 2.17356220e-01 1.87974885e-01
7.94320405e-02 -7.44133294e-02 -7.37450123e-02 -1.40696073e+00
-4.68333870e-01 -2.08039761e-01 -2.15401903e-01 -7.15316176e-01
-2.40959346e-01 -9.24844205e-01 3.41451496e-01 -1.99435067e+00
4.31349039e-01 -5.65024018e-01 -1.73723653e-01 1.10910702e+00
-5.81040323e-01 6.25363067e-02 -8.52881745e-02 3.60622764e-01
-6.96888745e-01 -1.14043899e-01 9.83401954e-01 8.45992565e-02
-6.58639252e-01 -4.61318046e-02 -7.65266836e-01 5.49185395e-01
4.38134551e-01 -1.05592656e+00 8.37996379e-02 -2.56666124e-01
2.24904418e-01 7.65688360e-01 -1.34753928e-01 -5.89716852e-01
5.11807203e-01 3.50177921e-02 3.13356817e-01 -4.52059478e-01
-5.11344485e-02 -5.83011270e-01 -2.83492710e-02 6.04461432e-01
-5.91162801e-01 4.47614312e-01 1.50342077e-01 3.67186278e-01
-3.66576493e-01 -3.79644573e-01 5.91010034e-01 -6.30382895e-01
-1.77436352e-01 1.87277660e-01 -8.70374680e-01 6.71256363e-01
7.45085835e-01 4.91813645e-02 -1.80168852e-01 -2.94923544e-01
-1.58220398e+00 3.99796098e-01 1.51138902e-01 1.35389464e-02
5.05735755e-01 -8.60272408e-01 -7.18638361e-01 -2.64587551e-01
3.27172518e-01 2.29078993e-01 1.84976757e-01 1.10833347e+00
-6.36082292e-01 9.31834161e-01 1.74654812e-01 -3.00542444e-01
-1.47281504e+00 5.11715591e-01 3.83994728e-01 -1.08368087e+00
-6.11322820e-01 7.98355579e-01 -5.16510718e-02 -7.70466387e-01
2.44757086e-01 -7.07297683e-01 -5.03331482e-01 1.52612925e-01
5.10364830e-01 -4.88300741e-01 3.08616191e-01 -4.09869075e-01
-4.11040902e-01 9.11376923e-02 -4.04972285e-01 -1.00081258e-01
1.42700422e+00 3.37410063e-01 -7.12204203e-02 1.40223816e-01
1.06178367e+00 3.99643511e-01 -1.86108604e-01 -3.36659819e-01
6.50632501e-01 3.85808051e-01 -4.72167403e-01 -1.14073062e+00
-5.59352040e-01 4.60167021e-01 2.50899136e-01 2.67194390e-01
1.01269388e+00 2.64238179e-01 9.78905082e-01 6.50959194e-01
2.11032685e-02 -7.83657253e-01 -3.00832361e-01 6.32142961e-01
2.47922912e-01 -1.09918833e+00 -8.19155872e-02 -6.05660141e-01
-9.22295928e-01 1.11126864e+00 2.64364600e-01 3.44176024e-01
7.68175662e-01 4.58936304e-01 6.74392343e-01 -5.25537312e-01
-7.92435706e-01 -4.19024885e-01 2.20512182e-01 5.43192208e-01
1.04418588e+00 9.98737589e-02 -5.18741012e-01 9.96566713e-01
-1.87659115e-01 3.12643200e-01 7.28658855e-01 1.28616059e+00
-8.01689178e-02 -1.48085558e+00 9.85448584e-02 5.32592595e-01
-1.25648963e+00 -9.02399600e-01 -5.34928799e-01 7.86100149e-01
6.58283681e-02 1.01110053e+00 -1.06592692e-01 -1.42819121e-01
4.96186078e-01 5.49998283e-01 -1.57666095e-02 -1.56226408e+00
-1.33028078e+00 3.07271868e-01 4.58791792e-01 -3.79422843e-01
-3.44971925e-01 -3.88204306e-01 -1.79707122e+00 5.43476343e-01
-3.92521441e-01 8.14134836e-01 2.55311191e-01 8.63411009e-01
7.27376461e-01 5.94343960e-01 -1.33732334e-01 2.87668973e-01
-6.23993099e-01 -1.01121914e+00 -3.23372811e-01 4.01023209e-01
3.51048052e-01 6.46262616e-02 1.64163962e-01 4.19529617e-01] | [8.45869255065918, 8.753642082214355] |
7ac30b45-d210-4be2-8c33-44f8e53e17b6 | borex-bayesian-optimization-based-refinement | 2210.17130 | null | https://arxiv.org/abs/2210.17130v1 | https://arxiv.org/pdf/2210.17130v1.pdf | BOREx: Bayesian-Optimization--Based Refinement of Saliency Map for Image- and Video-Classification Models | Explaining a classification result produced by an image- and video-classification model is one of the important but challenging issues in computer vision. Many methods have been proposed for producing heat-map--based explanations for this purpose, including ones based on the white-box approach that uses the internal information of a model (e.g., LRP, Grad-CAM, and Grad-CAM++) and ones based on the black-box approach that does not use any internal information (e.g., LIME, SHAP, and RISE). We propose a new black-box method BOREx (Bayesian Optimization for Refinement of visual model Explanation) to refine a heat map produced by any method. Our observation is that a heat-map--based explanation can be seen as a prior for an explanation method based on Bayesian optimization. Based on this observation, BOREx conducts Gaussian process regression (GPR) to estimate the saliency of each pixel in a given image starting from the one produced by another explanation method. Our experiments statistically demonstrate that the refinement by BOREx improves low-quality heat maps for image- and video-classification results. | ['Kohei Suenaga', 'Masaki Waga', 'Kotaro Uchida', 'Atsushi Kikuchi'] | 2022-10-31 | null | null | null | null | ['video-classification', 'gpr', 'gpr'] | ['computer-vision', 'computer-vision', 'miscellaneous'] | [ 1.57809481e-01 4.42316562e-01 1.16066495e-02 -5.00344813e-01
-3.52656603e-01 -3.32313664e-02 7.05271363e-01 2.18419135e-01
7.68948123e-02 3.82623792e-01 -3.39701353e-03 -3.65076929e-01
-1.24515086e-01 -5.06492555e-01 -8.64582777e-01 -8.06597412e-01
3.58014941e-01 3.67971003e-01 3.50093186e-01 -1.17357532e-02
8.20125699e-01 1.36403039e-01 -1.82446980e+00 2.01741949e-01
1.01450622e+00 8.10877025e-01 6.95137024e-01 6.23975158e-01
-3.15541357e-01 4.52923745e-01 -4.93186682e-01 -3.43763292e-01
8.02090764e-02 -6.73242629e-01 -6.45685673e-01 2.56021470e-01
1.47049159e-01 -8.98803174e-02 1.44816652e-01 1.25850916e+00
-8.50588754e-02 5.81588484e-02 9.78450775e-01 -1.54896915e+00
-8.37716579e-01 3.79924506e-01 -8.69846225e-01 -1.67097256e-01
1.31469592e-01 -4.66467887e-02 6.39597178e-01 -8.83008957e-01
3.83986115e-01 1.36775088e+00 3.63673747e-01 5.05259454e-01
-1.18750334e+00 -5.07915378e-01 4.25827146e-01 6.55295014e-01
-1.29321909e+00 1.05502941e-01 6.41283751e-01 -5.58863044e-01
4.50697064e-01 4.52882171e-01 8.88171613e-01 7.03399241e-01
4.78658080e-01 7.15457499e-01 1.28346241e+00 -6.07846200e-01
5.18307269e-01 4.02214766e-01 2.46325225e-01 8.03485930e-01
2.17355728e-01 4.58832085e-03 -5.87911010e-01 6.25996664e-02
7.82339156e-01 1.79950129e-02 -4.45023358e-01 -4.07329023e-01
-9.35375750e-01 9.76788342e-01 6.48507893e-01 -1.71056087e-03
-4.69551951e-01 2.13594601e-01 -2.91365206e-01 -2.21926793e-01
5.85494816e-01 6.71853364e-01 -3.88119072e-01 -1.98987238e-02
-1.09492862e+00 1.09038994e-01 3.84792000e-01 8.41873050e-01
9.72649097e-01 7.79399127e-02 -1.63115084e-01 5.39294839e-01
7.07374036e-01 4.20601308e-01 4.05976385e-01 -9.79831815e-01
-9.35490429e-02 5.34201980e-01 1.15991056e-01 -1.05442452e+00
-3.39367032e-01 -3.04813474e-01 -7.13101745e-01 7.86790311e-01
3.38806987e-01 -6.08198112e-04 -1.28851068e+00 1.41341698e+00
3.08586061e-01 3.64570439e-01 8.31542686e-02 1.12072730e+00
1.04625833e+00 9.01902258e-01 -2.28455514e-02 -1.92429766e-01
1.42527688e+00 -1.31949091e+00 -6.18520796e-01 -2.22392261e-01
1.35301471e-01 -6.30310416e-01 1.06045055e+00 6.85412705e-01
-7.65714526e-01 -8.86849940e-01 -1.13237369e+00 8.66053253e-02
-3.20664287e-01 1.87831104e-01 3.99793237e-01 5.00953197e-01
-1.11915052e+00 6.95443630e-01 -6.30377293e-01 -5.15479624e-01
1.44447163e-01 -1.10549610e-02 -2.01622367e-01 1.54051557e-01
-6.32300913e-01 1.11184597e+00 5.14346063e-01 1.20878085e-01
-9.10660625e-01 -4.70597535e-01 -7.18248188e-01 2.03554899e-01
3.02931517e-01 -6.31941259e-01 9.90770519e-01 -1.22629571e+00
-1.45497131e+00 5.58106184e-01 -4.90406871e-01 -5.02321243e-01
3.22234511e-01 -2.07636505e-01 -6.13613403e-04 2.34416634e-01
-2.79822312e-02 1.27963519e+00 1.38583088e+00 -1.77982795e+00
-7.33171523e-01 -2.82540739e-01 -3.10629234e-02 3.10144246e-01
8.16770121e-02 -1.36852384e-01 -5.89487314e-01 -5.15560031e-01
5.34654140e-01 -9.60908830e-01 -2.89385855e-01 -6.83100820e-02
-5.97792149e-01 -1.88978299e-01 8.40473533e-01 -9.73836720e-01
9.16372836e-01 -2.02102304e+00 2.21719593e-01 2.89762497e-01
3.71792734e-01 3.26800793e-02 1.36332020e-01 -1.09764546e-01
-2.87902176e-01 5.09227395e-01 -2.85712421e-01 -3.70346606e-01
-2.24803820e-01 1.74561396e-01 -2.34786138e-01 2.96497345e-01
2.21563146e-01 5.09557486e-01 -8.28391254e-01 -5.17134428e-01
4.95943457e-01 5.19887149e-01 -5.29854000e-01 3.12131494e-01
-3.29283416e-01 6.47653222e-01 -1.53369710e-01 1.71250075e-01
7.94989347e-01 -2.26425037e-01 -1.62356362e-01 -2.52984554e-01
-2.97324270e-01 -1.53769061e-01 -1.21407902e+00 1.36551821e+00
-1.65386051e-01 8.95270884e-01 -4.23359126e-01 -6.17739320e-01
1.38460076e+00 9.41038951e-02 -8.49714056e-02 -2.68413663e-01
3.16556729e-03 1.06346637e-01 -3.56324539e-02 -5.07606447e-01
6.45448208e-01 -1.18870735e-01 4.56187934e-01 2.70729035e-01
3.69249843e-02 -4.27138150e-01 -1.23217404e-01 2.24813446e-01
7.65498936e-01 4.74844426e-01 3.94784629e-01 -2.93121129e-01
3.16843152e-01 2.40994431e-02 5.25135458e-01 1.04779351e+00
-4.45700111e-03 1.19882381e+00 5.82017064e-01 -4.04181033e-01
-1.20954442e+00 -6.95629299e-01 -1.60860568e-01 6.02265835e-01
6.63871050e-01 -5.55621386e-01 -9.82524753e-01 -6.29066885e-01
-3.53178322e-01 1.12194133e+00 -8.25171351e-01 -1.86329886e-01
6.86148480e-02 -6.66547418e-01 -2.65450478e-01 4.41325903e-01
8.02633047e-01 -9.95677769e-01 -8.87927771e-01 -1.06859207e-01
-4.09630239e-01 -6.65131927e-01 -8.40657279e-02 2.84573972e-01
-8.32034826e-01 -1.00473821e+00 -8.03026378e-01 -4.68626678e-01
1.01842368e+00 3.53749692e-01 8.70230734e-01 4.07545388e-01
4.91608232e-02 1.77167445e-01 -6.38363659e-01 -6.00209713e-01
-3.17986876e-01 -4.30861235e-01 -3.08891892e-01 3.72028410e-01
2.00451523e-01 -3.31246853e-02 -5.09091020e-01 4.48140025e-01
-6.84042454e-01 8.04427445e-01 4.44926351e-01 7.53801703e-01
6.42213881e-01 2.49532208e-01 1.13513522e-01 -5.38126111e-01
3.74462008e-01 -3.25539857e-01 -6.09317124e-01 4.77042943e-01
-6.79942846e-01 4.82694566e-01 1.75471663e-01 -4.30433333e-01
-1.24631023e+00 2.13538170e-01 -9.99275893e-02 -5.58251739e-01
-3.32960308e-01 3.73025507e-01 -4.42706905e-02 1.46663889e-01
6.89744771e-01 2.12751657e-01 -4.60257381e-02 -3.08733582e-01
3.40504020e-01 5.93498111e-01 5.95866501e-01 -2.16351479e-01
7.62106061e-01 3.57275605e-01 2.11879183e-02 -7.98027277e-01
-9.06182110e-01 -3.39192331e-01 -5.59118330e-01 -6.43764377e-01
1.45516348e+00 -4.90951866e-01 -4.21767801e-01 4.24526542e-01
-1.33832920e+00 -2.40753278e-01 4.45849411e-02 6.64854407e-01
-5.17068982e-01 2.53013372e-01 -3.93996499e-02 -9.87432957e-01
-5.72464764e-02 -1.36847985e+00 9.98249471e-01 7.50379205e-01
-2.45237514e-01 -8.25035453e-01 -2.93429613e-01 4.11462903e-01
2.12105706e-01 1.42896265e-01 9.23142612e-01 -5.33154130e-01
-6.35473609e-01 1.55592963e-01 -3.77998292e-01 3.28291148e-01
-2.19249755e-01 3.19460928e-01 -1.03833532e+00 2.53853589e-01
3.23042840e-01 3.07949841e-01 8.17426503e-01 8.49838138e-01
1.30324209e+00 -2.83519179e-01 -3.00688714e-01 5.85368991e-01
1.42509174e+00 3.13339621e-01 9.96742249e-01 4.84015822e-01
6.26243830e-01 6.60069048e-01 9.17988241e-01 2.71564454e-01
2.82541335e-01 7.90239275e-01 8.93356323e-01 -4.14389491e-01
-1.51849017e-02 -2.75158048e-01 1.52585521e-01 1.94609314e-01
-2.34258220e-01 -1.74023151e-01 -8.92878592e-01 2.61381358e-01
-2.33578396e+00 -7.85129786e-01 -7.71963716e-01 2.27101636e+00
3.18959802e-01 1.62851065e-01 -1.22616038e-01 1.93176970e-01
1.00256193e+00 -7.93898404e-02 -3.11577111e-01 -4.73838598e-01
-1.92218367e-02 -2.45043650e-01 2.41886005e-01 6.68052673e-01
-9.82735097e-01 8.69903803e-01 5.73091984e+00 7.40343034e-01
-9.09738362e-01 -2.42334679e-02 7.83896863e-01 6.11861110e-01
-3.63005370e-01 4.57704514e-01 -6.79805160e-01 5.68274140e-01
4.75335330e-01 -4.69701588e-02 4.02352035e-01 9.58517551e-01
4.51007903e-01 -9.09000158e-01 -1.04526854e+00 1.22318435e+00
4.71765727e-01 -1.15128064e+00 1.16140671e-01 -1.17693059e-01
7.50591815e-01 -5.47077239e-01 1.17992699e-01 -1.73539780e-02
1.52563006e-01 -8.76604915e-01 1.11043096e+00 6.99773788e-01
2.07189098e-01 -4.90772903e-01 9.92988646e-01 3.76818627e-01
-8.94917727e-01 6.22236803e-02 -4.85593826e-01 -1.34565488e-01
1.38295799e-01 6.82303309e-01 -9.98797357e-01 5.40540636e-01
8.94274473e-01 5.90203524e-01 -9.62553740e-01 1.34818399e+00
-8.65454316e-01 7.41451919e-01 -7.17654899e-02 -8.47357735e-02
6.73796162e-02 -4.37449694e-01 5.61160803e-01 8.75894189e-01
4.76739496e-01 1.34455502e-01 -1.47604793e-01 1.28376234e+00
5.73345184e-01 -5.50438873e-02 -3.91149193e-01 6.44383788e-01
1.28344953e-01 1.27722394e+00 -1.16703176e+00 -6.53759420e-01
3.67193692e-03 1.09908104e+00 -1.02314949e-01 3.98966968e-01
-1.15967047e+00 -6.56044558e-02 1.90888137e-01 9.85871479e-02
3.59537870e-01 -4.81152209e-03 -7.78144419e-01 -9.17246521e-01
-2.17247859e-01 -5.14301479e-01 4.67199646e-02 -1.84370506e+00
-7.42273390e-01 6.94387913e-01 4.50992882e-01 -1.13034952e+00
-1.44552961e-01 -4.36917752e-01 -7.94506133e-01 9.48572576e-01
-1.28479588e+00 -7.73200989e-01 -6.95041537e-01 3.87739569e-01
8.06835413e-01 1.35270610e-01 5.21568120e-01 -3.84553313e-01
-2.88224339e-01 -4.28320281e-02 -1.65125415e-01 -3.05604666e-01
3.46346110e-01 -1.46465981e+00 -1.81339700e-02 1.09463096e+00
4.32767242e-01 3.55883569e-01 1.20909476e+00 -7.71100879e-01
-6.51255906e-01 -8.18398952e-01 8.11306000e-01 -4.25971359e-01
5.18181175e-02 -1.42507806e-01 -1.12494159e+00 4.41204399e-01
2.52172142e-01 -5.58862865e-01 1.32898003e-01 1.43316045e-01
-4.29211408e-02 -2.44691540e-02 -1.01792932e+00 8.07146132e-01
4.37328547e-01 -1.45132795e-01 -8.60121787e-01 2.50432432e-01
5.41293263e-01 -2.81704158e-01 -1.83043391e-01 3.60671759e-01
3.99060816e-01 -1.25041759e+00 6.85027003e-01 -1.36046752e-01
6.59446776e-01 -8.30097854e-01 1.90571249e-01 -1.55786848e+00
-3.32919538e-01 -1.97734341e-01 2.19099343e-01 1.25966179e+00
5.17922163e-01 -4.65511441e-01 4.97101247e-01 6.90062404e-01
-2.50859410e-01 -4.89991009e-01 -6.38523877e-01 -5.68246007e-01
-4.28790152e-01 -4.88447249e-01 4.81511444e-01 4.86900091e-01
-5.42962551e-03 2.19512030e-01 -3.69858712e-01 3.61316353e-01
6.87131405e-01 1.30380783e-02 9.32129920e-01 -1.28265965e+00
-2.21136361e-01 -2.46345669e-01 -5.22970140e-01 -1.03149903e+00
-1.00925468e-01 -5.01771152e-01 4.92676497e-01 -2.00072837e+00
5.98733664e-01 -2.20619425e-01 1.58478096e-02 5.24605513e-01
-4.76032764e-01 1.80227876e-01 5.43017626e-01 4.30790275e-01
-5.16744971e-01 5.77541471e-01 1.10870051e+00 -3.41223925e-02
-2.50330597e-01 -1.29260078e-01 -6.96153998e-01 1.07578397e+00
8.68140638e-01 -7.03021228e-01 -2.95010060e-01 -1.54926270e-01
1.61389574e-01 -1.05063811e-01 7.17781126e-01 -1.08968377e+00
9.00743678e-02 -6.79520145e-02 4.34701800e-01 -6.81405842e-01
3.32057506e-01 -7.99978971e-01 3.90276790e-01 4.71580744e-01
-2.42619738e-01 -1.71807222e-02 -4.77331020e-02 5.70473313e-01
-2.02601954e-01 -6.68890178e-01 7.81202912e-01 -1.02267778e-02
-9.79872882e-01 -3.21217179e-01 -4.15692300e-01 -4.96260643e-01
9.75794613e-01 -5.29565215e-01 -4.01752263e-01 -8.17207992e-01
-8.56770396e-01 1.15771309e-01 4.87819731e-01 4.22390401e-01
8.36872160e-01 -1.23091280e+00 -6.71819210e-01 1.31786749e-01
6.35564178e-02 -9.85028595e-02 2.05961660e-01 9.34531271e-01
-5.65925062e-01 1.74470752e-01 -2.65120775e-01 -1.06533349e+00
-1.46840024e+00 5.93155801e-01 2.44316414e-01 -8.26850235e-02
-6.96967542e-01 7.73130059e-01 6.74976408e-01 -1.13293186e-01
1.14677623e-01 -5.15441298e-01 -5.39706826e-01 -3.36985081e-01
4.80884135e-01 1.75471008e-01 -2.22095340e-01 -5.93331039e-01
-1.37723356e-01 7.34281123e-01 3.63939494e-01 -4.47849542e-01
1.09218812e+00 -2.91884840e-01 4.30841893e-02 6.25049412e-01
6.90178931e-01 -3.81610751e-01 -1.60006189e+00 2.55747557e-01
-2.06210390e-01 -6.17409348e-01 1.81456208e-01 -9.54111934e-01
-9.61573541e-01 1.00140893e+00 6.47439897e-01 2.75726765e-01
1.14014566e+00 1.68213084e-01 -2.16442019e-01 -6.93887174e-02
3.49228024e-01 -9.99487340e-01 2.04739630e-01 2.36507475e-01
1.30996561e+00 -1.24889910e+00 -2.58312207e-02 -5.17783225e-01
-9.12115812e-01 1.26132703e+00 6.86672866e-01 5.82102090e-02
6.02219760e-01 -2.83837825e-01 1.45190232e-03 -2.11024493e-01
-5.20093501e-01 -3.10061842e-01 5.68299651e-01 6.99342728e-01
-5.73713519e-02 2.19319668e-02 -2.00657502e-01 5.84168911e-01
-1.53455675e-01 -1.32767618e-01 6.48809612e-01 4.46645081e-01
-6.29122496e-01 -6.31860673e-01 -9.30037141e-01 3.78364235e-01
1.58433933e-02 -2.07275853e-01 -3.09274763e-01 6.82490587e-01
3.78797084e-01 1.38325369e+00 1.23656698e-01 -5.81978083e-01
-1.85079753e-01 1.53701246e-01 3.25611681e-01 -5.56932628e-01
-3.20085257e-01 1.67462736e-01 -2.22921476e-01 -3.93235832e-01
-4.80061442e-01 -5.94698429e-01 -1.38554120e+00 -1.41904041e-01
-7.21005142e-01 1.88596815e-01 1.13154972e+00 1.09652495e+00
9.07491818e-02 5.65701902e-01 4.03886139e-01 -1.13076007e+00
1.15556061e-01 -9.85525489e-01 -5.58585286e-01 3.98236096e-01
6.46979511e-02 -8.36573005e-01 -6.68147802e-01 3.20973396e-01] | [9.967728614807129, 2.18422794342041] |
f29bffa6-0407-4c00-8096-27a301e2f212 | synthesis-of-contrast-enhanced-breast-mri | 2307.00895 | null | https://arxiv.org/abs/2307.00895v1 | https://arxiv.org/pdf/2307.00895v1.pdf | Synthesis of Contrast-Enhanced Breast MRI Using Multi-b-Value DWI-based Hierarchical Fusion Network with Attention Mechanism | Magnetic resonance imaging (MRI) is the most sensitive technique for breast cancer detection among current clinical imaging modalities. Contrast-enhanced MRI (CE-MRI) provides superior differentiation between tumors and invaded healthy tissue, and has become an indispensable technique in the detection and evaluation of cancer. However, the use of gadolinium-based contrast agents (GBCA) to obtain CE-MRI may be associated with nephrogenic systemic fibrosis and may lead to bioaccumulation in the brain, posing a potential risk to human health. Moreover, and likely more important, the use of gadolinium-based contrast agents requires the cannulation of a vein, and the injection of the contrast media which is cumbersome and places a burden on the patient. To reduce the use of contrast agents, diffusion-weighted imaging (DWI) is emerging as a key imaging technique, although currently usually complementing breast CE-MRI. In this study, we develop a multi-sequence fusion network to synthesize CE-MRI based on T1-weighted MRI and DWIs. DWIs with different b-values are fused to efficiently utilize the difference features of DWIs. Rather than proposing a pure data-driven approach, we invent a multi-sequence attention module to obtain refined feature maps, and leverage hierarchical representation information fused at different scales while utilizing the contributions from different sequences from a model-driven approach by introducing the weighted difference module. The results show that the multi-b-value DWI-based fusion model can potentially be used to synthesize CE-MRI, thus theoretically reducing or avoiding the use of GBCA, thereby minimizing the burden to patients. Our code is available at \url{https://github.com/Netherlands-Cancer-Institute/CE-MRI}. | ['Ritse Mann', 'Tao Tan', 'Regina Beets-Tan', 'Jonas Teuwen', 'Chunyao Lu', 'Yuan Gao', 'Xin Wang', "Anna D'Angelo", 'Luyi Han', 'Tianyu Zhang'] | 2023-07-03 | null | null | null | null | ['breast-cancer-detection', 'breast-cancer-detection'] | ['knowledge-base', 'medical'] | [ 3.38330299e-01 -1.30660817e-01 -2.26638332e-01 -1.15051523e-01
-7.14620411e-01 -2.32353047e-01 5.55708230e-01 2.75933772e-01
-5.05527854e-01 4.90512371e-01 2.54483610e-01 -3.79242569e-01
-3.01320881e-01 -8.05167258e-01 -6.93862289e-02 -1.03042853e+00
-2.41325140e-01 4.06751454e-01 1.84767872e-01 -4.33733761e-02
-7.51382634e-02 8.15782964e-01 -9.82725263e-01 1.43752560e-01
1.06171119e+00 9.97228503e-01 7.53553748e-01 2.73010820e-01
-2.88200945e-01 6.87206805e-01 -2.39849791e-01 -2.80632585e-01
1.34331062e-01 -5.44417143e-01 -5.19799948e-01 6.98854327e-02
-9.83942971e-02 -3.42450559e-01 -3.59109044e-01 1.26340997e+00
4.61654931e-01 -2.73937490e-02 5.15853763e-01 -8.79091203e-01
-3.78951699e-01 5.20817280e-01 -8.13666821e-01 5.78606904e-01
-2.25286856e-01 2.36627162e-01 2.70339727e-01 -7.48977602e-01
6.69941068e-01 6.49434268e-01 3.48676115e-01 6.32893920e-01
-9.06767428e-01 -6.71862364e-01 -1.14548050e-01 3.86715055e-01
-9.85753894e-01 -1.99423715e-01 8.37124467e-01 -3.26912016e-01
3.91129047e-01 4.89530295e-01 8.24396253e-01 9.65758145e-01
6.14039481e-01 6.26219869e-01 1.35188305e+00 -1.86959118e-01
1.00834928e-01 -1.60506114e-01 -2.66378419e-03 7.15664148e-01
3.30886632e-01 2.11259305e-01 -3.73828448e-02 -1.39567748e-01
7.12057292e-01 4.43492144e-01 -5.95706284e-01 -5.10210514e-01
-1.27040076e+00 6.82682991e-01 6.05817378e-01 9.02486086e-01
-5.60411572e-01 -4.76026908e-02 3.91748607e-01 -1.61773399e-01
4.63909894e-01 1.76021144e-01 1.57492980e-02 1.55414745e-01
-1.01464450e+00 -7.00018480e-02 1.99448124e-01 1.60952628e-01
3.17370504e-01 -1.70767114e-01 -1.32403582e-01 8.46425414e-01
1.27328247e-01 4.72558051e-01 1.02728760e+00 -7.28392363e-01
1.08010866e-01 5.24929106e-01 -1.62733421e-01 -8.81594837e-01
-7.05270350e-01 -8.04675400e-01 -1.40303016e+00 4.37358767e-02
5.03738582e-01 1.61756501e-01 -1.01219857e+00 1.55929220e+00
4.59593564e-01 3.06526899e-01 -2.61172503e-01 1.23618758e+00
6.20360315e-01 8.85431394e-02 4.02549237e-01 -4.10009503e-01
1.48497653e+00 -7.91626811e-01 -7.36338437e-01 -6.91557378e-02
6.58684433e-01 -3.85586739e-01 6.83999717e-01 1.05458885e-01
-1.00183570e+00 -1.44920275e-01 -1.01285648e+00 2.63633370e-01
-1.98730826e-01 -1.70090392e-01 8.59117210e-01 6.01906538e-01
-7.49299884e-01 5.64582705e-01 -1.34091306e+00 -1.06535226e-01
5.33458710e-01 2.35682219e-01 -5.84335625e-01 -5.80940008e-01
-1.18838024e+00 1.20262289e+00 6.88579027e-03 2.56237090e-01
-9.20693696e-01 -1.05860138e+00 -7.59093046e-01 -2.18095317e-01
4.86167878e-01 -5.48214495e-01 7.26005673e-01 -7.27370799e-01
-1.13471818e+00 6.15661621e-01 -1.45047516e-01 -3.38736802e-01
6.48408234e-01 3.39896590e-01 -4.45520848e-01 6.09468281e-01
2.16190547e-01 5.18321514e-01 5.76257110e-01 -9.75500107e-01
-4.12790567e-01 -7.26834714e-01 -2.32383981e-01 1.58657879e-01
-1.95580915e-01 -5.06521668e-03 -3.61362576e-01 -6.24959528e-01
4.14738655e-01 -9.01596308e-01 -6.47577524e-01 2.92600822e-02
-5.25477007e-02 1.60092235e-01 5.53950310e-01 -1.00565505e+00
9.87121165e-01 -2.06581569e+00 5.67513295e-02 2.21786410e-01
6.67829156e-01 2.09407404e-01 -1.97663158e-01 -3.83384824e-01
-1.60398364e-01 1.53953835e-01 -4.63209361e-01 2.21065715e-01
-5.32771111e-01 -1.70115992e-01 4.83731031e-01 7.01384664e-01
1.68619379e-01 1.04396486e+00 -1.07581365e+00 -4.77404594e-01
3.39019507e-01 6.27027094e-01 -1.46809056e-01 -1.20638862e-01
2.53281415e-01 8.98719072e-01 -4.44527835e-01 7.47444391e-01
6.98584080e-01 -3.93419653e-01 5.61890781e-01 -4.81729597e-01
1.57555178e-01 -9.71435905e-02 -6.78007305e-01 1.66963542e+00
-4.60296214e-01 1.60118803e-01 3.05179715e-01 -1.20874095e+00
7.00913310e-01 2.64514357e-01 9.36530948e-01 -1.10396886e+00
1.90040186e-01 3.39845687e-01 4.93742377e-01 -5.56507647e-01
-2.38008589e-01 -3.46634388e-01 2.95238167e-01 3.14950824e-01
-2.99772352e-01 -4.60281922e-03 2.99259901e-01 3.40109505e-02
1.29277980e+00 -2.66007204e-02 2.02944577e-01 -4.40521926e-01
6.84948206e-01 -1.17856160e-01 6.63804054e-01 3.40719312e-01
-3.74890089e-01 3.40710461e-01 2.39994362e-01 -4.66404825e-01
-8.61483276e-01 -9.81224179e-01 -4.37265933e-01 5.10147750e-01
1.84282988e-01 6.57655746e-02 -5.20522296e-01 -6.65189743e-01
4.49752845e-02 3.88329268e-01 -7.15540886e-01 -1.38785690e-02
-6.28572583e-01 -1.21884048e+00 3.18653554e-01 4.62547332e-01
5.83496749e-01 -6.15139186e-01 -9.16934907e-01 4.19254035e-01
-5.05096674e-01 -8.16609323e-01 -3.47510874e-01 2.24283993e-01
-1.07163060e+00 -1.07965326e+00 -1.45898688e+00 -2.33396217e-01
7.68823624e-01 3.29089940e-01 7.70959616e-01 3.15975010e-01
-6.70090139e-01 1.67422071e-01 -2.56543428e-01 -1.90400168e-01
-5.03747165e-01 -7.14295879e-02 -2.74172693e-01 -6.57312050e-02
2.76172549e-01 -4.85426486e-01 -7.77269185e-01 2.19536677e-01
-1.06480038e+00 4.51887578e-01 7.28871584e-01 9.99621928e-01
5.69455266e-01 4.51586805e-02 5.55765986e-01 -7.76305318e-01
3.95016968e-01 -5.63055515e-01 -4.93442565e-01 1.59351110e-01
-3.26269567e-01 8.63487832e-03 2.05426902e-01 -5.26201367e-01
-9.30514872e-01 -1.82632789e-01 1.03329390e-01 -3.89065415e-01
5.76566439e-03 8.41010153e-01 -7.68970102e-02 -1.87820390e-01
4.56228077e-01 2.53331155e-01 4.42322195e-01 -2.05288291e-01
-1.59487769e-01 4.12693650e-01 3.54724705e-01 -1.39164865e-01
4.26393807e-01 7.25831807e-01 3.97723019e-01 -6.42298579e-01
-5.23986578e-01 -2.70688951e-01 -3.58163625e-01 -1.70344129e-01
7.75532842e-01 -7.16445863e-01 -3.65180552e-01 4.71958876e-01
-7.17832267e-01 -3.16577196e-01 9.55563337e-02 7.16820300e-01
-7.07071796e-02 5.69321811e-01 -8.29315364e-01 -2.88105279e-01
-3.95697176e-01 -1.57456958e+00 3.44432592e-01 1.87991902e-01
-1.81696311e-01 -8.99285316e-01 -3.17607999e-01 4.72598583e-01
9.25871432e-01 4.64137793e-01 1.13040864e+00 -4.44400311e-01
-4.08655405e-01 -9.41445157e-02 -4.07762676e-01 1.90692112e-01
4.89866287e-01 -5.58199465e-01 -5.28691232e-01 -1.84086293e-01
2.35491857e-01 1.18046261e-01 7.28654206e-01 7.38972068e-01
1.36611855e+00 2.00181648e-01 -5.25620162e-01 7.64267147e-01
1.33621347e+00 5.59828281e-01 4.19391572e-01 3.90152663e-01
4.90929365e-01 6.72651052e-01 3.91814590e-01 2.54329860e-01
3.86639565e-01 3.97269815e-01 3.93497944e-01 -3.22366714e-01
-3.13519150e-01 2.42403716e-01 -1.87958151e-01 6.63593531e-01
-2.92128146e-01 1.92267433e-01 -1.11687303e+00 6.63054466e-01
-1.26117992e+00 -7.93428957e-01 -3.46149541e-02 1.97073281e+00
7.83248544e-01 -1.26169071e-01 -1.64941654e-01 6.80916151e-03
7.19637990e-01 -1.11859210e-01 -7.63199270e-01 2.32511342e-01
5.51528931e-02 4.50194031e-01 5.94560802e-01 4.13860142e-01
-8.40546846e-01 1.58413768e-01 5.21160698e+00 6.04129076e-01
-1.70767653e+00 4.43926096e-01 8.44331861e-01 -1.68289050e-01
-5.16784310e-01 -2.11409450e-01 -3.07874322e-01 6.92546725e-01
8.12436938e-01 -2.27516621e-01 1.97273940e-01 5.28057277e-01
2.32948110e-01 -5.15296996e-01 -7.52511799e-01 8.82337153e-01
3.27773131e-02 -1.28310931e+00 -1.81283891e-01 2.66773969e-01
2.57691354e-01 1.79431707e-01 -1.78318899e-02 8.35023075e-02
-4.86331396e-02 -8.38941872e-01 3.13244343e-01 4.78563190e-01
8.90524924e-01 -6.58347249e-01 8.43589067e-01 1.44814953e-01
-9.05868769e-01 -2.56669447e-02 -1.43761352e-01 4.79050189e-01
7.99666494e-02 9.95529711e-01 -6.22685134e-01 7.30836689e-01
7.00815201e-01 4.22239929e-01 -5.54010510e-01 1.00004244e+00
2.61064377e-02 2.58255094e-01 -2.08237678e-01 1.53061748e-01
6.49458170e-02 -2.11294681e-01 3.67551267e-01 9.59270477e-01
6.32714212e-01 1.09422415e-01 2.12686852e-01 6.94820583e-01
1.39640048e-01 5.86003289e-02 -3.02171499e-01 1.28576696e-01
1.92135215e-01 1.67168462e+00 -1.11014307e+00 -5.04453182e-01
-4.91957307e-01 9.69965696e-01 -7.89924189e-02 2.26862609e-01
-8.26143682e-01 -1.45213038e-01 6.69403747e-02 4.90776867e-01
5.16395345e-02 -5.15629910e-02 -1.02580547e-01 -1.16949117e+00
-2.46210933e-01 -1.02488327e+00 2.57756323e-01 -4.01239395e-01
-1.10269415e+00 6.79383755e-01 5.82358502e-02 -1.03013730e+00
-9.24151987e-02 -3.58606189e-01 -3.55536133e-01 1.04957998e+00
-1.87170935e+00 -9.64279950e-01 -6.79776728e-01 3.94128889e-01
2.00436085e-01 1.12413473e-01 7.06953824e-01 4.68466729e-01
-4.94937152e-01 3.37285757e-01 -1.40604034e-01 5.11290208e-02
6.34007275e-01 -9.49864745e-01 -2.29098484e-01 5.88444948e-01
-4.73555148e-01 5.23396075e-01 3.34360659e-01 -7.30772734e-01
-1.19244015e+00 -9.48081136e-01 5.29201329e-01 -9.32355691e-03
5.45891762e-01 6.70090914e-02 -1.01696062e+00 2.18617529e-01
-1.67477950e-02 5.19884408e-01 9.47116256e-01 -5.40099442e-01
1.05374627e-01 -1.58205509e-01 -1.40771008e+00 5.50857186e-01
6.76953793e-01 -2.65393972e-01 -1.68499902e-01 6.20821454e-02
2.63976336e-01 -4.64989036e-01 -1.21547592e+00 6.88595235e-01
5.68303704e-01 -8.63782346e-01 8.51514280e-01 -1.71186805e-01
4.55387741e-01 -2.30711013e-01 1.40473440e-01 -1.28853714e+00
-5.99932969e-01 1.45841073e-02 3.42764147e-02 6.85772181e-01
2.26527929e-01 -6.95397854e-01 7.28078485e-01 8.04257035e-01
-5.98647892e-02 -8.41768920e-01 -9.49944019e-01 -5.97381294e-01
1.91169441e-01 -3.11842531e-01 6.55093312e-01 1.03534603e+00
2.05028504e-02 -3.79455298e-01 1.01592354e-01 6.94436394e-03
7.78153121e-01 -2.06129238e-01 1.08394794e-01 -1.19390607e+00
-2.21399829e-01 -5.91221631e-01 -2.80238569e-01 -3.74551088e-01
1.16118148e-01 -1.37189090e+00 -3.19588423e-01 -1.40795541e+00
3.98444653e-01 -5.59267759e-01 -6.57517314e-01 4.16349530e-01
-1.04790896e-01 4.05280739e-01 3.02399278e-01 2.27517605e-01
-8.21550563e-02 1.60232484e-01 1.77643371e+00 -3.67484242e-01
1.30957916e-01 -2.88278073e-01 -6.70412421e-01 7.45426476e-01
6.69046044e-01 -4.88461375e-01 -1.72636271e-01 -1.54844850e-01
-1.89230189e-01 5.39927661e-01 3.90785426e-01 -9.82074678e-01
2.29853630e-01 -1.91174254e-01 7.79203951e-01 -3.43846738e-01
7.84257948e-02 -8.80357802e-01 3.42137754e-01 1.05833375e+00
-4.70697470e-02 -3.52739207e-02 1.68418199e-01 1.60183817e-01
-2.15674356e-01 -3.92294466e-01 9.95211780e-01 -5.47418177e-01
-5.08327901e-01 4.34572369e-01 -7.65195966e-01 -5.04711390e-01
1.15311420e+00 -5.32502607e-02 -7.82046318e-02 -1.05583519e-01
-7.71430671e-01 1.54611468e-01 3.48111957e-01 1.66676000e-01
6.37474298e-01 -1.17784011e+00 -6.80919826e-01 2.14979783e-01
-1.17133699e-01 -1.65351182e-01 7.97032773e-01 1.47445786e+00
-5.65463424e-01 2.55970120e-01 -3.59631628e-01 -6.03627443e-01
-1.03029418e+00 6.43648446e-01 5.94227135e-01 -8.31196308e-01
-6.82770431e-01 5.18625677e-01 2.11622015e-01 -1.46162719e-01
-1.12635180e-01 -3.10724229e-01 -1.20612904e-01 2.23958939e-02
7.63177216e-01 3.04106772e-01 2.39706904e-01 -4.33124363e-01
-4.78618860e-01 2.40891650e-01 -4.19666409e-01 -8.01545307e-02
1.26174963e+00 -1.78243011e-01 -2.48771861e-01 -7.14939237e-02
9.21431959e-01 -2.32512489e-01 -1.00495017e+00 -2.36398280e-01
1.33193254e-01 -3.11580807e-01 5.88094413e-01 -8.77676249e-01
-1.49588883e+00 8.64182770e-01 8.70716274e-01 -1.97304875e-01
1.20179939e+00 -1.77425772e-01 7.57960141e-01 -2.52251595e-01
4.69352543e-01 -5.83331287e-01 4.91150171e-02 -2.15959549e-02
7.65777767e-01 -1.28876710e+00 3.43243703e-02 -4.04225349e-01
-6.46185160e-01 1.10671687e+00 5.42256117e-01 1.18741684e-01
7.16932237e-01 6.99696779e-01 2.36137941e-01 -1.63451195e-01
-3.81447434e-01 1.02132544e-01 7.13889822e-02 5.15355229e-01
6.45050704e-01 1.60639182e-01 -6.04723155e-01 6.99500501e-01
3.34555209e-01 2.36089870e-01 3.71545047e-01 9.10204351e-01
-1.31563246e-01 -1.16832554e+00 -2.71451265e-01 7.39947021e-01
-7.84247458e-01 -1.20982163e-01 2.81395704e-01 6.02055252e-01
1.13335036e-01 6.19465888e-01 -5.55851571e-02 1.24873400e-01
-2.97609381e-02 1.72576290e-02 6.03819549e-01 -3.36044520e-01
-4.23478097e-01 2.99973547e-01 -3.33703876e-01 -4.78739470e-01
-5.85332870e-01 -5.72898805e-01 -1.25256121e+00 -9.57643092e-02
-2.93402314e-01 -7.98492413e-03 8.44445586e-01 8.34677994e-01
2.61812896e-01 6.78137302e-01 5.88031948e-01 -7.20497370e-01
-3.49337459e-01 -9.84740019e-01 -6.70472324e-01 3.85011822e-01
1.65821642e-01 -6.48070455e-01 -2.87352532e-01 -3.37911338e-01] | [14.405495643615723, -2.4221982955932617] |
21078ba9-b9f1-4088-aca2-71c9ced730da | tsp-temporally-sensitive-pretraining-of-video | 2011.11479 | null | https://arxiv.org/abs/2011.11479v3 | https://arxiv.org/pdf/2011.11479v3.pdf | TSP: Temporally-Sensitive Pretraining of Video Encoders for Localization Tasks | Due to the large memory footprint of untrimmed videos, current state-of-the-art video localization methods operate atop precomputed video clip features. These features are extracted from video encoders typically trained for trimmed action classification tasks, making such features not necessarily suitable for temporal localization. In this work, we propose a novel supervised pretraining paradigm for clip features that not only trains to classify activities but also considers background clips and global video information to improve temporal sensitivity. Extensive experiments show that using features trained with our novel pretraining strategy significantly improves the performance of recent state-of-the-art methods on three tasks: Temporal Action Localization, Action Proposal Generation, and Dense Video Captioning. We also show that our pretraining approach is effective across three encoder architectures and two pretraining datasets. We believe video feature encoding is an important building block for localization algorithms, and extracting temporally-sensitive features should be of paramount importance in building more accurate models. The code and pretrained models are available on our project website. | ['Bernard Ghanem', 'Silvio Giancola', 'Humam Alwassel'] | 2020-11-23 | null | null | null | null | ['temporal-action-proposal-generation', 'dense-video-captioning'] | ['computer-vision', 'computer-vision'] | [ 5.57332158e-01 -5.31342268e-01 -6.90922797e-01 -4.08793241e-01
-1.09867489e+00 -4.78993863e-01 5.10928869e-01 -1.92697287e-01
-5.17915666e-01 6.60021126e-01 5.05782723e-01 1.42464027e-01
1.77629381e-01 -2.45944530e-01 -9.78383124e-01 -5.99885643e-01
-6.20865881e-01 -6.89124968e-03 6.56521440e-01 2.63000935e-01
2.93250591e-01 3.28417927e-01 -1.69478595e+00 8.65022600e-01
3.79820824e-01 1.33337188e+00 1.48904219e-01 8.77484620e-01
1.94273099e-01 1.30649185e+00 -4.65452254e-01 2.24161558e-02
2.09100321e-01 -7.03748584e-01 -7.67616808e-01 1.30962968e-01
6.98678553e-01 -6.20943964e-01 -7.32675910e-01 6.43246531e-01
1.85790867e-01 4.23750252e-01 4.39590901e-01 -1.24472129e+00
-4.58415478e-01 4.01653171e-01 -3.11827004e-01 7.13735878e-01
6.55925333e-01 1.00638680e-01 8.94934416e-01 -7.95211852e-01
6.95461154e-01 8.20597112e-01 7.36427605e-01 5.21419883e-01
-1.02800500e+00 -4.23537195e-01 2.53238320e-01 7.37831771e-01
-1.41894615e+00 -9.07038331e-01 6.42869532e-01 -4.97392654e-01
1.29380548e+00 9.98253152e-02 6.32883191e-01 1.36143732e+00
2.32610777e-01 1.01172495e+00 6.81770444e-01 -3.11870933e-01
2.01648980e-01 -2.29603261e-01 -4.42818642e-01 8.34570646e-01
-2.21676454e-01 1.96670130e-01 -1.00557303e+00 2.37025283e-02
9.72736239e-01 5.92333227e-02 -4.33858067e-01 -5.58685184e-01
-1.44043112e+00 6.11565530e-01 3.15325797e-01 5.92608333e-01
-2.76153862e-01 6.99643195e-01 5.81996679e-01 2.34025538e-01
3.74979168e-01 3.28492135e-01 -5.06120443e-01 -8.64349663e-01
-1.31136143e+00 -1.85544670e-01 3.20956439e-01 9.48514581e-01
6.84001207e-01 7.47643504e-03 -3.77592295e-01 5.18162072e-01
-4.64492217e-02 3.37510854e-01 7.14477241e-01 -1.32985568e+00
4.77141410e-01 3.65648210e-01 1.60094470e-01 -8.38695645e-01
-2.26065964e-01 2.73007974e-02 -2.18167409e-01 -2.13368177e-01
3.50836903e-01 -1.42770130e-02 -1.01221097e+00 1.70489430e+00
-4.20055650e-02 7.29679704e-01 -7.59240016e-02 9.90837991e-01
3.51832867e-01 6.95312560e-01 1.57928124e-01 -2.20473111e-01
9.64925647e-01 -1.18655694e+00 -6.84473097e-01 -2.34522671e-01
8.54890823e-01 -6.15845442e-01 6.86046004e-01 1.88426003e-01
-9.17665780e-01 -6.68098867e-01 -1.06828237e+00 -5.05370945e-02
-2.55488485e-01 4.30140316e-01 8.15959215e-01 2.72452563e-01
-1.07921576e+00 7.20796704e-01 -1.34472954e+00 -6.67456985e-01
5.01367033e-01 4.32515651e-01 -6.77775621e-01 -2.36978829e-01
-9.71005797e-01 7.84147143e-01 4.02331978e-01 4.29233760e-02
-1.07895815e+00 -2.45817408e-01 -9.90018606e-01 5.01309372e-02
4.03061122e-01 -2.51085490e-01 1.34210026e+00 -1.35271192e+00
-1.39986396e+00 4.12180126e-01 -3.33463371e-01 -6.69977725e-01
2.74282247e-01 -3.15242410e-01 -4.72803622e-01 8.54648709e-01
1.16784245e-01 9.41845834e-01 1.17242229e+00 -6.13055944e-01
-8.02720726e-01 5.22809811e-02 -5.90001838e-03 7.07736909e-02
-4.68092442e-01 2.43800148e-01 -5.81655204e-01 -7.11966872e-01
-1.51167974e-01 -8.64879608e-01 -1.19905904e-01 1.39868394e-01
2.61995018e-01 1.56541690e-02 9.78458405e-01 -6.33804917e-01
1.21642280e+00 -2.20035362e+00 2.70550419e-02 -1.08076483e-01
-2.84838796e-01 1.71554357e-01 -3.49831879e-01 3.70743752e-01
-5.64400153e-03 -1.74830586e-01 -1.43895007e-03 -1.88234150e-01
-1.00323804e-01 1.62395492e-01 -2.21763819e-01 6.00189805e-01
4.28287148e-01 9.03620780e-01 -1.03881514e+00 -7.10089684e-01
4.05890614e-01 5.88235974e-01 -5.40792584e-01 2.52178818e-01
-2.03148589e-01 5.60790002e-01 -4.56298083e-01 9.09045160e-01
5.36817722e-02 -3.09621602e-01 9.22500193e-02 -2.61185527e-01
2.36031394e-02 2.03428403e-01 -7.45363116e-01 2.21479368e+00
-2.55389333e-01 9.36129987e-01 -2.68293828e-01 -1.09100497e+00
4.58805025e-01 5.64910173e-01 1.04168117e+00 -7.66982973e-01
1.06249042e-01 -1.92085363e-03 -2.10368782e-01 -6.42655730e-01
5.35785913e-01 2.34855756e-01 -1.48534914e-02 3.69110256e-01
4.76756871e-01 4.69015330e-01 4.75083798e-01 8.66633803e-02
1.62722516e+00 7.30655670e-01 1.66838035e-01 1.63639769e-01
4.53149050e-01 7.03617632e-02 7.02687502e-01 6.74425364e-01
-5.80520213e-01 7.74492979e-01 3.39666784e-01 -6.46548033e-01
-8.54902387e-01 -8.34092200e-01 9.20788497e-02 1.42090690e+00
-4.54494022e-02 -7.64710426e-01 -6.12952054e-01 -9.83742297e-01
-1.78673521e-01 2.28591576e-01 -7.16632724e-01 -2.34887570e-01
-7.49121964e-01 -1.59390554e-01 5.48194528e-01 1.12500751e+00
3.98169726e-01 -1.06368113e+00 -8.82009387e-01 3.64069462e-01
-3.97085160e-01 -1.36357379e+00 -7.89748609e-01 2.66874582e-01
-8.95123601e-01 -1.15758204e+00 -6.34993494e-01 -7.18651414e-01
6.02564156e-01 3.90307426e-01 9.11334038e-01 -3.45572829e-02
-3.09458911e-01 6.75843298e-01 -8.15840960e-01 1.87221140e-01
-1.19375832e-01 5.61530963e-02 1.13374755e-01 1.95389211e-01
6.75360978e-01 -5.36665797e-01 -5.24381876e-01 4.64030415e-01
-8.79497111e-01 -4.13513221e-02 7.35956311e-01 6.63093507e-01
5.50217688e-01 -9.52707008e-02 2.42957413e-01 -2.18354076e-01
-8.63518864e-02 -3.22332591e-01 -4.64066625e-01 3.14667404e-01
-4.11448479e-02 2.18462154e-01 4.65083182e-01 -6.23372912e-01
-8.90819609e-01 6.18445754e-01 5.19373640e-02 -9.39621806e-01
-2.98727721e-01 3.28569561e-01 1.46304473e-01 -3.29619884e-01
5.20520449e-01 3.32782716e-01 -1.86246559e-01 -3.91223311e-01
9.27070081e-02 4.75208253e-01 5.72783709e-01 -4.08143193e-01
5.23022115e-01 5.49457431e-01 -8.31638202e-02 -6.68223977e-01
-7.48417854e-01 -7.19823241e-01 -1.00831330e+00 -4.03348416e-01
1.00264943e+00 -1.04694223e+00 -4.13952768e-01 2.00181171e-01
-1.04521787e+00 -4.25834000e-01 -1.49584159e-01 8.37744355e-01
-9.78080332e-01 3.38714659e-01 -5.56806803e-01 -4.79686618e-01
1.46209687e-01 -1.08099163e+00 1.26568592e+00 1.23823090e-02
-3.18980247e-01 -8.76936674e-01 1.31667748e-01 3.86535048e-01
4.25272048e-01 1.44717470e-01 3.02789360e-01 -5.14901340e-01
-9.20160592e-01 -3.24149400e-01 -4.35458086e-02 3.16838086e-01
1.54189363e-01 -1.12495512e-01 -9.02813196e-01 -3.21698606e-01
-3.03110063e-01 -5.92582643e-01 1.17761755e+00 4.62180436e-01
1.11746192e+00 -1.26891702e-01 -4.95060951e-01 6.62306547e-01
1.15300596e+00 2.13110700e-01 6.61860824e-01 3.01175714e-01
5.88826537e-01 7.67427832e-02 9.71804261e-01 5.72872579e-01
1.33039638e-01 8.86848867e-01 2.35133082e-01 2.15809092e-01
-1.10599056e-01 -3.28199387e-01 8.74811471e-01 5.53892493e-01
-2.89082468e-01 -1.66145027e-01 -6.58770919e-01 6.55750155e-01
-2.17150331e+00 -1.45747459e+00 4.92638797e-01 2.03323793e+00
7.74534762e-01 -3.61143686e-02 1.92955241e-01 -9.39403698e-02
5.67122817e-01 3.94742221e-01 -3.55784625e-01 -9.20149460e-02
1.77179068e-01 2.60766208e-01 5.83872437e-01 1.01043135e-01
-1.71972334e+00 9.56193984e-01 6.79821873e+00 5.00693321e-01
-1.14570725e+00 1.87674552e-01 2.36139685e-01 -3.68910789e-01
3.36418748e-01 6.98503107e-02 -6.07582927e-01 5.47013402e-01
1.24747908e+00 1.34800196e-01 2.62608409e-01 1.00098658e+00
2.70147592e-01 -3.49647582e-01 -1.51131558e+00 1.12992465e+00
4.17654783e-01 -1.35241640e+00 -2.21149191e-01 -4.77182455e-02
6.94176018e-01 1.27795279e-01 -9.95312631e-02 2.85591602e-01
-1.58703789e-01 -9.22819436e-01 7.10746884e-01 5.31013906e-01
8.34958434e-01 -5.33522487e-01 6.90448165e-01 -1.34415060e-01
-1.52792275e+00 -2.26289570e-01 -3.17419201e-01 -7.06942528e-02
3.90825212e-01 8.32424089e-02 -6.78136826e-01 1.62102968e-01
6.51953042e-01 1.25317490e+00 -6.17831945e-01 1.33787918e+00
-1.68128639e-01 6.41578019e-01 -1.42326832e-01 2.41682827e-01
3.27310264e-01 3.48217010e-01 1.63840994e-01 1.39320993e+00
4.86037105e-01 3.37992534e-02 4.81875539e-01 2.60586381e-01
-4.56285626e-02 -1.39240921e-01 -6.80073798e-01 -5.24742961e-01
1.95448026e-01 9.82213974e-01 -8.06565344e-01 -3.03140908e-01
-7.62457430e-01 1.32589316e+00 2.91294277e-01 3.49716991e-01
-1.23352492e+00 -3.40761930e-01 7.87450433e-01 1.07346848e-01
8.07026029e-01 -5.34783006e-01 4.28032517e-01 -1.44508982e+00
2.71465648e-02 -7.51381576e-01 3.82604659e-01 -7.68617749e-01
-8.21382403e-01 4.71102446e-01 -1.26478216e-02 -1.63874936e+00
-6.82917833e-01 -6.23235762e-01 -3.66498441e-01 1.40906528e-01
-1.23760092e+00 -1.21556497e+00 -2.95237839e-01 7.53919363e-01
6.97260439e-01 -6.28125742e-02 8.34348261e-01 3.91446263e-01
-4.05677795e-01 4.71373439e-01 3.65209803e-02 3.43036503e-01
9.85833704e-01 -8.45376730e-01 1.57173708e-01 1.06834936e+00
6.52014494e-01 3.90009433e-01 3.93361449e-01 -4.82939303e-01
-1.54792511e+00 -1.19819474e+00 7.10898638e-01 -5.03397465e-01
7.20829129e-01 -3.29077244e-01 -6.17996156e-01 9.96915281e-01
1.40442148e-01 4.10419106e-01 7.49898553e-01 -3.17023806e-02
-3.26522231e-01 -2.15639248e-01 -6.81000650e-01 1.90405071e-01
1.20324826e+00 -7.73703992e-01 -5.13638556e-01 4.61628556e-01
4.14707631e-01 -4.08125222e-01 -8.25101852e-01 2.79834300e-01
7.20584154e-01 -9.08752739e-01 7.83317983e-01 -6.47145629e-01
3.69961441e-01 -3.32194656e-01 -2.71692485e-01 -8.29477727e-01
-3.93810719e-01 -6.12927020e-01 -5.84013820e-01 8.64845693e-01
2.80707091e-01 -8.69141370e-02 9.40331519e-01 4.05916393e-01
-2.02859536e-01 -5.01034141e-01 -1.04132199e+00 -9.27968562e-01
-6.75896585e-01 -4.60782349e-01 9.37087163e-02 7.46735513e-01
2.52432466e-01 2.01490503e-02 -6.91281617e-01 -3.36457156e-02
2.08075464e-01 1.45054355e-01 5.34413218e-01 -6.71158433e-01
-4.11388129e-01 -9.88425408e-03 -1.03552377e+00 -1.25504863e+00
3.34235847e-01 -4.68536019e-01 3.34261686e-01 -1.33687854e+00
2.71433443e-01 -6.41307831e-02 -6.67531013e-01 8.70957375e-01
1.96437493e-01 6.36913598e-01 1.02579638e-01 1.70145914e-01
-1.26683235e+00 5.63242555e-01 8.32231581e-01 -1.03859730e-01
-8.14063400e-02 -2.17949644e-01 -2.13872671e-01 5.52009642e-01
6.06256425e-01 -4.67545450e-01 -5.73067188e-01 -5.23002565e-01
-4.21758384e-01 1.23018213e-02 4.48613167e-01 -1.56348252e+00
3.42967242e-01 -3.30119491e-01 8.04866672e-01 -4.14671957e-01
7.54745841e-01 -8.57828140e-01 -9.72416550e-02 1.36483461e-01
-3.60934913e-01 7.41889924e-02 2.47027472e-01 5.82635581e-01
-5.69487989e-01 -1.52699854e-02 5.84164739e-01 -1.44530997e-01
-1.37793434e+00 2.95216203e-01 -5.29479980e-01 4.07159608e-03
1.26997197e+00 -3.10537606e-01 -1.43872529e-01 -5.76284647e-01
-6.88269675e-01 -8.06638449e-02 6.21969104e-01 5.25595129e-01
7.10901678e-01 -1.48023653e+00 -2.76215225e-01 2.51440555e-01
1.57986194e-01 -7.17157423e-01 1.22325778e-01 1.10227096e+00
-4.52972680e-01 7.27422297e-01 -4.45109427e-01 -7.29797006e-01
-1.36594534e+00 6.89498782e-01 1.49625853e-01 -1.19478609e-02
-4.91365790e-01 9.16649938e-01 -9.02301371e-02 4.42696065e-01
4.36923146e-01 -5.13930738e-01 1.12647459e-01 -2.31824610e-02
8.23473692e-01 1.73981935e-01 -7.46563002e-02 -8.72406721e-01
-6.41277671e-01 5.40225089e-01 -1.87223226e-01 -1.28760159e-01
1.26150870e+00 8.74514580e-02 2.80972987e-01 2.68978864e-01
1.39600158e+00 -3.08921516e-01 -1.67506230e+00 -1.14668734e-01
9.25733522e-02 -7.98894286e-01 7.05942363e-02 -5.43527961e-01
-1.02395642e+00 7.78618276e-01 5.61916232e-01 -2.94445485e-01
1.40637767e+00 1.13427512e-01 8.82894814e-01 4.55401361e-01
6.24914527e-01 -1.27664864e+00 3.92225474e-01 5.16290069e-01
5.77681482e-01 -1.23959935e+00 6.16380246e-03 -1.03701115e-01
-6.66128278e-01 1.15057588e+00 7.06290185e-01 5.57708042e-03
4.38385308e-01 3.07644099e-01 -5.11683039e-02 9.01696831e-02
-9.51946974e-01 -3.29428732e-01 4.09454167e-01 5.99658847e-01
5.49556911e-01 -4.18547601e-01 8.67180079e-02 3.46161693e-01
4.77099717e-01 3.99703562e-01 2.62225509e-01 1.47047782e+00
-5.79031825e-01 -1.17807817e+00 -1.14551120e-01 4.16506767e-01
-4.84397382e-01 -5.52886026e-03 -3.04556102e-01 5.90530992e-01
2.58433610e-01 7.05261290e-01 2.03301758e-01 -5.54293394e-01
-2.32476778e-02 2.22261280e-01 7.73268640e-01 -5.11596322e-01
-1.95960775e-01 -2.86698574e-03 2.01232612e-01 -1.24106586e+00
-9.03293490e-01 -7.95871854e-01 -1.17337477e+00 3.51435021e-02
-2.14031726e-01 1.40590698e-01 4.11365479e-01 9.44805562e-01
6.40800059e-01 3.34201485e-01 4.17340785e-01 -1.27279103e+00
-3.03350866e-01 -9.14770126e-01 -3.11351329e-01 4.02211934e-01
4.17198032e-01 -9.93980527e-01 -1.09502368e-01 6.06735170e-01] | [8.513266563415527, 0.6222942471504211] |
1672c094-565a-4b29-a443-8baa60841e32 | norma-neighborhood-sensitive-maps-for | null | null | https://aclanthology.org/D18-1047 | https://aclanthology.org/D18-1047.pdf | NORMA: Neighborhood Sensitive Maps for Multilingual Word Embeddings | Inducing multilingual word embeddings by learning a linear map between embedding spaces of different languages achieves remarkable accuracy on related languages. However, accuracy drops substantially when translating between distant languages. Given that languages exhibit differences in vocabulary, grammar, written form, or syntax, one would expect that embedding spaces of different languages have different structures especially for distant languages. With the goal of capturing such differences, we propose a method for learning neighborhood sensitive maps, NORMA. Our experiments show that NORMA outperforms current state-of-the-art methods for word translation between distant languages. | ['Ndapa Nakashole'] | 2018-10-01 | null | null | null | emnlp-2018-10 | ['multilingual-word-embeddings'] | ['methodology'] | [-5.35039186e-01 -1.97428495e-01 -5.52405953e-01 -2.67862737e-01
-9.75419581e-01 -9.82130110e-01 8.04382503e-01 2.34331220e-01
-4.97372627e-01 6.16696000e-01 8.01209331e-01 -6.01527393e-01
1.70231298e-01 -7.71984041e-01 -6.46872461e-01 -1.40822038e-01
1.91088513e-01 5.89781642e-01 -1.33596748e-01 -6.17332637e-01
-2.76363548e-02 2.85315454e-01 -6.58507228e-01 1.88581899e-01
1.03992450e+00 9.81446076e-03 1.05102204e-01 1.44712567e-01
-5.18825829e-01 2.07026377e-01 -2.17909694e-01 -5.76027632e-01
3.36476237e-01 -4.79178041e-01 -6.93628013e-01 -4.29991692e-01
5.81380010e-01 1.76322907e-01 -4.99029785e-01 1.24940372e+00
3.94870967e-01 -6.74518049e-02 8.20691466e-01 -7.52952218e-01
-1.74413025e+00 9.16140795e-01 -1.35186717e-01 3.30720335e-01
4.44465786e-01 -1.30043209e-01 1.34252930e+00 -1.34285283e+00
7.85134971e-01 1.43517494e+00 8.33618820e-01 3.25298846e-01
-1.41093159e+00 -4.02562559e-01 2.93099672e-01 -7.61447474e-03
-1.66679621e+00 -2.90710121e-01 6.72882915e-01 -6.20778739e-01
9.88502324e-01 2.41620019e-02 3.74372512e-01 1.19890761e+00
4.11807239e-01 3.96821588e-01 1.12964773e+00 -6.60807848e-01
-1.39292642e-01 5.93469024e-01 1.05030805e-01 6.77445948e-01
4.96919066e-01 -8.08191150e-02 -4.64723825e-01 -1.40019953e-01
4.75440562e-01 -2.06752658e-01 -1.75622344e-01 -5.01648366e-01
-1.53395700e+00 1.11912394e+00 3.28518093e-01 7.39666760e-01
-1.64087731e-02 -1.16454504e-01 3.53646338e-01 6.47829056e-01
7.26370454e-01 6.99464262e-01 -3.98411363e-01 1.51735991e-01
-5.23757875e-01 -5.93637116e-02 6.15471005e-01 1.01835191e+00
8.31603706e-01 -7.83126280e-02 -8.26288611e-02 7.95542419e-01
2.65638322e-01 6.15967751e-01 5.47640681e-01 -3.25115919e-01
7.42877126e-01 4.97010797e-01 4.53545190e-02 -8.98324192e-01
-3.01802784e-01 -2.84463763e-01 -4.17290032e-01 -2.66490072e-01
4.00980920e-01 -2.66512875e-02 -2.79373288e-01 1.75528252e+00
-7.55998306e-03 -1.65081352e-01 3.75974119e-01 7.54082859e-01
4.83605713e-01 7.35865891e-01 -5.47506139e-02 1.28961831e-01
1.24860811e+00 -8.93504679e-01 -7.48914123e-01 -3.25123906e-01
8.67290258e-01 -1.00120533e+00 1.44058192e+00 -3.66869360e-01
-8.28909874e-01 -5.00483334e-01 -9.61879253e-01 -3.23056936e-01
-8.11868787e-01 8.62202123e-02 4.73029584e-01 4.17918324e-01
-1.09917080e+00 3.75587881e-01 -6.23807192e-01 -8.54055762e-01
-3.08586918e-02 1.14892181e-02 -5.45977771e-01 -3.99469063e-02
-1.54342926e+00 1.40384185e+00 1.96161896e-01 -3.03370893e-01
-4.12475169e-01 -8.98011982e-01 -1.23635268e+00 -1.38856784e-01
-4.73150194e-01 -4.97579873e-01 7.04882383e-01 -7.56325960e-01
-1.13771915e+00 1.02907372e+00 -1.45355329e-01 -1.30163923e-01
2.44501233e-01 -1.75058022e-01 -7.23023891e-01 -5.25267065e-01
2.99367845e-01 4.81774449e-01 4.19125944e-01 -7.67927229e-01
-4.16468054e-01 -4.05845970e-01 4.26834002e-02 3.89000624e-01
-7.59844899e-01 2.98116505e-01 -1.89534038e-01 -7.33480155e-01
-4.83029634e-02 -9.12735939e-01 -6.13584518e-02 6.13706037e-02
-1.01652227e-01 -4.75836545e-01 5.50901532e-01 -8.87995243e-01
1.18406069e+00 -2.22190285e+00 3.43261570e-01 2.33558714e-02
-1.37563571e-01 6.63099363e-02 -5.99188864e-01 5.33421874e-01
6.18674010e-02 1.53010026e-01 -1.63273096e-01 2.83190589e-02
4.05382127e-01 3.02469969e-01 -5.06767809e-01 7.15411305e-01
3.40876490e-01 1.03228486e+00 -1.09881091e+00 -3.18640441e-01
1.14742815e-01 7.60049343e-01 -4.20696944e-01 -1.44429699e-01
2.05133155e-01 2.09791094e-01 -1.64104819e-01 5.12253463e-01
4.20851320e-01 9.25646424e-02 3.79631519e-01 -1.17645279e-01
-2.15483636e-01 8.17640841e-01 -7.40429878e-01 2.06830263e+00
-8.31510603e-01 1.06359673e+00 -3.55416626e-01 -7.94122994e-01
1.07147682e+00 3.72304469e-01 4.45665605e-02 -6.74439847e-01
-2.77944729e-02 5.55175602e-01 8.62969682e-02 -1.86843455e-01
5.09432614e-01 -1.96915478e-01 -3.62364769e-01 5.13813138e-01
1.85667470e-01 -1.06246412e-01 1.08365171e-01 -1.71654150e-01
7.83100247e-01 -5.39098354e-03 3.18356127e-01 -7.65433908e-01
4.64120537e-01 -1.88663434e-02 5.03676534e-01 1.05799474e-01
-4.19182301e-01 4.17244226e-01 2.92065382e-01 -6.94996774e-01
-1.24694288e+00 -1.51578140e+00 -4.07447487e-01 1.24239540e+00
1.52154431e-01 -3.58425409e-01 -6.21238112e-01 -8.20338070e-01
1.41735047e-01 7.74037838e-01 -5.17459631e-01 -1.30973205e-01
-8.85706007e-01 -4.89517748e-01 6.16695702e-01 5.66095769e-01
-1.57710552e-01 -8.66903841e-01 2.61683077e-01 1.80995509e-01
-1.65534884e-01 -1.27000821e+00 -1.14932191e+00 3.64085585e-02
-7.10386992e-01 -7.05031335e-01 -6.93231583e-01 -1.51542723e+00
6.96931481e-01 1.76939055e-01 1.38035178e+00 -4.80287999e-01
1.02492221e-01 2.36977026e-01 -1.11126289e-01 -8.91416594e-02
-4.72610831e-01 3.51136625e-01 5.71868181e-01 -2.50227541e-01
1.05471742e+00 -4.74489421e-01 -2.06278607e-01 5.48711121e-02
-6.40650272e-01 -4.69389588e-01 3.62248033e-01 6.15717769e-01
5.44491112e-01 -5.82213938e-01 2.32345790e-01 -5.89744270e-01
9.79479969e-01 -4.70920980e-01 -5.56467354e-01 6.63465679e-01
-6.34926856e-01 5.77592134e-01 6.83149338e-01 -5.66826224e-01
-5.70922256e-01 -3.93728495e-01 1.65417820e-01 2.76752189e-02
2.27140442e-01 2.62146026e-01 -4.72949892e-02 -2.67933458e-02
8.15119922e-01 1.88669533e-01 -4.48716670e-01 -4.45900440e-01
7.75188088e-01 7.64264107e-01 2.03940064e-01 -5.23190975e-01
1.05667388e+00 2.58408755e-01 -5.71721375e-01 -6.15025282e-01
-6.13904417e-01 -2.35616192e-01 -9.87347722e-01 3.32762241e-01
9.81626332e-01 -1.09823275e+00 3.36598635e-01 -3.10568213e-01
-1.49923134e+00 7.11320192e-02 -3.41159582e-01 8.57461035e-01
-1.83968738e-01 -4.85455208e-02 -6.56893849e-01 2.50984868e-03
-2.20874950e-01 -1.03568709e+00 9.36634302e-01 -1.85675159e-01
-6.28784359e-01 -1.80267251e+00 9.02706861e-01 -3.78340334e-02
5.19410253e-01 -1.22916162e-01 1.19078970e+00 -7.50234365e-01
-1.56320691e-01 4.85894978e-02 -6.30215555e-02 3.05858940e-01
6.29141212e-01 -2.40519986e-01 -4.99300331e-01 -3.75621468e-01
-2.92121142e-01 -1.23417387e-02 4.70242918e-01 1.35508358e-01
1.26939625e-01 -3.52929264e-01 -3.14485192e-01 7.74634719e-01
1.48338842e+00 -1.45313635e-01 1.44121826e-01 5.19302785e-01
7.88731694e-01 5.43844640e-01 2.40740865e-01 -1.96808010e-01
4.87604380e-01 7.37106979e-01 -3.58629018e-01 -2.18595222e-01
-2.94762492e-01 -5.04947543e-01 9.59053576e-01 1.59247422e+00
4.64309871e-01 1.50662258e-01 -1.13100982e+00 1.13038516e+00
-1.50695503e+00 -6.18729830e-01 1.40196577e-01 1.99242187e+00
9.49119031e-01 -1.84491754e-01 -1.85940504e-01 -5.80452859e-01
7.89929330e-01 3.56080443e-01 -2.75277644e-01 -9.71310794e-01
-4.89551902e-01 2.25258976e-01 6.46285892e-01 9.34004843e-01
-8.58621240e-01 1.36062503e+00 7.17568207e+00 3.94658357e-01
-1.09709537e+00 4.65794891e-01 1.30001947e-01 7.38357455e-02
-9.91513431e-01 9.14833248e-02 -7.56054580e-01 1.60962403e-01
7.64925420e-01 -5.09312332e-01 5.09077966e-01 6.59302950e-01
5.64396828e-02 8.80219698e-01 -1.29727960e+00 8.96760166e-01
3.36550236e-01 -1.24252987e+00 3.46426725e-01 -8.37848932e-02
1.14229751e+00 4.40301478e-01 3.19555104e-01 3.66057515e-01
5.58752418e-01 -1.12872875e+00 4.87529665e-01 2.44149536e-01
8.85675371e-01 -8.63877833e-01 5.72812796e-01 8.01687501e-03
-1.27332544e+00 4.62680161e-01 -8.28601897e-01 -1.04202859e-01
4.71986867e-02 4.79079336e-01 -5.23125768e-01 2.60742784e-01
3.71911973e-01 8.15589309e-01 -4.80105609e-01 2.21504554e-01
-4.52104658e-01 4.41196293e-01 9.46483687e-02 1.19884545e-02
4.66656506e-01 -6.36279225e-01 5.18638790e-01 1.39070547e+00
5.96165359e-01 -5.54576457e-01 1.23246007e-01 1.00450408e+00
-2.30482891e-01 6.88302100e-01 -1.25009286e+00 -3.18821728e-01
6.46017849e-01 9.10354972e-01 -1.97130710e-01 -1.45955011e-01
-9.07390952e-01 1.22210789e+00 7.10754812e-01 5.07757604e-01
-8.77059758e-01 -4.59414780e-01 1.41940439e+00 1.97050959e-01
1.53207570e-01 -4.26724851e-01 -4.08766776e-01 -1.39317954e+00
3.70615542e-01 -7.49152780e-01 1.13981456e-01 -3.62395287e-01
-1.57003808e+00 7.90067554e-01 -3.53493750e-01 -1.21078575e+00
5.44199487e-03 -8.66379201e-01 -4.89477843e-01 1.12791765e+00
-1.48164046e+00 -1.17795241e+00 4.14098144e-01 4.87434357e-01
5.96817315e-01 -6.35618210e-01 1.06276572e+00 2.73342222e-01
-2.54608482e-01 8.52469683e-01 6.27990425e-01 4.34583217e-01
1.09468627e+00 -1.21422493e+00 8.84544015e-01 8.16904128e-01
9.43002522e-01 7.84034133e-01 5.08991063e-01 -3.80004376e-01
-1.38992310e+00 -1.33654583e+00 1.94273043e+00 -8.39252353e-01
1.19163430e+00 -6.50812805e-01 -8.93079102e-01 9.65645730e-01
6.48865461e-01 1.59429848e-01 9.04698253e-01 4.65345860e-01
-8.25988293e-01 -1.03165805e-01 -8.48548532e-01 9.26879883e-01
1.04487264e+00 -1.23664320e+00 -8.97080958e-01 6.29114091e-01
1.00472021e+00 7.84130245e-02 -9.87081230e-01 8.39833170e-03
4.13116395e-01 -3.16815346e-01 9.24594104e-01 -8.98615360e-01
5.24459593e-02 -2.72421211e-01 -5.39114833e-01 -1.78242469e+00
-5.32321036e-01 -5.20800114e-01 1.85458004e-01 1.12142277e+00
7.08027184e-01 -8.89607787e-01 2.80411094e-01 1.78705245e-01
1.21103324e-01 -2.71561384e-01 -9.82621074e-01 -1.24088562e+00
1.03479588e+00 -1.82419598e-01 6.67656302e-01 1.47369862e+00
4.07729238e-01 6.71777129e-01 -3.67762633e-02 2.18744755e-01
3.91094804e-01 6.67157993e-02 2.93913990e-01 -9.88901615e-01
2.45482735e-02 -5.27584732e-01 -6.45238280e-01 -8.01343679e-01
7.15422213e-01 -1.48574924e+00 -1.69579610e-01 -1.58512509e+00
1.71228513e-01 -2.39200369e-01 -5.00238478e-01 2.14575469e-01
-3.30919087e-01 3.42920095e-01 2.94678435e-02 2.51400441e-01
-3.97316009e-01 5.48901260e-01 9.21946406e-01 -4.25831020e-01
-8.74099135e-02 -7.14823484e-01 -7.71200657e-01 6.61239505e-01
8.26856136e-01 -5.65212846e-01 -2.27018252e-01 -1.07359433e+00
3.67921591e-01 -6.33276999e-01 -2.53806412e-01 -5.80288589e-01
-4.43048403e-02 -1.73619673e-01 -2.60632727e-02 -9.01344568e-02
-1.37933493e-01 -6.38032615e-01 -1.03033535e-01 4.14100200e-01
-4.64244336e-01 7.90418863e-01 1.49813578e-01 2.47015432e-01
-5.01204789e-01 9.58082005e-02 5.69639146e-01 5.96188605e-02
-5.91332912e-01 1.74481958e-01 -3.72385055e-01 4.58884627e-01
8.64355445e-01 9.29370970e-02 -1.15497015e-01 4.71902313e-03
-4.40578163e-01 -1.37324288e-01 7.24406004e-01 1.04824162e+00
3.08431178e-01 -2.10132670e+00 -1.19180024e+00 1.74798816e-01
5.40986955e-01 -7.32364178e-01 -5.46688259e-01 7.16214418e-01
-5.64508498e-01 7.51145482e-01 -1.50939539e-01 -3.43153477e-01
-1.09935772e+00 5.64323187e-01 2.67627507e-01 -3.27089369e-01
-4.36932296e-01 6.47110403e-01 3.58183742e-01 -1.29755723e+00
-5.85511103e-02 -4.49652195e-01 6.61637187e-02 7.09602982e-02
3.63828748e-01 2.11292878e-01 -1.54759781e-02 -9.55668330e-01
-6.04376256e-01 8.80803466e-01 -9.45238099e-02 -2.19607770e-01
1.11950588e+00 -1.35045677e-01 -2.54562616e-01 7.72812426e-01
1.49674344e+00 5.24822950e-01 -6.39477015e-01 -6.18430257e-01
3.00189674e-01 -3.55868250e-01 -2.33543679e-01 -2.81126469e-01
-6.60068333e-01 9.01502788e-01 6.83013022e-01 -2.65846938e-01
5.60129106e-01 2.19659284e-01 8.64069164e-01 2.91694164e-01
4.84071255e-01 -1.17174566e+00 -3.39057326e-01 9.99153197e-01
8.18047643e-01 -1.40190053e+00 -2.11334363e-01 3.78469564e-02
-4.78850484e-01 9.54546094e-01 2.93310851e-01 -4.15340871e-01
6.68431580e-01 1.77558124e-01 5.89937270e-01 8.14230368e-02
-4.98480886e-01 -3.21489200e-02 7.06532300e-01 7.77438283e-01
9.55813289e-01 3.97399545e-01 -6.66070998e-01 1.77667350e-01
-4.09343183e-01 -7.26156294e-01 1.41684681e-01 5.70484281e-01
-2.73747772e-01 -1.51739299e+00 -4.86508220e-01 -1.98975042e-01
-1.53292730e-01 -6.52745247e-01 -6.41518295e-01 9.00643349e-01
1.38579726e-01 6.22337162e-01 3.45368594e-01 -2.37127468e-01
2.94973254e-01 3.51704836e-01 5.88167846e-01 -7.03553617e-01
-4.87306207e-01 -3.57204258e-01 -1.35171160e-01 -4.10710990e-01
-6.79187849e-02 -7.32779443e-01 -9.80600417e-01 -4.21918958e-01
-1.17919050e-01 3.38311583e-01 6.60145938e-01 8.33375394e-01
4.03008312e-01 2.77788699e-01 4.72517669e-01 -2.11686835e-01
-6.06806755e-01 -9.58201885e-01 -4.17613477e-01 5.19958019e-01
2.53868252e-01 -2.81940699e-01 -2.86390603e-01 -1.52137220e-01] | [11.006823539733887, 9.988432884216309] |
cd95533e-756c-4c8f-8ff1-854f69692f22 | end-to-end-spoken-language-understanding-3 | 2207.08179 | null | https://arxiv.org/abs/2207.08179v1 | https://arxiv.org/pdf/2207.08179v1.pdf | End-to-End Spoken Language Understanding: Performance analyses of a voice command task in a low resource setting | Spoken Language Understanding (SLU) is a core task in most human-machine interaction systems. With the emergence of smart homes, smart phones and smart speakers, SLU has become a key technology for the industry. In a classical SLU approach, an Automatic Speech Recognition (ASR) module transcribes the speech signal into a textual representation from which a Natural Language Understanding (NLU) module extracts semantic information. Recently End-to-End SLU (E2E SLU) based on Deep Neural Networks has gained momentum since it benefits from the joint optimization of the ASR and the NLU parts, hence limiting the cascade of error effect of the pipeline architecture. However, little is known about the actual linguistic properties used by E2E models to predict concepts and intents from speech input. In this paper, we present a study identifying the signal features and other linguistic properties used by an E2E model to perform the SLU task. The study is carried out in the application domain of a smart home that has to handle non-English (here French) voice commands. The results show that a good E2E SLU performance does not always require a perfect ASR capability. Furthermore, the results show the superior capabilities of the E2E model in handling background noise and syntactic variation compared to the pipeline model. Finally, a finer-grained analysis suggests that the E2E model uses the pitch information of the input signal to identify voice command concepts. The results and methodology outlined in this paper provide a springboard for further analyses of E2E models in speech processing. | ['Michel Vacher', 'François Portet', 'Thierry Desot'] | 2022-07-17 | null | null | null | null | ['spoken-language-understanding', 'spoken-language-understanding'] | ['natural-language-processing', 'speech'] | [ 1.67849854e-01 3.60206753e-01 2.56303728e-01 -5.80007851e-01
-6.17708385e-01 -3.93450826e-01 5.04505157e-01 -7.58384690e-02
-2.97935754e-01 2.71213651e-01 5.44682264e-01 -4.23553884e-01
2.76819348e-01 -5.17872930e-01 -3.11129659e-01 -2.67540574e-01
2.80731115e-02 3.68242621e-01 -7.00135827e-02 -4.45527554e-01
-1.61106542e-01 2.79316783e-01 -1.66623735e+00 6.29763067e-01
4.93555009e-01 1.05987620e+00 5.55242121e-01 8.97614002e-01
-3.52250427e-01 8.28907251e-01 -7.12693036e-01 3.14103633e-01
1.28236614e-04 -5.98655581e-01 -8.83476496e-01 -5.00551797e-02
-8.65163952e-02 -3.16569567e-01 -6.91218674e-02 7.31960773e-01
6.63735151e-01 2.03070179e-01 3.94140273e-01 -9.17003274e-01
-2.13776350e-01 8.36117208e-01 2.68322021e-01 2.07608864e-01
8.98771107e-01 2.13658988e-01 1.19548678e+00 -8.78158867e-01
4.31186616e-01 1.48822403e+00 5.42290807e-01 7.04699993e-01
-1.18458724e+00 -3.73831034e-01 4.75304723e-02 9.17800292e-02
-1.32888186e+00 -8.09292614e-01 5.61224043e-01 -2.73915917e-01
1.76614630e+00 2.84465820e-01 2.50818938e-01 1.04297304e+00
-1.37798628e-02 1.14061892e+00 9.25006092e-01 -6.69230878e-01
5.05592942e-01 4.06135142e-01 4.68655229e-01 4.51611638e-01
-6.45690501e-01 1.43970430e-01 -7.01297343e-01 3.23775969e-02
2.93478608e-01 -4.05944318e-01 -4.89201128e-01 4.08086032e-01
-1.03138244e+00 8.02445889e-01 2.09979117e-01 6.83299541e-01
-5.55458307e-01 -1.23067178e-01 6.34207726e-01 3.62474591e-01
5.07783711e-01 3.53613824e-01 -7.59865224e-01 -5.22573948e-01
-9.05288696e-01 -2.34335318e-01 1.13181722e+00 7.69900322e-01
5.80759943e-01 2.24089816e-01 -1.00527227e-01 1.06852233e+00
5.64602137e-01 3.47542137e-01 8.68796229e-01 -5.64375579e-01
3.16720247e-01 5.08800626e-01 -2.05685109e-01 -3.79768550e-01
-5.38622916e-01 -1.24591421e-02 -3.70651841e-01 7.15105534e-02
3.50457966e-01 -3.59816223e-01 -7.27388322e-01 1.57931268e+00
-2.28070598e-02 -1.00013286e-01 3.56259108e-01 7.43518472e-01
8.71897995e-01 1.09068966e+00 2.39275157e-01 -8.81178491e-03
1.48003852e+00 -6.76862001e-01 -9.00717974e-01 -5.71565568e-01
7.84294963e-01 -6.89990759e-01 1.26753986e+00 4.23518628e-01
-8.06179047e-01 -6.91458941e-01 -9.19415295e-01 -6.88835755e-02
-4.14338022e-01 2.88686752e-01 2.70903617e-01 8.50508034e-01
-1.15281856e+00 3.82329673e-01 -7.11283028e-01 -8.07517111e-01
-1.05910534e-02 3.94950241e-01 -3.33974600e-01 1.93453372e-01
-1.53170514e+00 9.48499799e-01 7.23062754e-01 1.30088493e-01
-4.48805630e-01 -4.93978560e-01 -1.03761506e+00 3.08182657e-01
1.92760766e-01 -2.43075356e-01 1.58282852e+00 -1.02688944e+00
-2.00178647e+00 6.87886178e-01 -4.71802980e-01 -6.14893079e-01
1.38953865e-01 -3.30624700e-01 -3.24075729e-01 -4.34891041e-03
-2.16681242e-01 5.72614133e-01 6.04132354e-01 -8.39166105e-01
-7.79022872e-01 -3.35136175e-01 -3.99541199e-01 1.80505943e-02
1.16257519e-02 3.60019624e-01 1.85443889e-02 -3.97020340e-01
-2.82880478e-02 -6.10008597e-01 2.52476126e-01 -7.04570413e-01
-2.09766224e-01 -6.81887507e-01 7.88664222e-01 -1.04658115e+00
1.47072041e+00 -2.25734496e+00 -8.17356482e-02 1.57786310e-01
-3.09747785e-01 3.91645551e-01 9.41882608e-04 6.11996591e-01
-1.77475855e-01 9.90038738e-03 -3.32853734e-01 -4.79603171e-01
3.59362811e-01 2.66198933e-01 -4.09902632e-01 -5.87940589e-02
3.74341935e-01 8.45811188e-01 -8.35104406e-01 8.85787755e-02
5.60122073e-01 5.43345451e-01 -4.94859934e-01 4.93347347e-01
-2.63552040e-01 3.67981672e-01 -2.05916524e-01 2.69850463e-01
3.99330035e-02 3.45564365e-01 1.39957458e-01 -1.10873066e-01
-3.45714211e-01 1.01885307e+00 -9.99637961e-01 1.44446468e+00
-9.97174025e-01 7.63508022e-01 4.28808182e-01 -8.54207397e-01
1.08644128e+00 7.50017166e-01 1.94639236e-01 -5.96683264e-01
4.12406355e-01 5.10812581e-01 2.32937023e-01 -5.12613714e-01
1.48570940e-01 -2.95152664e-01 -8.89962390e-02 4.90453035e-01
3.41091305e-01 -3.06391269e-01 -1.54977500e-01 -1.35393232e-01
1.08878052e+00 8.36893171e-02 6.88537240e-01 -4.20276344e-01
8.91128838e-01 -2.03519434e-01 2.87242204e-01 6.11728549e-01
-3.42460096e-01 5.59968293e-01 1.69820681e-01 -2.47690976e-01
-8.31091940e-01 -7.50995874e-01 5.00578731e-02 1.18908346e+00
-6.54978275e-01 -5.33460677e-01 -1.13049126e+00 -3.55628788e-01
-3.96618813e-01 1.45920062e+00 -7.46202543e-02 -5.03602289e-02
-6.40578151e-01 -4.88254339e-01 5.76991737e-01 3.86336386e-01
3.23418289e-01 -1.51714766e+00 -8.18763494e-01 6.19239807e-01
-2.79039025e-01 -1.40774655e+00 -4.27083075e-01 4.66764361e-01
-6.07064307e-01 -6.67287707e-01 -3.40121955e-01 -7.08924055e-01
5.57330959e-02 -1.59257472e-01 9.03161108e-01 -2.95886070e-01
-1.52352482e-01 7.06420720e-01 -6.72042251e-01 -6.01605117e-01
-1.07206416e+00 1.94771096e-01 2.42980480e-01 5.66590577e-02
1.07307661e+00 -4.40446943e-01 -9.65839624e-02 1.50005460e-01
-8.16947043e-01 -7.95447454e-02 3.92555952e-01 6.76875114e-01
1.38415769e-01 9.28897634e-02 8.01395953e-01 -4.80821311e-01
9.15866077e-01 -3.49094242e-01 -3.04349095e-01 5.43438643e-02
-1.88885167e-01 2.09039435e-01 7.26634800e-01 -2.17774674e-01
-1.17307699e+00 3.37304085e-01 -8.58842373e-01 -5.04926313e-03
-7.20476508e-01 5.05252123e-01 -6.57010019e-01 6.31724775e-01
5.22970617e-01 3.24065745e-01 1.46342054e-01 -5.96213400e-01
3.30669820e-01 1.47013676e+00 3.38168651e-01 -2.23784879e-01
2.01877668e-01 1.18151745e-02 -8.33011627e-01 -1.84925640e+00
-5.91684401e-01 -6.27690732e-01 -7.10482776e-01 -1.50525153e-01
1.05855834e+00 -6.93539560e-01 -7.31790423e-01 5.87155998e-01
-1.58753574e+00 -5.03378749e-01 -2.93087393e-01 4.96136248e-01
-6.22856498e-01 3.30738723e-01 -5.28156996e-01 -1.23000026e+00
-5.40772974e-01 -1.25686443e+00 1.17585444e+00 6.35975823e-02
-9.11632657e-01 -9.12268758e-01 -3.27592939e-01 4.34074253e-01
4.52666283e-01 -4.83783692e-01 8.89446497e-01 -1.23958707e+00
-1.86906382e-01 -9.84127894e-02 8.66267532e-02 7.43051946e-01
3.16613823e-01 -4.24820513e-01 -1.52497602e+00 5.36797568e-02
5.53209782e-01 -1.14639245e-01 4.44215715e-01 3.85472178e-01
5.08785605e-01 -4.20000106e-01 1.80338636e-01 1.91463515e-01
8.71443450e-01 5.50501287e-01 5.91193438e-01 1.91030592e-01
3.20741773e-01 1.05412757e+00 2.48566955e-01 3.18341523e-01
2.66589522e-01 6.47884905e-01 2.86230650e-02 1.83279872e-01
-1.99564934e-01 -1.78735301e-01 1.05057299e+00 9.64356601e-01
5.59525788e-01 -1.35423750e-01 -1.12339342e+00 3.95993829e-01
-1.67423809e+00 -6.33853674e-01 2.73864400e-02 2.07905340e+00
6.99616909e-01 1.68120027e-01 -5.17188907e-02 3.16562444e-01
4.54830378e-01 9.41970646e-02 -2.42681414e-01 -7.79642820e-01
2.85735279e-02 2.83844799e-01 3.32052261e-02 9.11472797e-01
-9.54332411e-01 1.18376482e+00 5.96370173e+00 6.16088748e-01
-1.13514364e+00 4.28526998e-02 4.28796530e-01 2.28091910e-01
-1.74277667e-02 -4.62558061e-01 -9.35934126e-01 3.10651779e-01
1.54040408e+00 1.44971594e-01 6.52941763e-01 8.81596744e-01
7.75746703e-01 -4.16952908e-01 -1.54785502e+00 1.08141482e+00
5.26891090e-02 -7.81659901e-01 -1.01730511e-01 -2.80682683e-01
-5.99398687e-02 2.54688859e-01 -4.39438999e-01 3.82547349e-01
1.44908413e-01 -1.07613945e+00 8.37862492e-01 1.89425319e-01
4.93310690e-01 -5.14390647e-01 8.93134713e-01 5.82493842e-01
-1.11621702e+00 -1.15464225e-01 -8.53968710e-02 -2.35670626e-01
5.41506112e-01 2.98672050e-01 -1.48782003e+00 2.87858278e-01
5.93943894e-01 2.99519926e-01 -1.04543038e-01 3.44189852e-01
-4.97820556e-01 1.01835597e+00 -6.27967536e-01 -2.78942525e-01
3.76213461e-01 -2.34119773e-01 7.99829543e-01 1.64623094e+00
2.68225640e-01 2.94753581e-01 1.15500502e-01 8.89231026e-01
3.56983483e-01 3.15153420e-01 -6.32961512e-01 -2.21051678e-01
3.63727987e-01 7.86735415e-01 -6.35103345e-01 -1.79527000e-01
-4.50393856e-01 1.22011685e+00 -5.93058020e-02 1.89213008e-01
-1.79862767e-01 -2.92578697e-01 9.47278321e-01 6.79786457e-03
9.04246643e-02 -4.21821922e-01 -2.11964488e-01 -8.48058581e-01
-7.82380402e-02 -9.80550826e-01 6.06304295e-02 -8.16049993e-01
-1.12257040e+00 9.66224611e-01 -1.13490611e-01 -6.31012022e-01
-9.06946123e-01 -9.93676901e-01 -6.86709940e-01 1.19089603e+00
-1.24854577e+00 -9.25071120e-01 1.21817693e-01 2.09668681e-01
1.20333672e+00 -2.70291448e-01 1.07413995e+00 3.64439636e-02
-4.52831298e-01 9.53547880e-02 -1.43247634e-01 2.61717737e-01
2.87333012e-01 -1.21723771e+00 5.74818850e-01 8.20685506e-01
1.70986727e-01 6.39986694e-01 9.52145100e-01 -4.82691616e-01
-1.20064664e+00 -6.66914582e-01 1.34493124e+00 -4.95356649e-01
8.03942382e-01 -5.91668010e-01 -9.18189764e-01 6.58757746e-01
3.39590073e-01 -6.06101632e-01 7.86248088e-01 -8.56592432e-02
6.91362917e-02 6.67488351e-02 -9.17319000e-01 2.63906121e-01
5.95004797e-01 -9.65478182e-01 -1.06578302e+00 -7.66897053e-02
8.21271837e-01 5.06671406e-02 -6.03624582e-01 4.48842570e-02
4.03384417e-01 -9.80883837e-01 6.04359627e-01 -4.60029840e-01
5.22153862e-02 -1.41297519e-01 -5.31109810e-01 -1.54097605e+00
1.45945624e-01 -7.30150819e-01 1.70558497e-01 1.32881343e+00
4.76504534e-01 -7.21346557e-01 2.99812376e-01 7.84207642e-01
-2.43428245e-01 -2.89948970e-01 -1.08440650e+00 -5.77998340e-01
-3.01648676e-01 -1.06554222e+00 3.29834402e-01 3.05906117e-01
2.89615095e-01 7.79295564e-01 -4.17091139e-02 1.62347928e-01
1.15880527e-01 -5.30240476e-01 4.22764927e-01 -1.22078240e+00
-1.33902282e-01 -4.77814019e-01 -2.04188928e-01 -1.20299339e+00
4.71781462e-01 -9.25000846e-01 5.03577590e-01 -1.51844859e+00
-5.43833971e-01 -4.35644872e-02 -1.34805981e-02 4.86883700e-01
7.77886156e-03 -3.59194964e-01 3.42487216e-01 2.54991483e-02
1.18438885e-01 6.79714382e-01 4.00146097e-01 -9.27935913e-02
-6.32341862e-01 3.49285871e-01 -2.62835383e-01 7.63853788e-01
8.55246246e-01 -2.72736866e-02 -3.52217376e-01 -1.74011216e-01
-1.19053327e-01 -6.91081583e-02 -4.76327464e-02 -9.15940940e-01
1.41165957e-01 4.54262972e-01 -1.20383106e-01 -4.62538630e-01
3.26099664e-01 -9.24166739e-01 -2.63823479e-01 4.22259033e-01
-1.37458369e-01 -3.11683893e-01 3.30071181e-01 2.59125717e-02
-4.30012167e-01 -4.70569313e-01 7.68698752e-01 -2.14667797e-01
-9.52799380e-01 -3.12499106e-01 -1.15212083e+00 -3.74516428e-01
5.63296795e-01 -2.91308492e-01 3.21899861e-01 -6.27936065e-01
-9.26732779e-01 1.37606964e-01 -7.53037333e-02 7.15126276e-01
5.69830179e-01 -8.49367797e-01 -7.01625943e-01 5.80993056e-01
-6.84593245e-02 2.50850115e-02 4.74676527e-02 6.87246025e-01
-2.85793573e-01 8.76762033e-01 1.58693448e-01 -4.81788695e-01
-1.27571082e+00 1.96549118e-01 6.28254473e-01 -3.35756131e-02
-6.34717882e-01 8.69200528e-01 2.49328569e-01 -7.93911278e-01
3.82736504e-01 -5.00495195e-01 -3.80773216e-01 3.49743783e-01
8.48995805e-01 3.46135587e-01 2.86751747e-01 -8.41665506e-01
-4.61148173e-01 -6.24903385e-03 1.64270028e-01 -4.39166337e-01
1.35597634e+00 -3.35250586e-01 -1.83235317e-01 8.66476119e-01
1.20043778e+00 -1.91921189e-01 -8.04000199e-01 -6.30049929e-02
5.99164426e-01 4.35326010e-01 2.42620051e-01 -7.48832881e-01
-4.49202865e-01 1.05769515e+00 5.54421306e-01 3.30819935e-01
1.00455642e+00 6.33399189e-02 8.38330746e-01 5.42124391e-01
3.07417363e-01 -1.34638762e+00 -4.13253307e-01 1.08779466e+00
9.93112147e-01 -1.15790296e+00 -7.01761186e-01 -4.20768648e-01
-7.34136462e-01 1.40750825e+00 2.88051814e-01 2.04019636e-01
8.26930821e-01 3.03753406e-01 3.63902032e-01 -2.46052560e-03
-5.18584907e-01 -4.03090954e-01 1.67383760e-01 5.19458771e-01
7.21369624e-01 1.20918617e-01 -1.61349759e-01 9.11393046e-01
-6.01618350e-01 -5.64583912e-02 3.33646923e-01 6.72208726e-01
-6.95485294e-01 -1.15326500e+00 -5.82814872e-01 1.63242921e-01
-2.72969544e-01 -2.74475992e-01 -5.61341345e-01 5.73323309e-01
-1.12885885e-01 1.42356896e+00 -2.76727472e-02 -4.07484770e-01
5.93965113e-01 8.68271887e-01 -6.58978755e-03 -1.01689422e+00
-9.07040596e-01 2.05670446e-01 2.27042869e-01 -6.46162450e-01
-2.57244706e-01 -6.83934033e-01 -1.56087458e+00 3.01622331e-01
-1.15103908e-01 1.78811327e-01 8.88248205e-01 1.29926729e+00
2.15482175e-01 6.68705940e-01 5.78467727e-01 -9.89690661e-01
-5.43300331e-01 -1.34354997e+00 -4.59774703e-01 1.57026604e-01
3.04713041e-01 -1.13654204e-01 -4.60138083e-01 9.82668176e-02] | [14.095683097839355, 6.873390197753906] |
de8288b5-11bd-4630-9c85-faa8f47cc134 | metric-learning-vs-classification-for | 2008.03729 | null | https://arxiv.org/abs/2008.03729v2 | https://arxiv.org/pdf/2008.03729v2.pdf | Metric Learning vs Classification for Disentangled Music Representation Learning | Deep representation learning offers a powerful paradigm for mapping input data onto an organized embedding space and is useful for many music information retrieval tasks. Two central methods for representation learning include deep metric learning and classification, both having the same goal of learning a representation that can generalize well across tasks. Along with generalization, the emerging concept of disentangled representations is also of great interest, where multiple semantic concepts (e.g., genre, mood, instrumentation) are learned jointly but remain separable in the learned representation space. In this paper we present a single representation learning framework that elucidates the relationship between metric learning, classification, and disentanglement in a holistic manner. For this, we (1) outline past work on the relationship between metric learning and classification, (2) extend this relationship to multi-label data by exploring three different learning approaches and their disentangled versions, and (3) evaluate all models on four tasks (training time, similarity retrieval, auto-tagging, and triplet prediction). We find that classification-based models are generally advantageous for training time, similarity retrieval, and auto-tagging, while deep metric learning exhibits better performance for triplet-prediction. Finally, we show that our proposed approach yields state-of-the-art results for music auto-tagging. | ['Nicholas J. Bryan', 'Justin Salamon', 'Juhan Nam', 'Zeyu Jin', 'Jongpil Lee'] | 2020-08-09 | null | null | null | null | ['music-auto-tagging'] | ['music'] | [ 2.88770735e-01 -3.53596777e-01 -4.62963611e-01 -3.62665087e-01
-1.12486708e+00 -8.85742724e-01 5.44288039e-01 2.39206716e-01
-2.01523438e-01 3.52348506e-01 4.97601599e-01 2.11200207e-01
-7.85580873e-01 -5.91482759e-01 8.65942892e-03 -6.79939032e-01
-2.28346452e-01 5.58572888e-01 -3.52593511e-01 -3.18683863e-01
2.07827732e-01 3.80054951e-01 -1.67621422e+00 4.43493366e-01
3.94032717e-01 1.04975235e+00 -1.52869150e-01 5.98346412e-01
2.66636610e-01 7.48157024e-01 -5.42376757e-01 -6.05611026e-01
2.29731455e-01 -5.00506818e-01 -9.31573212e-01 -3.94265324e-01
5.38398564e-01 4.43869382e-02 -6.34768784e-01 7.94493020e-01
7.36488283e-01 4.09203351e-01 9.86760080e-01 -1.40052128e+00
-1.00427628e+00 7.36192763e-01 -3.71967703e-01 9.11247581e-02
3.82348806e-01 -4.66940492e-01 1.80603957e+00 -7.97030985e-01
2.51334488e-01 1.02277946e+00 7.30070591e-01 4.69965845e-01
-1.31807721e+00 -9.34536994e-01 -2.46032581e-01 5.93826294e-01
-1.49836314e+00 -3.49144489e-01 9.76822257e-01 -7.36129045e-01
7.10907698e-01 4.74602222e-01 4.05130386e-01 1.22810221e+00
-2.07029343e-01 7.24898219e-01 1.10593307e+00 -5.00856102e-01
-4.64793965e-02 -7.92826265e-02 2.68117815e-01 5.86218238e-01
8.05586278e-02 3.62283051e-01 -9.64607060e-01 -2.27169588e-01
4.07794267e-01 6.71646222e-02 -1.36806637e-01 -5.21602452e-01
-1.53968263e+00 1.02226222e+00 4.00425076e-01 5.24155617e-01
8.42153654e-02 3.67487311e-01 6.66668594e-01 5.53848505e-01
4.93588597e-01 1.01066041e+00 -4.22078520e-01 -4.20617133e-01
-9.95041370e-01 2.84287781e-01 8.11318278e-01 7.39700913e-01
8.08159173e-01 4.45593223e-02 -3.06206286e-01 1.18728971e+00
1.00907125e-01 3.17221999e-01 8.28672349e-01 -1.03463817e+00
2.77739108e-01 4.14100558e-01 -2.99900115e-01 -1.08578026e+00
-4.51712340e-01 -5.76718390e-01 -8.38613451e-01 1.72348674e-02
1.85409874e-01 1.99086547e-01 -4.15954053e-01 2.19027138e+00
-3.28016102e-01 4.79094386e-01 -3.22967116e-03 8.22034240e-01
7.86292911e-01 1.47590369e-01 -3.24157327e-02 8.21469128e-02
1.26471770e+00 -8.64806533e-01 -6.48995638e-01 7.97763243e-02
7.78451920e-01 -9.42404866e-01 9.27026331e-01 4.00253624e-01
-7.74456561e-01 -6.67077661e-01 -1.35331357e+00 -4.00117606e-01
-6.50149167e-01 2.57145703e-01 1.05435073e+00 5.34707427e-01
-8.72254252e-01 1.07168138e+00 -5.33874869e-01 -2.96966523e-01
1.94172427e-01 4.72899616e-01 -5.43396950e-01 2.43664961e-02
-1.39218986e+00 8.49578857e-01 3.37796241e-01 -3.28664333e-01
-5.04589140e-01 -5.90630651e-01 -8.63640547e-01 1.72091112e-01
-1.80473387e-01 -7.91265845e-01 1.09044194e+00 -4.71416235e-01
-1.40172589e+00 1.09769690e+00 2.52473980e-01 -1.19702555e-01
-2.59213269e-01 -1.35546491e-01 -5.89164078e-01 -1.60382316e-01
7.12300390e-02 4.07449603e-01 6.29659235e-01 -9.08822179e-01
-2.75098264e-01 -5.78683257e-01 2.47428656e-01 3.56692225e-01
-7.41259873e-01 -1.03031034e-02 1.08955719e-01 -1.14650679e+00
2.56409109e-01 -1.13249421e+00 1.43268496e-01 -4.64572534e-02
-2.27512553e-01 -3.79396647e-01 5.06497264e-01 -3.80282879e-01
1.46982038e+00 -2.47210073e+00 7.66096234e-01 -6.64857775e-02
3.60914558e-01 -9.98941213e-02 -4.41399843e-01 4.89411056e-01
-4.51560616e-01 6.93597198e-02 -6.05645664e-02 -5.70107579e-01
4.34087574e-01 3.48118812e-01 -3.41662824e-01 4.29346830e-01
-1.16497494e-01 9.53330994e-01 -1.13913822e+00 -2.07421467e-01
4.91074286e-02 3.75716001e-01 -6.87710345e-01 2.71435022e-01
1.56137854e-01 2.05996037e-01 -7.29785636e-02 6.02554858e-01
5.85201047e-02 -1.12601876e-01 1.94689885e-01 -5.07999659e-01
2.49007821e-01 5.65695584e-01 -1.17755830e+00 2.43781543e+00
-6.84409082e-01 9.35687244e-01 -4.30640489e-01 -1.19823086e+00
1.08314955e+00 4.26713884e-01 9.38980699e-01 -5.23954630e-01
-5.28884120e-02 1.90426111e-01 1.18075788e-01 -3.41299325e-01
7.19577491e-01 -3.92640352e-01 -5.08180380e-01 1.08978772e+00
6.30403042e-01 6.22426681e-02 1.04435056e-01 6.12937026e-02
1.03997993e+00 1.78468391e-01 4.40824211e-01 -1.51377633e-01
1.45510107e-01 -5.11072695e-01 4.38029408e-01 4.82787192e-01
-1.10097550e-01 6.64520860e-01 2.45707557e-01 -2.48117447e-01
-8.02423239e-01 -1.24214613e+00 -2.07536489e-01 1.68451059e+00
2.92404573e-02 -8.26569676e-01 -1.96777340e-02 -5.19937038e-01
2.11595461e-01 3.91990602e-01 -8.57764840e-01 -6.80385351e-01
-3.34943235e-01 -4.78670210e-01 8.61007154e-01 6.40043080e-01
-1.75493602e-02 -8.99716854e-01 -1.16531692e-01 -3.16194110e-02
-2.77689487e-01 -8.12802553e-01 -4.57033336e-01 4.89806175e-01
-8.62293303e-01 -1.16319418e+00 -4.24957812e-01 -7.58762658e-01
-7.73662776e-02 5.24224758e-01 1.42047942e+00 -1.70477912e-01
-2.79645115e-01 5.43301523e-01 -6.22330606e-01 -2.19120145e-01
-1.99766099e-01 3.33324373e-01 4.95536268e-01 -3.37866545e-02
5.30435205e-01 -1.05249143e+00 -4.85987127e-01 3.57255071e-01
-7.83418179e-01 -3.14538866e-01 3.56602192e-01 9.50336158e-01
4.94903445e-01 -3.79614860e-01 7.04714179e-01 -7.29499698e-01
8.66477966e-01 -6.65608406e-01 2.18439177e-01 3.04094285e-01
-7.50468910e-01 2.95926183e-01 1.55335292e-01 -5.84068894e-01
-3.46735686e-01 -2.53835827e-01 5.74187003e-02 -4.93424684e-01
-6.02266155e-02 5.20686805e-01 -9.99556780e-02 -4.89681475e-02
1.00871694e+00 -7.74299204e-02 -8.07893276e-02 -8.51997435e-01
8.80491257e-01 9.23491836e-01 4.96548682e-01 -8.32774103e-01
7.28124738e-01 3.14273506e-01 1.18770674e-01 -3.88243854e-01
-1.34671259e+00 -7.60599136e-01 -9.47710454e-01 4.15602960e-02
7.04374790e-01 -9.14101303e-01 -6.13128424e-01 -1.45559132e-01
-8.85294735e-01 9.69163626e-02 -7.79999495e-01 7.53049374e-01
-1.01022243e+00 2.78795272e-01 -4.29605603e-01 -3.86207879e-01
-2.10429177e-01 -8.07534873e-01 1.30588913e+00 -1.36950329e-01
-7.35000730e-01 -1.17214239e+00 7.55881429e-01 4.42755073e-01
3.21584702e-01 2.55583286e-01 1.24931347e+00 -1.00373662e+00
-1.13027208e-01 -1.56643644e-01 -1.75192267e-01 4.13915187e-01
4.24950808e-01 -5.57360768e-01 -1.25204194e+00 -3.52977097e-01
-2.70335585e-01 -6.20724618e-01 9.38256502e-01 -1.69963136e-01
1.29824066e+00 -1.67676378e-02 -1.07562244e-02 8.26856434e-01
1.11678064e+00 -1.60264552e-01 3.76599520e-01 2.71766692e-01
7.25797772e-01 5.08557022e-01 3.95335585e-01 4.49539602e-01
3.08066428e-01 1.24022007e+00 2.46038795e-01 9.15431157e-02
-5.44127047e-01 -2.84305155e-01 2.92473942e-01 1.30978823e+00
-2.53652483e-01 1.39094099e-01 -3.97813112e-01 3.31055641e-01
-1.96417272e+00 -1.22379625e+00 3.15463901e-01 2.22254181e+00
7.82476962e-01 -4.97693807e-01 3.05365354e-01 6.53066337e-01
3.93658191e-01 4.19237405e-01 -4.37974513e-01 -4.19163316e-01
-1.07048586e-01 6.76137030e-01 -1.28295600e-01 1.86268374e-01
-1.13065255e+00 7.78822660e-01 6.52993011e+00 7.50580788e-01
-9.19540226e-01 3.08639139e-01 -1.75869931e-02 -2.31851414e-01
-4.29210722e-01 -7.54631907e-02 -1.11158177e-01 1.17475346e-01
7.64795303e-01 -2.85413772e-01 6.90626204e-01 5.98103225e-01
-3.10080856e-01 6.22211516e-01 -1.76205862e+00 1.64619637e+00
3.20130974e-01 -1.11390984e+00 2.68871665e-01 7.39422739e-02
7.43871927e-01 -3.33549939e-02 3.79361898e-01 6.68037534e-01
5.77306375e-02 -1.14249170e+00 4.74380642e-01 5.90099216e-01
1.08149183e+00 -6.38978362e-01 5.11070430e-01 -6.17977865e-02
-1.30172777e+00 -1.45472035e-01 -3.64094496e-01 -4.17942971e-01
-1.93867013e-01 3.31403524e-01 -3.30095381e-01 8.49790573e-01
3.34292501e-01 1.32932663e+00 -7.19844043e-01 9.42499876e-01
-3.05753142e-01 3.46098423e-01 1.54523998e-01 1.37287840e-01
-1.29270121e-01 -3.19728613e-01 4.91564423e-01 1.20789611e+00
3.63107353e-01 -8.41164365e-02 8.50218982e-02 7.10835934e-01
-2.18903229e-01 1.52031004e-01 -7.40912557e-01 -2.23229066e-01
4.17537034e-01 1.19701684e+00 -2.73892671e-01 -1.75335202e-02
-2.75830507e-01 1.02646065e+00 6.15059555e-01 3.48282039e-01
-5.81629217e-01 -5.60998142e-01 1.27504909e+00 -2.18522996e-01
4.51346561e-02 -4.91508454e-01 -2.77922988e-01 -1.37383127e+00
-2.60229766e-01 -7.70997167e-01 6.85657322e-01 -7.07555234e-01
-1.67573261e+00 5.15797436e-01 -9.26418602e-02 -1.54999220e+00
-5.55039942e-01 -7.26871789e-01 -2.49483109e-01 6.76702917e-01
-1.32208288e+00 -1.15787780e+00 -1.55993775e-01 7.60961950e-01
2.97196150e-01 -5.73352456e-01 1.64729702e+00 5.94024241e-01
-2.25195915e-01 9.08401310e-01 3.72946799e-01 2.05011919e-01
1.02302086e+00 -1.51645112e+00 -1.49478495e-01 1.60626248e-01
1.40252852e+00 6.26864851e-01 1.79437160e-01 1.23576328e-01
-1.18083847e+00 -8.35450888e-01 9.04045939e-01 -7.38996565e-01
9.59204435e-01 -4.95165884e-01 -5.66206217e-01 4.35793996e-01
-6.25724718e-02 -4.47423421e-02 1.74053776e+00 9.38914895e-01
-1.12210441e+00 -2.10605755e-01 -6.55042112e-01 3.60980362e-01
1.26082182e+00 -1.30311048e+00 -6.85553670e-01 4.22442228e-01
6.79566622e-01 -1.43250182e-01 -1.27931094e+00 4.72924799e-01
1.03645957e+00 -9.66991425e-01 1.09775937e+00 -1.04607570e+00
3.12924713e-01 -8.76770318e-02 -8.31127703e-01 -1.47378862e+00
-8.30085278e-01 -5.86835444e-01 -5.54141343e-01 1.00415456e+00
3.10621232e-01 -2.02324659e-01 6.36807024e-01 8.36541355e-02
-9.58669111e-02 -1.04487860e+00 -9.17742968e-01 -9.75655556e-01
1.66309237e-01 -6.36655509e-01 6.16869092e-01 1.39720213e+00
4.26686347e-01 7.39551306e-01 -5.42205095e-01 -1.38276070e-01
4.49668050e-01 6.35355115e-01 4.63529050e-01 -1.67032433e+00
-6.37737095e-01 -6.94157779e-01 -1.08346868e+00 -8.88456345e-01
5.04645228e-01 -1.59545469e+00 -4.31530982e-01 -1.27368987e+00
4.18566406e-01 -4.94501621e-01 -1.05775690e+00 4.56440091e-01
-2.13172031e-03 6.80134058e-01 3.43938023e-01 4.30482984e-01
-7.47049689e-01 6.35294795e-01 9.19008434e-01 -3.79170537e-01
8.92040953e-02 1.35832533e-01 -1.05201840e+00 4.24719304e-01
7.50157297e-01 -5.96465111e-01 -5.20710289e-01 -5.16416252e-01
4.70374107e-01 -1.85773522e-01 2.83081651e-01 -9.50656295e-01
3.31504606e-02 2.38155827e-01 1.49253219e-01 -4.55098003e-02
6.67749584e-01 -6.82386518e-01 -1.30192831e-01 7.78997913e-02
-6.64750099e-01 9.17394608e-02 -1.22387394e-01 6.09735847e-01
-4.99770284e-01 -3.16498965e-01 4.33996528e-01 2.03214094e-01
-6.10067606e-01 4.28418279e-01 1.89963028e-01 2.73206025e-01
5.11867762e-01 -9.75204632e-02 -9.77300778e-02 -4.01191860e-01
-1.18728733e+00 -2.68971324e-01 4.93847020e-02 9.27794993e-01
5.28879166e-01 -1.96401322e+00 -7.32069254e-01 -4.17656498e-03
5.84154785e-01 -7.51031160e-01 -6.27517179e-02 4.62650299e-01
1.98749021e-01 3.84358287e-01 -2.64883965e-01 -4.42963153e-01
-1.06188142e+00 5.45177937e-01 1.40762687e-01 -3.38683128e-01
-1.49650022e-01 8.65268588e-01 -6.02981783e-02 -6.06302321e-01
3.36257964e-01 -6.76938668e-02 -2.55972981e-01 4.52134997e-01
3.24413002e-01 6.35729373e-01 7.28225932e-02 -7.39963591e-01
-2.69641668e-01 9.43927586e-01 1.07327327e-01 -2.00957730e-01
1.33335853e+00 2.36163363e-01 -1.73439562e-01 9.12072062e-01
1.61506021e+00 -1.55003346e-03 -5.84974766e-01 -5.07066607e-01
1.77896142e-01 -4.46927577e-01 3.37875895e-02 -7.24040985e-01
-9.87251520e-01 1.07512867e+00 7.19679236e-01 2.61483788e-01
9.95893836e-01 3.29048038e-01 5.63105047e-01 4.66295928e-01
3.31276417e-01 -7.74196327e-01 5.27192533e-01 5.22701085e-01
9.25850511e-01 -1.11065888e+00 2.41882563e-01 -2.25873329e-02
-3.18149835e-01 1.19073439e+00 3.47352356e-01 -1.50637537e-01
9.34570909e-01 -1.00576885e-01 -2.09125921e-01 -4.15300936e-01
-5.22639394e-01 -4.45028692e-01 8.98878217e-01 6.76048040e-01
8.85173261e-01 3.78961504e-01 -1.35172710e-01 9.21605229e-01
-6.39110386e-01 -4.97719467e-01 1.61531374e-01 5.48937738e-01
-8.26228335e-02 -1.47101629e+00 1.51240915e-01 4.33099687e-01
-1.44218236e-01 -1.42268971e-01 -6.41632676e-01 5.64874351e-01
3.46042514e-01 8.74347150e-01 -4.51065861e-02 -9.37495291e-01
4.26313668e-01 2.62237012e-01 8.46831024e-01 -1.08275175e+00
-4.88024265e-01 -4.05049682e-01 -6.47777319e-02 -6.22471631e-01
-5.63046038e-01 -5.65950572e-01 -6.47526920e-01 4.92645614e-02
-3.32105309e-01 3.01736116e-01 7.36995041e-01 9.38188255e-01
4.20285910e-01 5.16146839e-01 8.30299020e-01 -7.29166985e-01
-7.59474277e-01 -1.13619220e+00 -9.46613610e-01 6.95618570e-01
1.74488887e-01 -9.97547328e-01 -2.97749519e-01 -2.35945478e-01] | [15.74463176727295, 5.138858318328857] |
6619ed07-8961-4792-9259-ca6c70245efe | tri-axial-self-attention-for-concurrent | 1812.02817 | null | http://arxiv.org/abs/1812.02817v1 | http://arxiv.org/pdf/1812.02817v1.pdf | Tri-axial Self-Attention for Concurrent Activity Recognition | We present a system for concurrent activity recognition. To extract features
associated with different activities, we propose a feature-to-activity
attention that maps the extracted global features to sub-features associated
with individual activities. To model the temporal associations of individual
activities, we propose a transformer-network encoder that models independent
temporal associations for each activity. To make the concurrent activity
prediction aware of the potential associations between activities, we propose
self-attention with an association mask. Our system achieved state-of-the-art
or comparable performance on three commonly used concurrent activity detection
datasets. Our visualizations demonstrate that our system is able to locate the
important spatial-temporal features for final decision making. We also showed
that our system can be applied to general multilabel classification problems. | ['Kaixiang Huang', 'Xinyu Li', 'Ivan Marsic', 'Yehan Wang', 'Yanyi Zhang', 'Shuhong Chen'] | 2018-12-06 | null | null | null | null | ['concurrent-activity-recognition', 'activity-prediction', 'activity-prediction'] | ['computer-vision', 'computer-vision', 'time-series'] | [ 5.16273022e-01 -1.17237419e-01 -3.14006895e-01 -4.73326743e-01
-4.92427438e-01 -3.83784771e-01 7.77259052e-01 2.98066437e-01
-2.88057089e-01 5.07003903e-01 4.21715975e-01 -6.44959435e-02
-3.46444845e-01 -6.57942057e-01 -6.51332080e-01 -5.09141743e-01
-8.00354481e-01 2.81606734e-01 4.19372648e-01 1.53469056e-01
2.97633022e-01 7.81778872e-01 -1.79954815e+00 8.91906917e-01
4.16437984e-01 1.06086993e+00 2.49983981e-01 8.51792157e-01
-1.69679314e-01 1.29353321e+00 -7.24075735e-01 3.96742880e-01
2.27392670e-02 -6.35228395e-01 -8.80955935e-01 1.98380083e-01
3.16043735e-01 -1.84044801e-02 -3.82151812e-01 3.66126060e-01
1.35853216e-01 2.81286091e-01 6.18262470e-01 -1.48020196e+00
-3.84029299e-01 3.23180914e-01 -5.02441883e-01 7.31248438e-01
6.16690576e-01 7.91836902e-03 1.13937032e+00 -9.80627775e-01
5.05869508e-01 8.93250108e-01 4.74261940e-01 1.03939764e-01
-1.21677887e+00 -5.50750315e-01 5.95579326e-01 7.07071185e-01
-1.53867972e+00 -4.08855051e-01 4.46571290e-01 -5.60394466e-01
1.43048549e+00 3.67667139e-01 9.76090193e-01 9.19901192e-01
3.42481881e-01 9.96657729e-01 9.62824166e-01 -3.82829070e-01
1.05706565e-01 -2.44624972e-01 3.22613269e-01 1.05416012e+00
-2.09049895e-01 -1.51086167e-01 -1.04245543e+00 -1.50768131e-01
6.70392215e-01 1.92959607e-01 1.34519776e-02 -2.92467475e-01
-1.40965664e+00 5.53293049e-01 3.01035583e-01 5.73614538e-01
-3.79245639e-01 4.41022664e-01 7.81612545e-02 7.72784874e-02
4.23489630e-01 3.70352119e-01 -5.25931895e-01 -1.71636865e-01
-8.59869301e-01 -2.62571890e-02 5.46441853e-01 8.30650449e-01
8.41519654e-01 -4.33612108e-01 -6.82505667e-01 5.44879258e-01
1.27023354e-01 -8.09465721e-02 5.10368764e-01 -9.71208751e-01
1.83386490e-01 8.08049142e-01 4.61295843e-02 -6.85703695e-01
-6.74339473e-01 -5.53472638e-01 -5.30835330e-01 1.17840968e-01
2.03413874e-01 2.83091128e-01 -7.97675192e-01 1.57467449e+00
8.65854025e-02 4.83250529e-01 -1.66979134e-01 4.53194469e-01
2.19152853e-01 5.16902924e-01 1.59421861e-01 -2.16596797e-01
1.38652122e+00 -1.16663718e+00 -8.82758677e-01 -3.85628939e-01
8.26847911e-01 -2.70821005e-01 8.43880713e-01 2.41037622e-01
-7.83586025e-01 -6.49570882e-01 -1.06452525e+00 -4.16082777e-02
-6.16803050e-01 2.62547702e-01 7.92968631e-01 1.70331880e-01
-7.73603737e-01 7.86013544e-01 -1.16526663e+00 -5.40882826e-01
5.12763500e-01 4.50919896e-01 -4.45428461e-01 4.62683529e-01
-9.80930030e-01 9.78583694e-01 3.17978203e-01 -1.08202599e-01
-1.00477970e+00 -2.52461791e-01 -8.91668916e-01 2.55308717e-01
5.48773110e-01 -3.31740618e-01 1.14235365e+00 -1.09339750e+00
-1.24982166e+00 4.07301188e-01 -7.25240886e-01 -4.64657784e-01
1.03219010e-01 -2.89802313e-01 -6.43440843e-01 1.41651303e-01
1.89493462e-01 5.71531773e-01 5.08849740e-01 -6.22102022e-01
-9.43297029e-01 -8.75636488e-02 -1.81497142e-01 4.69263583e-01
-4.73948419e-01 1.86354592e-01 -6.16018713e-01 -6.72341883e-01
-1.07879259e-01 -8.71482432e-01 -1.27556846e-01 9.63326097e-02
-1.36558458e-01 -2.73388147e-01 9.26982403e-01 -5.43170333e-01
1.33807456e+00 -2.04980850e+00 1.77157834e-01 4.12720293e-01
7.67898634e-02 -2.48545095e-01 -2.34083474e-01 4.12114531e-01
-3.21419865e-01 -7.07269907e-02 -2.10976243e-01 -2.66428351e-01
-1.44001469e-01 3.16155195e-01 9.80990157e-02 4.26418364e-01
5.29320955e-01 9.53297377e-01 -9.34465110e-01 -5.58526814e-01
2.79375792e-01 4.27990705e-01 -3.01559299e-01 3.75542641e-01
-4.15779091e-02 2.57567883e-01 -1.79329321e-01 4.95914668e-01
-9.38176140e-02 -5.66151261e-01 4.08455521e-01 -4.12093513e-02
-6.06114492e-02 4.30869758e-01 -1.12265575e+00 1.75564837e+00
-4.28502142e-01 8.12088788e-01 -5.21174490e-01 -9.13763642e-01
5.49019575e-01 1.63983122e-01 9.26668048e-01 -7.51493752e-01
-3.49045008e-01 -1.52099788e-01 -4.50500287e-02 -3.51364106e-01
2.72271693e-01 4.62181389e-01 -8.58533531e-02 6.12189233e-01
3.06901962e-01 6.38744771e-01 4.62409317e-01 1.76585272e-01
1.57785618e+00 3.22675556e-01 7.87441254e-01 -2.49228910e-01
5.70473254e-01 -3.27869475e-01 4.95550871e-01 6.64757371e-01
-2.10276712e-02 3.08512449e-01 5.76144278e-01 -6.96629226e-01
-4.06603515e-01 -9.76968765e-01 3.23778540e-01 1.52638602e+00
1.03178397e-02 -9.20870125e-01 -3.53354841e-01 -1.15706253e+00
-3.27436090e-01 3.76475632e-01 -1.12338042e+00 -7.34411031e-02
-5.15084028e-01 -6.26843095e-01 3.60905647e-01 1.03090823e+00
3.08677286e-01 -1.28290272e+00 -9.09618556e-01 3.62662524e-01
-1.78748131e-01 -9.05156493e-01 -6.11599267e-01 9.24742162e-01
-6.22188210e-01 -1.29046082e+00 -2.52050787e-01 -6.55070484e-01
5.92888772e-01 3.88369197e-03 9.19611275e-01 -1.07250050e-01
-2.67414093e-01 2.72700131e-01 -1.81048810e-01 -1.44967705e-01
-1.26086026e-01 1.50149941e-01 -1.06894016e-01 3.15524578e-01
2.97844827e-01 -6.16170168e-01 -4.12718147e-01 5.45715094e-01
-7.44829774e-01 1.47281528e-01 6.59322619e-01 6.64192677e-01
6.60837770e-01 -1.29311010e-01 1.75241053e-01 -5.91750920e-01
5.40234506e-01 -3.63045394e-01 -3.27290922e-01 5.18810391e-01
-5.59303761e-01 5.95555067e-01 2.81225413e-01 -6.06534481e-01
-8.66891861e-01 6.24931097e-01 1.62796244e-01 -3.28905940e-01
-3.68703783e-01 5.81096888e-01 -2.15989962e-01 1.43656760e-01
5.11624932e-01 3.49173844e-01 -3.85606021e-01 -4.47852284e-01
2.27677628e-01 4.26598638e-01 3.91115338e-01 -3.60924333e-01
2.47396320e-01 3.72690529e-01 1.94719985e-01 -6.82113230e-01
-7.91679680e-01 -4.80971247e-01 -8.69697452e-01 -2.36399353e-01
9.95991766e-01 -8.44421685e-01 -6.80806875e-01 3.72627705e-01
-1.08392835e+00 -5.70708156e-01 -4.04462874e-01 2.58329570e-01
-6.03452563e-01 2.83965975e-01 -5.94105899e-01 -6.67376876e-01
-1.00260144e-02 -9.75900352e-01 1.10519159e+00 -9.10143331e-02
-7.35179007e-01 -9.19858634e-01 3.99978727e-01 4.62004766e-02
2.93958664e-01 -1.00134738e-01 6.30843520e-01 -7.77089894e-01
-5.82396507e-01 6.19273260e-02 -1.69748366e-02 -1.21015824e-01
4.84518439e-01 2.56858096e-02 -9.44624066e-01 1.72628239e-02
-6.92674875e-01 -5.13598435e-02 8.61658573e-01 1.42264187e-01
1.66238582e+00 -3.30054730e-01 -6.84615731e-01 4.52473998e-01
8.21578681e-01 5.27099013e-01 7.55207658e-01 1.42801866e-01
7.04477966e-01 4.77651507e-01 6.41222775e-01 5.09336293e-01
2.11402848e-01 1.12288272e+00 2.46182039e-01 -1.51235297e-01
-3.86934020e-02 -2.53247648e-01 6.06119275e-01 2.62413949e-01
-1.04828306e-01 -3.03807169e-01 -7.40588427e-01 6.80110157e-01
-2.15394473e+00 -1.15908015e+00 6.93184277e-03 1.92984927e+00
4.30226713e-01 8.56081173e-02 2.03976393e-01 5.37731536e-02
5.47838330e-01 1.57454923e-01 -2.77305454e-01 -2.31733620e-01
5.40460050e-02 4.00285482e-01 4.45736170e-01 4.21278387e-01
-1.31957126e+00 6.62913501e-01 7.71787310e+00 7.08710134e-01
-6.99405968e-01 1.57230079e-01 4.79437679e-01 -3.39080572e-01
1.28690019e-01 -1.38121024e-01 -6.33681834e-01 3.89907211e-01
1.02679265e+00 4.28993963e-02 2.16702655e-01 6.07874691e-01
1.89525485e-01 -3.29228789e-01 -1.46987689e+00 6.67982996e-01
1.34922728e-01 -1.30817020e+00 -2.67483993e-03 2.72957087e-01
3.90714496e-01 -5.34265041e-02 -2.99058825e-01 1.67545766e-01
4.68620658e-01 -9.21887338e-01 6.59288883e-01 7.56722391e-01
5.91072977e-01 -6.47677422e-01 4.27737474e-01 3.19677114e-01
-1.67811108e+00 -2.64071912e-01 3.06336194e-01 -2.95861751e-01
8.98056626e-02 4.25347537e-01 -9.03959990e-01 3.89527529e-01
7.04403937e-01 1.14619732e+00 -9.18211460e-01 7.54008055e-01
-3.28935117e-01 1.94498301e-01 -2.80840069e-01 -3.55884098e-02
1.84518546e-02 1.43905014e-01 4.95211817e-02 1.51418340e+00
5.05973339e-01 -2.55132318e-01 2.62276083e-01 6.33730710e-01
1.41480878e-01 4.80962358e-02 -6.47488356e-01 -1.25231594e-01
1.94384724e-01 1.13279665e+00 -9.70160007e-01 -7.35012770e-01
-4.98623282e-01 1.42239702e+00 6.92590892e-01 1.24298848e-01
-8.60405564e-01 -3.67869675e-01 7.53510833e-01 1.45999357e-01
4.84723657e-01 -1.73437461e-01 7.15404823e-02 -1.21318722e+00
3.57326530e-02 -4.82598156e-01 8.38316739e-01 -8.55127454e-01
-8.56933594e-01 4.38468128e-01 1.33092761e-01 -1.25941932e+00
-3.34580213e-01 -4.56128567e-01 -6.01071179e-01 6.18503690e-01
-8.94363701e-01 -1.05176950e+00 -3.41794580e-01 7.35743701e-01
5.36050439e-01 -2.55814046e-02 1.06266868e+00 1.52208984e-01
-4.18680340e-01 1.82467788e-01 -4.45111573e-01 2.69952923e-01
5.72343111e-01 -1.26608372e+00 5.18402278e-01 9.67840075e-01
8.42702568e-01 5.74959993e-01 2.29502901e-01 -5.78667045e-01
-8.04272294e-01 -1.09421515e+00 1.23335707e+00 -4.12111282e-01
6.33756042e-01 -5.64081907e-01 -8.37230682e-01 1.16275227e+00
2.94856906e-01 3.40865582e-01 1.05965698e+00 2.95664579e-01
-2.85334826e-01 -1.54197097e-01 -6.50952160e-01 4.86926436e-01
1.55007815e+00 -6.55740321e-01 -5.73636651e-01 1.61276713e-01
4.08946335e-01 -1.12686060e-01 -7.79013693e-01 2.06121117e-01
7.86462009e-01 -8.20349753e-01 8.12738538e-01 -5.63537180e-01
1.35850385e-01 -5.64601421e-01 -2.60635447e-02 -1.30497539e+00
-8.50074589e-01 -2.77091891e-01 -6.55737102e-01 7.58500159e-01
4.64008927e-01 -3.30248743e-01 7.01174200e-01 1.25951037e-01
-8.88856426e-02 -4.75238055e-01 -8.62032652e-01 -6.77222311e-01
-7.63180912e-01 -6.18185520e-01 4.36156929e-01 9.58943427e-01
4.33673561e-01 4.92000878e-01 -5.39289653e-01 1.86468333e-01
1.57045230e-01 2.32744843e-01 3.86994898e-01 -1.12120104e+00
-7.92737782e-01 -3.11413199e-01 -4.80740100e-01 -1.05340350e+00
2.01716378e-01 -8.57181668e-01 -7.09583014e-02 -1.60824883e+00
3.77786547e-01 1.16377406e-01 -8.99315894e-01 1.14518201e+00
-2.87944488e-02 4.31383222e-01 -2.73275506e-02 -1.59763843e-01
-9.72153127e-01 1.72129750e-01 8.36179912e-01 -3.86756510e-01
-4.45858359e-01 -8.86819959e-02 -5.16777992e-01 5.15414059e-01
6.76096380e-01 -5.89782298e-01 -3.33361983e-01 2.00766679e-02
1.47120193e-01 6.43941164e-02 3.12722027e-01 -1.41565287e+00
2.66211033e-01 -3.19650352e-01 7.46074498e-01 -7.89088309e-01
4.33513999e-01 -8.68170381e-01 4.31378603e-01 5.67804873e-01
-7.46607244e-01 -2.77139228e-02 1.99966505e-01 5.39517581e-01
-1.76660836e-01 2.07706004e-01 3.20774734e-01 3.40365469e-02
-9.56259906e-01 1.80082977e-01 -1.02524137e+00 -5.42449772e-01
1.12388873e+00 -8.47232491e-02 -7.74368122e-02 -4.00930256e-01
-1.21806109e+00 1.37853503e-01 1.31555542e-01 5.84681690e-01
4.29888874e-01 -1.49311352e+00 -2.62281328e-01 4.08504218e-01
5.86220443e-01 -5.04454792e-01 -3.98541801e-02 8.05622756e-01
-1.57712743e-01 3.78194869e-01 -3.90469015e-01 -5.99051535e-01
-1.66468883e+00 6.40034795e-01 4.65563029e-01 -7.07618654e-01
-4.76150215e-01 5.19995511e-01 1.57852724e-01 9.55261663e-02
2.66512096e-01 -7.03448892e-01 -3.04385930e-01 2.12782711e-01
7.19639599e-01 4.04714733e-01 1.68576613e-01 -4.22632009e-01
-7.98220873e-01 4.30532366e-01 9.51043963e-02 -3.05077910e-01
1.33686602e+00 8.33109841e-02 1.09906971e-01 5.85772634e-01
1.12432098e+00 -8.88098627e-02 -1.58979642e+00 -1.46465197e-01
3.64599437e-01 -2.41385460e-01 -1.12928890e-01 -9.30039465e-01
-1.06413651e+00 7.61731505e-01 6.87568128e-01 8.68128240e-02
1.22216272e+00 1.63523450e-01 1.26670048e-01 5.48478782e-01
2.00030372e-01 -9.65161502e-01 3.76217246e-01 5.85628748e-01
6.56864941e-01 -9.04041111e-01 6.39042109e-02 -3.13922822e-01
-5.45512259e-01 1.05257952e+00 4.92419451e-01 1.87030196e-01
5.27625740e-01 6.24939799e-01 -2.58356035e-01 -3.63209128e-01
-1.14867234e+00 -4.45547104e-01 5.48646927e-01 5.09772956e-01
5.78863144e-01 -2.18484402e-02 -4.77945283e-02 3.46312553e-01
4.18688297e-01 2.08381757e-01 2.68461049e-01 1.22029591e+00
-3.90127987e-01 -1.20085859e+00 -1.56421185e-01 5.67939758e-01
-3.40674341e-01 9.39805880e-02 -5.74879408e-01 6.66196644e-01
3.20986181e-01 7.36877263e-01 4.85078901e-01 -4.91790682e-01
3.12258899e-01 5.94438195e-01 6.27296925e-01 -6.69610798e-01
-6.50814772e-01 2.19929188e-01 2.89922744e-01 -1.10283983e+00
-6.73955381e-01 -8.61146033e-01 -1.21765196e+00 2.14130372e-01
9.18877721e-02 2.53621619e-02 2.74458885e-01 9.94522989e-01
6.01730049e-01 9.52762485e-01 4.17138964e-01 -7.74572790e-01
3.62844229e-01 -1.05299151e+00 -6.62506282e-01 4.18601573e-01
2.54366040e-01 -7.54148841e-01 -9.43495333e-02 2.34079242e-01] | [8.278986930847168, 0.6972343921661377] |
8b6becde-f8f6-4b1d-90d2-57a288817f55 | empowering-relational-network-by-self | null | null | https://www.ecva.net/papers/eccv_2020/papers_ECCV/html/410_ECCV_2020_paper.php | https://www.ecva.net/papers/eccv_2020/papers_ECCV/papers/123460069.pdf | Empowering Relational Network by Self-Attention Augmented Conditional Random Fields for Group Activity Recognition | This paper presents a novel relational network for group activity recognition. The core of our network is to augment the conditional random fields (CRF), amenable to learning inter-dependency of correlated observations, with the newly devised temporal and spatial self-attention to learn the temporal evolution and spatial relational contexts of every actor in videos. Such a combination utilizes the global receptive fields of self-attention to construct a spatio-temporal graph topology to address the temporal dependency and non-local relationships of the actors. The network first uses the temporal self-attention along with the spatial self-attention, which considers multiple cliques with different scales of locality to account for the diversity of the actors' relationships in group activities, to model the pairwise energy of CRF. Afterward, to accommodate the distinct characteristics of each video, a new mean-field inference algorithm with dynamic halting is also addressed. Finally, a bidirectional universal transformer encoder (UTE), which combines both of the forward and backward temporal context information, is used to aggregate the relational contexts and scene information for group activity recognition. Simulations show that the proposed approach surpasses the state-of-the-art methods on the widespread Volleyball and Collective Activity datasets. | ['Wen Hsien Fang', 'Yie Tarng Chen', 'Rizard Renanda Adhi Pramono'] | null | null | null | null | eccv-2020-8 | ['group-activity-recognition'] | ['computer-vision'] | [-7.96631072e-03 -1.36536717e-01 -1.80614606e-01 -4.93145198e-01
-1.66985840e-01 -2.30642408e-01 8.55559051e-01 1.06687360e-01
-3.01459253e-01 5.68729579e-01 4.73650336e-01 1.01411752e-01
-5.77481687e-01 -7.23048270e-01 -8.32508206e-01 -8.99119556e-01
-5.12863219e-01 3.07866275e-01 3.51740569e-01 -6.79030791e-02
1.85245097e-01 5.66460371e-01 -1.36311686e+00 2.82414824e-01
6.73797607e-01 9.86981571e-01 4.62213159e-01 5.65177739e-01
-6.31934479e-02 1.44048667e+00 -1.98448479e-01 -1.46770686e-01
-1.84159905e-01 -5.63813508e-01 -7.24583685e-01 3.92235398e-01
2.76604772e-01 2.36240756e-02 -7.90550292e-01 5.11022031e-01
1.73602477e-01 7.07252681e-01 4.30752814e-01 -1.02261436e+00
-7.22587705e-01 4.21102643e-01 -6.42280221e-01 7.63582945e-01
3.61776292e-01 1.31145895e-01 1.00785255e+00 -5.69433630e-01
7.24902868e-01 1.08680856e+00 2.92930335e-01 -1.84638258e-02
-1.03284729e+00 -4.66726601e-01 5.21057367e-01 6.65471852e-01
-1.47649622e+00 -1.00367144e-01 9.79912281e-01 -4.67196375e-01
9.91093040e-01 -5.76087348e-02 1.00211310e+00 1.20446002e+00
4.03958857e-01 5.14466465e-01 9.40863609e-01 -2.61292338e-01
2.22257361e-01 -3.79786372e-01 4.25688624e-02 7.09353447e-01
-4.07654792e-01 -9.15199369e-02 -1.00937986e+00 7.21592382e-02
8.87166381e-01 3.17427993e-01 1.47840708e-01 -2.75074571e-01
-1.33835244e+00 4.04066950e-01 6.03708446e-01 5.99865019e-01
-5.69434047e-01 2.84494460e-01 4.65929918e-02 -1.56180143e-01
5.20966768e-01 -5.56258671e-02 -1.06156543e-01 -9.55581218e-02
-8.58971238e-01 -2.31267512e-02 5.40739954e-01 9.17076349e-01
9.78484869e-01 -1.75976649e-01 -4.41784531e-01 3.79134059e-01
4.28098649e-01 1.02662899e-01 2.03085437e-01 -9.44968641e-01
5.10688901e-01 7.42362916e-01 -2.24055320e-01 -1.28956687e+00
-2.32958004e-01 -7.32574999e-01 -1.06980360e+00 -3.61257017e-01
1.55454516e-01 1.49024934e-01 -6.90771103e-01 1.78719795e+00
4.57893223e-01 9.10378218e-01 -2.88132161e-01 7.31323242e-01
3.22058052e-01 6.17606103e-01 4.00400788e-01 -3.74966294e-01
1.24641919e+00 -9.33354497e-01 -8.60916376e-01 -1.51231930e-01
2.51801401e-01 -2.67868608e-01 5.53921044e-01 -4.63617174e-03
-1.11015236e+00 -8.04667115e-01 -7.16513455e-01 -7.04511479e-02
-2.62807786e-01 -1.36672944e-01 7.24450052e-01 -8.11922550e-02
-9.46578801e-01 5.23994923e-01 -1.10014081e+00 -3.93023759e-01
4.31909233e-01 3.52921128e-01 -4.48928237e-01 -1.62341986e-02
-1.11527789e+00 6.14530444e-01 1.76311031e-01 4.00574118e-01
-7.74217546e-01 -4.10262257e-01 -7.86532283e-01 1.51751980e-01
5.79973698e-01 -6.72633171e-01 5.13887584e-01 -1.01665699e+00
-1.43424273e+00 4.45817173e-01 -3.99275303e-01 -3.00779253e-01
2.45273948e-01 2.97977272e-02 -4.96435046e-01 2.55717039e-01
1.32766828e-01 3.52916628e-01 5.54677248e-01 -9.11141574e-01
-4.89310622e-01 -4.27267432e-01 2.37834714e-02 6.69020295e-01
-8.44686702e-02 7.92856738e-02 -9.20007944e-01 -6.36217296e-01
1.43673867e-01 -8.66267383e-01 -3.51454645e-01 -3.35243225e-01
-1.82072341e-01 -3.30264360e-01 7.30511189e-01 -5.90022326e-01
1.35373092e+00 -2.32654309e+00 7.01065302e-01 3.85431737e-01
5.94378300e-02 -3.33886743e-01 4.48948741e-02 6.67884111e-01
1.82301668e-03 -2.17471421e-01 -2.31835067e-01 -5.99489212e-01
-2.90316433e-01 6.92964375e-01 -1.46854660e-02 6.54524863e-01
2.08418548e-01 8.34732354e-01 -1.01185071e+00 -8.67852867e-01
4.03055966e-01 6.75501943e-01 -4.74823654e-01 3.71335506e-01
-1.49320781e-01 1.05953467e+00 -5.67029774e-01 2.86533475e-01
4.42218244e-01 -5.10921538e-01 6.04971111e-01 -3.41332220e-02
-2.65107572e-01 1.45410076e-01 -1.25419044e+00 1.96632671e+00
-2.32174635e-01 2.84886658e-01 -6.84057549e-02 -1.23024309e+00
7.06605434e-01 2.46420056e-01 9.14755821e-01 -7.13322163e-01
-1.12656720e-01 -2.76063055e-01 -2.36420214e-01 -6.71517670e-01
2.75486887e-01 2.34972924e-01 -6.27365261e-02 2.64279157e-01
4.82063234e-01 5.44189155e-01 3.46775949e-01 5.57544351e-01
1.18210804e+00 4.90731001e-01 1.97873905e-01 -2.01683134e-01
6.17198050e-01 -5.22204816e-01 8.08150053e-01 5.58336914e-01
5.10757901e-02 2.56278813e-01 4.40142155e-01 -5.67390740e-01
-6.21955752e-01 -1.14622879e+00 2.72187293e-01 1.30049288e+00
4.17640150e-01 -3.81461442e-01 -4.28472579e-01 -5.93120158e-01
-3.35762799e-01 1.57346755e-01 -9.19927716e-01 5.91838099e-02
-6.79730058e-01 -5.90528905e-01 3.04741729e-02 6.42069936e-01
6.61137223e-01 -1.23943913e+00 -3.76560479e-01 3.41829091e-01
-3.66939604e-01 -1.19382787e+00 -5.39211571e-01 2.44803250e-01
-6.90700948e-01 -1.00245917e+00 -1.89265624e-01 -6.24233425e-01
5.63903928e-01 7.92070925e-02 9.43541288e-01 -1.89299136e-02
-4.32702713e-02 4.48135316e-01 -3.61186564e-01 3.23388129e-01
1.04688041e-01 3.36678661e-02 -3.84441502e-02 7.56597757e-01
1.50311261e-01 -1.16036904e+00 -6.31834447e-01 3.54738593e-01
-8.66270959e-01 1.46873102e-01 5.38263023e-01 7.74705648e-01
8.37793648e-01 7.32310340e-02 1.93454996e-01 -5.38166583e-01
1.54941201e-01 -8.50544631e-01 -3.29455227e-01 5.46994448e-01
-1.44986466e-01 2.80968193e-02 4.06073868e-01 -6.25215888e-01
-1.37768710e+00 1.53086916e-01 3.31344128e-01 -6.32097661e-01
-1.96059480e-01 4.99145031e-01 -2.33032659e-01 1.15606114e-01
2.24564642e-01 2.82673717e-01 -3.25709820e-01 -4.31979597e-01
3.88549745e-01 2.43636653e-01 6.34912968e-01 -5.83881021e-01
6.25306129e-01 7.68516719e-01 2.90827870e-01 -7.73062229e-01
-7.89086878e-01 -6.17837608e-01 -1.02702439e+00 -4.95915681e-01
1.37446237e+00 -1.00891733e+00 -8.64696741e-01 5.11964321e-01
-1.06853449e+00 -3.08433414e-01 -3.97979617e-01 6.88149035e-01
-4.89157021e-01 5.13105571e-01 -7.41088808e-01 -8.10395658e-01
3.09743613e-01 -7.78914332e-01 9.85215724e-01 2.87917346e-01
1.94563508e-01 -1.11159110e+00 3.27946126e-01 4.67800617e-01
-4.44759941e-03 3.16929102e-01 5.46618998e-01 -3.00963342e-01
-1.09425735e+00 1.67482913e-01 -1.36976968e-02 9.59266424e-02
4.04936112e-02 1.89655777e-02 -6.17962480e-01 -5.42807318e-02
-1.67273030e-01 1.00660868e-01 8.05922508e-01 5.02984107e-01
1.24501920e+00 -2.24983692e-01 -4.43451911e-01 7.43774772e-01
1.27274978e+00 1.79877430e-01 6.56343699e-01 -3.97381140e-03
1.03169668e+00 5.85018575e-01 4.89561349e-01 6.29108667e-01
7.78455675e-01 8.30674112e-01 4.46028382e-01 -5.83086461e-02
8.84459987e-02 -4.09191608e-01 1.96417168e-01 1.04201829e+00
-8.42796028e-01 -4.32929158e-01 -7.01329231e-01 5.10572493e-01
-2.22623491e+00 -1.51500142e+00 6.34013303e-03 1.88384330e+00
1.96289897e-01 -2.43401468e-01 -1.11626443e-02 -2.40117252e-01
8.48117530e-01 6.20433688e-01 -3.02582055e-01 -5.58062568e-02
-2.11823478e-01 9.13550183e-02 4.25448328e-01 3.21362406e-01
-9.99432087e-01 6.98142707e-01 5.80457306e+00 7.65324831e-01
-6.32204711e-01 3.70999366e-01 3.94993842e-01 1.83666090e-03
-1.88357085e-01 3.00231814e-01 -4.59948748e-01 6.16724730e-01
8.10030818e-01 3.36549640e-01 7.14356005e-01 3.64593565e-01
2.93843329e-01 -2.82720327e-01 -8.18285286e-01 6.28748536e-01
5.67120835e-02 -1.37186706e+00 -2.41185352e-01 3.70442897e-01
7.17926025e-01 8.09993595e-02 -1.49539113e-01 2.17831090e-01
4.37273264e-01 -7.66483128e-01 7.86421835e-01 1.15218675e+00
3.89159411e-01 -6.58357620e-01 5.26665866e-01 5.77708542e-01
-1.74750257e+00 -4.20895666e-01 5.72288707e-02 -1.45436555e-01
4.57378328e-01 4.50371563e-01 -4.18399513e-01 9.09206986e-01
9.07032430e-01 1.17351305e+00 -5.36770225e-01 5.53588152e-01
-2.18964309e-01 4.00001585e-01 -3.49902183e-01 2.30656892e-01
2.27769673e-01 -5.27046978e-01 5.18840611e-01 1.15543997e+00
1.49738178e-01 2.04678670e-01 3.97509545e-01 6.79119051e-01
6.51032627e-02 -1.37426525e-01 -4.45416212e-01 1.31360203e-01
2.94741541e-01 1.07784045e+00 -6.37700856e-01 -3.29000354e-01
-5.61062396e-01 9.35459852e-01 7.60816276e-01 5.79588473e-01
-1.01884413e+00 2.61897057e-01 3.87820542e-01 2.34902367e-01
4.96888757e-01 -6.25954151e-01 1.75535589e-01 -1.21268201e+00
1.29424557e-01 -3.45421463e-01 6.83627784e-01 -7.73073673e-01
-1.32653499e+00 4.35201257e-01 2.50593692e-01 -1.13432276e+00
-7.08470121e-02 3.20040504e-03 -7.42957592e-01 6.81799471e-01
-1.22977579e+00 -1.55141640e+00 -3.77249688e-01 1.01525462e+00
4.18128192e-01 4.92944121e-02 3.94311726e-01 2.90032774e-01
-6.03898406e-01 1.91739369e-02 -2.19041720e-01 9.44679156e-02
3.56435448e-01 -1.15388119e+00 6.71246871e-02 1.00905931e+00
4.00759965e-01 5.70533276e-01 2.17390463e-01 -8.37176919e-01
-1.22723210e+00 -1.00577903e+00 1.10255384e+00 -3.42747927e-01
8.54695022e-01 -4.34206426e-01 -9.18426991e-01 9.80480492e-01
2.77365595e-01 6.18065715e-01 5.45440257e-01 2.64263868e-01
-1.90407798e-01 -2.35770822e-01 -6.36701226e-01 3.91631603e-01
1.56351185e+00 -8.06719720e-01 -4.77012098e-01 2.83163488e-01
5.78544080e-01 -2.47223958e-01 -1.00202656e+00 3.42280954e-01
3.29859495e-01 -1.05865073e+00 9.54610288e-01 -4.42416519e-01
2.65961260e-01 -4.58644658e-01 -1.10211484e-01 -8.77749860e-01
-8.11370492e-01 -6.54942334e-01 -3.66957188e-01 1.27367759e+00
-3.26476209e-02 -3.17392170e-01 5.17665863e-01 2.72581637e-01
-1.13715656e-01 -6.38410032e-01 -1.26024318e+00 -4.37092781e-01
-5.47815621e-01 -3.65021080e-01 4.04436290e-01 1.05596137e+00
1.31124437e-01 3.74074370e-01 -7.00297058e-01 2.78981179e-01
6.56128347e-01 9.75638703e-02 4.46484864e-01 -9.86670852e-01
-6.10260904e-01 -4.87184711e-03 -6.29662931e-01 -1.10280025e+00
2.97026455e-01 -7.30469584e-01 -1.09306790e-01 -1.33884132e+00
3.48795772e-01 -3.17167014e-01 -6.91946328e-01 1.75593317e-01
-7.19731301e-02 2.15388983e-02 2.17077523e-01 2.53512532e-01
-1.14830077e+00 6.54547095e-01 1.33659422e+00 5.98539896e-02
-3.73835504e-01 -4.53882664e-03 -1.40172020e-01 4.85858202e-01
4.17052358e-01 -4.13851708e-01 -4.88646448e-01 -3.41150910e-01
2.07252473e-01 4.29918766e-01 6.64802730e-01 -1.21005511e+00
6.69519722e-01 -4.43554938e-01 2.85977781e-01 -5.79702258e-01
4.49313104e-01 -8.86377692e-01 4.68449295e-01 -7.76155367e-02
-4.37391371e-01 1.69339642e-01 -4.27851737e-01 1.22105706e+00
-3.59007090e-01 4.25183982e-01 5.03528953e-01 -2.21353069e-01
-6.12003148e-01 7.67263234e-01 -4.64306235e-01 -1.87055483e-01
1.19435954e+00 -2.35459849e-01 -1.81609765e-02 -2.49940529e-01
-1.15644193e+00 2.39973933e-01 1.98231325e-01 2.99742073e-01
3.63105536e-01 -1.31069613e+00 -3.59686255e-01 1.84375376e-01
-1.27923757e-01 1.83189765e-01 9.60887253e-01 1.18096173e+00
-9.63120759e-02 5.96527532e-02 -3.19480121e-01 -7.56263494e-01
-1.04783976e+00 6.89892650e-01 3.01367611e-01 -8.14801753e-01
-5.97159386e-01 6.41444504e-01 2.90025443e-01 8.19329321e-02
8.81587267e-02 -2.43271381e-01 -4.91834939e-01 4.79994379e-02
1.68189839e-01 3.58381033e-01 -3.11173230e-01 -1.10572422e+00
-4.92840677e-01 6.85867369e-01 3.13159704e-01 -8.11385289e-02
1.43691325e+00 -4.01416540e-01 -1.34382904e-01 5.30875742e-01
9.43420053e-01 -5.35842851e-02 -1.67182195e+00 -5.00915706e-01
-1.41743913e-01 -3.94462675e-01 2.56845932e-02 -3.49614829e-01
-1.11661577e+00 8.24480116e-01 2.47996822e-01 9.36817527e-02
1.23794484e+00 2.58639604e-01 3.26145887e-01 3.71479802e-02
4.65645194e-01 -1.18503201e+00 2.87981659e-01 5.06475806e-01
5.90998650e-01 -8.96126628e-01 1.65049043e-02 -4.03976470e-01
-8.08367014e-01 7.26935744e-01 4.75384295e-01 -2.72666097e-01
8.48362803e-01 1.93991438e-01 -6.20768487e-01 -3.46748143e-01
-8.47919822e-01 -3.54596913e-01 3.38652045e-01 5.09066761e-01
7.58369714e-02 2.21976507e-02 -1.30002260e-01 3.22179675e-01
4.44142669e-01 1.24704428e-01 -6.37168884e-02 9.16206717e-01
-2.54619997e-02 -9.11348999e-01 -2.76560802e-02 7.12196752e-02
-3.21315259e-01 1.94527224e-01 1.28402701e-02 6.05006814e-01
7.57551014e-01 8.11347365e-01 4.85819072e-01 -4.52935040e-01
8.79993066e-02 -1.37704238e-01 4.47158575e-01 -4.62653488e-01
-6.93856835e-01 1.49959728e-01 -9.73720029e-02 -8.37836802e-01
-1.05776215e+00 -7.95955598e-01 -1.10968709e+00 -1.63030952e-01
4.32585664e-02 6.42935783e-02 4.41621602e-01 1.22620344e+00
4.77361619e-01 6.35010481e-01 7.57755935e-01 -9.37454164e-01
2.44112179e-01 -9.65803862e-01 -7.82505155e-01 6.46061301e-01
2.97477841e-01 -8.73349786e-01 -2.84055740e-01 3.35410058e-01] | [8.399045944213867, 0.7013359069824219] |
c4be4a34-8c1e-4a23-af11-ea9cabde8c07 | a-critical-analysis-of-image-based-camera | 2201.05816 | null | https://arxiv.org/abs/2201.05816v1 | https://arxiv.org/pdf/2201.05816v1.pdf | A Critical Analysis of Image-based Camera Pose Estimation Techniques | Camera, and associated with its objects within the field of view, localization could benefit many computer vision fields, such as autonomous driving, robot navigation, and augmented reality (AR). In this survey, we first introduce specific application areas and the evaluation metrics for camera localization pose according to different sub-tasks (learning-based 2D-2D task, feature-based 2D-3D task, and 3D-3D task). Then, we review common methods for structure-based camera pose estimation approaches, absolute pose regression and relative pose regression approaches by critically modelling the methods to inspire further improvements in their algorithms such as loss functions, neural network structures. Furthermore, we summarise what are the popular datasets used for camera localization and compare the quantitative and qualitative results of these methods with detailed performance metrics. Finally, we discuss future research possibilities and applications. | ['Pengfei Xu', 'Stefan Poslad', 'Jian Ren', 'Jun Zhang', 'Bin Xu', 'Youchen Wang', 'Meng Xu'] | 2022-01-15 | null | null | null | null | ['camera-localization'] | ['computer-vision'] | [-7.06744269e-02 -1.17970422e-01 -4.13813412e-01 -4.03173178e-01
-9.26450908e-01 -7.94569016e-01 6.08423293e-01 -1.24738842e-01
-5.83650231e-01 3.73290658e-01 1.95387322e-02 -8.08590874e-02
-2.16493279e-01 -7.66895562e-02 -6.99345350e-01 -5.20530045e-01
2.67980583e-02 4.79569703e-01 1.58536926e-01 1.47247612e-01
6.27499044e-01 9.83452737e-01 -1.49812281e+00 -5.93598425e-01
2.27548540e-01 1.22067130e+00 3.01638871e-01 6.46806419e-01
3.75036240e-01 4.75243300e-01 -3.50383341e-01 -2.08450064e-01
3.38305682e-01 1.34213030e-01 -2.24985942e-01 1.48029134e-01
1.07200122e+00 -2.08815709e-01 -3.03178519e-01 1.01849210e+00
5.05938530e-01 3.70108038e-01 8.57424200e-01 -1.31878459e+00
-2.50919789e-01 -1.88454449e-01 -5.48736572e-01 1.14474848e-01
9.92216706e-01 -8.80497396e-02 6.00240290e-01 -1.18069506e+00
5.82665443e-01 1.15780711e+00 1.22740269e+00 3.11873168e-01
-6.60881281e-01 -3.60711128e-01 1.94299445e-01 1.02329679e-01
-1.53989911e+00 -2.42886528e-01 7.18001902e-01 -7.24503934e-01
1.07569766e+00 -7.31021836e-02 3.32934529e-01 1.13575256e+00
3.66760045e-01 6.81722701e-01 9.51118886e-01 -4.56484050e-01
-3.87839414e-02 2.33428627e-01 -2.82152393e-03 8.90760481e-01
3.14261734e-01 2.32874006e-01 -5.22911012e-01 1.17183685e-01
6.72322690e-01 -1.38369843e-01 -2.04276349e-02 -1.26585579e+00
-1.46938086e+00 8.52467358e-01 6.69473588e-01 -2.27993920e-01
4.00521867e-02 3.64553094e-01 3.24866474e-01 -1.00722380e-01
2.33426705e-01 6.29197180e-01 -7.13717759e-01 1.87860001e-02
-6.07914686e-01 1.86131522e-01 3.91627669e-01 1.71781611e+00
8.32244098e-01 -1.61596492e-01 3.52830857e-01 6.28455937e-01
5.82550049e-01 8.58630180e-01 2.83367783e-01 -1.11369240e+00
5.70281625e-01 2.67589748e-01 2.33580783e-01 -1.14304435e+00
-8.21066439e-01 -1.52895465e-01 -5.32022357e-01 -4.51907516e-03
-1.18126990e-02 -4.05507572e-02 -5.18421829e-01 1.16865551e+00
1.26359776e-01 6.17469065e-02 -7.31683671e-02 9.41314638e-01
1.02313387e+00 3.01973313e-01 -3.26304317e-01 7.13299289e-02
1.29552579e+00 -1.31621218e+00 -3.75989348e-01 -8.05090487e-01
6.30166769e-01 -8.01728487e-01 7.14642465e-01 3.50002170e-01
-7.70008683e-01 -7.00794399e-01 -1.24215198e+00 -1.98150635e-01
-5.84491491e-01 7.18603492e-01 5.00556231e-01 6.02412462e-01
-9.42312062e-01 7.53512839e-03 -7.90300965e-01 -7.57386804e-01
1.60258055e-01 5.67095995e-01 -6.43456638e-01 -6.31045625e-02
-7.26726711e-01 1.55221903e+00 1.95147127e-01 4.21097577e-02
-7.37278700e-01 -2.46482700e-01 -1.35427535e+00 -5.78850985e-01
2.61614919e-01 -6.53348684e-01 1.23348260e+00 -1.17506452e-01
-1.30542457e+00 1.21558607e+00 -1.12544768e-01 -3.65306169e-01
4.45090890e-01 -5.31423151e-01 -1.15152746e-01 -2.00039342e-01
2.11469904e-01 1.11289132e+00 6.31362915e-01 -1.25491405e+00
-1.03597879e+00 -5.39269269e-01 1.76656637e-02 4.85593975e-01
3.09885293e-01 -5.15376665e-02 -8.50994110e-01 1.03604998e-02
5.11861920e-01 -1.24195135e+00 -3.96477699e-01 1.79639548e-01
-1.81483194e-01 -6.75438121e-02 8.43932092e-01 -2.34363750e-01
7.48808503e-01 -2.22625566e+00 3.49915415e-01 -8.44815969e-02
4.82811156e-05 -1.98747873e-01 7.18306750e-02 1.65044770e-01
2.55703241e-01 -3.84315670e-01 2.09956393e-01 -6.90133870e-01
-8.41737092e-02 2.43564602e-02 -6.71777204e-02 9.72633958e-01
-1.76663160e-01 7.70239413e-01 -8.05949152e-01 -3.34623665e-01
9.89974976e-01 5.52764118e-01 -5.20631671e-02 1.02958053e-01
3.53525639e-01 4.22958016e-01 -2.96517551e-01 8.81529808e-01
6.70770109e-01 1.58181787e-01 -4.00768876e-01 -5.60239375e-01
-2.48276800e-01 1.11863250e-02 -9.67020571e-01 1.60046089e+00
-7.65111208e-01 1.19013727e+00 -4.76293825e-02 -4.97820109e-01
1.23167574e+00 -1.91750750e-01 3.65629196e-01 -5.68936110e-01
2.97571063e-01 3.57681841e-01 -6.49400711e-01 -3.24473739e-01
8.79290760e-01 4.82711494e-01 -4.93461102e-01 -3.12067300e-01
1.37770116e-01 -4.92209226e-01 -1.70030653e-01 -3.54711831e-01
7.55967379e-01 3.98168832e-01 7.20014691e-01 5.60683720e-02
8.60941648e-01 1.71697482e-01 -7.45390728e-02 5.41360259e-01
-5.20859599e-01 9.30824697e-01 6.61950335e-02 -6.46298230e-01
-1.03560722e+00 -9.42021966e-01 -2.58620232e-01 8.27377319e-01
9.11573410e-01 -1.56166762e-01 -4.40311253e-01 -7.06944823e-01
2.16751754e-01 4.30377364e-01 -4.64061171e-01 -1.81082189e-01
-6.19354188e-01 -3.53486329e-01 3.77577603e-01 8.79589021e-01
3.87878925e-01 -6.51530206e-01 -9.03787076e-01 -3.85013759e-01
-1.31385654e-01 -1.53127396e+00 -4.53948230e-01 5.63457489e-01
-8.74638796e-01 -1.03410578e+00 -5.50238252e-01 -9.17533934e-01
7.75819421e-01 6.78883791e-01 1.01802135e+00 -6.36861980e-01
-8.20569992e-02 9.78587747e-01 -2.04399601e-01 -3.75794977e-01
1.83664098e-01 1.57270193e-01 4.48068798e-01 -6.19460881e-01
7.29481101e-01 1.02526434e-01 -3.57094496e-01 7.18524575e-01
-2.17616907e-03 -3.77151281e-01 7.13460684e-01 3.95434558e-01
7.40167618e-01 -4.22084510e-01 -3.84744436e-01 -3.09017211e-01
4.01295066e-01 -2.41302580e-01 -1.17497492e+00 2.67145205e-02
-6.42043293e-01 -2.15008751e-01 -6.89691231e-02 -4.48289216e-01
-4.61275965e-01 8.13282371e-01 1.63605571e-01 -7.86730707e-01
-1.87313423e-01 2.46141687e-01 -1.72898918e-01 -7.25375533e-01
8.03523242e-01 7.08812177e-02 3.48522142e-02 -2.53982216e-01
3.39763045e-01 6.13091648e-01 7.61557341e-01 -5.87535985e-02
8.00658822e-01 3.81333381e-01 3.40854138e-01 -8.34424734e-01
-9.34178114e-01 -1.12121546e+00 -1.21315801e+00 -4.62067038e-01
1.00579655e+00 -1.37808764e+00 -7.17105091e-01 2.96006739e-01
-1.32442164e+00 1.06237322e-01 5.32542318e-02 8.84451151e-01
-9.87309039e-01 2.47154430e-01 2.10314821e-02 -7.95122325e-01
1.27961740e-01 -1.69244695e+00 1.70924997e+00 2.30483145e-01
-1.95412770e-01 -1.06386614e+00 1.60640657e-01 4.69416529e-01
7.22540542e-02 2.58666217e-01 1.96927369e-01 -3.16850424e-01
-7.33935714e-01 -7.19193816e-01 -3.19448203e-01 2.42023692e-01
-2.52510995e-01 -2.15430722e-01 -8.58574152e-01 -3.99302781e-01
-2.59427369e-01 -2.91668326e-01 5.37918687e-01 7.91671753e-01
7.02936292e-01 2.78044671e-01 -8.83126855e-01 1.00562048e+00
1.51089048e+00 1.44537836e-01 3.86609465e-01 8.01500320e-01
8.49497557e-01 5.56329072e-01 1.01051033e+00 5.52514791e-02
4.68053311e-01 9.56207097e-01 7.55889297e-01 1.41913995e-01
2.23305151e-02 -3.63915175e-01 5.33195794e-01 7.01319873e-01
4.45682146e-02 -1.88790604e-01 -7.36606777e-01 2.58305937e-01
-1.82714081e+00 -2.32331619e-01 -2.89777994e-01 2.19248271e+00
-1.78899780e-01 1.34247318e-01 1.45275220e-01 -2.59109408e-01
7.22095430e-01 1.32631764e-01 -4.22483057e-01 -1.27401143e-01
-7.12315664e-02 -5.59565306e-01 1.32002532e+00 5.35921574e-01
-1.71670341e+00 9.46135581e-01 7.35073853e+00 4.53830034e-01
-1.02443588e+00 3.73099968e-02 1.88770354e-01 2.23107487e-01
4.92820382e-01 -5.71788549e-02 -1.54327285e+00 2.65211940e-01
5.57410061e-01 4.13156718e-01 1.57857358e-01 1.48772252e+00
-6.92843497e-02 -4.94279265e-01 -1.39269650e+00 1.58638072e+00
6.45759046e-01 -1.14180446e+00 -3.51798326e-01 6.92679062e-02
6.27097905e-01 5.23584127e-01 1.91279024e-01 2.87420779e-01
-2.41118103e-01 -8.21149290e-01 9.56642210e-01 3.75981778e-01
1.03014386e+00 -6.08347356e-01 1.10579288e+00 3.13903064e-01
-1.24354315e+00 -2.41794676e-01 -5.02770245e-01 2.89247204e-02
1.43273652e-01 9.18676630e-02 -7.04630077e-01 3.38235497e-01
7.85775900e-01 1.04066586e+00 -9.02495444e-01 1.29194033e+00
-1.95494875e-01 -2.47035280e-01 -3.75110418e-01 -3.25980186e-01
4.25389290e-01 -2.30462939e-01 5.26256859e-01 1.11749506e+00
3.97121072e-01 -4.52595264e-01 2.38826901e-01 4.68208909e-01
2.21527487e-01 -3.04327488e-01 -9.98238623e-01 5.17175913e-01
6.83292627e-01 1.35461473e+00 -7.79248059e-01 3.49606633e-01
-6.69369280e-01 6.96506739e-01 4.37646024e-02 1.12448864e-01
-9.93520916e-01 -3.65153313e-01 4.66332227e-01 2.42237762e-01
2.94594586e-01 -6.99900508e-01 -1.09090798e-01 -8.39441061e-01
3.42733599e-02 -4.31507677e-01 -6.71851113e-02 -1.22640312e+00
-1.07345498e+00 4.94235486e-01 4.63361144e-01 -1.70652664e+00
-1.87596887e-01 -1.33363295e+00 -9.90888625e-02 4.31810379e-01
-1.22210145e+00 -1.26791346e+00 -6.09966218e-01 1.36157617e-01
6.29784822e-01 -4.93085504e-01 3.39466900e-01 2.33591899e-01
-1.60201311e-01 5.25650442e-01 2.32116476e-01 1.38219684e-01
7.36941397e-01 -1.09373176e+00 3.87528926e-01 5.54241598e-01
1.34754702e-01 3.16627234e-01 5.24861276e-01 -2.72923946e-01
-1.54781687e+00 -1.05887270e+00 7.53171682e-01 -1.34603500e+00
3.60565156e-01 -5.83378732e-01 3.07898633e-02 8.98564100e-01
-1.22784212e-01 9.85876843e-02 1.48787946e-02 3.68927233e-02
-1.12998508e-01 -2.49424368e-01 -1.13762248e+00 3.53443354e-01
9.02211726e-01 -4.32360262e-01 -3.52003068e-01 4.72144336e-01
5.64699233e-01 -9.09372866e-01 -5.10112345e-01 5.39883256e-01
5.44507384e-01 -1.02146900e+00 1.35847425e+00 1.08910585e-02
-1.05058625e-01 -4.42926228e-01 -5.72965384e-01 -8.66224468e-01
-4.32575434e-01 -2.38821521e-01 -7.98369199e-02 6.27444386e-01
3.63656998e-01 -4.39601421e-01 9.68632460e-01 6.89498335e-03
-3.60524565e-01 -6.67728364e-01 -1.04183006e+00 -8.76128376e-01
-2.99729258e-01 -5.45984030e-01 9.93594676e-02 3.39061856e-01
-4.69871879e-01 6.18073463e-01 -2.77082682e-01 4.40552264e-01
5.66508591e-01 -3.72359842e-01 1.05519676e+00 -1.25472021e+00
3.57729971e-01 -4.78356481e-01 -1.06718087e+00 -1.53968918e+00
2.91460246e-01 -4.50031877e-01 2.64786303e-01 -1.37720001e+00
1.02389053e-01 -2.08351225e-01 2.77920842e-01 -1.40673727e-01
3.34067076e-01 3.20569694e-01 -2.91737206e-02 3.27811509e-01
-1.00177658e+00 1.72658309e-01 9.02609587e-01 -1.57961026e-01
7.27901757e-02 4.31933284e-01 -2.21665248e-01 9.22378600e-01
2.93287307e-01 -3.15566570e-01 -4.55376267e-01 -4.97056395e-01
4.69969928e-01 -4.26844507e-02 5.32947898e-01 -1.34145463e+00
6.18326426e-01 2.55541801e-02 6.17253602e-01 -1.20199883e+00
7.50950456e-01 -1.16875577e+00 -1.42150849e-01 3.57130438e-01
-1.26796305e-01 5.33147037e-01 7.21002417e-03 5.52206099e-01
-3.49350542e-01 -5.31074166e-01 9.23780620e-01 -2.17569247e-01
-1.21148062e+00 2.80165404e-01 -4.41105872e-01 -1.21115230e-01
1.35752821e+00 -9.10686553e-01 7.41861984e-02 -5.41639566e-01
-5.01104712e-01 2.98739314e-01 5.85287690e-01 7.19971418e-01
7.92995453e-01 -1.39751244e+00 -2.43915230e-01 1.54299691e-01
5.82014680e-01 2.08843365e-01 -1.99817836e-01 1.02892745e+00
-9.32628036e-01 1.01483309e+00 -3.05115953e-02 -1.22442245e+00
-1.27208722e+00 7.96692848e-01 5.25463283e-01 2.41209507e-01
-8.22958574e-02 1.01968598e+00 1.04092762e-01 -1.07906508e+00
6.04407132e-01 -4.97162610e-01 -2.55436778e-01 -1.71273321e-01
7.10940883e-02 6.23638511e-01 2.36014411e-01 -1.05759227e+00
-8.91163111e-01 1.36519825e+00 2.40103900e-01 9.65512022e-02
8.14588010e-01 -6.34289980e-01 1.26014277e-01 4.55175668e-01
1.36632156e+00 -1.88436553e-01 -1.36673176e+00 -4.27896418e-02
1.40637636e-01 -4.23124522e-01 2.98281640e-01 -5.15497029e-01
-7.97811687e-01 8.25141549e-01 1.01252127e+00 -4.03423399e-01
8.09514105e-01 3.80296528e-01 1.76722661e-01 7.80723810e-01
7.60246813e-01 -9.41657841e-01 2.21359119e-01 7.74437726e-01
7.08219826e-01 -1.60934186e+00 4.36517715e-01 -3.27537745e-01
-5.90746403e-01 1.10118294e+00 8.13836873e-01 -1.99490681e-01
7.56906152e-01 2.87891775e-01 1.60040304e-01 -1.26971662e-01
-1.96196795e-01 -1.21220738e-01 5.31733274e-01 1.17202652e+00
2.61091232e-01 -1.66404739e-01 4.34125125e-01 9.22952816e-02
-2.31695369e-01 -6.75366819e-01 2.65033215e-01 8.10012817e-01
-4.59726512e-01 -5.82077920e-01 -6.48913801e-01 9.21446481e-04
2.52614766e-01 2.31294170e-01 -6.63588107e-01 1.16673291e+00
3.23702723e-01 6.77695572e-01 1.75965413e-01 -5.49363911e-01
6.45114481e-01 -4.31895763e-01 8.69596660e-01 -4.96252239e-01
-1.37975728e-02 4.71005291e-02 -2.14057490e-01 -5.38658500e-01
-5.71673036e-01 -8.02103937e-01 -6.87448263e-01 -4.90594171e-02
-7.93707311e-01 -2.90848076e-01 1.16312993e+00 6.33693159e-01
1.95359409e-01 5.23425341e-02 6.20532990e-01 -1.41334820e+00
-3.65966618e-01 -8.61096442e-01 -2.57208735e-01 -3.52748215e-01
5.54857671e-01 -1.18730903e+00 -5.69984078e-01 -4.21439484e-02] | [7.565536022186279, -2.1579835414886475] |
d1609fb7-e952-4e50-afec-80b0ce7329c3 | representation-power-of-graph-convolutions | 2210.09809 | null | https://arxiv.org/abs/2210.09809v2 | https://arxiv.org/pdf/2210.09809v2.pdf | Analysis of Convolutions, Non-linearity and Depth in Graph Neural Networks using Neural Tangent Kernel | The fundamental principle of Graph Neural Networks (GNNs) is to exploit the structural information of the data by aggregating the neighboring nodes using a `graph convolution' in conjunction with a suitable choice for the network architecture, such as depth and activation functions. Therefore, understanding the influence of each of the design choice on the network performance is crucial. Convolutions based on graph Laplacian have emerged as the dominant choice with the symmetric normalization of the adjacency matrix as the most widely adopted one. However, some empirical studies show that row normalization of the adjacency matrix outperforms it in node classification. Despite the widespread use of GNNs, there is no rigorous theoretical study on the representation power of these convolutions, that could explain this behavior. Similarly, the empirical observation of the linear GNNs performance being on par with non-linear ReLU GNNs lacks rigorous theory. In this work, we theoretically analyze the influence of different aspects of the GNN architecture using the Graph Neural Tangent Kernel in a semi-supervised node classification setting. Under the population Degree Corrected Stochastic Block Model, we prove that: (i) linear networks capture the class information as good as ReLU networks; (ii) row normalization preserves the underlying class structure better than other convolutions; (iii) performance degrades with network depth due to over-smoothing, but the loss in class information is the slowest in row normalization; (iv) skip connections retain the class information even at infinite depth, thereby eliminating over-smoothing. We finally validate our theoretical findings numerically and on real datasets such as Cora and Citeseer. | ['Debarghya Ghoshdastidar', 'Pascal Esser', 'Mahalakshmi Sabanayagam'] | 2022-10-18 | null | null | null | null | ['stochastic-block-model'] | ['graphs'] | [ 5.11225313e-02 2.85481513e-01 -1.14138857e-01 -8.29976276e-02
2.89317310e-01 -6.23444021e-01 6.08838022e-01 3.54377091e-01
-5.12925327e-01 4.76372242e-01 -3.92314158e-02 -5.02472937e-01
-4.90657032e-01 -1.05200982e+00 -8.20727170e-01 -1.13425994e+00
-3.76459688e-01 2.25303739e-01 3.18077922e-01 -3.27072978e-01
-1.04325332e-01 6.66380584e-01 -1.18347287e+00 -5.77285513e-02
8.37040663e-01 8.58888268e-01 -8.74571651e-02 5.57101846e-01
-1.02058314e-01 7.33834565e-01 -2.66326517e-01 -5.03410876e-01
5.29649198e-01 -4.84072089e-01 -5.73133707e-01 -2.45844543e-01
3.07724059e-01 1.45661384e-01 -6.94557250e-01 1.35144627e+00
2.72642702e-01 5.31172566e-02 8.30442607e-01 -1.25855768e+00
-6.25494599e-01 1.11258507e+00 -4.49031383e-01 2.29507983e-01
-2.62954384e-01 1.08455323e-01 1.22591007e+00 -3.94152492e-01
5.91158628e-01 1.03371668e+00 9.35555756e-01 3.28585595e-01
-1.58514345e+00 -4.22551453e-01 9.87147726e-03 -8.41108486e-02
-1.74851584e+00 -1.00188799e-01 8.22982132e-01 -2.93341577e-01
6.79503441e-01 1.08582675e-01 6.21410728e-01 9.93990362e-01
2.06840903e-01 2.99676061e-01 9.12935793e-01 -4.46085781e-01
9.13332328e-02 2.17434436e-01 5.34691334e-01 8.19304943e-01
8.76336873e-01 1.84749160e-02 -1.92242056e-01 -2.68533565e-02
8.41106176e-01 7.82889053e-02 -4.21499282e-01 -6.61037624e-01
-9.03832138e-01 8.51870477e-01 1.09720874e+00 6.24905467e-01
-3.93948555e-01 4.87946570e-01 2.71244198e-01 6.42170787e-01
3.87174636e-01 3.60839009e-01 -1.05605401e-01 3.17917794e-01
-7.78426945e-01 -1.79379448e-01 7.93154001e-01 6.43604517e-01
9.88899112e-01 1.82519123e-01 1.31181762e-01 4.73025590e-01
2.60833502e-01 3.69876027e-01 1.89295694e-01 -6.19829059e-01
2.52486259e-01 9.28468466e-01 -5.95901012e-01 -1.13691020e+00
-7.83527076e-01 -1.00695300e+00 -1.50565517e+00 2.15737686e-01
9.57317770e-01 -2.04753652e-01 -7.40566790e-01 1.97410643e+00
-2.38672808e-01 -1.78437099e-01 5.26832324e-03 6.63508475e-01
6.73192739e-01 2.75472075e-01 5.11662737e-02 8.05680454e-02
1.06916916e+00 -4.03594017e-01 -3.71377766e-01 -2.24102408e-01
8.07603657e-01 -1.78129107e-01 8.14136386e-01 -6.67802291e-03
-8.26448262e-01 -2.79099733e-01 -1.14383996e+00 1.99405849e-01
-5.25271356e-01 -6.86143264e-02 8.62328827e-01 9.00876582e-01
-1.45513892e+00 9.69771981e-01 -8.09856832e-01 -6.56560302e-01
4.95294243e-01 6.14170194e-01 -5.56558132e-01 -1.04531661e-01
-1.24222720e+00 6.49099410e-01 5.56241572e-01 2.04156637e-01
-3.63639802e-01 -5.86216509e-01 -6.48748100e-01 4.47520196e-01
1.15467988e-01 -6.99181497e-01 3.71847123e-01 -1.23789752e+00
-1.13238871e+00 6.68458223e-01 3.29369277e-01 -8.44930708e-01
5.16081095e-01 3.60489815e-01 -1.57679692e-01 1.10695973e-01
-4.26745832e-01 6.46183550e-01 5.64567387e-01 -1.06490743e+00
-7.94405192e-02 -5.51551342e-01 2.72993445e-01 9.17699784e-02
-5.90695739e-01 -5.65658033e-01 -8.53115544e-02 -4.34689522e-01
3.15321594e-01 -1.00478029e+00 -2.26618424e-01 -1.56657949e-01
-5.64194143e-01 -3.89329158e-02 4.58614886e-01 -2.79707998e-01
1.33543897e+00 -2.42641544e+00 1.21695697e-01 7.23780215e-01
6.66834772e-01 7.72248283e-02 -2.07123622e-01 7.27335989e-01
-3.10573131e-01 1.69517234e-01 -2.51161307e-01 8.55443161e-03
-9.72072035e-02 2.23300502e-01 4.98332456e-03 8.48208666e-01
-3.18810833e-03 1.01429904e+00 -7.31379330e-01 -1.03002064e-01
-1.29245184e-02 7.36077428e-01 -4.63728309e-01 -3.17778975e-01
1.80932209e-01 4.40594517e-02 -3.12889814e-01 7.94653594e-02
5.67385137e-01 -6.00575924e-01 3.93565685e-01 -3.16561639e-01
1.34829745e-01 1.75223276e-01 -1.04317617e+00 1.37020946e+00
-6.05673008e-02 8.30657661e-01 2.17783213e-01 -1.22702432e+00
8.59743357e-01 8.68410245e-02 4.23421979e-01 -5.12294590e-01
3.08443189e-01 1.98824197e-01 6.81649327e-01 1.12085328e-01
9.30885077e-02 8.80773365e-02 2.56984055e-01 5.05523086e-01
1.88771755e-01 5.02700686e-01 3.30567479e-01 6.01608098e-01
1.40752912e+00 -5.58449745e-01 1.20476931e-01 -8.00523460e-01
2.87832439e-01 -2.58816093e-01 1.39298037e-01 9.78164792e-01
-4.66696126e-03 4.74611878e-01 1.06495178e+00 -3.19668829e-01
-8.96203756e-01 -1.02855301e+00 -2.15702638e-01 9.31566536e-01
7.20143840e-02 -2.02379212e-01 -9.61081803e-01 -5.44102490e-01
5.83885424e-02 3.61675531e-01 -9.62755740e-01 -5.06882191e-01
-3.20151240e-01 -1.12395072e+00 7.91637063e-01 2.57557660e-01
5.88545382e-01 -8.04601073e-01 -3.44444513e-01 -1.09925151e-01
3.27477068e-01 -9.67543662e-01 -2.85493970e-01 5.38056314e-01
-9.33707058e-01 -1.10931814e+00 -5.96466184e-01 -6.26862228e-01
1.01309669e+00 3.01973790e-01 9.99939740e-01 4.74741310e-01
1.45613074e-01 3.21015745e-01 -1.99473411e-01 3.98507416e-02
-4.97445613e-01 7.25272477e-01 1.31934490e-02 1.89679042e-01
1.66560277e-01 -9.98525798e-01 -5.30470669e-01 3.20581347e-02
-1.01514697e+00 -6.23281710e-02 9.90108430e-01 6.36679769e-01
-6.70761243e-02 3.32163036e-01 3.28749269e-01 -1.25007975e+00
8.65199327e-01 -4.21888024e-01 -4.53320563e-01 1.61873221e-01
-8.25967133e-01 4.89253908e-01 6.52639210e-01 -3.49860579e-01
-4.55147564e-01 -8.00059512e-02 6.88152090e-02 -1.77093640e-01
1.12254471e-01 7.14270473e-01 -5.50273657e-02 -4.72315818e-01
8.97312999e-01 7.65897706e-02 2.92666793e-01 -3.04632753e-01
3.07246685e-01 1.62508085e-01 3.19421202e-01 -3.28804702e-01
1.00378239e+00 5.41008592e-01 4.45534855e-01 -1.07493746e+00
-3.95781547e-01 -8.81115422e-02 -7.76440501e-01 -1.75086066e-01
8.00059080e-01 -5.03278673e-01 -8.87893319e-01 3.88837487e-01
-9.72514272e-01 -4.18109059e-01 -3.30655485e-01 4.31127906e-01
7.62722865e-02 3.10868770e-01 -7.86780238e-01 -7.64667153e-01
-2.93115437e-01 -9.51851547e-01 2.78210521e-01 2.55360126e-01
1.34974435e-01 -1.32296455e+00 -1.86810002e-01 -4.07465935e-01
6.45758212e-01 1.15163729e-01 1.29435301e+00 -8.01428080e-01
-5.28754890e-01 -2.63014525e-01 -6.22201383e-01 3.41215312e-01
-7.87173286e-02 5.92631996e-02 -8.08797061e-01 -4.04108495e-01
-4.58551079e-01 2.56151795e-01 1.20948923e+00 6.21322989e-01
7.32498527e-01 -2.98478246e-01 -2.30998218e-01 7.05180228e-01
1.52095962e+00 -2.60395467e-01 7.32194126e-01 2.92364322e-02
9.37055171e-01 6.25019014e-01 -6.36165261e-01 -7.29763657e-02
1.70483842e-01 2.99555331e-01 6.45014465e-01 -2.17344627e-01
-2.09614277e-01 -1.62359029e-01 4.02104199e-01 9.16506231e-01
-3.80966425e-01 -3.63543719e-01 -9.05520678e-01 2.50369072e-01
-1.70277309e+00 -8.54399562e-01 -3.13301265e-01 2.45765734e+00
3.78323168e-01 4.15668726e-01 1.88699558e-01 2.52033561e-01
6.96486235e-01 2.31740832e-01 -3.40022385e-01 -1.50728732e-01
-5.18331230e-01 7.81723633e-02 1.12471545e+00 3.99003863e-01
-8.78840566e-01 5.78119099e-01 5.85767555e+00 6.88588679e-01
-1.15757132e+00 -6.79401234e-02 6.50897741e-01 1.97936669e-01
-4.26644713e-01 6.42401278e-02 -4.26857293e-01 2.75006026e-01
8.68878782e-01 -6.34668097e-02 6.89772487e-01 4.55752164e-01
-1.54939413e-01 1.32066756e-01 -1.00042856e+00 7.35723376e-01
-1.14468560e-01 -1.29294288e+00 8.85495171e-02 4.98215020e-01
5.27763009e-01 2.56588846e-01 -2.91779395e-02 2.58457065e-01
4.52772409e-01 -1.07525885e+00 5.07937491e-01 3.66214305e-01
5.15225351e-01 -7.73208141e-01 7.88377881e-01 2.87282586e-01
-1.02726269e+00 -1.04437716e-01 -4.62082922e-01 -1.13460973e-01
-2.78612018e-01 8.20925057e-01 -4.44388568e-01 4.71158475e-01
4.46001261e-01 5.81109464e-01 -9.78668213e-01 9.06604528e-01
-1.30348355e-01 8.28422070e-01 -5.39412320e-01 -9.18872505e-02
3.91050935e-01 -5.78347683e-01 5.99485874e-01 1.11157680e+00
9.94654521e-02 -2.04022303e-01 -2.92023629e-01 1.05329967e+00
-3.30147117e-01 3.40926880e-03 -7.54975677e-01 -2.81168878e-01
2.96291202e-01 1.28573799e+00 -1.15775430e+00 2.07031086e-01
-4.04739648e-01 7.40915716e-01 5.84205031e-01 8.45700443e-01
-5.21934688e-01 -4.85679060e-01 6.11656368e-01 4.49933201e-01
2.73184776e-01 -2.73008317e-01 -3.82920474e-01 -7.66075790e-01
-1.16938025e-01 -4.38670784e-01 3.03654552e-01 -3.36151749e-01
-1.26883459e+00 6.46803081e-01 -2.10701302e-01 -8.07840466e-01
7.03307241e-02 -8.08411121e-01 -5.97579002e-01 7.70552456e-01
-1.25317883e+00 -8.69773567e-01 -2.08582789e-01 4.27659929e-01
-4.97745275e-01 -9.22288150e-02 5.22929549e-01 4.82483089e-01
-7.59076834e-01 7.70769060e-01 1.60119742e-01 5.75984418e-01
2.48146713e-01 -1.30197644e+00 3.06450218e-01 9.07738984e-01
3.62021148e-01 9.49713409e-01 6.65247321e-01 -4.80830044e-01
-1.49107993e+00 -8.88455153e-01 5.22701204e-01 -1.48910791e-01
7.87693381e-01 -5.49793959e-01 -1.15616274e+00 5.76601148e-01
1.84219941e-01 2.32351139e-01 3.03037137e-01 1.76186591e-01
-5.71980417e-01 -2.43428007e-01 -9.04628873e-01 6.28758788e-01
1.20810091e+00 -5.28171480e-01 6.73894286e-02 2.77777482e-02
5.89150310e-01 1.51337858e-03 -8.58541846e-01 3.07277590e-01
5.45802295e-01 -1.13305330e+00 9.22714591e-01 -5.60502946e-01
2.08945408e-01 -7.51208514e-02 -1.48556624e-02 -1.28663003e+00
-5.62524557e-01 -4.08896089e-01 1.51100114e-01 1.11885440e+00
6.38146102e-01 -1.09917033e+00 9.29397225e-01 4.79087502e-01
1.61445871e-01 -6.78044498e-01 -8.99283469e-01 -7.33551204e-01
2.11974755e-01 -1.64321661e-01 2.80802280e-01 9.99824047e-01
-2.17758209e-01 5.94088495e-01 8.39748885e-03 1.52769879e-01
6.24419630e-01 -1.89381510e-01 4.92777944e-01 -1.56613135e+00
-1.44289479e-01 -9.96336341e-01 -7.96515703e-01 -8.16837192e-01
1.40689060e-01 -1.31146884e+00 -4.00334328e-01 -1.49564004e+00
3.60651724e-02 -5.77541888e-01 -5.38856626e-01 3.41157585e-01
4.88235764e-02 2.99104869e-01 8.19504410e-02 1.87186375e-01
-2.81579047e-01 1.44763276e-01 1.04059327e+00 -1.10718936e-01
-2.41137341e-01 3.73845585e-02 -7.61007488e-01 5.35375059e-01
7.20977128e-01 -4.60277021e-01 -5.03262281e-01 -4.35114563e-01
7.22998917e-01 -2.78678983e-01 4.95268703e-01 -1.08515871e+00
3.76672566e-01 3.02455068e-01 1.49326220e-01 -1.04395144e-01
-7.30895773e-02 -9.34559882e-01 1.71565041e-01 6.22793972e-01
-2.95370221e-01 1.00449115e-01 2.28964947e-02 6.78237617e-01
9.51531231e-02 -1.15585819e-01 7.69548595e-01 8.02479163e-02
-2.82783121e-01 3.32008839e-01 -3.54924083e-01 3.48549597e-02
5.83250403e-01 -3.20018470e-01 -4.33817118e-01 -5.23428917e-01
-4.56107616e-01 -1.21139558e-02 5.57699859e-01 1.31119579e-01
4.91751097e-02 -1.17482305e+00 -6.55828834e-01 3.04253817e-01
-1.74423248e-01 -2.38369837e-01 1.73920572e-01 1.19581628e+00
-5.51097989e-01 3.17606926e-01 -9.86526310e-02 -5.46995044e-01
-8.01065624e-01 4.62504506e-01 5.60499549e-01 -3.49233896e-01
-6.39735401e-01 6.65410221e-01 4.57438022e-01 -3.33966464e-01
2.41527125e-01 -3.44655931e-01 -1.91133872e-01 2.18832064e-02
7.90532604e-02 2.92054325e-01 2.09676966e-01 -6.92543030e-01
-2.94649005e-01 3.21714908e-01 1.27849609e-01 2.14820430e-01
1.30251384e+00 -1.63563509e-02 -2.99977213e-01 3.92172277e-01
1.18320167e+00 -8.09488297e-02 -1.02061152e+00 -3.27549666e-01
8.07956234e-03 1.56933576e-01 1.22845225e-01 -2.43340358e-01
-1.57216167e+00 1.00881970e+00 3.90126348e-01 7.54557908e-01
8.56397390e-01 -1.47217996e-02 2.74477988e-01 3.69809419e-01
2.18638390e-01 -6.17351651e-01 -3.29328805e-01 5.11547565e-01
5.65213442e-01 -8.32835019e-01 -2.03690119e-02 -4.60963279e-01
-1.81248590e-01 1.06899023e+00 2.76104361e-01 -4.20717567e-01
9.26359177e-01 1.08903676e-01 -2.28180751e-01 -5.73010921e-01
-4.50191796e-01 -3.21013808e-01 2.31250420e-01 4.67359602e-01
2.94409454e-01 1.30838230e-01 -2.33861953e-01 5.03075004e-01
-2.95528710e-01 -5.68343639e-01 4.91946459e-01 2.86345810e-01
-1.96348548e-01 -8.09937298e-01 8.15870762e-02 7.25835204e-01
-3.52408588e-01 -3.23267996e-01 -6.33672476e-01 1.02407169e+00
-2.40647033e-01 6.12798393e-01 1.50787920e-01 -4.47745204e-01
1.82096198e-01 5.38545027e-02 5.42576969e-01 -3.97393227e-01
-7.09865928e-01 -3.68020326e-01 -2.00392380e-01 -2.51363963e-01
-1.85925663e-01 -3.92491251e-01 -1.19077587e+00 -6.95514619e-01
-4.57047343e-01 2.19073623e-01 5.17553627e-01 8.70320499e-01
4.25742149e-01 6.34669900e-01 3.06912214e-01 -5.51143885e-01
-4.85098243e-01 -8.94607186e-01 -9.10553157e-01 5.03271759e-01
2.18867242e-01 -5.13255298e-01 -7.85296977e-01 -4.78881359e-01] | [6.823948860168457, 6.033977031707764] |
aaaf68bf-bbfc-4212-b5ac-625bd8f67ccf | 3d-human-pose-estimation-under-limited | 1908.05293 | null | https://arxiv.org/abs/1908.05293v3 | https://arxiv.org/pdf/1908.05293v3.pdf | Multiview-Consistent Semi-Supervised Learning for 3D Human Pose Estimation | The best performing methods for 3D human pose estimation from monocular images require large amounts of in-the-wild 2D and controlled 3D pose annotated datasets which are costly and require sophisticated systems to acquire. To reduce this annotation dependency, we propose Multiview-Consistent Semi Supervised Learning (MCSS) framework that utilizes similarity in pose information from unannotated, uncalibrated but synchronized multi-view videos of human motions as additional weak supervision signal to guide 3D human pose regression. Our framework applies hard-negative mining based on temporal relations in multi-view videos to arrive at a multi-view consistent pose embedding. When jointly trained with limited 3D pose annotations, our approach improves the baseline by 25% and state-of-the-art by 8.7%, whilst using substantially smaller networks. Lastly, but importantly, we demonstrate the advantages of the learned embedding and establish view-invariant pose retrieval benchmarks on two popular, publicly available multi-view human pose datasets, Human 3.6M and MPI-INF-3DHP, to facilitate future research. | ['Nitesh B. Gundavarapu', 'Rahul Mitra', 'Arjun Jain', 'Abhishek Sharma'] | 2019-08-14 | multiview-consistent-semi-supervised-learning | http://openaccess.thecvf.com/content_CVPR_2020/html/Mitra_Multiview-Consistent_Semi-Supervised_Learning_for_3D_Human_Pose_Estimation_CVPR_2020_paper.html | http://openaccess.thecvf.com/content_CVPR_2020/papers/Mitra_Multiview-Consistent_Semi-Supervised_Learning_for_3D_Human_Pose_Estimation_CVPR_2020_paper.pdf | cvpr-2020-6 | ['pose-retrieval'] | ['computer-vision'] | [-1.27921402e-01 -1.88069552e-01 -3.27701807e-01 -4.77711827e-01
-8.83773386e-01 -6.08116269e-01 3.80087316e-01 -4.13630605e-01
-6.70979857e-01 5.24047971e-01 4.90747273e-01 7.59645641e-01
1.73666865e-01 -3.91686447e-02 -8.06193769e-01 -2.51796961e-01
-1.36041820e-01 8.49689960e-01 1.15395218e-01 -2.85046011e-01
-3.29278022e-01 4.05305773e-01 -1.58272064e+00 5.62537760e-02
-1.39949387e-02 8.61529827e-01 1.51737249e-02 7.03371763e-01
9.05410886e-01 5.49076915e-01 -3.84013653e-01 -2.65241981e-01
6.05577767e-01 -8.83397684e-02 -6.49195433e-01 3.00105423e-01
1.02722204e+00 -6.76075339e-01 -6.77256644e-01 5.19755125e-01
1.05256677e+00 1.68789789e-01 3.59423071e-01 -1.33098316e+00
-6.60479963e-02 -1.50441498e-01 -7.07984507e-01 7.46736899e-02
9.86723244e-01 2.37158060e-01 9.73021328e-01 -1.11191249e+00
1.14150023e+00 1.42538226e+00 7.06314862e-01 6.82471931e-01
-1.20727277e+00 -4.85205919e-01 2.20248356e-01 -2.68454198e-02
-1.31678402e+00 -4.07243073e-01 7.92597115e-01 -3.98898840e-01
1.09077656e+00 2.36216456e-01 1.06347334e+00 1.58260691e+00
2.28635237e-01 9.54457760e-01 9.13977742e-01 -1.73923388e-01
-3.59048545e-01 -3.25595021e-01 -2.81134009e-01 9.82478917e-01
2.57721961e-01 8.25617090e-02 -1.05149913e+00 -9.60925221e-02
8.90821338e-01 2.92239845e-01 -1.94498062e-01 -1.18007469e+00
-1.39257538e+00 5.50053179e-01 4.63546991e-01 -2.15234473e-01
-2.40314171e-01 3.48363787e-01 6.60351157e-01 1.60488829e-01
6.85429811e-01 2.97647446e-01 -8.00654888e-01 -2.08492801e-01
-6.41876340e-01 6.79566562e-01 4.09084827e-01 1.18112242e+00
3.37752223e-01 -2.67931044e-01 3.70339490e-03 6.38380170e-01
2.74233997e-01 8.27749193e-01 1.94423005e-01 -1.09431565e+00
8.14774513e-01 6.50853634e-01 1.79278478e-01 -1.04636502e+00
-7.17991829e-01 -3.20936471e-01 -5.19952238e-01 -2.65686214e-01
4.14547950e-01 7.47816414e-02 -8.61932397e-01 1.84853280e+00
6.47388101e-01 -3.31344634e-01 -4.30955172e-01 1.41332984e+00
7.68653810e-01 1.51303038e-01 -1.53133407e-01 1.17752783e-01
1.25583470e+00 -1.14759481e+00 -4.38283116e-01 -5.76662540e-01
6.25116467e-01 -4.52408940e-01 9.18403506e-01 4.00792509e-01
-1.01490247e+00 -7.42912114e-01 -1.02006531e+00 -3.48977387e-01
1.92239154e-02 4.84650254e-01 6.03458762e-01 1.61225066e-01
-6.32294953e-01 3.97870988e-01 -1.24697316e+00 -3.29770178e-01
1.68176562e-01 4.44161177e-01 -1.15092289e+00 -1.80202842e-01
-1.02648842e+00 9.33411002e-01 9.19113234e-02 2.04148740e-01
-9.91708398e-01 -6.20315492e-01 -1.35624671e+00 -5.98597467e-01
8.36379886e-01 -1.09108484e+00 9.56257343e-01 -2.63781488e-01
-1.06289399e+00 1.35419953e+00 -6.66163629e-03 -2.74454504e-01
9.54642236e-01 -1.19274509e+00 8.68765563e-02 4.69957054e-01
3.25360119e-01 1.00524378e+00 7.71872222e-01 -1.09750426e+00
-3.09820890e-01 -9.69510555e-01 -6.39941022e-02 6.45565212e-01
-9.52169150e-02 -2.99189121e-01 -1.04673481e+00 -5.82755923e-01
3.82101238e-01 -1.51394653e+00 -1.98042572e-01 3.52926344e-01
-5.46031773e-01 -1.28595993e-01 5.05697429e-01 -8.18260252e-01
8.05135369e-01 -1.72698879e+00 9.94764030e-01 -1.55309103e-02
3.36016625e-01 -1.90086707e-01 -1.74896885e-02 2.60829985e-01
-6.11286722e-02 -5.06393075e-01 3.79584908e-01 -8.02959263e-01
4.99985777e-02 2.61728436e-01 2.58748591e-01 9.19899583e-01
2.20054369e-02 1.08550584e+00 -8.46155643e-01 -5.95225930e-01
4.61307675e-01 5.61817765e-01 -7.55567491e-01 5.29265583e-01
3.78945693e-02 8.19535017e-01 -1.80077538e-01 8.35333109e-01
7.19485953e-02 -5.63355803e-01 1.83821902e-01 -6.71750665e-01
3.64825159e-01 -8.09561461e-02 -1.14985764e+00 2.70106149e+00
-1.18995063e-01 3.34288657e-01 -1.93792522e-01 -7.89431751e-01
4.64404553e-01 3.38506490e-01 9.01695967e-01 -3.36186737e-01
2.26269737e-01 -2.87572555e-02 -4.81860548e-01 -4.91104692e-01
4.19927388e-01 2.79274970e-01 -4.74429816e-01 2.05332845e-01
4.42715406e-01 3.74516435e-02 -1.18435649e-02 1.87426776e-01
1.01842725e+00 8.31247091e-01 4.41199422e-01 2.03391075e-01
3.05542499e-01 -2.04541057e-01 6.35241449e-01 2.84257740e-01
-5.30117214e-01 8.35614264e-01 1.73019588e-01 -7.84383059e-01
-1.17304063e+00 -1.27315152e+00 2.94997156e-01 1.23253202e+00
9.95149761e-02 -5.44662416e-01 -4.36365515e-01 -8.23221743e-01
2.24309400e-01 -2.67491430e-01 -6.91403210e-01 1.38120160e-01
-9.09314513e-01 -2.15504736e-01 4.24917936e-01 7.73441613e-01
3.73553336e-01 -5.78639328e-01 -9.64463830e-01 -2.56181091e-01
-5.69274783e-01 -1.51727402e+00 -8.34525108e-01 8.88816044e-02
-6.88945174e-01 -1.15254056e+00 -1.06891322e+00 -6.22182369e-01
6.92120194e-01 2.32149109e-01 1.23481274e+00 -2.32713729e-01
-6.37711763e-01 9.08646166e-01 -2.12644354e-01 7.19106421e-02
3.40606272e-01 1.94842011e-01 7.34020054e-01 -2.98231423e-01
2.15377808e-01 -4.91730809e-01 -6.75262570e-01 5.44147670e-01
-2.15953827e-01 7.98169225e-02 5.55898547e-01 1.00891232e+00
8.58085513e-01 -7.90014207e-01 1.51379287e-01 -5.72686255e-01
-5.46457656e-02 3.22473310e-02 -3.68067712e-01 1.63466603e-01
-2.73005664e-01 -3.02822199e-02 1.09249614e-01 -4.33060676e-01
-7.42748618e-01 7.33786583e-01 2.24412739e-01 -9.71263707e-01
-1.54066429e-01 2.84926835e-02 -1.22643404e-01 -5.28144129e-02
7.68096507e-01 -2.37916596e-03 2.77228683e-01 -3.29738379e-01
4.90726411e-01 2.14970395e-01 8.81710649e-01 -4.35620189e-01
1.00504458e+00 7.25155652e-01 1.11393139e-01 -4.31979895e-01
-1.23859859e+00 -8.76320362e-01 -1.28129971e+00 -4.55322504e-01
1.10557914e+00 -1.76237261e+00 -6.95844173e-01 3.30254167e-01
-1.07208443e+00 1.20999813e-02 1.03498340e-01 6.78171933e-01
-9.22791779e-01 4.49759543e-01 -7.47213066e-01 -6.66709960e-01
-3.97502065e-01 -1.04314494e+00 1.73571932e+00 -3.65418524e-01
-7.43009508e-01 -5.50389349e-01 2.24069640e-01 9.18272972e-01
-2.92253673e-01 7.32450962e-01 1.59903616e-01 -4.74889696e-01
-3.94741625e-01 -5.25961101e-01 1.58267319e-01 2.86675483e-01
1.01653859e-02 -7.01522648e-01 -7.03977704e-01 -7.76094258e-01
-3.63166839e-01 -1.09894407e+00 6.48254454e-01 3.57126892e-01
6.55401886e-01 3.63714136e-02 -4.19376373e-01 6.38141036e-01
9.23036635e-01 -6.42658055e-01 1.64160311e-01 1.34052873e-01
1.13270700e+00 7.76732147e-01 9.48818564e-01 6.84035659e-01
5.90572536e-01 1.07367313e+00 2.50503838e-01 8.74982029e-02
3.64802182e-02 -5.54793835e-01 3.48993301e-01 1.00312614e+00
-3.44796330e-01 9.81827453e-02 -6.31861806e-01 3.27719748e-01
-1.97945893e+00 -9.09296513e-01 3.97567183e-01 2.02581859e+00
8.16007316e-01 1.75836802e-01 4.88894224e-01 4.65002805e-02
3.92814577e-01 3.34794343e-01 -7.32502401e-01 5.93616128e-01
8.22851732e-02 -8.59532133e-02 4.95649159e-01 2.68220514e-01
-1.54617000e+00 8.34807396e-01 5.48323917e+00 3.15602183e-01
-8.19561541e-01 -1.03481654e-02 3.20376933e-01 -8.64692390e-01
2.30710045e-01 -5.09837091e-01 -8.54323387e-01 -4.01842706e-02
4.10889030e-01 4.13952470e-01 1.26973346e-01 9.93567884e-01
1.45923927e-01 -8.11186433e-02 -1.66452777e+00 1.69934189e+00
5.35831988e-01 -7.99588263e-01 -1.59457162e-01 9.91181955e-02
7.05824971e-01 1.39496207e-01 -9.94413570e-02 1.87795088e-01
-8.77077281e-02 -8.06722879e-01 7.63625741e-01 5.32445848e-01
8.36106122e-01 -7.93202758e-01 5.31307220e-01 4.47498769e-01
-1.42983758e+00 1.19215719e-01 -9.83415097e-02 -1.38182729e-01
4.31228876e-01 1.59649834e-01 -4.11688238e-01 6.35305226e-01
9.37811792e-01 1.05061698e+00 -7.07050681e-01 4.64533240e-01
-2.07722917e-01 -1.69005066e-01 -4.35094446e-01 1.56609833e-01
-1.76126480e-01 3.32039356e-01 5.00100970e-01 6.26321614e-01
-1.50776833e-01 1.09454673e-02 6.99656904e-01 3.82764041e-01
-2.39664949e-02 -2.27906108e-01 -7.10878730e-01 1.61776245e-01
2.72933275e-01 1.24284339e+00 -4.77327287e-01 -1.69512659e-01
-2.77605772e-01 1.48779547e+00 4.88687366e-01 4.97119203e-02
-9.41991210e-01 1.37220010e-01 5.63091457e-01 1.28067896e-01
2.13533178e-01 -6.35803103e-01 1.70628756e-01 -1.49978137e+00
4.23330247e-01 -9.71967399e-01 3.52425039e-01 -7.11386502e-01
-1.31577420e+00 3.22076768e-01 3.99165958e-01 -1.51186717e+00
-7.90342689e-01 -5.33032358e-01 4.06752765e-01 4.73642379e-01
-9.41531539e-01 -1.53521872e+00 -4.71978426e-01 6.17799819e-01
5.45730412e-01 -1.42812371e-01 7.49583840e-01 4.37401652e-01
-3.36855024e-01 7.05233753e-01 -7.51498699e-01 3.49105686e-01
1.27626288e+00 -1.08076108e+00 5.78283072e-01 4.00614589e-01
4.12358850e-01 6.13872707e-01 6.37660921e-01 -6.71882331e-01
-1.80979192e+00 -9.27394927e-01 8.47773135e-01 -1.27789903e+00
1.11448497e-01 -7.68447161e-01 -2.04325140e-01 9.19678748e-01
-3.61656666e-01 4.56875086e-01 6.20873809e-01 1.79762959e-01
-5.15384734e-01 4.70536873e-02 -9.51005340e-01 5.49618065e-01
1.64742649e+00 -5.51388383e-01 -7.31031120e-01 4.50877249e-01
7.90346384e-01 -9.49341357e-01 -1.14631379e+00 5.56800246e-01
1.06873643e+00 -7.75016904e-01 1.44688153e+00 -7.26846457e-01
3.78882051e-01 -1.77425772e-01 -5.19350767e-01 -9.96760011e-01
-1.84247062e-01 -4.76360619e-01 -1.89603388e-01 4.01477814e-01
6.47465810e-02 -5.63834198e-02 1.18583143e+00 4.70832348e-01
1.14215046e-01 -8.70290458e-01 -1.02316880e+00 -8.70316386e-01
-4.07218546e-01 -2.78961509e-01 -1.02776960e-01 6.24681771e-01
-9.90835279e-02 4.70159769e-01 -9.89015639e-01 1.00594349e-01
9.02373433e-01 -7.60769099e-02 1.47938561e+00 -1.22381115e+00
-5.03783107e-01 4.29728895e-01 -7.70397007e-01 -1.43309748e+00
2.55650014e-01 -5.52502811e-01 8.66460353e-02 -1.00400937e+00
4.73475367e-01 2.46105582e-01 1.14454843e-01 4.53135610e-01
-5.88715598e-02 6.76441312e-01 3.16063285e-01 3.41296375e-01
-1.21428311e+00 7.26048708e-01 1.26072502e+00 7.06556812e-02
5.24552390e-02 -2.14297444e-01 -1.02718160e-01 8.23264539e-01
1.68010592e-01 -3.22431803e-01 -4.96968716e-01 -3.59550834e-01
2.03531861e-01 2.39222676e-01 8.20797920e-01 -1.06789982e+00
1.48857534e-01 1.50065079e-01 9.27612007e-01 -1.03713512e+00
1.06356871e+00 -6.70770288e-01 1.03849374e-01 5.69800079e-01
-3.75370175e-01 5.19748628e-01 -5.30474074e-02 8.19114268e-01
1.43089639e-02 5.87213635e-01 4.30249512e-01 -4.72088873e-01
-7.61416614e-01 6.46313727e-01 3.40079755e-01 5.57138503e-01
9.74173963e-01 -5.80708943e-02 1.47986010e-01 -6.16119206e-01
-9.54018056e-01 3.49906147e-01 5.12226641e-01 7.08410561e-01
7.25296378e-01 -1.68547213e+00 -4.59077537e-01 9.19449627e-02
4.59070444e-01 1.75123185e-01 3.59885484e-01 7.50709176e-01
-4.80093807e-01 7.32519209e-01 -3.92856538e-01 -1.12847316e+00
-1.55213284e+00 2.57400423e-01 1.59600958e-01 -3.37512076e-01
-6.52519941e-01 8.59017193e-01 -6.37763292e-02 -8.63062203e-01
4.58516985e-01 -1.39036968e-01 1.92506924e-01 -1.49754379e-02
2.57837057e-01 4.55583811e-01 -4.05461974e-02 -1.06788039e+00
-6.19325697e-01 7.79483616e-01 6.16717935e-02 -2.90689915e-01
1.32613862e+00 -1.28565982e-01 2.96949029e-01 6.17735267e-01
1.59969556e+00 -2.11881608e-01 -1.57368982e+00 -3.14855337e-01
-3.84500742e-01 -5.16816914e-01 -4.23629850e-01 -5.63063800e-01
-8.20004702e-01 6.89715147e-01 6.71380937e-01 -8.46019149e-01
6.96273029e-01 1.66950613e-01 8.59068274e-01 7.99173415e-01
8.09835494e-01 -1.34128892e+00 6.78623676e-01 3.45938832e-01
1.13930798e+00 -1.53847349e+00 4.90906596e-01 -4.78340238e-01
-6.42388225e-01 7.27994382e-01 9.69479620e-01 -2.31339335e-01
6.06199622e-01 -5.75034842e-02 -3.25916447e-02 -3.68317783e-01
-7.72149801e-01 -8.31350535e-02 7.74571717e-01 5.27208447e-01
5.30357659e-01 -7.48404711e-02 -1.80719495e-02 4.69918847e-01
-2.72413492e-01 -1.24847449e-01 -1.55749962e-01 1.12419152e+00
-1.19934581e-01 -8.27356100e-01 -2.59903580e-01 2.11939394e-01
-2.80597717e-01 3.14267606e-01 -5.47454953e-01 1.02409792e+00
-4.05376069e-02 5.49976051e-01 -1.11041754e-01 -7.32301950e-01
7.13926971e-01 1.88519303e-02 1.01226652e+00 -7.24111199e-01
-3.51088762e-01 2.76918888e-01 1.75317273e-01 -1.20522678e+00
-7.43943512e-01 -8.72414112e-01 -1.03219974e+00 -6.53490350e-02
-1.76887706e-01 -4.78094488e-01 1.79731995e-01 1.03026617e+00
4.43799913e-01 1.80328473e-01 2.49220625e-01 -1.49360800e+00
-7.49856293e-01 -9.13147688e-01 -3.43823969e-01 8.31680298e-01
2.20233679e-01 -1.16458726e+00 -1.80444509e-01 -8.13828781e-03] | [7.028799533843994, -0.8625867962837219] |
1e019abc-ec62-4834-88da-82d9ac5b1ab7 | cross-attention-guided-dense-network-for | 2109.11393 | null | https://arxiv.org/abs/2109.11393v2 | https://arxiv.org/pdf/2109.11393v2.pdf | Cross Attention-guided Dense Network for Images Fusion | In recent years, various applications in computer vision have achieved substantial progress based on deep learning, which has been widely used for image fusion and shown to achieve adequate performance. However, suffering from limited ability in modeling the spatial correspondence of different source images, it still remains a great challenge for existing unsupervised image fusion models to extract appropriate feature and achieves adaptive and balanced fusion. In this paper, we propose a novel cross-attention-guided image fusion network, which is a unified and unsupervised framework for multi-modal image fusion, multi-exposure image fusion, and multi-focus image fusion. Different from the existing self-attention module, our cross-attention module focus on modeling the cross-correlation between different source images. Using the proposed cross attention module as a core block, a densely connected cross attention-guided network is built to dynamically learn the spatial correspondence to derive better alignment of important details from different input images. Meanwhile, an auxiliary branch is also designed to model the long-range information, and a merging network is attached to finally reconstruct the fusion image. Extensive experiments have been carried out on publicly available datasets, and the results demonstrate that the proposed model outperforms the state-of-the-art quantitatively and qualitatively. | ['Jiangyu Wang', 'Yulian Li', 'Zaiyu Pan', 'Jun Wang', 'Zhengwen Shen'] | 2021-09-23 | null | null | null | null | ['multi-exposure-image-fusion'] | ['computer-vision'] | [ 1.54419914e-01 -3.34826678e-01 3.19953039e-02 -3.38265091e-01
-1.03581250e+00 8.24977681e-02 5.02808988e-01 -1.61116961e-02
-3.28823030e-01 4.54026818e-01 3.89138281e-01 1.60377860e-01
-3.04742426e-01 -7.26215661e-01 -6.22754097e-01 -9.64090288e-01
5.18740058e-01 -9.58245061e-03 2.61377960e-01 -3.06928456e-01
1.86848670e-01 3.15494150e-01 -1.62564826e+00 -4.57139648e-02
1.29608607e+00 1.08205092e+00 6.08035445e-01 3.03872973e-01
-1.81506038e-01 6.91932380e-01 -3.94928753e-01 -2.87681669e-01
-7.32456818e-02 -4.50361282e-01 -5.10264814e-01 4.15811449e-01
2.33821258e-01 -2.75907695e-01 -5.41668892e-01 1.35215139e+00
5.87029517e-01 1.98099405e-01 5.76659143e-01 -1.12883651e+00
-9.01331902e-01 5.24176598e-01 -8.82287443e-01 4.12956178e-01
2.70056799e-02 1.14165977e-01 6.85055137e-01 -8.50252271e-01
3.80735956e-02 1.28987288e+00 2.69363016e-01 5.22769652e-02
-9.68832195e-01 -8.77028644e-01 2.84210503e-01 3.76795828e-01
-1.33303916e+00 -5.11900783e-01 1.18086553e+00 -2.03278616e-01
4.62428182e-01 -2.32681725e-02 5.47774911e-01 4.96474981e-01
3.77035707e-01 8.97319019e-01 1.02378047e+00 -3.95055354e-01
-3.31008196e-01 -7.64815435e-02 9.07390863e-02 7.40420580e-01
6.23942576e-02 1.61584005e-01 -2.47623503e-01 2.28788450e-01
7.49874115e-01 5.60220063e-01 -6.42037690e-01 -1.57078728e-01
-1.32452595e+00 6.72297359e-01 1.07630515e+00 8.56259346e-01
-5.85381627e-01 -1.00266598e-01 -5.66246063e-02 -2.80878004e-02
5.97517967e-01 1.55472904e-01 -2.03191608e-01 5.10728657e-01
-9.99044478e-01 4.97368984e-02 2.81702373e-02 5.61385095e-01
1.12687874e+00 -1.43892355e-02 -2.50089556e-01 8.52513313e-01
5.44003963e-01 5.30179262e-01 6.39772773e-01 -8.35083425e-01
3.43786091e-01 9.58752513e-01 4.08096384e-04 -1.39052820e+00
-3.23870420e-01 -7.19266534e-01 -1.41655183e+00 1.76897794e-01
-1.25960931e-01 -3.40243019e-02 -9.57872272e-01 1.80715084e+00
3.33521515e-01 4.85690057e-01 2.55662084e-01 1.02664983e+00
1.06569076e+00 7.21659184e-01 1.52066991e-01 -4.61870402e-01
1.34075344e+00 -1.22962618e+00 -9.75419343e-01 -1.59084976e-01
1.34729236e-01 -1.02434504e+00 5.70385695e-01 2.05149472e-01
-1.25625336e+00 -1.09618163e+00 -1.14897752e+00 -2.80715168e-01
-3.96823585e-01 6.93289936e-02 5.22628367e-01 7.48413876e-02
-1.02523053e+00 1.34992063e-01 -5.70738494e-01 -2.60459661e-01
5.66629231e-01 2.89242834e-01 -5.25173306e-01 -4.11718458e-01
-1.09537041e+00 7.72635937e-01 5.76528192e-01 3.57292742e-01
-7.35067248e-01 -4.84058440e-01 -1.05231738e+00 2.64226198e-01
4.27716315e-01 -8.44972372e-01 8.75078321e-01 -9.00040567e-01
-1.21924198e+00 5.25201678e-01 -1.81024641e-01 -2.04979971e-01
1.90575086e-02 -1.23813584e-01 -6.21373057e-01 4.31472026e-02
2.75432199e-01 8.85911047e-01 8.40778172e-01 -1.55269229e+00
-8.47049415e-01 -4.32073236e-01 -4.54239845e-02 4.43175197e-01
-4.33766097e-01 3.99953574e-02 -7.79735148e-01 -8.56959343e-01
2.17026487e-01 -3.89895260e-01 -3.36699516e-01 -3.65995586e-01
-2.62343168e-01 -3.09803206e-02 1.11967289e+00 -7.14803934e-01
1.19055569e+00 -2.07398605e+00 4.93640840e-01 -1.54513747e-01
3.38806897e-01 3.50259334e-01 -2.39187866e-01 7.44606629e-02
-9.68544558e-02 -1.88795343e-01 -5.99433303e-01 -5.62100828e-01
-3.42746824e-01 7.43941888e-02 -2.54983723e-01 2.98970193e-01
1.74891770e-01 1.11516559e+00 -7.11493254e-01 -7.16146350e-01
7.37502575e-01 7.00597525e-01 -2.86676317e-01 4.97794569e-01
1.92721188e-02 7.50396729e-01 -3.78046751e-01 6.95256352e-01
9.92187023e-01 -5.03653944e-01 -2.83955932e-01 -6.71093047e-01
-7.95119405e-02 -4.75379020e-01 -9.79235888e-01 2.08742619e+00
-4.20345277e-01 1.92270041e-01 1.28756478e-01 -1.22671914e+00
9.62305784e-01 1.51065275e-01 6.07813418e-01 -8.20161045e-01
5.90814352e-01 1.69446561e-02 3.98160471e-03 -5.04629195e-01
4.30670500e-01 -9.75314453e-02 -3.92559581e-02 2.47374922e-01
2.97411948e-01 -2.42892504e-02 1.29903555e-01 2.30891630e-01
5.29537559e-01 -1.57161534e-01 1.83076188e-01 -1.71730250e-01
1.19497716e+00 -2.68152893e-01 6.01820946e-01 2.87774384e-01
-6.47455603e-02 8.03472281e-01 -2.79842407e-01 -2.80376047e-01
-6.53491139e-01 -8.72273743e-01 7.68013597e-02 6.76449120e-01
7.24796355e-01 -2.44835779e-01 -7.27723300e-01 -5.11097550e-01
-2.85652846e-01 3.04987431e-01 -5.78099012e-01 -3.77696067e-01
-3.58038783e-01 -8.62975121e-01 3.68216559e-02 4.75339919e-01
1.18391538e+00 -1.12122405e+00 -2.49856010e-01 3.16053480e-01
-6.29674315e-01 -1.09485340e+00 -6.82629049e-01 -2.21323773e-01
-6.14834547e-01 -1.16303658e+00 -9.50150669e-01 -9.54778194e-01
6.90843701e-01 9.98776913e-01 8.14570010e-01 3.87634426e-01
-1.61392555e-01 3.13441306e-01 -3.29621851e-01 -4.22489405e-01
4.77046184e-02 -1.48866922e-02 -3.29339594e-01 6.30695283e-01
1.53775543e-01 -6.53194427e-01 -6.92356825e-01 2.86443323e-01
-1.33573580e+00 3.23625207e-01 8.56984138e-01 1.05449677e+00
6.59388065e-01 3.95560533e-01 5.68742990e-01 -3.36279780e-01
4.14687485e-01 -5.12895465e-01 -5.28687537e-01 5.03488481e-01
-3.82503003e-01 3.17913927e-02 4.45392340e-01 -1.17187828e-01
-1.44932878e+00 -1.91987917e-01 -2.40724787e-01 -7.23417878e-01
-1.44117907e-01 5.80651104e-01 -7.37960637e-01 -2.39721730e-01
-1.36434332e-01 4.79581922e-01 1.79866761e-01 -4.96745169e-01
3.95495564e-01 6.20330632e-01 9.05666590e-01 -2.37197995e-01
7.69675076e-01 6.08227789e-01 -2.27946997e-01 -4.62571383e-01
-1.00917661e+00 -2.71686912e-01 -6.14039063e-01 -1.49089277e-01
1.22110164e+00 -1.12631011e+00 -5.81421614e-01 9.35366154e-01
-1.11137617e+00 6.76853284e-02 5.81941605e-02 4.66145426e-01
-3.28432441e-01 4.92888480e-01 -2.66565800e-01 -4.75743771e-01
-5.07064998e-01 -1.60610032e+00 1.26956773e+00 8.62313807e-01
5.11690259e-01 -1.00536776e+00 -5.13859652e-02 6.37918413e-01
4.96408731e-01 1.01010323e-01 5.80918789e-01 -1.88190803e-01
-8.44181657e-01 8.07572976e-02 -6.12931728e-01 3.34309489e-01
5.31847537e-01 -1.88905939e-01 -7.67622650e-01 -3.11321557e-01
1.77675247e-01 -9.00287554e-02 1.22853899e+00 4.80562389e-01
1.21290338e+00 5.31524494e-02 -4.49962437e-01 8.07858646e-01
1.46254015e+00 2.83498734e-01 6.13999903e-01 1.50782362e-01
9.08968151e-01 4.11050588e-01 6.52433217e-01 5.02961762e-02
6.53560936e-01 5.23618579e-01 5.53975761e-01 -5.78768134e-01
-1.70096636e-01 -8.47550705e-02 -2.09520422e-02 9.73137259e-01
1.04183711e-01 -2.16416284e-01 -5.36589682e-01 5.34813523e-01
-2.03359079e+00 -9.64044929e-01 2.51979113e-01 1.85420704e+00
4.64118391e-01 -8.73357579e-02 -2.04036742e-01 1.18947960e-01
9.01443481e-01 5.03576994e-01 -5.53448558e-01 3.83497179e-01
-4.72243398e-01 2.09099837e-02 9.33262482e-02 6.27913952e-01
-1.27281535e+00 8.05486798e-01 5.50138998e+00 1.06747532e+00
-1.19683349e+00 3.64451021e-01 8.71298015e-01 2.94368029e-01
-3.20180535e-01 -4.71954048e-02 -5.29583454e-01 7.18817949e-01
2.35650703e-01 -3.12765017e-02 2.19120070e-01 3.46936762e-01
-8.07015300e-02 -1.72640786e-01 -3.51309329e-01 1.20101237e+00
4.62423295e-01 -1.29771245e+00 1.66433305e-01 -3.01298350e-02
1.01247883e+00 -1.78649470e-01 1.10423520e-01 1.62321836e-01
3.21234286e-01 -8.24597001e-01 4.16724771e-01 1.04536867e+00
5.11976719e-01 -9.62252378e-01 1.04935932e+00 3.88859570e-01
-1.52700806e+00 -2.47453332e-01 -2.63820827e-01 1.73201114e-01
3.95827621e-01 7.54730105e-01 1.54384539e-01 1.24518096e+00
7.81556845e-01 1.06492352e+00 -7.25213587e-01 9.95481551e-01
-1.01670623e-01 6.37602434e-02 -1.17116600e-01 6.39735818e-01
2.33190581e-01 -3.26412588e-01 3.56452256e-01 7.63342977e-01
4.41962838e-01 2.88516581e-01 3.01343948e-01 6.60945952e-01
-7.24799186e-02 -1.81346759e-02 -3.44258994e-01 2.93074161e-01
1.93931922e-01 1.55160642e+00 -6.88610673e-01 -6.36918664e-01
-5.96127093e-01 8.27841580e-01 2.69987464e-01 3.72525692e-01
-8.64301383e-01 -2.90770441e-01 5.54663241e-01 -2.66838700e-01
4.77776319e-01 -8.15995932e-02 -7.63592869e-02 -1.47994924e+00
-1.46036655e-01 -8.08243811e-01 5.10844588e-01 -1.14451134e+00
-1.26642191e+00 8.15835059e-01 -9.32762101e-02 -1.20776618e+00
1.45481259e-01 -1.41187951e-01 -9.32798684e-01 1.04557312e+00
-1.80342722e+00 -1.47828650e+00 -7.47818530e-01 9.25341725e-01
5.21474719e-01 -2.88479388e-01 3.83843780e-01 4.37100768e-01
-7.53464639e-01 4.87466991e-01 1.79646574e-02 1.16923144e-02
8.03866923e-01 -8.28818917e-01 -1.35357261e-01 1.18371654e+00
1.86126888e-01 5.48904121e-01 3.07836980e-01 -6.04291677e-01
-1.25514674e+00 -1.21871555e+00 4.35274214e-01 4.31549810e-02
2.66427010e-01 1.51475534e-01 -1.16488326e+00 4.96374995e-01
7.60660648e-01 2.22240329e-01 3.27562302e-01 -3.55107218e-01
-1.03291035e-01 -5.68436623e-01 -9.14141476e-01 3.21199298e-01
8.06455672e-01 -3.17045480e-01 -6.59381092e-01 3.78774665e-02
9.47021484e-01 -3.58905792e-01 -9.05668080e-01 7.60893345e-01
1.34760320e-01 -1.17028010e+00 1.09740925e+00 -3.94839868e-02
3.83145303e-01 -7.69638836e-01 -1.40477553e-01 -1.37067330e+00
-5.56350529e-01 -3.32836777e-01 4.33990359e-02 1.66053402e+00
-2.27636367e-01 -6.79826677e-01 2.58806318e-01 1.82104707e-01
-2.93834955e-01 -7.78473139e-01 -6.08983159e-01 -9.65953022e-02
-2.11258397e-01 -9.01727825e-02 7.84548104e-01 7.42499709e-01
-4.80258405e-01 5.69337547e-01 -4.35108632e-01 3.74439687e-01
8.27243447e-01 4.79289562e-01 7.14216411e-01 -1.10896778e+00
-1.52557716e-01 -7.45124459e-01 -3.42580259e-01 -1.16023374e+00
2.56550550e-01 -7.03138828e-01 -1.74500234e-02 -1.71999574e+00
4.58912820e-01 -2.00412318e-01 -8.11146080e-01 2.47493997e-01
-7.24552870e-01 4.38784719e-01 2.65653551e-01 2.29366362e-01
-8.02785158e-01 1.03436673e+00 1.42205012e+00 -4.75140542e-01
1.13138922e-01 -3.69034380e-01 -1.22446692e+00 6.85594678e-01
5.86666644e-01 -1.45818830e-01 -3.30674648e-01 -6.07210755e-01
-3.41165066e-01 2.98220515e-01 5.17785609e-01 -1.22692275e+00
6.12168491e-01 1.06763551e-02 5.98059297e-01 -1.00193715e+00
3.34994763e-01 -9.74769592e-01 1.13945156e-01 2.24935591e-01
-9.41147879e-02 -2.65659355e-02 1.20343842e-01 5.95660567e-01
-8.18183601e-01 1.08114727e-01 8.15241754e-01 -6.87153935e-02
-7.78663635e-01 8.29964161e-01 1.37391984e-01 -3.19698632e-01
1.02661419e+00 9.46086347e-02 -2.66151041e-01 -1.92438781e-01
-5.51339447e-01 4.61236060e-01 4.90178913e-01 5.75663805e-01
7.59043932e-01 -1.58930445e+00 -7.09566236e-01 4.43326682e-01
1.84081718e-01 2.27392197e-01 9.70088124e-01 8.90680015e-01
9.88201723e-02 2.39141494e-01 -3.53868842e-01 -6.45177484e-01
-1.16807413e+00 8.93918753e-01 2.49784023e-01 -3.71772528e-01
-4.08973426e-01 6.58522785e-01 7.10065067e-01 -2.13995323e-01
-7.28378594e-02 -1.94198668e-01 -4.39443409e-01 -4.68143448e-02
7.86993682e-01 8.79458338e-02 -8.54160637e-02 -1.19027114e+00
-1.17720447e-01 1.14093673e+00 -2.52161860e-01 1.21400505e-01
1.15074444e+00 -5.47316134e-01 -3.41595411e-01 1.44535795e-01
1.25479949e+00 -2.34905019e-01 -1.30532861e+00 -7.21264184e-01
-6.60008311e-01 -5.14349997e-01 3.90990525e-01 -5.22173405e-01
-1.73902965e+00 1.09353113e+00 6.30376697e-01 7.84966871e-02
1.78579438e+00 -4.77543622e-02 9.82893884e-01 -9.99170542e-02
2.42363855e-01 -4.54036862e-01 1.90620035e-01 2.25526258e-01
8.77844155e-01 -1.59052372e+00 2.08180845e-02 -2.62766570e-01
-3.30084860e-01 8.06572855e-01 8.70456159e-01 -2.09183052e-01
8.59511554e-01 1.16333894e-01 -5.06580137e-02 -1.43274277e-01
-4.05504733e-01 -4.17883933e-01 4.14073437e-01 4.64223891e-01
1.53837457e-01 -2.26072714e-01 -1.11774571e-01 7.30089545e-01
1.70455679e-01 1.05249219e-01 7.91250914e-03 6.05881870e-01
-5.54855585e-01 -9.90169883e-01 -6.16380930e-01 3.06521088e-01
-3.17789912e-01 -8.66256189e-03 1.84279367e-01 6.17380857e-01
4.20625001e-01 1.14329779e+00 1.80774048e-01 -5.41391194e-01
9.08962265e-02 -4.02990878e-01 4.20618415e-01 -2.15554729e-01
-4.49596107e-01 3.38041842e-01 -7.04377890e-01 -3.70239019e-01
-1.00799453e+00 -3.47038209e-01 -1.00234330e+00 -9.44773853e-02
-4.98649478e-01 2.76576310e-01 2.76006788e-01 1.18502367e+00
5.63535154e-01 8.96116138e-01 7.24447787e-01 -1.19563472e+00
2.00633854e-01 -1.09940338e+00 -3.84620517e-01 4.76144940e-01
4.78894085e-01 -1.02710438e+00 -2.11792976e-01 1.08390100e-01] | [10.550753593444824, -1.83033287525177] |
77ff703c-cff0-4108-a004-6a2a61425c6c | t-ner-an-all-round-python-library-for-1 | 2209.12616 | null | https://arxiv.org/abs/2209.12616v1 | https://arxiv.org/pdf/2209.12616v1.pdf | T-NER: An All-Round Python Library for Transformer-based Named Entity Recognition | Language model (LM) pretraining has led to consistent improvements in many NLP downstream tasks, including named entity recognition (NER). In this paper, we present T-NER (Transformer-based Named Entity Recognition), a Python library for NER LM finetuning. In addition to its practical utility, T-NER facilitates the study and investigation of the cross-domain and cross-lingual generalization ability of LMs finetuned on NER. Our library also provides a web app where users can get model predictions interactively for arbitrary text, which facilitates qualitative model evaluation for non-expert programmers. We show the potential of the library by compiling nine public NER datasets into a unified format and evaluating the cross-domain and cross-lingual performance across the datasets. The results from our initial experiments show that in-domain performance is generally competitive across datasets. However, cross-domain generalization is challenging even with a large pretrained LM, which has nevertheless capacity to learn domain-specific features if fine-tuned on a combined dataset. To facilitate future research, we also release all our LM checkpoints via the Hugging Face model hub. | ['Jose Camacho-Collados', 'Asahi Ushio'] | 2022-09-09 | t-ner-an-all-round-python-library-for | https://aclanthology.org/2021.eacl-demos.7 | https://aclanthology.org/2021.eacl-demos.7.pdf | eacl-2021-2 | ['face-model'] | ['computer-vision'] | [-3.51087362e-01 1.09058106e-02 -1.40580878e-01 -5.17882645e-01
-1.36006749e+00 -1.26168442e+00 3.66353571e-01 3.13820802e-02
-7.29510248e-01 7.56904662e-01 4.53208864e-01 -4.90182549e-01
1.76808108e-02 -4.77384597e-01 -5.30881584e-01 -1.84069723e-02
2.35609010e-01 5.68575442e-01 1.42197143e-02 -2.58956611e-01
-1.42309174e-01 5.86415291e-01 -9.77077425e-01 5.68754792e-01
8.87254953e-01 4.24420536e-01 8.68554041e-02 5.56206524e-01
-5.16365528e-01 5.77144980e-01 -5.40347517e-01 -6.15025818e-01
1.39451012e-01 7.05960207e-03 -1.15245044e+00 -7.51977921e-01
6.16174340e-01 1.61746591e-01 1.66962203e-03 6.61406815e-01
8.55727613e-01 3.35926414e-01 4.24151093e-01 -9.40647244e-01
-9.93682504e-01 9.94687140e-01 7.55124614e-02 1.61873713e-01
3.29966158e-01 1.33214980e-01 8.96111369e-01 -9.61527526e-01
8.99701476e-01 1.16050684e+00 1.26880705e+00 8.55879009e-01
-1.27146649e+00 -8.94993782e-01 -1.79327056e-01 -2.60563135e-01
-1.48825634e+00 -7.36669242e-01 1.47832751e-01 -2.55188555e-01
1.52046156e+00 7.28128329e-02 -1.78624526e-01 1.21926033e+00
-7.17499331e-02 6.90028429e-01 1.05814838e+00 -3.79055321e-01
-6.63725613e-03 3.93449277e-01 3.07613574e-02 4.12246972e-01
-9.35349986e-02 -6.60571158e-02 -4.23102707e-01 -2.12328345e-01
3.78607035e-01 -4.72796768e-01 -6.22879937e-02 1.08079001e-01
-1.17889321e+00 4.80897337e-01 3.15877110e-01 5.98032534e-01
-1.43359497e-01 -2.45261788e-01 5.53359866e-01 4.22196805e-01
4.08146709e-01 9.24598455e-01 -1.39130747e+00 -4.91100550e-01
-9.39858198e-01 -1.56274021e-01 1.21313131e+00 1.07736099e+00
5.21263063e-01 6.34579584e-02 -2.15867966e-01 1.11349118e+00
-5.56932129e-02 3.65404218e-01 7.64707863e-01 -6.91076398e-01
6.97903335e-01 6.56526506e-01 1.09304711e-01 -4.13790286e-01
-6.50844276e-01 -2.01304495e-01 -5.53980529e-01 -2.44960800e-01
5.42453885e-01 -6.43154263e-01 -8.62784624e-01 1.95933330e+00
1.87435746e-01 4.66747284e-02 4.02638644e-01 3.25603604e-01
1.32163370e+00 5.22299170e-01 9.22814250e-01 2.57242471e-01
1.24945521e+00 -7.25969732e-01 -3.58009905e-01 -2.54721344e-01
1.49096572e+00 -7.87705123e-01 1.35716474e+00 6.08680807e-02
-7.21835673e-01 -5.49063981e-01 -3.33160996e-01 -4.64491367e-01
-1.08949840e+00 3.34181368e-01 4.47429717e-01 7.91070879e-01
-1.20256591e+00 7.05681562e-01 -9.12356794e-01 -9.91213322e-01
1.74330458e-01 5.04505455e-01 -9.12803531e-01 -8.60936865e-02
-1.45423508e+00 1.26069498e+00 7.43332386e-01 -2.01985970e-01
-5.45691073e-01 -1.24917150e+00 -9.63310838e-01 1.29052684e-01
-1.10813245e-01 -4.57497567e-01 1.48162580e+00 -5.69074035e-01
-1.27859879e+00 1.05991578e+00 -2.01696545e-01 -1.16653860e-01
3.19465905e-01 -2.34512389e-01 -7.74020851e-01 -5.05425572e-01
2.71646231e-01 9.45660591e-01 -1.51063904e-01 -8.56166899e-01
-6.40728474e-01 -1.75123692e-01 -9.94573683e-02 2.09491104e-01
-6.60888910e-01 4.39574569e-01 -2.14750841e-01 -5.95112801e-01
-4.27691519e-01 -7.14274645e-01 -1.58821493e-01 -7.03716516e-01
-3.65121990e-01 -5.88824570e-01 6.89348698e-01 -8.90380502e-01
1.36998022e+00 -1.95907784e+00 -3.76238108e-01 -2.18471065e-01
-1.80511490e-01 5.70638478e-01 -4.67754722e-01 6.92233145e-01
-3.81357551e-01 8.06804180e-01 -2.06223037e-02 -3.91093671e-01
1.95265338e-01 3.22896503e-02 -8.55339244e-02 -2.82891449e-02
3.33200246e-01 1.08830333e+00 -8.65979373e-01 -4.03788120e-01
4.95196804e-02 7.10195541e-01 -3.38847816e-01 1.94493055e-01
6.16794713e-02 5.98430216e-01 -3.59358311e-01 5.33801377e-01
5.42360485e-01 -1.52019560e-01 2.90624648e-01 -1.62882000e-01
-4.06722963e-01 6.61147952e-01 -1.11050272e+00 1.75725639e+00
-7.62915790e-01 5.82289875e-01 -7.04568624e-02 -4.45544571e-01
8.34610939e-01 5.49848497e-01 8.11328888e-02 -5.33584058e-01
-2.18558788e-01 1.86670318e-01 -4.62643296e-01 -3.56895238e-01
6.49218619e-01 -2.45023727e-01 -4.87943202e-01 4.87118542e-01
4.13536936e-01 1.88708872e-01 2.82673180e-01 3.58356759e-02
1.17027831e+00 2.22817704e-01 4.28347409e-01 -3.48588228e-01
2.71508902e-01 4.07617122e-01 6.92687988e-01 7.27883101e-01
-1.51329905e-01 2.46446356e-01 9.37852710e-02 -2.55575001e-01
-9.46743071e-01 -9.48881209e-01 -4.34655339e-01 1.82228529e+00
-5.71974277e-01 -5.45710146e-01 -8.45674574e-01 -9.99233484e-01
-1.03336930e-01 1.18283856e+00 -4.21542823e-01 -1.61869396e-02
-6.30070269e-01 -6.20855093e-01 1.43393540e+00 9.06643271e-01
4.94502991e-01 -1.29635859e+00 1.64024755e-01 2.05676675e-01
-3.55548829e-01 -1.17313886e+00 -5.57315588e-01 5.87665260e-01
-8.57276797e-01 -6.55417919e-01 -5.31117737e-01 -1.14556038e+00
4.13203239e-01 -3.41808707e-01 1.30989802e+00 -2.65651047e-01
-8.94675404e-02 5.81045687e-01 -2.76403576e-01 -3.90888602e-01
-6.44293368e-01 7.81109869e-01 1.96477711e-01 -8.38940680e-01
7.50006557e-01 -3.22240084e-01 -1.09929137e-01 4.24985021e-01
-8.30015123e-01 -2.82022536e-01 5.63172340e-01 7.45878816e-01
4.83257979e-01 4.37244251e-02 5.86792171e-01 -1.14811027e+00
8.24401677e-01 -5.95077336e-01 -3.30683440e-01 6.23488069e-01
-4.54688132e-01 1.94057211e-01 7.95372248e-01 -4.46844816e-01
-1.47009170e+00 1.51196584e-01 -3.73092055e-01 1.07473377e-02
-7.53468275e-01 7.56713748e-01 -4.58422393e-01 4.26035263e-02
9.44196224e-01 -3.79562378e-02 -8.27925503e-01 -1.05839920e+00
6.38522327e-01 8.69696796e-01 5.67080796e-01 -7.86740243e-01
5.34467638e-01 -1.76036045e-01 -5.53899527e-01 -8.19733262e-01
-9.55505610e-01 -5.29098868e-01 -1.01653588e+00 3.14564556e-01
8.46871972e-01 -1.11245978e+00 -5.23802340e-01 1.34627938e-01
-1.02840877e+00 -8.72476518e-01 -1.62801757e-01 3.63369673e-01
-1.30830199e-01 -4.86467779e-02 -8.68992209e-01 -3.64267468e-01
-4.62019742e-01 -8.12886178e-01 9.33045268e-01 4.98747349e-01
-5.30390441e-01 -1.58698809e+00 4.11602914e-01 3.13384742e-01
4.96271521e-01 -2.41840426e-02 9.71492112e-01 -1.50420773e+00
2.38820940e-01 -1.45837471e-01 -2.28002846e-01 3.85344118e-01
1.25159947e-02 -1.42651752e-01 -1.17542589e+00 -1.95655823e-01
-5.82060158e-01 -5.14013529e-01 4.87817228e-01 4.43119789e-03
8.20068777e-01 -7.21786693e-02 -5.62215567e-01 7.26257563e-01
1.24636972e+00 -1.09813549e-01 4.45214123e-01 6.75901175e-01
7.98791409e-01 4.25619394e-01 4.54057276e-01 9.47476178e-02
6.84844732e-01 3.00812364e-01 -2.40142062e-01 -4.12027121e-01
-9.16015506e-02 -5.89832604e-01 5.13204098e-01 6.01697981e-01
2.06307545e-01 -4.76824999e-01 -1.46914017e+00 6.33190572e-01
-1.42977071e+00 -9.18176115e-01 -9.91530344e-02 2.04559612e+00
1.08436537e+00 -1.77844107e-01 -1.02531217e-01 -6.34214401e-01
7.70813346e-01 -4.03020352e-01 -5.91452718e-01 -6.98942244e-01
-2.70375282e-01 4.73037243e-01 6.84470296e-01 2.31455490e-01
-1.29488420e+00 1.64433801e+00 7.10018301e+00 9.51800525e-01
-1.16225874e+00 2.55131960e-01 4.39195693e-01 3.31146330e-01
-9.57839787e-02 4.76078968e-03 -1.44708991e+00 1.06137015e-01
1.55903423e+00 -3.59024942e-01 2.81223267e-01 1.00297952e+00
2.22700477e-01 4.76945817e-01 -1.18859708e+00 5.71311116e-01
-3.90874714e-01 -1.07871401e+00 2.44664401e-02 -1.65658757e-01
6.88463032e-01 7.27079570e-01 -1.54046431e-01 1.09073615e+00
8.08655739e-01 -1.23201609e+00 3.02642345e-01 3.70570630e-01
1.18591845e+00 -7.54480064e-01 7.93518424e-01 4.90316570e-01
-1.29383302e+00 5.36954105e-02 -3.99748802e-01 2.70124584e-01
1.61650598e-01 1.93355694e-01 -1.30745840e+00 4.88314718e-01
8.96338046e-01 4.89901334e-01 -7.72089183e-01 9.98690248e-01
-1.97627425e-01 7.88015544e-01 -5.21715939e-01 2.30891854e-01
-3.31362560e-02 3.49155337e-01 1.95957005e-01 1.97561765e+00
2.94336408e-01 1.59363464e-01 1.36743516e-01 6.83116794e-01
-4.76677448e-01 4.73023832e-01 -5.10064185e-01 -5.16723573e-01
6.64645374e-01 1.45863652e+00 -4.89106268e-01 -3.10258418e-01
-3.85756731e-01 1.11871362e+00 7.92959213e-01 3.92881840e-01
-6.12240970e-01 -7.34152794e-01 8.08742702e-01 -1.50351420e-01
2.66034901e-01 -1.28873199e-01 -3.06581885e-01 -1.29543459e+00
-3.94178361e-01 -9.03934896e-01 8.70928586e-01 -8.49553049e-01
-1.68388236e+00 6.96942866e-01 -1.41417250e-01 -8.11093032e-01
-1.41464651e-01 -8.16526532e-01 -6.04420602e-01 1.07454550e+00
-1.29730248e+00 -1.51200116e+00 -1.79480985e-02 6.42347455e-01
3.12902391e-01 1.23554304e-01 1.41082776e+00 4.56654638e-01
-7.20570445e-01 1.12595582e+00 4.05501276e-01 7.87212729e-01
1.34699631e+00 -1.33328974e+00 7.88970768e-01 5.65597951e-01
-2.46903878e-02 1.15685797e+00 3.34722698e-01 -8.79532993e-01
-9.26001728e-01 -1.40576780e+00 1.45691741e+00 -1.12625039e+00
7.25935459e-01 -3.55423927e-01 -9.23287570e-01 1.29370975e+00
1.54905960e-01 -6.53090179e-02 1.25100684e+00 6.00995243e-01
-5.29612660e-01 2.15832353e-01 -1.36099339e+00 5.24117112e-01
1.10001862e+00 -6.88818872e-01 -8.35734367e-01 2.80639827e-01
7.57398546e-01 -4.50895816e-01 -1.58245039e+00 3.66212040e-01
4.27738607e-01 -5.72304308e-01 7.27261901e-01 -1.33972716e+00
2.40178732e-03 -1.66394904e-01 -9.44639221e-02 -1.43812394e+00
-4.11527693e-01 -3.68974119e-01 5.21666288e-01 2.01265454e+00
9.99728143e-01 -5.56120276e-01 5.16483188e-01 1.10306990e+00
-2.45148465e-01 -1.78161964e-01 -6.61418557e-01 -8.55324805e-01
6.88831151e-01 -7.98592627e-01 6.68720424e-01 1.55995381e+00
3.29170644e-01 3.73538792e-01 1.24066032e-01 3.85759413e-01
9.82395485e-02 -3.74734700e-01 6.71844006e-01 -1.19580114e+00
-6.05027601e-02 -3.01638782e-01 -5.23824953e-02 -8.54563594e-01
4.80085552e-01 -1.48375738e+00 3.67818698e-02 -1.56521654e+00
1.77214414e-01 -6.92799926e-01 -4.68130678e-01 1.23786473e+00
-1.29876539e-01 7.02077150e-02 1.10808581e-01 1.87095940e-01
-7.49899626e-01 1.45807952e-01 7.62421787e-01 2.67858267e-01
-4.92307782e-01 -3.47481161e-01 -9.13390994e-01 6.46377623e-01
8.17607760e-01 -8.80344570e-01 3.25443819e-02 -8.23548138e-01
7.02914670e-02 -2.34441400e-01 -7.89179131e-02 -9.31663275e-01
2.75488973e-01 -9.48821008e-02 6.63604319e-01 -2.98776299e-01
-4.25890349e-02 -4.44982648e-01 1.68628395e-01 -9.93557498e-02
-6.56880081e-01 1.88243702e-01 8.45798850e-01 3.38808075e-02
3.34782004e-02 -1.78532273e-01 7.08801508e-01 -2.19767898e-01
-9.96371746e-01 1.37647271e-01 -3.02818686e-01 6.30648553e-01
6.39358759e-01 -2.11011060e-02 -4.73083526e-01 -1.11558795e-01
-8.95856321e-01 3.62161011e-01 5.90895176e-01 4.63892251e-01
-1.09422460e-01 -1.15572286e+00 -6.69844806e-01 3.43576819e-02
3.53373706e-01 -2.63563722e-01 2.46783540e-01 5.81748247e-01
-2.33954743e-01 6.83521807e-01 -2.64314525e-02 -2.33149290e-01
-1.08877826e+00 2.88289219e-01 5.21239519e-01 -6.11420453e-01
-2.05714047e-01 1.01874375e+00 7.85120353e-02 -1.54871273e+00
2.62651946e-02 -2.39484638e-01 -2.28100225e-01 -8.94744620e-02
6.67692900e-01 5.49597919e-01 2.56392568e-01 -6.78433597e-01
-4.85217690e-01 2.42857695e-01 -5.16362041e-02 -4.90807779e-02
1.48099601e+00 -9.74514335e-02 3.61463949e-02 4.01612848e-01
1.25749993e+00 2.66455382e-01 -7.70707905e-01 -7.50940889e-02
4.71940279e-01 3.22393268e-01 -2.97637861e-02 -1.39382982e+00
-5.70807993e-01 6.56995535e-01 4.98036981e-01 -7.45017827e-02
8.83539617e-01 2.62792055e-02 8.15332294e-01 8.18072796e-01
2.18749270e-01 -1.08924592e+00 -7.07409263e-01 9.70866740e-01
4.31741923e-01 -1.19799995e+00 -4.64950353e-01 -4.40223925e-02
-8.14336181e-01 1.12952590e+00 7.81074107e-01 3.41602743e-01
6.95761621e-01 4.41519797e-01 5.53988934e-01 1.21788852e-01
-7.05880404e-01 -1.36235848e-01 3.32221597e-01 8.63369226e-01
1.08899891e+00 1.54730156e-02 -5.80205955e-02 1.04508471e+00
-5.24227679e-01 1.01144426e-01 2.11127833e-01 6.74430490e-01
-1.56851634e-01 -1.45144737e+00 -3.04840744e-01 1.92083627e-01
-7.02312827e-01 -6.81483150e-01 -5.42523682e-01 9.37869370e-01
1.42777026e-01 1.07538819e+00 -1.58155486e-01 -4.11854148e-01
7.34031022e-01 6.70588791e-01 -1.10995835e-02 -9.38590944e-01
-1.07222700e+00 -4.56199318e-01 4.94426012e-01 -4.72360104e-01
1.00298122e-01 -5.41890264e-01 -1.50855076e+00 -4.14516479e-01
-2.56663799e-01 3.86815608e-01 7.84432471e-01 8.83931518e-01
7.99246371e-01 4.78254892e-02 8.39478374e-02 -6.09723032e-01
-5.08360982e-01 -1.13004935e+00 -4.44309413e-01 3.80482078e-01
-2.59048492e-01 -2.66400009e-01 -4.31714803e-01 1.97705537e-01] | [10.063929557800293, 9.659943580627441] |
692f717b-5631-4405-8292-6878e05358c3 | srlgrn-semantic-role-labeling-graph-reasoning | 2010.03604 | null | https://arxiv.org/abs/2010.03604v2 | https://arxiv.org/pdf/2010.03604v2.pdf | SRLGRN: Semantic Role Labeling Graph Reasoning Network | This work deals with the challenge of learning and reasoning over multi-hop question answering (QA). We propose a graph reasoning network based on the semantic structure of the sentences to learn cross paragraph reasoning paths and find the supporting facts and the answer jointly. The proposed graph is a heterogeneous document-level graph that contains nodes of type sentence (question, title, and other sentences), and semantic role labeling sub-graphs per sentence that contain arguments as nodes and predicates as edges. Incorporating the argument types, the argument phrases, and the semantics of the edges originated from SRL predicates into the graph encoder helps in finding and also the explainability of the reasoning paths. Our proposed approach shows competitive performance on the HotpotQA distractor setting benchmark compared to the recent state-of-the-art models. | ['Parisa Kordjamshidi', 'Chen Zheng'] | 2020-10-07 | null | https://aclanthology.org/2020.emnlp-main.714 | https://aclanthology.org/2020.emnlp-main.714.pdf | emnlp-2020-11 | ['multi-hop-question-answering'] | ['knowledge-base'] | [ 2.32581180e-02 9.11207318e-01 -2.91974485e-01 -5.06945312e-01
-8.26663911e-01 -7.74096668e-01 2.86243230e-01 8.81115377e-01
6.79417849e-02 7.95831382e-01 6.43046260e-01 -5.33216238e-01
-5.77363968e-01 -1.11674154e+00 -9.64213252e-01 -6.32507205e-02
-1.27860963e-01 9.20648813e-01 9.93075430e-01 -6.55338407e-01
3.96475457e-02 -1.14205018e-01 -9.34639513e-01 9.10705924e-01
8.73130143e-01 8.63558054e-01 -1.55114681e-01 6.60726726e-01
-7.64303148e-01 1.72989535e+00 -6.09199643e-01 -1.08201873e+00
-2.17352882e-01 -5.01428843e-01 -1.60074270e+00 -2.07486004e-01
6.11949325e-01 4.61901799e-02 -5.27291477e-01 9.58841801e-01
-2.05446407e-02 1.53864175e-01 3.54749054e-01 -1.32066095e+00
-9.28714514e-01 9.91423428e-01 -2.40317971e-01 5.27018845e-01
7.38169253e-01 -2.87204504e-01 1.93845034e+00 -1.30126715e-01
8.61297965e-01 1.50433123e+00 2.97903746e-01 5.97338617e-01
-5.07791638e-01 -3.43785211e-02 2.44209334e-01 9.21450794e-01
-6.69852674e-01 1.70641631e-01 7.12768435e-01 -5.16096503e-02
1.27947259e+00 1.74103826e-01 2.79102981e-01 8.51449072e-01
1.37479812e-01 8.72867703e-01 4.68218118e-01 -3.49500507e-01
8.17993283e-02 2.28451509e-02 9.90583539e-01 1.45361888e+00
1.09160267e-01 -7.83959687e-01 -4.22535211e-01 -4.07734156e-01
3.96271795e-02 -4.79078591e-01 -2.69214779e-01 -2.92220950e-01
-7.68376827e-01 9.55796778e-01 7.54718006e-01 1.50564045e-01
-3.11916947e-01 3.38704675e-01 5.15737653e-01 4.11792994e-01
1.64142966e-01 5.02133787e-01 -8.08811605e-01 2.67259151e-01
-2.33991459e-01 3.25990468e-01 1.14944124e+00 1.01525629e+00
4.42883372e-01 -6.15454197e-01 -6.93948746e-01 5.99645555e-01
4.37819690e-01 6.52219951e-02 -1.25228334e-03 -1.02253866e+00
1.18820381e+00 1.23413026e+00 -8.48119557e-02 -1.01802754e+00
-4.38590348e-01 -3.57346356e-01 -2.17767462e-01 -8.23257983e-01
3.21213275e-01 -1.06890902e-01 -6.06464148e-01 1.69608510e+00
6.30936682e-01 2.00052515e-01 6.05931222e-01 6.63610876e-01
1.48183823e+00 7.33008265e-01 3.37932587e-01 1.36386991e-01
1.83290076e+00 -1.53976238e+00 -6.35983527e-01 -1.87462747e-01
7.13847756e-01 -1.48330241e-01 8.69000077e-01 -3.80594343e-01
-9.91898358e-01 -2.26889580e-01 -8.24677050e-01 -6.76045716e-01
-3.51762086e-01 -2.50351042e-01 4.43873644e-01 -7.07507432e-02
-8.44945014e-01 2.48260140e-01 -1.71566039e-01 -2.08996832e-01
3.06177884e-01 -3.03129889e-02 -3.13591629e-01 -6.01901233e-01
-1.82006645e+00 8.62911165e-01 5.14059484e-01 -2.17718378e-01
-7.73403943e-01 -8.35097134e-01 -1.19265544e+00 5.32519221e-01
8.21818411e-01 -1.34922624e+00 1.14083755e+00 -2.44579062e-01
-9.15759504e-01 7.52053320e-01 -3.91669720e-01 -8.32047045e-01
-3.11335195e-02 -7.61969164e-02 -4.81333315e-01 7.55445063e-01
5.82044840e-01 5.15349567e-01 3.64417642e-01 -9.35648203e-01
-7.22702801e-01 -4.74242330e-01 9.27688599e-01 3.93454015e-01
2.15319455e-01 -1.91028923e-01 -6.32755160e-01 5.58836311e-02
-3.55284177e-02 -4.04015422e-01 9.28988829e-02 -3.06334257e-01
-7.13293314e-01 -9.48947668e-01 7.67897964e-01 -1.02184629e+00
1.04019368e+00 -1.86587489e+00 2.52045810e-01 1.76955003e-03
3.09863329e-01 -1.89986423e-01 -1.85839519e-01 6.36943698e-01
1.19941525e-01 -4.43388782e-02 -2.17948720e-01 2.07267761e-01
1.65685341e-01 5.84237814e-01 -4.14626122e-01 -1.55395254e-01
1.92311317e-01 1.15919530e+00 -1.18051720e+00 -6.56705618e-01
-4.69164193e-01 6.88053444e-02 -4.05233979e-01 2.60839075e-01
-9.87053216e-01 8.56613442e-02 -7.68809140e-01 4.30086970e-01
2.30703145e-01 -7.06282258e-01 2.44410530e-01 -3.53637576e-01
6.61487937e-01 8.69264781e-01 -6.49237752e-01 1.56453168e+00
-3.09108227e-01 3.06291968e-01 -7.03575015e-02 -8.26686263e-01
6.70912862e-01 3.15762520e-01 7.34817609e-02 -6.74444735e-01
-1.70878321e-01 -4.35053138e-03 1.09507456e-01 -8.67861032e-01
1.80993780e-01 1.04259267e-01 -4.09294851e-02 2.06087872e-01
4.03328329e-01 -6.12596273e-02 6.86689913e-01 9.52893078e-01
1.43317699e+00 1.64876450e-02 6.27874658e-02 -1.03825063e-01
1.15572608e+00 4.33651030e-01 1.08341597e-01 5.81725061e-01
1.61499217e-01 3.93325882e-03 1.17162251e+00 -2.05516949e-01
-6.28696561e-01 -9.62313056e-01 2.92907476e-01 9.94267881e-01
3.60059351e-01 -6.75581634e-01 -6.18447483e-01 -1.39557362e+00
2.71338969e-01 9.86334801e-01 -6.39577389e-01 -7.04171360e-02
-7.30257392e-01 -1.76273599e-01 6.21503472e-01 5.82057595e-01
8.07171941e-01 -1.01695299e+00 5.46029992e-02 8.40847716e-02
-6.63098395e-01 -1.63392019e+00 -1.27655476e-01 -1.56458288e-01
-6.28207266e-01 -1.74201000e+00 -5.79114025e-03 -9.53159690e-01
5.44142306e-01 -3.84761840e-02 1.58603358e+00 3.95991087e-01
1.30919605e-01 5.22175848e-01 -4.01484162e-01 -7.53291920e-02
-3.79739940e-01 2.20909398e-02 -8.55110466e-01 -1.19186528e-01
1.87267274e-01 -2.33188435e-01 -5.50477564e-01 -3.02463230e-02
-6.16616726e-01 -4.77899751e-03 1.46101698e-01 6.55413866e-01
6.50939167e-01 4.56792861e-01 6.10983849e-01 -1.44438541e+00
8.60308826e-01 -1.01753390e+00 -4.33379740e-01 9.81761575e-01
-3.02199095e-01 5.34486473e-01 7.67067790e-01 4.63959306e-01
-1.30392742e+00 -7.24683106e-01 -2.86842793e-01 1.78664953e-01
4.35525663e-02 6.04476988e-01 -2.77857661e-01 3.12302053e-01
4.54529881e-01 4.48271958e-03 -2.71782279e-01 -3.15356165e-01
7.65963137e-01 1.02624632e-01 3.81365359e-01 -7.94514477e-01
5.52162349e-01 3.59477937e-01 4.02217597e-01 -2.40899682e-01
-1.62682819e+00 -7.29148090e-01 -2.61750400e-01 1.13375276e-01
1.09678936e+00 -7.03163564e-01 -8.91808808e-01 -3.49810600e-01
-1.39325392e+00 7.29522482e-02 -3.53196025e-01 -1.14586815e-01
-3.71454507e-01 4.03785139e-01 -9.62439954e-01 -3.41701061e-01
-6.60071015e-01 -7.92871535e-01 9.87998486e-01 2.27356046e-01
5.61611615e-02 -1.35726380e+00 7.79663352e-03 1.29851413e+00
3.57627645e-02 1.06195189e-01 1.89560544e+00 -1.00102127e+00
-7.58656621e-01 -2.71402462e-03 -5.44985235e-01 9.92084444e-02
-7.24292323e-02 -4.84117001e-01 -4.22554523e-01 2.42605537e-01
-2.15947747e-01 -3.96986187e-01 7.89273560e-01 7.83440471e-02
9.70729470e-01 -5.80683947e-01 -3.40392232e-01 -1.22917973e-01
1.36623204e+00 -1.17029570e-01 3.61809134e-01 3.05138856e-01
7.20831215e-01 8.62649143e-01 3.71771365e-01 -2.68586487e-01
1.09342861e+00 3.56368095e-01 6.36552215e-01 2.05800146e-01
-4.43489820e-01 -7.06000507e-01 3.71191837e-02 5.13414145e-01
3.55584711e-01 -7.06766963e-01 -8.39061022e-01 7.46799946e-01
-1.91496253e+00 -1.06237769e+00 -7.34244466e-01 1.26133001e+00
5.80619216e-01 7.52761737e-02 -1.30328164e-01 -1.08460970e-02
5.77870786e-01 4.14018273e-01 -4.12140220e-01 -4.23194647e-01
-2.21163318e-01 1.36308268e-01 1.67345986e-01 8.56522739e-01
-6.90870225e-01 8.82574379e-01 5.42048407e+00 4.27974194e-01
5.28309904e-02 3.61998379e-01 2.49426574e-01 5.20046473e-01
-8.37039888e-01 2.06964657e-01 -8.59393060e-01 3.60426567e-02
7.28241503e-01 -6.60816729e-02 1.42711222e-01 6.67160869e-01
-3.67867559e-01 -4.24360707e-02 -1.01135254e+00 2.06169724e-01
3.82472277e-01 -1.66197526e+00 4.88825768e-01 -6.65901303e-01
4.33645934e-01 -8.65066275e-02 -4.13845956e-01 5.84333539e-01
3.91066045e-01 -6.25583410e-01 4.00899023e-01 2.53939629e-01
8.55179429e-02 -5.49093366e-01 1.10402048e+00 4.15164053e-01
-1.27933955e+00 -3.26313555e-01 -2.83247501e-01 2.42321432e-01
1.78712130e-01 3.34059834e-01 -1.05546403e+00 1.31414008e+00
6.64934039e-01 7.54745841e-01 -5.83792865e-01 6.03651285e-01
-1.10418880e+00 8.17786515e-01 1.77636236e-01 -3.36986125e-01
6.08595729e-01 -2.06203107e-02 4.81750965e-01 9.48270798e-01
-1.58064127e-01 4.22616571e-01 -7.50764683e-02 6.98375762e-01
-5.95835209e-01 1.66787133e-01 -3.73045832e-01 -1.17975131e-01
2.57098824e-01 9.71551597e-01 -3.91332984e-01 -4.16915566e-01
-7.20592439e-01 8.74393225e-01 8.01825643e-01 2.32195497e-01
-6.36821210e-01 -2.41722763e-01 3.74340892e-01 1.03032880e-01
2.03745663e-01 1.00771315e-01 1.03234187e-01 -1.00529933e+00
2.86366701e-01 -7.49882340e-01 1.43282831e+00 -1.14260828e+00
-1.57639682e+00 6.11964405e-01 1.09674104e-01 -3.46713036e-01
1.59042119e-03 -5.35658300e-01 -6.06564105e-01 8.19001555e-01
-1.95142388e+00 -1.12997997e+00 -2.57445157e-01 8.49859536e-01
6.63560688e-01 -1.16533056e-01 7.27434516e-01 1.87950544e-02
-1.28658831e-01 1.90293789e-01 -5.97724378e-01 4.02265608e-01
5.28785698e-02 -1.35078812e+00 3.37770194e-01 5.94428420e-01
4.77999359e-01 3.94661784e-01 5.14643073e-01 -6.83176041e-01
-1.22456157e+00 -9.02822256e-01 1.39150620e+00 -7.46112347e-01
9.98934329e-01 -1.80491544e-02 -1.12376535e+00 8.53012741e-01
6.18522525e-01 -1.67323843e-01 5.66176236e-01 2.38672778e-01
-6.75466418e-01 4.67549078e-02 -1.15725601e+00 2.25676447e-01
1.11779225e+00 -4.86107141e-01 -1.32676113e+00 8.47564340e-01
1.47117424e+00 -6.14535630e-01 -6.54954314e-01 4.41348940e-01
-1.94423586e-01 -5.45396328e-01 1.02337837e+00 -1.31665134e+00
7.82176614e-01 -2.61805356e-01 -1.43773094e-01 -1.04454434e+00
-2.45994702e-01 -3.08206175e-02 -5.24259984e-01 1.06406057e+00
1.07584465e+00 -2.95899063e-01 7.36088455e-01 2.60949910e-01
-2.48795643e-01 -8.70760679e-01 -8.96192968e-01 -2.91648775e-01
-2.95978487e-01 -7.03751743e-02 6.03699982e-01 6.88235879e-01
8.01535323e-02 1.19480419e+00 1.31016284e-01 6.60620511e-01
6.71676934e-01 5.23726225e-01 1.27380177e-01 -1.21924257e+00
-4.39675778e-01 -4.77564186e-02 -2.40539953e-01 -9.97698545e-01
6.89943314e-01 -1.32552969e+00 -3.80131185e-01 -2.50893593e+00
2.43265718e-01 -1.68673247e-01 -5.89784645e-02 5.86629629e-01
-4.70454216e-01 -5.59043646e-01 1.45327961e-02 -1.57418802e-01
-9.48591828e-01 4.36962873e-01 1.56519568e+00 -5.38551629e-01
3.41683328e-01 -5.72223850e-02 -7.70210266e-01 4.21579391e-01
5.49385846e-01 -5.05834460e-01 -1.02795577e+00 -8.88155162e-01
6.33170962e-01 5.77156603e-01 3.61637026e-01 -4.43495005e-01
4.72941041e-01 1.22854680e-01 -1.88794255e-01 -3.57344955e-01
2.66111672e-01 -8.42881680e-01 -4.61218387e-01 3.53302270e-01
-8.40035260e-01 2.54961282e-01 -8.72346610e-02 9.65224683e-01
-6.24209940e-01 -4.01733011e-01 1.74153745e-01 -4.03402835e-01
-7.21201301e-01 2.98628241e-01 3.96106452e-01 8.11010361e-01
7.66926825e-01 4.87513095e-01 -1.12366557e+00 -5.31062663e-01
-7.77229011e-01 8.72428358e-01 -4.05426413e-01 5.93156815e-01
6.88518405e-01 -8.56113374e-01 -8.18290830e-01 -4.15878862e-01
1.87762350e-01 2.78160572e-02 4.52061415e-01 6.46847785e-01
-5.46929717e-01 6.09525561e-01 8.93645212e-02 -4.95976619e-02
-1.25960004e+00 5.38541794e-01 4.43548650e-01 -8.78480136e-01
-6.46285236e-01 1.05777454e+00 -2.32097998e-01 -4.14684147e-01
1.59641713e-01 -1.35515973e-01 -8.17654729e-01 -1.69642359e-01
3.16868335e-01 2.23271072e-01 1.46216333e-01 -3.51096720e-01
-3.59999388e-01 2.81359226e-01 -5.47743887e-02 1.16514757e-01
1.08029854e+00 -3.48055139e-02 -7.28393674e-01 8.13595802e-02
1.03741682e+00 1.30804675e-02 -3.41623336e-01 -4.17629182e-01
3.08500081e-01 -3.80839333e-02 -3.92246306e-01 -9.93418217e-01
-7.39848077e-01 6.91655397e-01 -3.04029167e-01 4.42776203e-01
6.49089217e-01 8.78603220e-01 1.02021718e+00 4.41966027e-01
2.75307357e-01 -6.48023248e-01 3.02154034e-01 7.71133065e-01
9.27480161e-01 -9.60840464e-01 -3.29804301e-01 -1.05709577e+00
-7.95576751e-01 1.00230443e+00 7.97297895e-01 1.32936835e-01
3.88929725e-01 -2.51982599e-01 -1.67458773e-01 -9.74397242e-01
-1.03745031e+00 -3.90802026e-01 5.41829050e-01 3.30983728e-01
2.38423645e-01 -7.64446110e-02 -5.30452967e-01 5.90690553e-01
-2.10573509e-01 -4.96647030e-01 1.64905101e-01 6.36710167e-01
-4.36673015e-01 -9.11619842e-01 2.50603527e-01 2.57513583e-01
-4.44182307e-01 -2.81615973e-01 -6.17031813e-01 7.19311178e-01
1.37387246e-01 1.31415880e+00 -1.12072341e-01 -6.46706074e-02
6.11039460e-01 4.60728019e-01 4.76131499e-01 -7.66697943e-01
-7.15678930e-01 -1.05554855e+00 8.55289996e-01 -4.28444088e-01
-3.78353715e-01 -1.48627963e-02 -1.88608432e+00 1.07322007e-01
2.01475620e-03 8.56175005e-01 3.27515036e-01 1.25378227e+00
4.60167259e-01 8.25713754e-01 2.43055508e-01 8.77790749e-01
-4.16636944e-01 -7.16065943e-01 -2.96473354e-01 6.66929722e-01
2.27263704e-01 -3.20889473e-01 -2.15981096e-01 -3.07886064e-01] | [10.686893463134766, 7.906129360198975] |
255fc505-0707-4f78-b917-808403bdfa47 | few-shot-human-motion-prediction-for | 2212.11771 | null | https://arxiv.org/abs/2212.11771v2 | https://arxiv.org/pdf/2212.11771v2.pdf | Few-shot human motion prediction for heterogeneous sensors | Human motion prediction is a complex task as it involves forecasting variables over time on a graph of connected sensors. This is especially true in the case of few-shot learning, where we strive to forecast motion sequences for previously unseen actions based on only a few examples. Despite this, almost all related approaches for few-shot motion prediction do not incorporate the underlying graph, while it is a common component in classical motion prediction. Furthermore, state-of-the-art methods for few-shot motion prediction are restricted to motion tasks with a fixed output space meaning these tasks are all limited to the same sensor graph. In this work, we propose to extend recent works on few-shot time-series forecasting with heterogeneous attributes with graph neural networks to introduce the first few-shot motion approach that explicitly incorporates the spatial graph while also generalizing across motion tasks with heterogeneous sensors. In our experiments on motion tasks with heterogeneous sensors, we demonstrate significant performance improvements with lifts from 10.4% up to 39.3% compared to best state-of-the-art models. Moreover, we show that our model can perform on par with the best approach so far when evaluating on tasks with a fixed output space while maintaining two magnitudes fewer parameters. | ['Lars Schmidt-Thieme', 'Lukas Brinkmeyer', 'Rafael Rego Drumond'] | 2022-12-22 | null | null | null | null | ['motion-prediction'] | ['computer-vision'] | [ 3.04123461e-01 1.97416708e-01 -3.96042347e-01 8.02513584e-02
-5.58050215e-01 -2.99530178e-02 6.93291485e-01 2.24611517e-02
-1.48527801e-01 3.78040701e-01 5.59124887e-01 -3.33732069e-02
9.11756456e-02 -7.43838966e-01 -8.22012544e-01 -5.37137568e-01
-3.09907943e-01 2.84152508e-01 9.10864592e-01 -3.66051108e-01
-8.48510414e-02 1.26115128e-01 -1.44999707e+00 1.39952168e-01
2.27466717e-01 9.22540069e-01 3.15634608e-01 1.18139839e+00
3.67893130e-01 1.37211299e+00 -1.38316229e-01 -1.50266021e-01
1.37133285e-01 -2.86930442e-01 -6.13509059e-01 2.72076428e-01
4.64308262e-01 -3.18433046e-01 -7.37578273e-01 4.99298453e-01
5.33594966e-01 7.62385130e-01 6.32324457e-01 -1.42476332e+00
-3.84212941e-01 3.73003900e-01 -3.36574942e-01 2.93784052e-01
4.08624798e-01 3.67608309e-01 9.81829405e-01 -6.99109375e-01
9.44781840e-01 1.16920328e+00 8.60828757e-01 7.19057024e-01
-9.79591787e-01 -1.41897708e-01 4.73587036e-01 6.08109355e-01
-9.88719165e-01 -4.06934142e-01 8.73850584e-01 -6.84788585e-01
1.40413415e+00 2.59440276e-03 7.16334581e-01 1.34591007e+00
4.16919857e-01 8.72636974e-01 3.24231863e-01 -1.32049456e-01
4.60380524e-01 -6.38944685e-01 2.32097089e-01 6.90651894e-01
-1.93688571e-02 -1.33415490e-01 -6.54623210e-01 -2.21371464e-02
2.98981249e-01 4.92673844e-01 -1.07887611e-01 -7.11539805e-01
-1.44347334e+00 8.29720497e-01 3.92736256e-01 2.31300786e-01
-5.63841224e-01 6.70000851e-01 4.67377216e-01 3.26657519e-02
7.13317037e-01 2.52890885e-02 -1.34875521e-01 -3.59823972e-01
-9.54669714e-01 2.55472988e-01 9.25963521e-01 1.10540974e+00
7.30821908e-01 3.70682269e-01 -4.09361631e-01 4.23021764e-01
-2.79351529e-02 4.18899983e-01 6.45418942e-01 -1.16301370e+00
5.79801917e-01 3.33011627e-01 1.11692511e-01 -1.13416386e+00
-6.41242027e-01 -7.63844922e-02 -9.36351180e-01 7.70174786e-02
2.65092283e-01 -5.14739394e-01 -1.10374832e+00 1.64403081e+00
2.56873548e-01 8.90216291e-01 -2.60416530e-02 8.70856941e-01
6.93839431e-01 7.39605844e-01 1.39255464e-01 -1.19289413e-01
9.53268349e-01 -1.42940390e+00 -5.83113432e-01 -5.37961960e-01
1.00151706e+00 -1.40246198e-01 1.04809177e+00 1.58057690e-01
-7.77522564e-01 -7.41852403e-01 -1.10103667e+00 4.20076437e-02
-3.78079355e-01 -2.66575217e-01 4.83471990e-01 3.11384767e-01
-1.26711428e+00 8.97292733e-01 -1.18083966e+00 -8.03732276e-01
2.33938143e-01 2.04816431e-01 -2.70599097e-01 -2.39165187e-01
-1.08709872e+00 8.26180398e-01 2.66952991e-01 -2.92427987e-01
-9.74627197e-01 -5.33413947e-01 -1.16075289e+00 4.48773354e-02
8.32358360e-01 -1.09799516e+00 1.22185838e+00 -5.82361042e-01
-1.47206247e+00 1.57907054e-01 -3.21217716e-01 -7.73135960e-01
4.59415734e-01 -3.37552011e-01 -3.16875368e-01 6.62349612e-02
1.01729177e-01 7.33862281e-01 7.96844304e-01 -7.83167005e-01
-5.93937099e-01 -2.53084570e-01 1.09456182e-01 2.49467134e-01
-3.79340053e-01 -3.97312015e-01 -4.60802972e-01 -6.44846320e-01
-2.55405873e-01 -1.29850340e+00 -6.57526255e-01 -9.70034860e-03
-2.09687129e-01 -1.92158490e-01 1.04895973e+00 -3.91597360e-01
1.21956503e+00 -1.99665570e+00 5.28167367e-01 -3.53118479e-01
2.55515128e-01 5.44542260e-02 -1.03541501e-01 5.14489770e-01
2.92453885e-01 -2.30824098e-01 -1.83565557e-01 -7.53889620e-01
1.02856658e-01 2.32842267e-01 -2.61240572e-01 5.06547689e-01
-3.87678668e-02 1.17419505e+00 -1.06765938e+00 -4.05462146e-01
4.92614001e-01 4.17815626e-01 -4.57901537e-01 6.43792301e-02
-4.56608027e-01 3.76133144e-01 -3.87991428e-01 4.34582055e-01
-8.11485574e-02 -8.23043823e-01 -1.81639828e-02 2.25489214e-01
3.23650628e-01 -1.56501144e-01 -1.04395771e+00 2.13302541e+00
-2.84309566e-01 8.18331003e-01 -5.32095492e-01 -1.00402343e+00
5.85278451e-01 4.35075074e-01 8.77700984e-01 -2.89887398e-01
-4.57442589e-02 -1.85926095e-01 -1.91252410e-01 -6.33869588e-01
7.36304462e-01 -2.13209689e-01 -2.77446210e-01 2.85275698e-01
-1.66298188e-02 6.49377406e-02 1.00546725e-01 2.51497924e-01
1.67635202e+00 3.27279538e-01 3.99619818e-01 2.25361720e-01
1.77095369e-01 2.11875156e-01 4.07200933e-01 8.29439461e-01
-3.29436243e-01 8.82670522e-01 2.47914210e-01 -5.56909323e-01
-1.01753521e+00 -7.06847191e-01 7.41799474e-01 1.42160201e+00
2.76014596e-01 -6.50931299e-01 -5.41334510e-01 -5.10361671e-01
-1.10989504e-01 7.46211052e-01 -7.14755774e-01 -3.75648178e-02
-7.22953916e-01 -5.05894423e-01 2.21866146e-01 8.81599545e-01
2.14143008e-01 -1.10844696e+00 -1.06158280e+00 3.66232425e-01
-1.00591771e-01 -1.48709798e+00 -2.63809472e-01 -1.44876644e-01
-9.03542101e-01 -8.70745003e-01 -1.09500706e+00 -4.14487660e-01
1.15562707e-01 4.27684993e-01 1.08044434e+00 -1.32057145e-01
-1.16081282e-01 8.17751825e-01 -5.24478137e-01 -3.41648251e-01
-1.15171522e-01 2.04738528e-01 -8.19015354e-02 1.22839317e-01
1.64404109e-01 -5.51639020e-01 -7.35895991e-01 7.66132474e-02
-6.49398923e-01 1.63194552e-01 2.73066819e-01 6.64284885e-01
5.43328643e-01 -2.28819489e-01 2.75188178e-01 -9.21629488e-01
1.88769832e-01 -9.28001463e-01 -7.32923746e-02 1.15356036e-01
-2.93846667e-01 -1.92788430e-02 7.25930274e-01 -8.20058763e-01
-7.44275630e-01 4.34371799e-01 2.98654079e-01 -1.05495048e+00
-2.66235054e-01 4.11068708e-01 2.08646417e-01 1.13323860e-01
6.48259223e-01 1.09402165e-01 -1.12904787e-01 -1.14456296e-01
5.56879997e-01 1.30505547e-01 4.20611292e-01 1.03867002e-01
4.60565954e-01 7.72668362e-01 5.57877183e-01 -1.08208990e+00
-6.52169883e-01 -7.81273186e-01 -7.17542648e-01 -5.10241747e-01
1.24156034e+00 -1.02314115e+00 -6.02356195e-01 4.61994082e-01
-9.49902594e-01 -9.40719068e-01 -3.53629082e-01 6.42127335e-01
-1.16313362e+00 4.89967763e-01 -6.71101451e-01 -9.49947894e-01
-1.22167550e-01 -9.68238592e-01 1.43015313e+00 -3.55476022e-01
-4.93017137e-01 -1.17939210e+00 2.93697089e-01 8.90073553e-02
3.30752999e-01 5.62876701e-01 4.89091694e-01 -5.21817982e-01
-6.35408700e-01 -2.05178559e-01 3.68736535e-01 -3.26673090e-01
-1.23603418e-02 -3.70121598e-01 -8.89172733e-01 -2.78974295e-01
-6.04604110e-02 -2.10046411e-01 1.17310250e+00 7.61871338e-01
8.84924829e-01 -3.74059714e-02 -6.65769398e-01 4.17255014e-01
1.24960768e+00 -3.05090612e-03 5.65591455e-01 1.79023147e-01
1.23588836e+00 6.14076614e-01 8.07131469e-01 6.19010091e-01
5.37213147e-01 9.04084086e-01 5.88110328e-01 2.18326315e-01
-2.17154413e-01 -3.40928108e-01 5.00394583e-01 7.42677271e-01
-2.43424103e-01 -7.54620492e-01 -1.03985953e+00 7.89739251e-01
-2.55857658e+00 -1.17261720e+00 -1.87847927e-01 1.99540973e+00
-7.16969892e-02 1.48500577e-01 4.69525218e-01 1.12003312e-01
6.42172635e-01 8.31043065e-01 -7.09258914e-01 7.10727051e-02
1.44078359e-01 -2.12009892e-01 7.30933487e-01 5.30332267e-01
-1.42066336e+00 1.07095456e+00 6.29962969e+00 5.95074952e-01
-1.04694915e+00 2.52759993e-01 4.56974417e-01 -4.73333716e-01
2.20278483e-02 5.41735627e-02 -7.83202767e-01 4.46311414e-01
1.17301393e+00 -4.00752686e-02 1.30009562e-01 1.05540848e+00
1.56869069e-01 -4.62577157e-02 -1.22639203e+00 9.83500004e-01
3.41367126e-01 -1.31439090e+00 -8.43541995e-02 -1.07685849e-03
7.50712633e-01 2.85490125e-01 -1.22867078e-01 3.29838783e-01
4.18973446e-01 -1.08459151e+00 5.97524941e-01 8.67489040e-01
3.81473929e-01 -2.93413103e-01 5.01214862e-01 7.19526470e-01
-1.58097553e+00 -2.51678407e-01 -4.29012954e-01 -4.99797344e-01
6.96021855e-01 2.38099828e-01 -7.97450066e-01 4.83854383e-01
5.97519815e-01 1.38212037e+00 -3.86533946e-01 8.13706815e-01
1.74896061e-01 5.13866186e-01 -1.59053922e-01 -2.10490748e-01
3.21048707e-01 3.02408040e-01 6.95793808e-01 1.07684207e+00
7.24998534e-01 2.48433933e-01 4.88006443e-01 2.67393500e-01
2.49736056e-01 -4.40817364e-02 -1.21228027e+00 1.73760399e-01
1.13098845e-01 8.76755357e-01 -7.27023363e-01 -6.31738305e-01
-7.78682590e-01 1.13785350e+00 2.73207188e-01 4.31233913e-01
-8.01891387e-01 6.70181662e-02 4.40441459e-01 3.36507946e-01
7.90185928e-01 -5.35447478e-01 -2.38100090e-03 -1.12572157e+00
3.16907801e-02 -3.01831901e-01 5.48075676e-01 -8.06916833e-01
-1.22276413e+00 3.67675662e-01 1.14020750e-01 -1.65222955e+00
-9.96103227e-01 -5.02122045e-01 -6.08192027e-01 4.13229555e-01
-1.25478399e+00 -1.37616432e+00 -4.45704699e-01 7.28679359e-01
1.00710428e+00 -1.14829443e-01 6.45788372e-01 -4.19591628e-02
-2.77908504e-01 3.26252282e-02 -1.54328212e-01 -1.46101192e-01
5.02119780e-01 -1.09869385e+00 8.81191194e-01 9.48504269e-01
2.22979292e-01 -5.24061583e-02 8.12016010e-01 -9.72862661e-01
-1.81382549e+00 -1.16889215e+00 7.91790545e-01 -6.55900896e-01
9.59136128e-01 -3.08523029e-01 -7.97799885e-01 9.88226175e-01
-1.71517897e-02 5.10597646e-01 3.29101980e-01 -8.05061981e-02
6.43889979e-02 2.61875600e-01 -5.14301836e-01 8.14929605e-01
1.49130559e+00 -3.28947127e-01 -3.31363767e-01 3.61982435e-01
9.49755847e-01 -3.86065930e-01 -7.99979329e-01 4.43958789e-01
4.79127556e-01 -8.08578908e-01 1.03789473e+00 -7.53038883e-01
4.14930224e-01 -1.45987853e-01 -4.29916680e-01 -1.26617086e+00
-5.58724761e-01 -6.66289091e-01 -6.51422024e-01 5.13355196e-01
1.89374655e-01 -2.16345280e-01 1.09828317e+00 6.39777660e-01
-2.19491750e-01 -5.79677224e-01 -8.26609612e-01 -9.06704605e-01
-2.16948822e-01 -7.24017501e-01 1.97305143e-01 7.64557123e-01
1.88269056e-02 4.55075532e-01 -1.07902634e+00 -5.16660232e-03
4.74209279e-01 1.98436994e-02 1.17556930e+00 -1.15464652e+00
-6.97337389e-01 -1.73523098e-01 -8.52817118e-01 -1.27559280e+00
3.81684124e-01 -5.92661142e-01 2.34537393e-01 -1.67392230e+00
1.85608461e-01 2.18381301e-01 -1.67898431e-01 2.78017193e-01
-2.48438254e-01 8.99821147e-02 5.20004034e-01 2.88880616e-01
-1.01551199e+00 6.20523512e-01 1.09391439e+00 -2.88853645e-01
-1.85462654e-01 9.65159163e-02 -8.54760334e-02 8.09048831e-01
5.86176753e-01 -3.10980588e-01 -8.24874043e-01 -2.91524410e-01
7.13832602e-02 5.06161690e-01 4.98346210e-01 -1.47421420e+00
6.21936023e-01 -2.92496026e-01 9.60714146e-02 -5.56679904e-01
8.86672020e-01 -8.64854276e-01 4.94746357e-01 4.34766859e-01
-2.81246841e-01 1.04882255e-01 -7.47752935e-03 1.29355001e+00
-1.99555587e-02 1.38844475e-01 2.02390760e-01 -2.42590234e-01
-1.51707637e+00 6.18905067e-01 -4.43951249e-01 -5.22231907e-02
1.42254138e+00 -4.18311238e-01 -3.70515943e-01 -8.78190935e-01
-1.13960230e+00 1.26406506e-01 5.46513140e-01 5.42789400e-01
6.77374661e-01 -1.34832263e+00 -4.88694638e-01 -2.59126902e-01
2.88086355e-01 -1.01851873e-01 5.27026594e-01 8.53111207e-01
-1.85071334e-01 3.09579521e-01 -1.88132569e-01 -8.23022962e-01
-1.24583650e+00 9.44515467e-01 1.03044830e-01 -3.30592155e-01
-1.12693405e+00 4.80869800e-01 2.20553309e-01 5.63002042e-02
2.72585452e-01 -4.47403550e-01 -2.89020419e-01 -4.04962637e-02
2.78411806e-01 8.41856837e-01 -2.72358447e-01 -8.79004657e-01
-3.00853342e-01 7.61138439e-01 5.11958778e-01 -1.62273943e-01
1.33882368e+00 -2.35304639e-01 5.29723883e-01 1.19510567e+00
1.02489853e+00 -4.26744878e-01 -1.58619583e+00 -2.34066010e-01
1.65274721e-02 -1.69330165e-01 -2.16163784e-01 -9.83382761e-02
-9.76653934e-01 1.01749575e+00 3.87023866e-01 2.32028291e-01
8.56705606e-01 1.18193701e-01 1.01807213e+00 5.64981461e-01
6.17178977e-01 -9.76882935e-01 3.28969002e-01 5.93891799e-01
5.74615955e-01 -1.46479619e+00 -2.18835403e-03 -3.77310187e-01
-1.00785148e+00 9.19130802e-01 4.09826159e-01 -4.19312954e-01
9.08100545e-01 1.03344955e-01 -4.11172003e-01 -3.12074065e-01
-1.11565995e+00 -3.55617434e-01 4.84660923e-01 6.47494435e-01
2.54767120e-01 1.98299780e-01 1.30714059e-01 3.95703912e-01
2.84564439e-02 2.90888429e-01 4.75304782e-01 1.06485152e+00
-5.14615476e-01 -6.39487743e-01 -1.27200754e-02 4.88975674e-01
-3.14673722e-01 8.20855498e-02 -1.70536771e-01 8.34824383e-01
-3.23673099e-01 1.15562856e+00 1.42610192e-01 -7.25570023e-01
2.25357413e-01 8.32410902e-02 2.07614198e-01 -7.54967868e-01
-2.81315982e-01 -6.99409619e-02 2.13220403e-01 -9.88203347e-01
-6.78913891e-01 -6.22875512e-01 -1.10474360e+00 -3.36162418e-01
1.52327299e-01 -5.08378029e-01 2.30606318e-01 9.91857588e-01
6.09242857e-01 6.32607043e-01 1.06009200e-01 -1.34795237e+00
-3.13477039e-01 -1.07012606e+00 -5.38403928e-01 6.04224026e-01
4.94908571e-01 -7.63756216e-01 -2.26543143e-01 2.72740632e-01] | [7.529101371765137, 0.024860821664333344] |
34403cd6-2ada-46e3-97bb-190ab65737ff | luminance-attentive-networks-for-hdr-image | 2109.06688 | null | https://arxiv.org/abs/2109.06688v1 | https://arxiv.org/pdf/2109.06688v1.pdf | Luminance Attentive Networks for HDR Image and Panorama Reconstruction | It is very challenging to reconstruct a high dynamic range (HDR) from a low dynamic range (LDR) image as an ill-posed problem. This paper proposes a luminance attentive network named LANet for HDR reconstruction from a single LDR image. Our method is based on two fundamental observations: (1) HDR images stored in relative luminance are scale-invariant, which means the HDR images will hold the same information when multiplied by any positive real number. Based on this observation, we propose a novel normalization method called " HDR calibration " for HDR images stored in relative luminance, calibrating HDR images into a similar luminance scale according to the LDR images. (2) The main difference between HDR images and LDR images is in under-/over-exposed areas, especially those highlighted. Following this observation, we propose a luminance attention module with a two-stream structure for LANet to pay more attention to the under-/over-exposed areas. In addition, we propose an extended network called panoLANet for HDR panorama reconstruction from an LDR panorama and build a dualnet structure for panoLANet to solve the distortion problem caused by the equirectangular panorama. Extensive experiments show that our proposed approach LANet can reconstruct visually convincing HDR images and demonstrate its superiority over state-of-the-art approaches in terms of all metrics in inverse tone mapping. The image-based lighting application with our proposed panoLANet also demonstrates that our method can simulate natural scene lighting using only LDR panorama. Our source code is available at https://github.com/LWT3437/LANet. | ['Chunxia Xiao', 'Qin Zou', 'Bo Dong', 'Chengjiang Long', 'Wentao Liu', 'Hanning Yu'] | 2021-09-14 | null | null | null | null | ['hdr-reconstruction', 'tone-mapping', 'inverse-tone-mapping'] | ['computer-vision', 'computer-vision', 'computer-vision'] | [ 3.19934398e-01 -3.11364055e-01 -2.02165898e-02 -1.41684696e-01
-2.99096912e-01 -2.22017527e-01 3.40702504e-01 -6.50684297e-01
-2.44051903e-01 8.02626669e-01 1.67578965e-01 -2.58805603e-01
1.18488364e-01 -1.14674556e+00 -7.56040931e-01 -6.02806151e-01
3.11450213e-01 -2.64962137e-01 2.47679651e-01 -5.87655365e-01
3.54301557e-02 6.82695568e-01 -1.68898165e+00 -3.26803252e-02
9.12438750e-01 9.06270802e-01 6.09574258e-01 8.75087619e-01
8.55800435e-02 8.34880769e-01 -5.56626022e-01 -9.61637273e-02
6.93843067e-01 -6.46168888e-01 -3.74922335e-01 6.18191548e-02
6.13594770e-01 -7.31338024e-01 -7.26385772e-01 1.28163028e+00
6.72810078e-01 6.14477471e-02 2.16472358e-01 -1.21745968e+00
-9.87288356e-01 1.78528517e-01 -9.56111789e-01 1.38495728e-01
4.62993264e-01 2.87437469e-01 6.36931300e-01 -6.74879313e-01
5.44667482e-01 1.38859618e+00 5.49931765e-01 3.77212912e-01
-1.13885581e+00 -9.25884366e-01 -1.61776215e-01 3.21093410e-01
-1.48743832e+00 -3.90068144e-01 8.28567863e-01 3.01345170e-01
6.41210079e-01 5.48427701e-01 8.37764144e-01 6.92542315e-01
2.78329849e-01 2.91847497e-01 1.89314497e+00 -5.06950557e-01
-1.70593649e-01 4.30752300e-02 -2.89248109e-01 4.41338629e-01
1.42812833e-01 4.82987493e-01 -3.20579141e-01 4.58221972e-01
1.32940924e+00 -4.52503003e-02 -8.21390152e-01 6.42955005e-02
-1.19913208e+00 4.47283387e-01 5.81349969e-01 2.77549386e-01
-8.74760598e-02 2.03957081e-01 -7.73813054e-02 5.50063133e-01
2.84291983e-01 1.40742525e-01 -1.51033312e-01 1.76654965e-01
-6.42873287e-01 -2.38986880e-01 4.16198283e-01 9.55285847e-01
9.81079996e-01 1.83329567e-01 1.11635841e-01 1.14250219e+00
1.95455223e-01 9.37711775e-01 2.97470838e-01 -9.65050876e-01
3.07175070e-01 2.13972569e-01 -8.43288675e-02 -9.64563310e-01
-2.43986920e-01 -1.28793508e-01 -1.31997073e+00 7.49940991e-01
1.15230739e-01 9.35216099e-02 -9.61358190e-01 1.65186477e+00
1.58151716e-01 2.83991136e-02 2.62823850e-01 1.17514479e+00
9.23504472e-01 1.08371270e+00 -1.37570262e-01 -3.01182240e-01
1.33103871e+00 -7.28117347e-01 -9.91580904e-01 -4.97045778e-02
-1.61121517e-01 -1.12058640e+00 1.49407673e+00 4.60231036e-01
-1.19552767e+00 -9.64750767e-01 -1.26538277e+00 -5.27874470e-01
-2.76119202e-01 1.35905072e-01 3.31891030e-01 6.61821485e-01
-1.32587123e+00 3.69496793e-01 -6.57341555e-02 -4.36174780e-01
-1.26979351e-02 -1.35806063e-02 -3.58440280e-01 -4.88823473e-01
-1.36277771e+00 1.05041981e+00 3.76144022e-01 2.66134977e-01
-7.43303299e-01 -7.44832695e-01 -8.24507236e-01 -1.45258933e-01
2.24568203e-01 -5.16541600e-01 6.97348833e-01 -1.05609739e+00
-1.72127771e+00 7.83696294e-01 1.65660530e-02 -2.18070671e-01
7.00652301e-01 -5.83144501e-02 -8.30150127e-01 1.90014020e-01
-1.68018445e-01 6.87626719e-01 8.20399463e-01 -1.60315585e+00
-6.47138119e-01 -3.28021199e-02 3.97836529e-02 5.37609696e-01
2.73062419e-02 2.46088617e-02 -5.71127474e-01 -7.93631971e-01
2.38840744e-01 -6.17992699e-01 1.67460322e-01 2.87757963e-01
-4.35621232e-01 3.34560663e-01 1.26547277e+00 -8.13174009e-01
1.14864349e+00 -2.02344298e+00 -2.80780584e-01 1.78026989e-01
1.63235292e-01 2.43980363e-01 -2.60010749e-01 1.31627675e-02
-4.47197229e-01 -1.41996399e-01 -2.83281565e-01 1.40331462e-01
-2.26369008e-01 -1.71134863e-02 -5.05600572e-01 6.35811090e-01
-1.13699958e-01 8.26222718e-01 -5.75834095e-01 -4.87884820e-01
7.57478714e-01 8.65974009e-01 5.17805805e-03 4.41068977e-01
1.35734826e-01 5.12928963e-01 2.42544413e-01 5.43238997e-01
1.35126317e+00 1.58319250e-01 -6.43302174e-03 -7.60038197e-01
-4.44403976e-01 -3.17725748e-01 -1.29052305e+00 1.33422267e+00
-8.08824539e-01 9.57675159e-01 -4.04274017e-02 -3.16004932e-01
1.31588900e+00 7.72631587e-03 1.44772217e-01 -1.37014520e+00
1.63573518e-01 2.72361040e-01 -1.30555078e-01 -4.43330437e-01
6.15955353e-01 -2.93905407e-01 1.57215267e-01 3.88520479e-01
-3.31417650e-01 -5.29199779e-01 6.95023239e-02 -1.89732295e-02
6.26054585e-01 1.71698734e-01 4.69500273e-01 -5.13672009e-02
8.26143920e-01 -5.26026249e-01 5.08808017e-01 5.78232229e-01
-9.19281617e-02 9.80894089e-01 2.21141636e-01 -4.37822551e-01
-1.49930584e+00 -1.26025105e+00 -3.10017794e-01 7.37097263e-01
6.66938365e-01 3.70678827e-02 -4.69368696e-01 -3.72831151e-02
-4.54742640e-01 6.58067703e-01 -5.12944698e-01 9.25000831e-02
-7.37272501e-01 -6.80140018e-01 4.72901881e-01 1.80628508e-01
1.48046565e+00 -1.13595426e+00 -5.98573744e-01 -1.92569643e-01
-3.60414386e-01 -1.07248855e+00 -5.28336108e-01 2.25027949e-02
-4.40486550e-01 -1.01409721e+00 -1.19217992e+00 -8.23640823e-01
5.62958539e-01 7.28988051e-01 1.11650193e+00 5.58285825e-02
-2.90729403e-01 2.02631354e-01 -1.96954861e-01 -6.28978610e-02
-5.45787573e-01 -4.64348286e-01 -1.96275100e-01 1.08916007e-01
-3.17068142e-03 -5.75480342e-01 -9.56734776e-01 6.55127764e-01
-1.18236387e+00 6.24650478e-01 6.53223991e-01 4.95116800e-01
7.99338579e-01 4.54460919e-01 3.73679280e-01 -5.07299602e-01
5.17643154e-01 -1.37111535e-02 -7.78993785e-01 3.64681691e-01
-5.14349341e-01 -2.51589775e-01 8.08552861e-01 -4.97118384e-01
-1.46882319e+00 -2.67421305e-01 -1.39015689e-01 -4.98177111e-01
-2.42309421e-01 -4.92600352e-01 -4.55860347e-01 -3.26892495e-01
4.55225140e-01 3.89174968e-01 2.18952224e-02 -1.31075561e-01
6.26022756e-01 7.21619070e-01 9.22818005e-01 -2.78718591e-01
1.06500196e+00 7.17763484e-01 9.02377591e-02 -9.07705188e-01
-3.73719305e-01 4.31660516e-03 -1.87433615e-01 -4.99808520e-01
1.00967646e+00 -1.03289700e+00 -8.87763202e-01 8.24647129e-01
-9.22176600e-01 -6.53489530e-01 -4.11757797e-01 3.27327490e-01
-6.76165640e-01 4.37604934e-01 -6.62972808e-01 -4.78172958e-01
-3.39812458e-01 -1.15340054e+00 8.06286514e-01 6.00540042e-01
2.59940237e-01 -8.01099241e-01 4.37733978e-02 4.32896651e-02
7.76123941e-01 4.32661138e-02 7.98814893e-01 4.32841599e-01
-9.26354587e-01 3.22568953e-01 -8.97177160e-01 5.60207903e-01
1.86993539e-01 -3.42167765e-02 -1.10508120e+00 -2.48978093e-01
1.18561022e-01 -1.08899832e-01 7.73997843e-01 5.26996195e-01
1.26989257e+00 -9.12896171e-02 2.06012726e-01 1.19786251e+00
1.90593600e+00 2.69270658e-01 1.45485759e+00 5.65586567e-01
7.40114748e-01 4.70196784e-01 8.09470713e-01 1.13335975e-01
2.56128103e-01 7.68968105e-01 3.53152037e-01 -8.74586463e-01
-9.42327201e-01 -3.51673424e-01 3.28852206e-01 5.59893072e-01
-1.04102999e-01 -2.90132910e-01 -3.63221794e-01 1.18101545e-01
-1.22203505e+00 -8.38948727e-01 -2.34129876e-01 2.26942325e+00
9.34044838e-01 -2.07737029e-01 -2.06834376e-01 1.19202353e-01
1.05319226e+00 4.56413269e-01 -4.92984682e-01 -4.17199939e-01
-8.94816458e-01 2.22800106e-01 8.79918873e-01 5.56634128e-01
-7.87137330e-01 7.82734156e-01 5.75055838e+00 1.12299132e+00
-1.42656910e+00 7.09743798e-02 9.20077026e-01 2.38354430e-01
-4.17750835e-01 -1.11532234e-01 -4.88721400e-01 5.29930532e-01
4.55323279e-01 -2.07967907e-01 8.03362072e-01 3.72922927e-01
3.70855927e-01 -4.81959105e-01 -4.96860623e-01 1.34206510e+00
3.59863311e-01 -8.97206128e-01 -3.78870703e-02 -4.30507623e-02
8.73812377e-01 -2.29186028e-01 5.79234302e-01 -1.24213189e-01
3.09854060e-01 -9.14157987e-01 6.21032119e-01 3.44517231e-01
1.62526655e+00 -7.02845693e-01 4.90975738e-01 -1.79070652e-01
-1.40340805e+00 -1.36990482e-02 -6.08736992e-01 4.22554284e-01
3.13809574e-01 4.27459836e-01 -3.79829139e-01 5.09805858e-01
1.01771867e+00 7.67937124e-01 -6.68133259e-01 9.22969162e-01
-5.12772679e-01 2.03517191e-02 -1.30141824e-01 5.80934346e-01
-2.90061295e-01 -4.56383616e-01 3.89585614e-01 1.04966688e+00
5.45317709e-01 1.95108891e-01 -5.01158297e-01 9.91523683e-01
-1.18769623e-01 2.25371663e-02 -7.26789832e-01 6.90957844e-01
2.57481366e-01 1.32971525e+00 -7.56244004e-01 -3.15790564e-01
-3.70178103e-01 1.13171470e+00 -3.24182421e-01 6.78046763e-01
-1.13200641e+00 -7.35924363e-01 2.84394354e-01 6.16771053e-04
1.23444676e-01 1.09468922e-01 -2.70891964e-01 -1.10261548e+00
-1.50064752e-01 -1.00369930e+00 1.10494614e-01 -1.65535378e+00
-1.04474795e+00 7.78165638e-01 -6.58322796e-02 -1.52136779e+00
5.13014436e-01 -3.75655651e-01 -5.91857135e-01 9.62246835e-01
-2.01327872e+00 -1.02341807e+00 -1.00531197e+00 8.77788961e-01
4.44311827e-01 8.68749171e-02 4.37809110e-01 4.89728302e-01
-3.70904058e-01 3.21439534e-01 1.87922642e-02 -1.18444182e-01
1.02074969e+00 -1.08713806e+00 2.61638761e-01 1.13744140e+00
-3.37499410e-01 2.85312951e-01 7.26519108e-01 -5.98469973e-01
-1.18339455e+00 -1.13662171e+00 3.27608734e-01 1.19354434e-01
2.76887029e-01 -2.97465324e-01 -9.60306287e-01 4.75801617e-01
4.37971979e-01 2.34001856e-02 1.70275122e-02 -6.09023511e-01
-4.40106302e-01 -6.02802336e-01 -1.33569074e+00 7.41510510e-01
9.10557866e-01 -5.19421756e-01 -2.04543605e-01 2.11074948e-02
9.50921059e-01 -4.70866144e-01 -8.28021526e-01 4.67338920e-01
5.83389819e-01 -1.35133815e+00 1.21528399e+00 5.43108523e-01
6.07605159e-01 -8.37043405e-01 -3.01843792e-01 -1.19419897e+00
-7.56500438e-02 -6.43656909e-01 2.37206519e-01 1.13177073e+00
-5.53821363e-02 -9.04315710e-01 2.40151346e-01 1.01537153e-01
7.35703409e-02 -3.81044328e-01 -6.81002021e-01 -6.28276050e-01
-2.42444679e-01 -2.74305373e-01 7.00615883e-01 8.95278394e-01
-7.62214720e-01 6.67137355e-02 -7.93895483e-01 1.59070656e-01
8.50617349e-01 1.20452017e-01 8.53023589e-01 -6.24463201e-01
-1.03126854e-01 -1.74377710e-01 -2.77239859e-01 -1.02665365e+00
-1.03063494e-01 -4.80465472e-01 1.22318447e-01 -1.54889655e+00
2.40070194e-01 -6.63149416e-01 -9.14419889e-02 1.53281316e-01
-4.68383990e-02 9.02436197e-01 4.48117226e-01 2.07487836e-01
-3.60860020e-01 5.33064485e-01 1.64229608e+00 3.07106748e-02
-2.27694258e-01 -4.33769315e-01 -5.17327607e-01 5.65634906e-01
8.04316878e-01 3.37180041e-04 -3.54458958e-01 -3.07679385e-01
3.70208435e-02 1.29338458e-01 4.90302503e-01 -1.22393346e+00
6.96795955e-02 -1.33913875e-01 6.77894831e-01 -6.85564578e-01
3.71193677e-01 -1.03563714e+00 6.39838398e-01 3.53914529e-01
-2.10690111e-01 1.20052971e-01 9.16698128e-02 3.02893490e-01
-1.75195992e-01 2.09885165e-01 1.32701874e+00 -3.18672359e-02
-1.02145994e+00 1.55724913e-01 -2.01926939e-02 -1.65926576e-01
1.07512391e+00 -3.90589386e-01 -1.02611864e+00 -6.50575817e-01
-2.34293774e-01 -2.21589863e-01 8.40028346e-01 3.21767777e-01
9.45575774e-01 -1.40791309e+00 -6.00197077e-01 5.75597644e-01
-3.08487769e-02 -2.45286092e-01 7.58470178e-01 7.42114544e-01
-1.08312416e+00 1.54916465e-01 -7.55021572e-01 -3.96454215e-01
-1.37601984e+00 6.24469340e-01 5.42118371e-01 4.00336795e-02
-9.17801082e-01 2.93605626e-01 5.88018596e-01 -1.74484536e-01
6.86159208e-02 -1.95635676e-01 -6.95341751e-02 -3.29728931e-01
5.86215615e-01 5.08480668e-01 -2.18485519e-01 -8.58948112e-01
1.45724982e-01 1.15424025e+00 2.38279581e-01 -1.67560428e-01
1.16623199e+00 -7.35480011e-01 -3.20440948e-01 2.28539899e-01
1.34572089e+00 1.08777136e-01 -1.29294121e+00 -1.13586269e-01
-9.56165910e-01 -1.02996457e+00 1.57182604e-01 -8.20908487e-01
-1.42535412e+00 7.81688094e-01 1.12325096e+00 3.14320214e-02
1.91369069e+00 -1.92877606e-01 8.43520939e-01 -6.90128654e-02
2.69416153e-01 -9.54439342e-01 1.16170026e-01 8.66088867e-02
1.01529157e+00 -1.19126630e+00 1.20687880e-01 -4.26596642e-01
-6.03321135e-01 1.27611458e+00 7.19955206e-01 -8.47854614e-02
2.43715271e-01 2.11705565e-01 3.41086119e-01 1.18975595e-01
-3.58468801e-01 -3.14173937e-01 3.50235514e-02 7.65337527e-01
1.51365340e-01 -1.81345880e-01 -1.34109393e-01 -3.16024154e-01
-2.25808501e-01 4.39133635e-03 9.37552810e-01 4.15952444e-01
-4.35878128e-01 -7.33922958e-01 -7.21943498e-01 -8.60739872e-02
-2.51729012e-01 -1.74398124e-01 2.01916799e-01 9.98103023e-01
2.53532141e-01 8.91563833e-01 2.40584567e-01 -3.44397664e-01
4.26951170e-01 -6.14126682e-01 4.87535149e-01 4.51090597e-02
-8.71080458e-02 1.99409842e-01 -2.70875424e-01 -7.21714616e-01
-5.88768959e-01 1.25996051e-02 -9.14555192e-01 -7.29455173e-01
-7.57434592e-02 -4.75678414e-01 7.31427610e-01 1.49963468e-01
-7.50674680e-02 7.17127919e-01 8.54399204e-01 -7.14364648e-01
8.77791420e-02 -6.91492438e-01 -8.82076740e-01 5.17062068e-01
4.85128731e-01 -3.89362961e-01 -5.76999426e-01 5.27431592e-02] | [10.897172927856445, -2.230468988418579] |
1ce02ac1-1e75-4686-810e-30ad1439e33d | dadagger-disagreement-augmented-dataset | 2301.01348 | null | https://arxiv.org/abs/2301.01348v1 | https://arxiv.org/pdf/2301.01348v1.pdf | DADAgger: Disagreement-Augmented Dataset Aggregation | DAgger is an imitation algorithm that aggregates its original datasets by querying the expert on all samples encountered during training. In order to reduce the number of samples queried, we propose a modification to DAgger, known as DADAgger, which only queries the expert for state-action pairs that are out of distribution (OOD). OOD states are identified by measuring the variance of the action predictions of an ensemble of models on each state, which we simulate using dropout. Testing on the Car Racing and Half Cheetah environments achieves comparable performance to DAgger but with reduced expert queries, and better performance than a random sampling baseline. We also show that our algorithm may be used to build efficient, well-balanced training datasets by running with no initial data and only querying the expert to resolve uncertainty. | ['Samarendra Chandan Bindu Dash', 'Karim Hamadeh', 'Akash Haridas'] | 2023-01-03 | null | null | null | null | ['carracing-v0'] | ['playing-games'] | [-1.98906228e-01 4.39448118e-01 -4.82405007e-01 -4.20552760e-01
-1.30968463e+00 -8.74430835e-01 4.91123110e-01 -1.60073414e-01
-6.43943906e-01 1.05909801e+00 -1.35417342e-01 -3.91394049e-01
-8.07364136e-02 -5.50163329e-01 -1.25561607e+00 -4.96257216e-01
-8.14512521e-02 1.13894129e+00 4.13574159e-01 3.33087772e-01
-6.91788197e-02 3.62319082e-01 -1.57314932e+00 4.66740988e-02
8.13324273e-01 6.51561856e-01 -1.07673891e-01 8.61473382e-01
3.19231391e-01 1.07197452e+00 -7.61504829e-01 -1.67053252e-01
2.10126951e-01 -4.94382679e-01 -6.93144262e-01 -1.65078163e-01
3.68584633e-01 -7.42198169e-01 -4.53513265e-01 1.05313540e+00
2.74879187e-01 5.03721476e-01 3.32754850e-01 -1.36563480e+00
2.91394331e-02 8.22139680e-01 3.79746500e-03 8.17554370e-02
1.53587550e-01 6.30577266e-01 7.55404353e-01 -2.78107226e-01
6.91397607e-01 1.47771084e+00 3.06033790e-01 6.89505577e-01
-1.56187987e+00 -8.22470367e-01 4.72414196e-01 6.51609898e-02
-9.35825944e-01 -5.00466406e-01 3.41849059e-01 -1.62609830e-01
1.05999720e+00 6.26383498e-02 5.13612807e-01 1.58308661e+00
1.28930271e-01 1.19525468e+00 1.00931787e+00 5.89764267e-02
9.08272624e-01 -1.53461713e-02 1.57821193e-01 5.95597386e-01
2.74901628e-01 6.52270675e-01 -4.51473653e-01 -5.19810677e-01
5.13628900e-01 -1.81180984e-01 7.52570108e-02 -5.27312338e-01
-9.48340714e-01 7.14168012e-01 -1.67451307e-01 -4.55229193e-01
-4.07489717e-01 6.01538002e-01 2.32361048e-01 5.30273616e-01
1.86669379e-01 5.33307672e-01 -6.64425194e-01 -6.39024854e-01
-6.48777544e-01 8.22435915e-01 1.00801957e+00 1.06276071e+00
8.90729368e-01 1.57245696e-01 -4.14437145e-01 1.82587519e-01
6.29779696e-02 6.41719580e-01 4.00597423e-01 -1.78709543e+00
5.39443612e-01 3.36908817e-01 7.62868941e-01 1.49814278e-01
-1.12769492e-01 -2.81074792e-01 3.49436030e-02 4.55995917e-01
3.56005579e-01 -7.30880857e-01 -1.10462141e+00 2.03619123e+00
2.47774035e-01 5.67133486e-01 2.37382293e-01 4.56669688e-01
-7.72536397e-02 4.29562896e-01 7.86198489e-03 3.97488363e-02
5.50373912e-01 -6.36638880e-01 -4.14519310e-01 -6.19792998e-01
6.99197173e-01 3.36425528e-02 1.07532001e+00 5.92966557e-01
-1.15250421e+00 -2.82394260e-01 -1.01308870e+00 3.99508357e-01
8.78146570e-03 -3.78794745e-02 3.70591938e-01 4.43487853e-01
-8.32814336e-01 1.10776567e+00 -1.68329990e+00 1.16019025e-01
4.26298976e-01 4.64917839e-01 -6.50932789e-02 9.50240530e-03
-1.08587766e+00 9.54446316e-01 4.53426868e-01 -1.43435746e-01
-1.91784871e+00 -6.07233107e-01 -7.94888616e-01 -1.06618434e-01
7.46314347e-01 -3.12789083e-01 1.89515626e+00 -7.31814265e-01
-1.75710499e+00 1.73445001e-01 -1.83267400e-01 -8.74291718e-01
7.32754827e-01 -4.67270941e-01 -1.30062029e-01 -1.40057027e-01
-2.35235691e-03 6.34391487e-01 6.30501330e-01 -1.07169390e+00
-6.89496338e-01 -4.27622527e-01 -5.78652658e-02 1.81106612e-01
4.12659258e-01 -3.07599068e-01 -3.69414508e-01 3.96414027e-02
-2.09926844e-01 -1.22900105e+00 -2.79671907e-01 -4.61334050e-01
-5.02277970e-01 -3.09383839e-01 7.30893970e-01 -2.87747949e-01
8.72883916e-01 -1.95019722e+00 8.73400494e-02 1.89924106e-01
-2.56879419e-01 7.85599202e-02 -3.02470565e-01 4.61526662e-01
3.36750656e-01 -2.15418749e-02 -1.86284259e-01 -5.98369658e-01
2.42032841e-01 6.02305830e-01 -5.52027822e-01 3.70578527e-01
2.26453736e-01 7.37486124e-01 -1.17655396e+00 -3.24080229e-01
1.54103473e-01 -9.03748795e-02 -7.13863850e-01 4.09271389e-01
-7.65984535e-01 6.28992796e-01 -6.98750854e-01 3.51431876e-01
3.06295276e-01 1.40851783e-02 3.99948269e-01 5.53909183e-01
1.46265224e-01 5.04717410e-01 -1.28663027e+00 1.43785143e+00
-3.48577529e-01 3.91848147e-01 3.86585146e-02 -8.52317810e-01
5.73660731e-01 2.58974433e-01 9.13421810e-02 -3.79965663e-01
2.04955295e-01 1.17259651e-01 -1.11629711e-02 -3.64614904e-01
1.30678073e-01 1.83959454e-01 -3.06955278e-01 5.57066083e-01
1.40779614e-01 -3.53272200e-01 1.70107424e-01 1.56541124e-01
1.41459811e+00 4.93381172e-01 -2.11460873e-01 5.54869957e-02
-2.38321662e-01 1.34140134e-01 8.96182597e-01 1.12168229e+00
-2.95915574e-01 4.61943187e-02 9.80394065e-01 -2.83714645e-02
-8.86858881e-01 -1.41524577e+00 2.04048485e-01 8.03742588e-01
1.49982080e-01 -9.84825939e-02 -9.37089682e-01 -8.18745017e-01
3.05423796e-01 1.52575159e+00 -6.90111160e-01 -6.19327784e-01
-3.78171295e-01 -4.12240952e-01 5.69108963e-01 6.29407823e-01
5.18669784e-01 -1.11335921e+00 -8.67440701e-01 2.35386088e-01
6.67587295e-02 -8.20722818e-01 -4.70892310e-01 5.93573153e-01
-1.01893568e+00 -1.11890316e+00 -1.06132306e-01 -3.16838920e-01
4.96033430e-01 -6.38997138e-01 1.13987446e+00 -6.08573139e-01
1.64273068e-01 3.04064214e-01 9.37149581e-03 -5.27642965e-01
-8.42876673e-01 1.34890333e-01 1.27489999e-01 -3.11299086e-01
2.99157798e-01 -4.52324718e-01 -3.81059885e-01 2.50483096e-01
-5.08461297e-01 -3.40830117e-01 5.48096836e-01 8.52560341e-01
5.78166902e-01 3.01150363e-02 5.05757868e-01 -1.06402171e+00
5.93489766e-01 -5.74098110e-01 -1.13552928e+00 2.56781757e-01
-6.46408260e-01 7.14274943e-01 6.19695783e-01 -6.41271770e-01
-1.08762574e+00 1.61209121e-01 1.99027494e-01 -8.36030483e-01
-2.76755571e-01 1.63676187e-01 -1.40870348e-01 5.71204484e-01
5.87267220e-01 7.47298598e-02 1.82709724e-01 -6.05571747e-01
2.80433595e-01 4.30656612e-01 6.50059998e-01 -9.32885051e-01
5.54802477e-01 1.54524043e-01 -2.32780412e-01 -1.98478654e-01
-8.20948303e-01 -2.59779468e-02 -1.05829604e-01 4.05274592e-02
7.08375335e-01 -9.84027803e-01 -1.09166968e+00 5.52528381e-01
-6.53097391e-01 -1.13780963e+00 -6.66854501e-01 6.59796357e-01
-8.58935237e-01 -1.02781802e-01 -5.29328048e-01 -1.17450738e+00
2.47051477e-01 -1.19226766e+00 1.02382898e+00 2.87131786e-01
-3.23634863e-01 -5.37421823e-01 3.73107702e-01 -3.85515997e-03
1.39995933e-01 8.49473290e-03 7.95845509e-01 -1.01760209e+00
-7.46444821e-01 -3.73744369e-01 5.34933209e-01 4.80832219e-01
-1.87211975e-01 -1.87204741e-02 -9.69010711e-01 -4.93050098e-01
-2.65395045e-01 -9.23397124e-01 7.77297974e-01 3.63518804e-01
1.13114285e+00 -2.66935766e-01 -3.70264500e-01 1.44452006e-01
9.34550047e-01 4.74905968e-01 3.53307784e-01 2.71132797e-01
-4.03518081e-02 3.27967614e-01 7.73308516e-01 5.35951853e-01
1.66566372e-01 3.66834193e-01 5.38387239e-01 4.67632145e-01
3.83259267e-01 -8.63961697e-01 8.51967156e-01 -4.66624610e-02
5.43764353e-01 -2.92687178e-01 -6.46224618e-01 6.25786483e-01
-1.98380291e+00 -1.01816463e+00 5.34750581e-01 2.68123078e+00
1.01362026e+00 5.66394031e-01 2.77172148e-01 -4.12284583e-01
3.68055254e-01 -1.30560314e-02 -1.45889890e+00 -3.41592968e-01
2.21738368e-01 3.41980845e-01 8.41804385e-01 8.35045040e-01
-8.46562445e-01 1.05277765e+00 7.28319168e+00 5.10578036e-01
-6.80709064e-01 -4.90786098e-02 6.95926130e-01 -4.25542653e-01
-2.79795855e-01 3.43078375e-01 -1.04837322e+00 5.58643222e-01
1.58297992e+00 -7.00550750e-02 7.40378678e-01 1.18422616e+00
2.43022308e-01 -4.91483241e-01 -1.51270831e+00 5.01083612e-01
-3.68568808e-01 -1.01299131e+00 -4.51178849e-01 -1.18539751e-01
8.94176841e-01 4.89696264e-01 1.08566850e-01 7.77975976e-01
1.60724783e+00 -7.05931187e-01 8.15591037e-01 5.36524355e-01
5.32883227e-01 -8.90851974e-01 5.62793136e-01 8.08431566e-01
-5.66376090e-01 -2.68616915e-01 -7.06400201e-02 1.00356124e-01
-5.07127903e-02 5.37650362e-02 -9.83651817e-01 4.96056415e-02
4.84435618e-01 2.00354308e-01 -3.18231523e-01 8.45551670e-01
-3.96428555e-01 1.03252888e+00 -6.21961832e-01 -9.24037918e-02
2.00304240e-01 -2.00427279e-01 4.89467055e-01 4.43942398e-01
1.67885020e-01 -3.44120443e-01 4.50249791e-01 1.21913123e+00
2.70162690e-02 -8.06716740e-01 -5.31601489e-01 -5.96232489e-02
1.06120074e+00 5.79855442e-01 -1.76252306e-01 -6.65880859e-01
6.13258630e-02 9.88627195e-01 3.96685600e-01 5.87986887e-01
-1.00451267e+00 -3.16321433e-01 8.42926562e-01 -2.82190263e-01
3.63698363e-01 -3.03697754e-02 2.43614525e-01 -9.82987404e-01
-1.20659649e-01 -9.75485742e-01 3.93787205e-01 -6.09986305e-01
-9.29941654e-01 3.08934391e-01 5.37075818e-01 -8.97636235e-01
-1.00064087e+00 -2.79320568e-01 -6.28912568e-01 7.54811466e-01
-9.79735553e-01 -3.15697700e-01 1.60980538e-01 3.47944468e-01
5.07236540e-01 8.47376511e-02 6.01922214e-01 -2.28302255e-01
-1.00403309e+00 6.33584023e-01 3.86994928e-01 -2.48079747e-02
5.14128149e-01 -1.56213152e+00 7.72117257e-01 6.37371540e-01
-6.23700060e-02 3.84096265e-01 9.60907757e-01 -8.32995117e-01
-1.38654506e+00 -1.21448493e+00 2.91429013e-01 -7.70100713e-01
5.34527481e-01 -3.38005126e-01 -6.83557987e-01 1.39011347e+00
-1.14295974e-01 -1.02834240e-01 -3.20948251e-02 -8.12164694e-03
-8.54740292e-02 -7.44370818e-02 -1.14583790e+00 6.91019416e-01
8.81765187e-01 -5.48562944e-01 -6.15725100e-01 1.48122534e-01
6.70178413e-01 -6.49084687e-01 -5.27520001e-01 8.21172446e-02
2.60567665e-01 -8.53058338e-01 4.43885744e-01 -9.92098868e-01
-1.59023613e-01 -1.49395898e-01 -3.04207969e-02 -1.50263143e+00
1.89830601e-01 -1.00212967e+00 -3.45217913e-01 8.72699142e-01
6.04414940e-01 -9.42997336e-01 1.19481325e+00 1.01489508e+00
-7.44610801e-02 -6.05573654e-01 -1.10166645e+00 -1.10354137e+00
2.44948819e-01 -4.98990089e-01 7.16074467e-01 1.52701572e-01
-2.29105592e-01 1.24306105e-01 -2.91544497e-01 3.05392653e-01
8.17488968e-01 1.21832609e-01 8.48808587e-01 -1.01156151e+00
-7.03989446e-01 -2.38615200e-02 1.41207427e-01 -1.16823757e+00
5.84703088e-01 -4.58352923e-01 4.68299627e-01 -9.85876024e-01
1.07641995e-01 -3.01718116e-01 -2.13759333e-01 5.49304426e-01
-2.90760081e-02 -4.75049913e-01 -1.13551587e-01 -1.17313951e-01
-8.59184265e-01 6.40818059e-01 9.00120497e-01 9.12177935e-02
-2.90153742e-01 3.88792664e-01 -2.77784854e-01 5.78409255e-01
7.58376420e-01 -5.88595569e-01 -7.58176744e-01 -1.69264928e-01
-3.01548876e-02 3.75534654e-01 4.24625456e-01 -1.07362700e+00
5.27156936e-03 -2.57783264e-01 2.17081383e-01 -7.95430303e-01
4.26499873e-01 -6.24866903e-01 4.43701625e-01 6.46385312e-01
-8.76398802e-01 2.24462375e-02 2.56467134e-01 8.74962866e-01
1.78061649e-02 -2.75468558e-01 7.96749711e-01 -2.24624544e-01
-5.39792359e-01 2.77661681e-01 -6.58811271e-01 4.18969005e-01
1.06722903e+00 1.71358138e-01 -3.01727116e-01 -6.07376635e-01
-7.79378355e-01 7.15370357e-01 5.46089530e-01 8.93263593e-02
2.33071551e-01 -9.10183191e-01 -3.41195673e-01 2.34980896e-01
-1.43773645e-01 3.04253817e-01 7.84231797e-02 4.51158583e-01
-1.19388640e-01 2.90710300e-01 2.72957027e-01 -4.02282655e-01
-7.77649581e-01 3.80018771e-01 7.81411350e-01 -3.95027667e-01
-3.73351783e-01 6.68332577e-01 -1.79365858e-01 -7.39745080e-01
5.42309642e-01 -6.04583085e-01 4.29000109e-01 -2.49041378e-01
3.13409299e-01 6.35033131e-01 -8.65558535e-02 2.06783488e-01
-3.06161672e-01 -6.99334964e-02 -9.50484574e-02 -5.42491198e-01
1.04105282e+00 2.28699580e-01 4.73686397e-01 9.05682623e-01
9.26295161e-01 -2.21163720e-01 -2.20442295e+00 3.96160446e-02
1.77799866e-01 -2.46314913e-01 6.36326447e-02 -1.05068421e+00
-6.77993476e-01 5.40574908e-01 6.28318012e-01 -1.37711301e-01
5.45524359e-01 5.57759963e-02 5.21149457e-01 8.51509213e-01
5.82401693e-01 -1.30145562e+00 -4.42321971e-02 4.03564781e-01
3.34260494e-01 -1.13276744e+00 -3.49077731e-01 3.14917684e-01
-8.26411963e-01 5.98089814e-01 9.19541717e-01 -3.72186154e-01
3.09097350e-01 5.14339566e-01 -5.27076907e-02 9.31210965e-02
-1.45083320e+00 -7.43994564e-02 -2.87033945e-01 4.67706412e-01
-3.34543854e-01 1.62406951e-01 5.39682865e-01 4.69415694e-01
5.28708845e-02 3.32257986e-01 4.24138486e-01 1.09630632e+00
-5.77067614e-01 -1.17878294e+00 -9.49723050e-02 6.51945651e-01
-2.37739325e-01 2.17462301e-01 -4.23578322e-01 8.53544235e-01
-3.12880009e-01 8.98282766e-01 4.25577521e-01 -3.74912739e-01
4.61915463e-01 4.86530542e-01 3.88981134e-01 -5.76202095e-01
-3.01688194e-01 -2.27349088e-01 3.37833673e-01 -1.05974913e+00
2.74681866e-01 -8.80606115e-01 -1.28169215e+00 -2.29566783e-01
-1.90141842e-01 4.53908622e-01 2.21727490e-01 8.02894473e-01
6.33375108e-01 3.53369385e-01 6.98862791e-01 -6.34768069e-01
-1.62456489e+00 -9.86397743e-01 -6.67356014e-01 1.96841463e-01
5.04461408e-01 -7.15260983e-01 -7.14067996e-01 -4.15606469e-01] | [4.098577499389648, 2.1370291709899902] |
61766279-5f07-475d-a9e1-ffcec1822d95 | proseqo-projection-sequence-networks-for-on | null | null | https://aclanthology.org/D19-1402 | https://aclanthology.org/D19-1402.pdf | ProSeqo: Projection Sequence Networks for On-Device Text Classification | We propose a novel on-device sequence model for text classification using recurrent projections. Our model ProSeqo uses dynamic recurrent projections without the need to store or look up any pre-trained embeddings. This results in fast and compact neural networks that can perform on-device inference for complex short and long text classification tasks. We conducted exhaustive evaluation on multiple text classification tasks. Results show that ProSeqo outperformed state-of-the-art neural and on-device approaches for short text classification tasks such as dialog act and intent prediction. To the best of our knowledge, ProSeqo is the first on-device long text classification neural model. It achieved comparable results to previous neural approaches for news article, answers and product categorization, while preserving small memory footprint and maintaining high accuracy. | ['Zornitsa Kozareva', 'Sujith Ravi'] | 2019-11-01 | null | null | null | ijcnlp-2019-11 | ['product-categorization'] | ['miscellaneous'] | [ 1.49660736e-01 6.59757331e-02 -6.02199912e-01 -6.71827853e-01
-4.92002547e-01 -5.10021746e-01 6.81858480e-01 4.63178307e-01
-5.71199417e-01 2.62631088e-01 5.27873039e-01 -8.08748960e-01
2.14407235e-01 -5.08822203e-01 -4.38570708e-01 -4.34903428e-02
5.33082187e-01 6.54026031e-01 4.09475043e-02 -1.80343971e-01
3.29625636e-01 -1.34218052e-01 -1.31375301e+00 7.21729934e-01
1.88964739e-01 1.29875386e+00 8.84199589e-02 9.49682474e-01
-4.83158499e-01 1.24225199e+00 -5.07986307e-01 -5.81275940e-01
-2.30549604e-01 3.08409780e-01 -9.81435537e-01 -6.67387545e-01
3.37436199e-01 -6.34310603e-01 -3.75799507e-01 4.14164960e-01
8.51909220e-01 1.25940979e-01 4.69610274e-01 -7.45991528e-01
-9.89707172e-01 1.11128402e+00 -1.51441144e-02 1.10396497e-01
6.42333984e-01 2.90467106e-02 1.30163574e+00 -1.22194254e+00
4.98083770e-01 1.12014747e+00 1.11843348e+00 5.40376127e-01
-1.31592178e+00 -6.10917032e-01 3.54608893e-01 -3.22149426e-01
-9.25390601e-01 -8.06396604e-01 3.50849509e-01 -3.63804102e-01
2.06638694e+00 2.07750484e-01 2.79836059e-01 1.62315178e+00
7.91081131e-01 1.05931222e+00 6.37564063e-01 -3.05055976e-01
3.19773406e-01 3.02179068e-01 1.17156243e+00 6.79279089e-01
-9.36851054e-02 -4.06085432e-01 -6.82098508e-01 -4.72048789e-01
-6.77651092e-02 3.76329362e-01 5.12563474e-02 2.91302800e-01
-9.14411724e-01 8.28446507e-01 2.32471600e-02 2.74295419e-01
-3.37961435e-01 3.23026568e-01 9.16065454e-01 4.22598630e-01
8.31668615e-01 4.42935616e-01 -1.03997767e+00 -7.39521384e-01
-7.33523607e-01 -1.31354645e-01 1.40595353e+00 7.80695200e-01
1.52635410e-01 4.87940088e-02 -2.70719081e-01 1.10104334e+00
9.02273357e-02 3.68390471e-01 9.44445431e-01 -4.12224382e-01
8.52660000e-01 5.79243541e-01 -1.24242589e-01 -6.33029878e-01
-7.69979656e-01 -2.20298335e-01 -6.68428957e-01 -3.21513027e-01
4.08581309e-02 -3.88220549e-01 -8.07870090e-01 1.18901908e+00
-2.72939652e-01 -2.36014083e-01 2.25658253e-01 1.11686647e-01
1.10199201e+00 1.08967769e+00 6.92692026e-02 5.93928210e-02
1.31797493e+00 -1.24468815e+00 -8.07579637e-01 -3.13931733e-01
1.14145529e+00 -7.56177604e-01 1.50911343e+00 2.88411587e-01
-9.11825836e-01 -6.37429774e-01 -1.13712728e+00 -4.96050477e-01
-7.89074719e-01 4.45068270e-01 7.14970648e-01 1.04704988e+00
-1.02427077e+00 5.41282952e-01 -9.56033826e-01 -4.14833069e-01
3.42030644e-01 7.05879569e-01 1.03787504e-01 3.16480041e-01
-1.00858915e+00 7.92853177e-01 3.89610946e-01 6.72975555e-02
-2.82092333e-01 -7.82678664e-01 -6.10409677e-01 3.06313306e-01
1.38351366e-01 -5.22284329e-01 1.85344124e+00 -6.65698647e-01
-2.25362587e+00 6.26179636e-01 -3.15885603e-01 -7.72516429e-01
-4.05432209e-02 -7.06942379e-01 -6.21760428e-01 -3.56169969e-01
-3.73957932e-01 6.66002989e-01 6.20396495e-01 -2.65551984e-01
-2.07567126e-01 -2.25231707e-01 -5.98516176e-03 1.11877225e-01
-9.14491951e-01 2.76939631e-01 -1.76137149e-01 -5.26766717e-01
-4.10629511e-01 -1.13612294e+00 2.86676195e-02 -7.77350292e-02
-5.34251034e-01 -7.61483371e-01 1.10573685e+00 -6.61823630e-01
1.48696625e+00 -1.97600424e+00 -4.16969687e-01 -4.02286202e-01
2.65169710e-01 4.06955779e-01 -3.13655706e-03 5.70665896e-01
3.79790068e-02 2.65152425e-01 5.51223636e-01 -7.37361610e-01
3.29522729e-01 6.69044852e-02 -7.56644726e-01 8.03338811e-02
-2.02054933e-01 1.32512712e+00 -2.96498060e-01 -1.65164739e-01
4.06288594e-01 3.54055494e-01 -4.84501213e-01 6.89316168e-02
-6.65992737e-01 -3.85452747e-01 -3.45897228e-01 4.04467463e-01
1.86327383e-01 -7.20184147e-01 2.31240168e-01 -1.07555371e-02
1.04479924e-01 1.22964180e+00 -4.17417794e-01 1.89867425e+00
-1.10863757e+00 8.78798068e-01 -2.52382159e-01 -6.37086809e-01
7.96984732e-01 5.39975822e-01 3.82480234e-01 -7.67786443e-01
5.27809620e-01 7.88416713e-02 -4.41711903e-01 -3.77390146e-01
9.47270751e-01 2.72728533e-01 -5.00005484e-01 9.46440876e-01
3.52308542e-01 2.81737357e-01 -2.65158534e-01 1.75488308e-01
1.04286957e+00 -1.22724868e-01 2.92355686e-01 -1.10834040e-01
2.25233227e-01 -3.09360325e-01 -1.21442869e-01 8.27944517e-01
2.36909255e-01 1.93572566e-01 2.32772544e-01 -6.41519010e-01
-7.04719484e-01 -8.42044711e-01 -1.60270091e-02 1.81689680e+00
-3.46794277e-01 -9.56313133e-01 -4.56817329e-01 -8.52632165e-01
-2.17660051e-02 9.29943919e-01 -4.68026847e-01 1.41379200e-02
-5.55146873e-01 -6.54492855e-01 8.74911368e-01 8.74620736e-01
2.97870815e-01 -9.99778867e-01 -3.77377212e-01 3.78173262e-01
-1.40518127e-02 -1.16960657e+00 -7.45796740e-01 6.38951719e-01
-9.93663907e-01 -3.73579919e-01 -4.17105526e-01 -9.84239340e-01
1.09621167e-01 -6.95725307e-02 1.08112288e+00 -5.41519642e-01
9.97283459e-02 2.00806007e-01 -2.81275213e-01 -3.60179454e-01
-4.00527894e-01 7.98448503e-01 5.30218109e-02 -3.94377500e-01
7.57071018e-01 -5.84334671e-01 -3.04061443e-01 3.66150290e-01
-4.21061367e-01 9.37684476e-02 4.70891446e-01 1.01201320e+00
1.89653158e-01 -1.48148879e-01 5.10815084e-01 -1.03023672e+00
1.11549938e+00 -4.45683777e-01 -4.47136402e-01 3.19151670e-01
-9.75278616e-01 4.03371662e-01 9.87359524e-01 -7.94002831e-01
-9.82706606e-01 7.32938573e-02 -7.83574760e-01 -1.48350345e-02
-3.88496369e-02 7.31586277e-01 2.60471255e-01 3.74433249e-01
7.11255789e-01 1.18965194e-01 -1.36833265e-01 -7.48533726e-01
4.18556750e-01 1.34216630e+00 -1.81571674e-02 -1.59351781e-01
-1.81774557e-01 9.85438153e-02 -4.66776341e-01 -8.78831625e-01
-7.22273111e-01 -5.59100270e-01 -5.36986351e-01 3.01159352e-01
9.22198355e-01 -8.20606470e-01 -1.14916050e+00 6.18129015e-01
-1.30699468e+00 -6.29790664e-01 -1.21015817e-01 2.52127975e-01
-3.32984388e-01 7.15744197e-02 -1.24750936e+00 -7.57611990e-01
-1.11269677e+00 -8.14810276e-01 1.18429530e+00 -2.53832459e-01
-7.70788729e-01 -1.23024118e+00 -6.21949397e-02 3.38733107e-01
6.13303065e-01 -7.38283575e-01 7.90402353e-01 -1.24316382e+00
-1.24191619e-01 -4.69089746e-01 -2.91900914e-02 2.28985965e-01
-4.86718081e-02 -5.00978947e-01 -1.13567567e+00 -1.94714323e-01
8.11435580e-02 -7.50921011e-01 8.62382352e-01 2.97924101e-01
1.29813504e+00 -5.45066178e-01 -5.30998111e-01 3.27301264e-01
8.50541532e-01 3.50712687e-01 3.40714127e-01 -1.86859488e-01
7.47624576e-01 2.87201464e-01 1.28004998e-01 2.67347991e-01
2.39625707e-01 8.10769081e-01 -2.60613412e-01 2.29416981e-01
1.40721947e-01 -4.94601637e-01 8.37481081e-01 1.14251804e+00
7.67350018e-01 -8.21793199e-01 -1.01464820e+00 2.21165672e-01
-1.95464587e+00 -7.30996072e-01 3.33731249e-02 1.65462136e+00
6.37342095e-01 4.34644490e-01 -2.12848280e-02 3.42710242e-02
3.22778821e-01 3.99420232e-01 -6.48385286e-01 -8.37823212e-01
3.55852246e-01 5.42848825e-01 4.65880275e-01 4.50727642e-01
-1.07333505e+00 8.93678308e-01 7.19228220e+00 9.57239151e-01
-1.26349783e+00 5.87120950e-01 6.88869119e-01 -4.57003295e-01
-1.75385009e-02 -3.67496610e-01 -1.41169727e+00 4.89489585e-01
1.80682802e+00 2.03406855e-01 1.66985333e-01 1.10602748e+00
-1.94258511e-01 2.14347020e-01 -1.35518003e+00 1.11743546e+00
1.80333748e-01 -1.68263614e+00 -5.98798767e-02 -9.62925702e-02
3.33132863e-01 5.33356786e-01 2.26328179e-01 5.87398231e-01
5.51181138e-01 -1.10058165e+00 4.51429009e-01 1.68759555e-01
8.97148132e-01 -3.99362266e-01 8.22286665e-01 3.60884637e-01
-1.05543458e+00 -3.25651407e-01 -4.16393764e-02 -2.71763176e-01
2.91114539e-01 5.61367393e-01 -9.54410970e-01 -4.09371294e-02
4.45598841e-01 1.03838122e+00 -4.89369363e-01 -3.67925353e-02
2.48030648e-01 8.39038730e-01 -5.25885820e-01 -8.55768442e-01
2.55337328e-01 3.99285823e-01 2.20450442e-02 1.34688544e+00
1.38192803e-01 -2.08193675e-01 3.01483274e-01 3.23488623e-01
-4.12970662e-01 1.95418507e-01 -6.12755537e-01 -4.89539236e-01
1.24033973e-01 7.36521363e-01 -5.80084085e-01 -4.23189849e-01
-5.46352506e-01 1.27755690e+00 2.35341907e-01 1.95411250e-01
-6.62694156e-01 -4.64276940e-01 4.48870212e-01 -9.59508717e-02
3.64357948e-01 -2.78718263e-01 -6.44926846e-01 -1.31691849e+00
3.06650639e-01 -5.36204875e-01 4.44130421e-01 -6.91084445e-01
-1.19751513e+00 8.09884787e-01 -2.95050561e-01 -7.26275086e-01
-6.42904460e-01 -9.49624777e-01 -3.38073403e-01 6.70237362e-01
-8.32580566e-01 -1.25695455e+00 2.08787411e-01 3.19038272e-01
9.80545223e-01 -5.07755458e-01 1.36040175e+00 1.43911272e-01
-6.51394188e-01 9.30058241e-01 4.54052687e-01 1.62697554e-01
3.56004119e-01 -9.33036923e-01 1.07852519e+00 2.59711385e-01
1.12048730e-01 8.66994143e-01 3.31920922e-01 -6.11558139e-01
-1.57337987e+00 -9.37829018e-01 1.13115966e+00 -7.47141957e-01
9.48589683e-01 -1.13584614e+00 -5.51901698e-01 1.03665781e+00
4.89417881e-01 -1.82566136e-01 9.77611661e-01 7.81726956e-01
-5.08726656e-01 -1.44861266e-01 -6.46493256e-01 5.83801568e-01
9.50022638e-01 -1.28135526e+00 -4.86750543e-01 4.95429307e-01
1.24126136e+00 -2.13435844e-01 -8.89945805e-01 -1.91475581e-02
1.09037745e+00 -4.68187600e-01 8.26516092e-01 -4.81940269e-01
4.66138333e-01 4.65743631e-01 -3.01135182e-01 -9.48173821e-01
2.64820568e-02 -8.99499655e-01 -5.68367660e-01 1.20125568e+00
9.12315488e-01 -8.60490024e-01 1.03618443e+00 5.47601223e-01
-1.54768154e-01 -1.02210259e+00 -6.81095004e-01 -5.69700599e-01
1.92162797e-01 -7.19078124e-01 2.91510910e-01 7.07195997e-01
5.55389404e-01 1.16222489e+00 -6.54207468e-01 -4.40033257e-01
-1.54185638e-01 8.91349390e-02 5.51950395e-01 -1.14207494e+00
-6.09608710e-01 -5.00918090e-01 -1.34927869e-01 -1.79922676e+00
5.05516410e-01 -1.05281222e+00 -1.12260900e-01 -1.31785679e+00
-9.41037908e-02 -3.61275196e-01 -1.08117312e-01 5.86256266e-01
2.45864362e-01 -1.58661921e-02 -3.26407477e-02 1.22036077e-01
-7.67065406e-01 6.06094182e-01 3.10530156e-01 -3.87170911e-01
-4.33511555e-01 2.87998736e-01 -7.29389131e-01 3.81350547e-01
1.24210787e+00 -5.33136129e-01 -5.60599446e-01 -4.76982504e-01
4.83428568e-01 9.43982527e-02 -1.41239896e-01 -9.74766910e-01
3.81635219e-01 3.95507663e-01 1.44554853e-01 -9.67601538e-01
7.62585461e-01 -6.66702569e-01 -3.33666831e-01 4.26624060e-01
-9.11737740e-01 3.12198699e-01 4.26467806e-01 4.41478878e-01
1.17677085e-01 -1.58833861e-01 2.70128429e-01 2.85328835e-01
-4.07534838e-01 -3.78293241e-03 -9.85576689e-01 -2.96725035e-01
6.23246133e-01 4.82164435e-02 -5.42615712e-01 -4.40867156e-01
-6.47706926e-01 6.29301295e-02 -7.10615069e-02 8.66838217e-01
5.03412545e-01 -8.62292349e-01 -9.95038003e-02 2.08476514e-01
1.11968972e-01 -6.10050559e-01 4.70286757e-02 4.24167186e-01
-1.45501778e-01 1.20258772e+00 2.05931574e-01 -4.61426735e-01
-1.44914711e+00 6.76552653e-01 2.72078544e-01 -8.05474877e-01
-6.60148144e-01 6.94750249e-01 -1.88031912e-01 -8.75357687e-01
6.27993643e-01 -8.55394363e-01 -7.01669306e-02 4.11431631e-03
6.53870523e-01 2.18260556e-01 4.24062073e-01 -1.84932724e-02
-1.34531885e-01 2.48972163e-01 -6.15787804e-01 -3.15823674e-01
1.14287460e+00 -1.20450832e-01 1.60468847e-01 1.09350395e+00
1.62895048e+00 -1.82405517e-01 -6.03346884e-01 -3.07799727e-01
-1.05773948e-01 1.37070745e-01 3.30382466e-01 -1.23115575e+00
-4.91671592e-01 1.13265240e+00 7.36679971e-01 6.86598420e-02
8.02976608e-01 -1.34130895e-01 1.39277709e+00 1.24390113e+00
1.35202318e-01 -1.23905468e+00 3.18332583e-01 1.10500515e+00
5.62387228e-01 -9.43245053e-01 -1.50710672e-01 -2.37534776e-01
-5.57020068e-01 1.17726636e+00 4.15055275e-01 2.17143387e-01
1.21694720e+00 6.88963711e-01 -1.98491924e-02 -3.44706237e-01
-1.29433584e+00 3.22151005e-01 2.11601287e-01 1.40033886e-01
8.09946299e-01 -8.56087580e-02 -1.04257111e-02 6.06185555e-01
-3.32656622e-01 2.72876412e-01 1.25398561e-01 1.20437908e+00
-1.03401341e-01 -1.21392465e+00 3.50406051e-01 9.53624606e-01
-7.19086111e-01 -4.95787710e-01 -5.06425321e-01 1.73986912e-01
-4.47536618e-01 1.04063952e+00 2.55831599e-01 -9.66953695e-01
6.00122139e-02 7.15510607e-01 -4.90088668e-03 -7.36826956e-01
-1.17568398e+00 -3.69425505e-01 6.63360476e-01 -4.98651832e-01
-3.68091464e-02 -4.10756379e-01 -1.09378743e+00 -5.46161473e-01
-6.66743815e-01 -2.33939029e-02 8.50435674e-01 8.87482643e-01
9.07327473e-01 4.52186584e-01 4.09376264e-01 -7.57311285e-01
-7.37147510e-01 -1.26305413e+00 -1.46471620e-01 -1.21652726e-02
3.62430364e-01 -1.60402492e-01 -2.91583091e-02 -1.39569625e-01] | [10.776817321777344, 7.841664791107178] |
65c50443-02b5-4bdb-a100-b589afc4057c | dynasp25-dynamic-programming-on-tree | 1706.09370 | null | http://arxiv.org/abs/1706.09370v1 | http://arxiv.org/pdf/1706.09370v1.pdf | DynASP2.5: Dynamic Programming on Tree Decompositions in Action | A vibrant theoretical research area are efficient exact parameterized
algorithms. Very recent solving competitions such as the PACE challenge show
that there is also increasing practical interest in the parameterized
algorithms community. An important research question is whether dedicated
parameterized exact algorithms exhibit certain practical relevance and one can
even beat well-established problem solvers. We consider the logic-based
declarative modeling language and problem solving framework Answer Set
Programming (ASP). State-of-the-art ASP solvers rely considerably on Sat-based
algorithms. An ASP solver (DynASP2), which is based on a classical dynamic
programming on tree decompositions, has been published very recently.
Unfortunately, DynASP2 can outperform modern ASP solvers on programs of small
treewidth only if the question of interest is to count the number of solutions.
In this paper, we describe underlying concepts of our new implementation
(DynASP2.5) that shows competitive behavior to state-of-the-art ASP solvers
even for finding just one solution when solving problems as the Steiner tree
problem that have been modeled in ASP on graphs with low treewidth. Our
implementation is based on a novel approach that we call multi-pass dynamic
programming (M-DPSINC). | ['Johannes K. Fichte', 'Stefan Woltran', 'Markus Hecher', 'Michael Morak'] | 2017-06-28 | null | null | null | null | ['steiner-tree-problem'] | ['graphs'] | [-2.81592906e-02 7.96238720e-01 -3.42233390e-01 -3.36825222e-01
-4.83709544e-01 -7.44073272e-01 -2.34555732e-02 3.63230556e-01
4.19382453e-02 9.54427898e-01 -1.80499434e-01 -5.16762257e-01
-7.34462142e-01 -1.33469188e+00 -1.07040405e+00 -1.80019438e-01
-2.75389373e-01 1.38878107e+00 8.12607765e-01 -4.52892363e-01
2.98338141e-02 4.88678753e-01 -1.61065769e+00 7.26619363e-01
9.37147319e-01 8.58199477e-01 -2.37639531e-01 4.05803472e-01
-7.02016175e-01 5.20370245e-01 -3.04878294e-01 -8.10814500e-01
6.17048383e-01 -1.03460446e-01 -1.38974452e+00 -1.43003434e-01
3.78503591e-01 -1.50933415e-01 -1.14958622e-01 1.05189478e+00
-1.47013634e-01 -3.83311242e-01 -3.34373802e-01 -1.77123499e+00
9.77275148e-02 1.21792543e+00 -4.59282249e-01 4.27244371e-03
6.49271548e-01 -6.76739439e-02 1.36999178e+00 -3.24446619e-01
9.75260735e-01 1.17738533e+00 5.72683573e-01 3.27888578e-01
-1.30978119e+00 -1.56164050e-01 3.61037195e-01 8.70930612e-01
-1.23831534e+00 4.47416455e-02 4.92364079e-01 3.06754895e-02
1.36142242e+00 8.11097085e-01 1.12833452e+00 1.89297825e-01
-2.20506474e-01 7.61339247e-01 1.00252521e+00 -4.27119881e-01
5.22854149e-01 1.74972534e-01 5.60108423e-01 8.25562298e-01
8.33666086e-01 -3.66365165e-02 -6.38856471e-01 -3.67808551e-01
4.84877020e-01 -6.69363618e-01 -3.79645079e-01 -7.66879141e-01
-9.81603205e-01 1.10133147e+00 8.69291350e-02 1.83703825e-01
-1.55603051e-01 3.82172525e-01 4.83182311e-01 5.03859401e-01
1.89641491e-02 6.54695511e-01 -7.99275041e-01 -2.68911332e-01
-1.00774360e+00 9.08476830e-01 1.76024210e+00 1.24448752e+00
5.41328549e-01 -2.27537960e-01 3.19758952e-01 2.25257978e-01
9.07398239e-02 3.66328657e-02 -4.39779431e-01 -1.35969055e+00
6.93955481e-01 9.78822172e-01 2.80525535e-01 -8.53603065e-01
-3.49276900e-01 -2.46461093e-01 -3.32270451e-02 1.77560315e-01
4.77204084e-01 4.41535003e-02 -2.16693595e-01 1.41373384e+00
6.40769482e-01 -1.71379708e-02 4.81748022e-02 8.24970067e-01
6.46610200e-01 1.03096700e+00 -3.92478079e-01 -5.08900106e-01
1.11751008e+00 -1.25935388e+00 -4.79123712e-01 -1.55556083e-01
8.71186197e-01 -2.32365653e-01 4.48261231e-01 8.73304129e-01
-1.72303748e+00 4.13719118e-01 -1.16288161e+00 1.75886646e-01
-4.09092426e-01 -5.84801018e-01 1.30611348e+00 8.48704994e-01
-1.22304463e+00 3.95312697e-01 -7.43996561e-01 -5.29159784e-01
2.45715216e-01 7.88742721e-01 -2.29606479e-01 -4.90460664e-01
-8.65919530e-01 1.01554990e+00 8.29559445e-01 -1.80999488e-01
-3.27314645e-01 -1.09449697e+00 -5.61474442e-01 3.76395404e-01
1.24769378e+00 -9.24636960e-01 1.43711054e+00 -8.54625285e-01
-1.52346873e+00 1.12142086e+00 8.13579112e-02 -1.09755433e+00
5.63119352e-01 2.18879998e-01 -1.84345871e-01 3.89106631e-01
-4.10143226e-01 4.07939255e-01 2.30554994e-02 -9.02330160e-01
-5.77690303e-01 -2.75649518e-01 1.00757551e+00 3.90919447e-02
1.01117283e-01 3.86812598e-01 -1.42023340e-01 6.66298345e-02
9.54038650e-02 -7.93446422e-01 -3.75380546e-01 3.79788548e-01
-7.88164213e-02 -3.17278713e-01 5.64040184e-01 -1.22529179e-01
1.45178676e+00 -1.62554026e+00 6.61197186e-01 2.02502981e-01
5.74462339e-02 2.07588717e-01 -9.61919278e-02 7.94785023e-01
-3.00198402e-02 1.49616838e-01 -4.11529005e-01 5.30917466e-01
4.08542573e-01 7.30194986e-01 -3.78303707e-01 2.65391529e-01
1.02573834e-01 7.24465132e-01 -7.11267054e-01 -6.92618132e-01
-1.71660986e-02 -4.70030457e-01 -1.21549046e+00 -1.33358851e-01
-1.28279531e+00 -5.96296132e-01 -4.57759291e-01 7.47054040e-01
1.10938978e+00 -1.61146879e-01 5.58080018e-01 4.85026501e-02
-4.25997972e-01 2.32367426e-01 -1.62601662e+00 1.81536460e+00
-2.26627097e-01 2.34286785e-01 7.17429340e-01 -1.39962411e+00
6.61530972e-01 1.26409486e-01 6.49768531e-01 -3.49948317e-01
-2.13281751e-01 5.49145162e-01 2.64158044e-02 -3.42824340e-01
3.75119239e-01 6.84602261e-02 2.99035776e-02 3.59328151e-01
-2.40421072e-01 -6.94288969e-01 9.03042257e-01 2.69694507e-01
1.54004717e+00 1.72313273e-01 2.21111894e-01 -6.03859127e-01
6.25118852e-01 9.76546228e-01 6.41873956e-01 5.10251760e-01
3.42060804e-01 2.37846941e-01 1.08229303e+00 -9.37678337e-01
-8.93142343e-01 -7.63202250e-01 -1.66214198e-01 6.35769427e-01
1.85394108e-01 -1.11065888e+00 -8.52394283e-01 -4.98088866e-01
-6.04801960e-02 9.72261548e-01 -2.19909638e-01 2.02114895e-01
-8.76987576e-01 -7.63023436e-01 2.68919945e-01 3.46864641e-01
3.05566132e-01 -7.64587700e-01 -1.02818882e+00 5.86419225e-01
1.89080592e-02 -1.36194861e+00 5.15969753e-01 3.05326283e-01
-1.15789616e+00 -1.33644092e+00 -1.57821059e-01 -4.83718693e-01
3.68069649e-01 -4.96200919e-02 1.42607832e+00 9.75125432e-02
-2.85497874e-01 4.93786454e-01 -3.52894485e-01 -6.38114631e-01
-2.12687984e-01 1.06630340e-01 -4.01518762e-01 -5.82000434e-01
3.28354895e-01 -6.48755968e-01 -5.40167326e-03 3.44085574e-01
-7.94298232e-01 3.59439552e-01 2.46736437e-01 4.84595358e-01
8.09702516e-01 4.70939130e-01 -1.07435547e-01 -1.23476362e+00
1.56347647e-01 -4.07483339e-01 -1.34080136e+00 5.32809079e-01
-4.13988680e-01 1.65330231e-01 5.97800553e-01 1.41335085e-01
-6.64459169e-01 1.24610394e-01 -1.02768447e-02 -1.09297544e-01
2.92019218e-01 1.03986597e+00 -1.52776927e-01 -1.37126729e-01
4.15729672e-01 -1.03385486e-01 -2.88168788e-01 9.00174864e-03
2.12039530e-01 -1.03649564e-01 3.23711604e-01 -1.24513209e+00
7.71594942e-01 4.04482454e-01 6.02175772e-01 -4.43078548e-01
-8.77807260e-01 -2.59424865e-01 -7.13516073e-03 1.03837416e-01
3.42460930e-01 -4.04732972e-01 -6.21529818e-01 1.18766293e-01
-1.58218968e+00 -4.63899851e-01 -8.36589456e-01 -5.02185635e-02
-9.68250871e-01 -6.71096072e-02 -1.59822419e-01 -6.00997031e-01
-1.45545512e-01 -9.37829256e-01 7.46974170e-01 -4.85107861e-02
-7.86721334e-02 -6.58636510e-01 3.79718930e-01 4.00585592e-01
3.71352196e-01 5.35774052e-01 1.18712294e+00 -6.88464999e-01
-1.10645318e+00 -2.07884580e-01 -3.18365425e-01 -2.56030500e-01
-7.98206389e-01 3.73965681e-01 -2.85541385e-01 5.93008026e-02
-1.41421273e-01 -1.35153040e-01 3.06809358e-02 4.63989615e-01
1.34959674e+00 -6.45414412e-01 -3.60397786e-01 6.62352502e-01
1.94733143e+00 -2.32451126e-01 7.97015131e-01 6.50588751e-01
7.74099827e-02 7.03744471e-01 6.51624322e-01 4.85466063e-01
4.41688418e-01 8.86945784e-01 9.57102001e-01 5.75910568e-01
2.80784428e-01 1.02333009e-01 -5.05639520e-03 5.43496072e-01
-2.87710637e-01 -9.35633034e-02 -1.09465325e+00 5.25630176e-01
-2.18233323e+00 -1.03790128e+00 -6.97325349e-01 1.98277020e+00
8.95751894e-01 4.54526424e-01 2.68137902e-01 3.91699404e-01
5.01218736e-01 -3.98484757e-03 -2.34927520e-01 -1.35318983e+00
-1.56431213e-01 8.05123627e-01 5.80689788e-01 4.29297894e-01
-5.77397585e-01 8.06541264e-01 6.06544018e+00 6.14341915e-01
-7.28945374e-01 -5.75079061e-02 5.42977229e-02 -3.83176446e-01
-4.54631895e-01 4.15469021e-01 -7.90000856e-01 1.39078060e-02
1.35864758e+00 -9.01201844e-01 7.11188853e-01 1.28777361e+00
-1.84453934e-01 -2.12847725e-01 -1.48723888e+00 7.96190023e-01
-2.02289239e-01 -1.91589880e+00 -4.51866128e-02 1.80608422e-01
7.11994469e-01 -1.70508206e-01 -2.78599083e-01 3.52818877e-01
4.19826567e-01 -9.96384382e-01 8.93212259e-01 1.30026445e-01
2.92141557e-01 -7.57088125e-01 6.94303930e-01 3.70724350e-01
-1.06424987e+00 -2.90287524e-01 -6.07672811e-01 -2.11586297e-01
3.45739543e-01 6.61492109e-01 -3.74723375e-01 9.49872315e-01
7.08778560e-01 3.94895315e-01 -6.71744049e-02 1.64123118e+00
-6.37710616e-02 4.27387178e-01 -1.12567186e+00 -9.56397876e-02
4.03008401e-01 -1.82070687e-01 6.69054151e-01 9.50944185e-01
3.53619725e-01 6.34360790e-01 1.42413834e-02 9.99655545e-01
1.01652727e-01 3.63927037e-02 -2.98879713e-01 -1.65839583e-01
4.10365880e-01 1.08600485e+00 -6.30082071e-01 -4.28263936e-03
-5.42671144e-01 2.98514485e-01 2.81272203e-01 -9.91971418e-02
-1.17075431e+00 -8.51128176e-02 3.26051772e-01 5.35644531e-01
5.23459077e-01 -2.10938171e-01 -4.57402915e-01 -9.31833088e-01
3.18655282e-01 -1.10714030e+00 7.07188964e-01 -8.39139819e-01
-6.61488295e-01 3.89302969e-01 4.75086451e-01 -6.79767489e-01
-3.44745182e-02 -8.63257349e-01 -5.86311519e-01 3.55564594e-01
-1.49148095e+00 -9.37990785e-01 -4.17001694e-01 5.02837181e-01
4.33156580e-01 3.69444400e-01 1.02726662e+00 1.25107229e-01
-4.65686321e-01 3.18012118e-01 -3.42067152e-01 -7.56613135e-01
-9.22918320e-03 -1.28637636e+00 -1.41431749e-01 6.75442278e-01
1.04951374e-01 1.25639871e-01 1.05729318e+00 -3.70228365e-02
-2.27104306e+00 -7.03389168e-01 8.12825859e-01 -1.05770238e-01
1.02503157e+00 -2.81704683e-02 -6.34589970e-01 7.90688038e-01
-8.24812613e-03 2.89770782e-01 2.15654790e-01 4.59627137e-02
-3.12482268e-01 -3.79408777e-01 -1.31891835e+00 3.38869333e-01
1.28451383e+00 1.69024631e-01 -6.79904163e-01 7.07191765e-01
8.38997543e-01 -8.15370858e-01 -6.98012829e-01 3.59140873e-01
3.56766284e-01 -1.25941634e+00 7.57737875e-01 -6.37646496e-01
6.04302824e-01 -2.42867619e-01 -4.41157073e-01 -8.78921330e-01
1.78524870e-02 -9.70403612e-01 -7.16377437e-01 7.38190472e-01
3.67109001e-01 -7.37896264e-01 1.22644567e+00 8.30008507e-01
-3.30387771e-01 -1.29219413e+00 -1.05766046e+00 -1.10999048e+00
1.70610040e-01 -4.44831192e-01 9.34533417e-01 6.80330753e-01
5.14157116e-01 -8.55442882e-02 4.66755450e-01 2.43967995e-01
5.42827368e-01 7.15499997e-01 7.52644300e-01 -1.57949650e+00
-8.03497493e-01 -5.99722624e-01 -5.34970105e-01 -7.90373862e-01
1.04460359e-01 -8.50759327e-01 -5.89988530e-01 -1.78529119e+00
6.45968989e-02 -6.08931601e-01 2.39062518e-01 5.02024114e-01
7.86842465e-01 -3.67450714e-01 5.21678850e-02 -2.49211535e-01
-9.09412503e-01 3.23637645e-03 9.90950406e-01 -1.82280704e-01
-5.30141853e-02 6.51705936e-02 -4.98484850e-01 6.86369717e-01
5.62415838e-01 -5.67250788e-01 -2.78042138e-01 -6.07413054e-01
1.12374198e+00 4.71343875e-01 1.69762164e-01 -1.02204943e+00
5.40480077e-01 -7.18021035e-01 -9.20421779e-01 -5.34157395e-01
3.47443551e-01 -1.30895543e+00 7.28826702e-01 7.45609641e-01
2.78477352e-02 6.33854121e-02 4.61706430e-01 2.07603619e-01
-3.56314093e-01 -9.70432997e-01 3.85002702e-01 -4.04709041e-01
-5.72742820e-01 4.73177165e-01 -1.77431330e-01 2.66842902e-01
1.67943609e+00 -4.42211777e-01 -5.36354542e-01 -1.42474234e-01
-6.43241286e-01 5.05639434e-01 5.49395502e-01 -3.96235585e-01
2.64805436e-01 -6.51994288e-01 -8.66181791e-01 -4.97668445e-01
5.99715561e-02 2.90974587e-01 -1.75624024e-02 9.52799380e-01
-1.38103437e+00 8.58881116e-01 -2.51433849e-01 -3.14769536e-01
-1.26930356e+00 8.33212256e-01 4.22817826e-01 -7.08793461e-01
-6.09605253e-01 8.14081907e-01 -2.60317206e-01 -1.93172336e-01
2.03534365e-01 -5.43116927e-01 4.93111908e-01 -2.50385165e-01
2.46183783e-01 8.01285744e-01 3.26127857e-01 1.89254776e-01
-6.67628050e-01 4.14672852e-01 -3.64755243e-02 1.38187945e-01
2.03578734e+00 4.43917632e-01 -7.57056832e-01 -8.60694945e-02
4.33607936e-01 -3.78704220e-02 -5.49302042e-01 1.82693768e-02
1.19361937e-01 -2.95678735e-01 6.78674038e-03 -9.68132854e-01
-1.12282515e+00 4.30215061e-01 -2.00897947e-01 7.39298820e-01
1.26729918e+00 1.11110590e-01 6.26045108e-01 8.57296526e-01
1.20265329e+00 -1.09663665e+00 -5.76227963e-01 4.33809072e-01
1.02792871e+00 -6.38600707e-01 6.67650938e-01 -1.38138556e+00
-1.62829816e-01 1.50794077e+00 6.19739652e-01 -3.55401605e-01
3.87621760e-01 9.12980437e-01 -7.72761941e-01 -4.22835648e-01
-1.25920165e+00 1.06793649e-01 -4.87619042e-01 4.55500126e-01
-1.99645400e-01 8.11148807e-02 -7.95912802e-01 8.88771892e-01
-4.92819548e-01 1.92890555e-01 8.25052798e-01 1.10895264e+00
-5.44959009e-01 -1.57440054e+00 -3.87510031e-01 2.68508673e-01
-3.62691194e-01 2.41117198e-02 -5.45629621e-01 1.11087084e+00
-7.90936500e-02 6.30014479e-01 -8.82920623e-02 1.59223944e-01
3.48976463e-01 -1.07253730e-01 1.32575512e+00 -8.01175654e-01
-7.38311648e-01 -8.09549987e-01 8.62758100e-01 -9.56941605e-01
-4.64230865e-01 -6.51607811e-01 -1.35218191e+00 -7.17674971e-01
-4.00268734e-01 4.35745180e-01 7.38236845e-01 6.84580743e-01
7.46813118e-02 1.37909532e-01 8.87703374e-02 -1.36129871e-01
-7.45670915e-01 -1.35095222e-02 -5.25774181e-01 -4.74013388e-01
-4.57921743e-01 -3.96838158e-01 -2.99856886e-02 -2.18635619e-01] | [8.532633781433105, 6.632782936096191] |
c2693c27-59f9-4e28-8086-9a1153ef2f72 | contrastive-audio-language-learning-for-music | 2208.12208 | null | https://arxiv.org/abs/2208.12208v1 | https://arxiv.org/pdf/2208.12208v1.pdf | Contrastive Audio-Language Learning for Music | As one of the most intuitive interfaces known to humans, natural language has the potential to mediate many tasks that involve human-computer interaction, especially in application-focused fields like Music Information Retrieval. In this work, we explore cross-modal learning in an attempt to bridge audio and language in the music domain. To this end, we propose MusCALL, a framework for Music Contrastive Audio-Language Learning. Our approach consists of a dual-encoder architecture that learns the alignment between pairs of music audio and descriptive sentences, producing multimodal embeddings that can be used for text-to-audio and audio-to-text retrieval out-of-the-box. Thanks to this property, MusCALL can be transferred to virtually any task that can be cast as text-based retrieval. Our experiments show that our method performs significantly better than the baselines at retrieving audio that matches a textual description and, conversely, text that matches an audio query. We also demonstrate that the multimodal alignment capability of our model can be successfully extended to the zero-shot transfer scenario for genre classification and auto-tagging on two public datasets. | ['György Fazekas', 'Elio Quinton', 'Emmanouil Benetos', 'Ilaria Manco'] | 2022-08-25 | null | null | null | null | ['genre-classification', 'music-information-retrieval'] | ['computer-vision', 'music'] | [ 2.74715543e-01 -3.28186862e-02 -1.67146355e-01 -2.36510932e-01
-1.51961672e+00 -6.86874866e-01 6.61110818e-01 3.74105334e-01
-3.58306885e-01 2.36475408e-01 5.91368139e-01 5.40073104e-02
-2.78618515e-01 -5.04953384e-01 -6.12275541e-01 -4.11372542e-01
-7.67257437e-02 5.73440731e-01 3.33927423e-02 -3.41828614e-01
9.92536917e-02 8.13668743e-02 -1.94568777e+00 7.85989285e-01
8.40753317e-02 1.15251124e+00 1.08294629e-01 8.89752746e-01
-2.22630322e-01 8.04510236e-01 -2.85100549e-01 -4.19827521e-01
-8.43948051e-02 -2.01091304e-01 -1.13904130e+00 -2.96212316e-01
6.42868936e-01 -1.69047341e-01 -2.48428181e-01 6.00138724e-01
8.49601328e-01 3.41729373e-01 7.50929296e-01 -1.17673075e+00
-4.87014025e-01 8.78006339e-01 -1.89811245e-01 -2.58847147e-01
9.01384711e-01 -2.97788203e-01 1.57345843e+00 -1.01300335e+00
5.45346260e-01 1.43531549e+00 6.25580013e-01 4.14679080e-01
-1.46553540e+00 -6.53254151e-01 -2.66528666e-01 3.06919008e-01
-1.49231696e+00 -7.66473770e-01 7.79741704e-01 -5.41077018e-01
7.95619607e-01 3.45019698e-01 5.51709414e-01 1.21679688e+00
-2.33908653e-01 9.68728244e-01 4.30226088e-01 -6.70017064e-01
-7.92222694e-02 1.20079227e-01 -2.12685674e-01 3.52943659e-01
-5.61429501e-01 -1.97643518e-01 -1.25675535e+00 -2.06428453e-01
3.67721051e-01 -2.32242361e-01 -2.28717759e-01 -4.23760653e-01
-1.29980183e+00 7.83570588e-01 3.03745687e-01 6.30334198e-01
-8.41560587e-02 4.27287936e-01 8.88539612e-01 5.70365548e-01
2.62592852e-01 7.57417142e-01 -2.58519411e-01 -3.27566177e-01
-1.00810564e+00 1.85415983e-01 7.74522066e-01 8.65711451e-01
5.38080752e-01 -1.52781174e-01 -2.74110466e-01 1.09567618e+00
2.61699587e-01 4.09079671e-01 6.03301167e-01 -9.54843104e-01
2.01160729e-01 1.40609384e-01 -1.51473626e-01 -7.23307848e-01
-1.60380661e-01 -1.45098418e-01 -5.08704662e-01 -2.65039623e-01
2.48498898e-02 2.78790295e-01 -1.74809620e-01 1.85160673e+00
7.30578005e-02 2.39106238e-01 3.64098921e-02 7.71189213e-01
9.69685256e-01 7.27994561e-01 2.89608408e-02 7.95519128e-02
1.54018795e+00 -7.64542699e-01 -5.95357597e-01 7.27121979e-02
5.20835757e-01 -1.10846210e+00 1.47736299e+00 3.53835344e-01
-1.11595285e+00 -6.36887372e-01 -9.28842247e-01 -4.47030306e-01
-4.03114021e-01 2.26101428e-02 4.08269286e-01 1.55881241e-01
-9.76472318e-01 6.11126065e-01 -5.44338942e-01 -6.20639265e-01
1.96633324e-01 1.96687952e-01 -5.36097407e-01 1.90755665e-01
-1.15628982e+00 4.72885281e-01 4.97874647e-01 -3.52105260e-01
-7.62891173e-01 -8.70478511e-01 -9.41304028e-01 2.68098921e-01
2.53918648e-01 -7.83119082e-01 1.58017516e+00 -9.77518916e-01
-1.45070875e+00 9.44509864e-01 -1.04317982e-02 -3.49698991e-01
-1.10194907e-01 -4.07896310e-01 -4.49214220e-01 4.44253266e-01
2.58279562e-01 1.01169181e+00 9.45976079e-01 -1.00078881e+00
-4.17298675e-01 -2.15997592e-01 1.32481068e-01 2.27825969e-01
-8.50538850e-01 1.60792619e-01 -5.38626313e-01 -7.59819865e-01
-2.78979033e-01 -8.78095567e-01 3.76181632e-01 6.25245795e-02
-4.47043270e-01 -4.16917264e-01 4.55261290e-01 -2.31164560e-01
1.31892800e+00 -2.48451304e+00 4.85236406e-01 -2.08789669e-03
-1.53125837e-01 -1.07634850e-01 -4.92183864e-01 1.04549742e+00
-3.16327624e-02 9.51911695e-03 1.02628712e-02 -5.26060760e-01
5.18398404e-01 -1.61210567e-01 -7.14024305e-01 8.96884575e-02
1.66512534e-01 1.00656760e+00 -9.85703886e-01 -6.80169344e-01
-5.12089301e-03 6.55446947e-01 -6.45636976e-01 4.22985584e-01
-2.36647740e-01 3.73830140e-01 -1.41273648e-01 4.74284053e-01
-1.80362031e-01 -3.51921357e-02 2.76074558e-02 -3.25700551e-01
6.89077452e-02 5.56424439e-01 -9.03514802e-01 2.55091262e+00
-9.34587836e-01 9.16979611e-01 3.46542187e-02 -7.98437417e-01
5.65955698e-01 7.98517585e-01 7.20812857e-01 -6.12489283e-01
-5.10323308e-02 2.81243831e-01 -3.00888091e-01 -4.87247080e-01
6.34985447e-01 -3.56017470e-01 -5.10082126e-01 7.88691282e-01
5.52647173e-01 -3.64310354e-01 6.69176364e-03 2.36456484e-01
9.52286243e-01 1.89719260e-01 1.42876431e-01 8.60517099e-02
5.44080853e-01 -3.01060915e-01 -2.13310167e-01 6.46897733e-01
3.35603505e-01 4.89280462e-01 9.75482762e-02 -1.15626834e-01
-9.88055170e-01 -1.20050597e+00 -2.44774029e-01 1.94810224e+00
-2.62443900e-01 -9.21125889e-01 -4.35444444e-01 -1.05262056e-01
2.46435944e-02 4.71478254e-01 -2.98079252e-01 -3.12841386e-01
-1.83184326e-01 2.39627033e-01 1.03584445e+00 4.18282211e-01
-1.21781819e-01 -1.23073912e+00 -3.92453969e-01 2.74114460e-01
-3.93767685e-01 -9.83726084e-01 -7.27598310e-01 1.53805494e-01
-4.97715831e-01 -7.96504974e-01 -7.43945062e-01 -1.05269206e+00
-2.86565930e-01 1.64546132e-01 1.37530816e+00 -2.28382885e-01
-4.41579878e-01 1.07680476e+00 -5.89799643e-01 -5.01670659e-01
-3.67603123e-01 3.89646769e-01 1.19192950e-01 1.69943690e-01
2.93404281e-01 -9.21511412e-01 -4.47937608e-01 9.63076204e-02
-1.22417963e+00 -1.86765492e-01 3.77884656e-01 8.69060755e-01
4.79962051e-01 -3.86507601e-01 8.94848585e-01 -4.82736439e-01
7.88934231e-01 -3.43656659e-01 -3.22662760e-03 1.73269808e-01
-4.33880091e-02 8.85770991e-02 4.24631715e-01 -6.81076825e-01
-4.95186985e-01 2.25753576e-01 -2.04388589e-01 -6.28916740e-01
-2.12066546e-01 7.71022260e-01 -1.74526080e-01 2.21831456e-01
4.99895632e-01 1.07401296e-01 -1.33021310e-01 -6.87745988e-01
8.06386948e-01 1.15133929e+00 8.41126800e-01 -7.17151344e-01
7.05462933e-01 2.93845266e-01 -1.12636641e-01 -1.08858573e+00
-9.41161752e-01 -9.10939574e-01 -6.20057821e-01 -3.34238499e-01
1.05845571e+00 -1.09926093e+00 -8.77540410e-01 -1.51633769e-01
-1.15161657e+00 -1.31502688e-01 -4.44930017e-01 4.81768131e-01
-1.01432252e+00 2.16409236e-01 -6.45457685e-01 -9.62369800e-01
-4.56439227e-01 -7.53144801e-01 1.71429110e+00 -2.96793461e-01
-6.94316983e-01 -8.14126730e-01 3.69700193e-01 3.19478780e-01
3.67887497e-01 -3.06251317e-01 1.21366215e+00 -8.61760497e-01
-3.37335765e-01 -2.37566590e-01 7.48054609e-02 1.67038757e-02
-2.02744529e-01 -4.29641247e-01 -1.57593679e+00 -3.78431231e-01
-4.20184702e-01 -1.11016130e+00 9.44911122e-01 -8.39869082e-02
1.12801456e+00 -1.47826910e-01 3.64538729e-02 2.92339027e-01
1.06430304e+00 -9.70881134e-02 5.81136584e-01 2.08811447e-01
5.68954825e-01 6.68972015e-01 5.74266672e-01 4.97977495e-01
2.96596497e-01 1.22201240e+00 2.53439069e-01 7.21897781e-02
-1.59986719e-01 -6.24961257e-01 4.68318760e-01 1.18942297e+00
4.00356174e-01 -3.22534181e-02 -7.27069795e-01 6.64471805e-01
-1.74645889e+00 -1.29463232e+00 3.96509200e-01 2.22677898e+00
1.13186038e+00 -3.11725557e-01 3.40001404e-01 3.68096769e-01
3.66862625e-01 1.09422229e-01 -1.41487211e-01 -3.62998515e-01
2.80191809e-01 5.48346162e-01 -3.40600997e-01 2.87533313e-01
-1.38705373e+00 8.24634075e-01 5.62609720e+00 1.00679433e+00
-1.07341218e+00 1.89668491e-01 -1.01889163e-01 -3.23682815e-01
-3.14591855e-01 -2.58549809e-01 -2.98725635e-01 4.47200201e-02
1.21827638e+00 -1.58038735e-01 6.76554739e-01 5.14014602e-01
-6.17993176e-02 3.91771406e-01 -1.89154458e+00 1.30246472e+00
2.73600399e-01 -1.09206223e+00 3.29533458e-01 -1.48251936e-01
2.84049124e-01 -2.13536352e-01 2.25668520e-01 6.04440928e-01
-3.92902672e-01 -1.15176392e+00 8.52845967e-01 5.56378365e-01
1.07710767e+00 -6.56517386e-01 5.20312786e-01 1.86621428e-01
-1.53509915e+00 -5.75705804e-02 -9.27633792e-02 1.05014130e-01
2.18530044e-01 9.04329568e-02 -9.46366608e-01 6.26496732e-01
6.90981030e-01 8.65466595e-01 -3.09091270e-01 1.06067705e+00
1.39511988e-01 3.16940635e-01 -8.50991532e-02 3.48863490e-02
-4.44798395e-02 2.14526117e-01 6.71068311e-01 1.42058945e+00
4.85451460e-01 -4.01292205e-01 3.53729844e-01 7.27088273e-01
-2.85125703e-01 5.94588637e-01 -1.01053047e+00 -5.06178916e-01
6.29615247e-01 9.76653755e-01 -1.09760471e-01 -2.45311767e-01
-3.08328658e-01 1.12060666e+00 2.06518680e-01 1.62697002e-01
-5.19955337e-01 -7.38127828e-01 4.82878447e-01 6.16324656e-02
3.27146769e-01 -7.68260732e-02 2.72275925e-01 -1.01819897e+00
2.37196498e-03 -9.60186064e-01 3.15046817e-01 -1.04610705e+00
-1.60808313e+00 5.28031349e-01 -8.65795389e-02 -1.62855148e+00
-5.59204400e-01 -3.98452312e-01 -3.81105453e-01 5.65245628e-01
-1.23525882e+00 -1.38420129e+00 1.21539243e-01 7.78526723e-01
6.01568520e-01 -3.29350263e-01 1.49239576e+00 7.28448987e-01
1.37526631e-01 6.70561910e-01 -1.68255821e-01 2.16099173e-01
1.30370688e+00 -1.16440058e+00 -1.56805113e-01 6.84770644e-02
1.01530015e+00 6.56851828e-01 5.02538204e-01 -3.41985072e-03
-1.57265139e+00 -9.87937212e-01 9.99255896e-01 -3.60893965e-01
1.06920862e+00 -6.19947076e-01 -7.66522169e-01 5.70309699e-01
5.42600155e-01 -3.45851392e-01 1.36292851e+00 5.04949510e-01
-8.12132120e-01 -2.60720313e-01 -3.42647672e-01 5.04944146e-01
6.98161662e-01 -1.60373080e+00 -7.45315075e-01 2.78652966e-01
9.54579294e-01 -1.06609583e-01 -1.11628616e+00 1.12184197e-01
1.00176692e+00 -4.79636878e-01 1.18509150e+00 -9.31392729e-01
6.14811361e-01 -1.29346520e-01 -5.67346513e-01 -1.15676641e+00
2.83803921e-02 -7.12017238e-01 1.45008132e-01 1.58726835e+00
5.10739565e-01 7.39551112e-02 8.74983296e-02 -1.64407089e-01
-2.81419545e-01 -3.26907665e-01 -1.18236566e+00 -7.18264461e-01
5.52162789e-02 -9.16991830e-01 3.77757609e-01 8.05506408e-01
5.88310778e-01 1.02156901e+00 -5.20580232e-01 -9.68350023e-02
1.86397314e-01 3.32141340e-01 9.47168410e-01 -1.59121263e+00
-7.47479439e-01 -4.86891538e-01 -6.05191350e-01 -1.05309474e+00
4.85880077e-01 -1.42533267e+00 2.50408024e-01 -1.12592793e+00
2.88037479e-01 -1.16570614e-01 -3.77144992e-01 5.18678188e-01
4.27846998e-01 6.18445039e-01 5.07172227e-01 1.40675709e-01
-1.02094889e+00 6.64519668e-01 9.08057153e-01 -5.80762029e-01
-1.38516381e-01 -1.74388930e-01 -5.63680649e-01 4.84853595e-01
4.42559153e-01 -4.07879472e-01 -2.92383641e-01 -3.95843178e-01
4.99547571e-01 3.74005079e-01 3.58875662e-01 -1.02975142e+00
3.07755709e-01 3.71255100e-01 -1.18462741e-01 -3.90832573e-01
8.97594631e-01 -9.25261080e-01 -5.05115725e-02 -1.39866486e-01
-1.00825763e+00 -1.12752527e-01 3.69095474e-01 6.31062984e-01
-6.89961135e-01 -3.10528398e-01 3.41457218e-01 2.10283443e-01
-2.78747737e-01 2.17215531e-02 -4.98204798e-01 1.98206007e-01
4.41428035e-01 3.15007746e-01 -1.95490029e-02 -8.57945740e-01
-8.74581099e-01 3.35542038e-02 1.90456122e-01 6.32804990e-01
5.70688486e-01 -1.80253828e+00 -5.34424663e-01 2.13634633e-02
7.64376581e-01 -5.81544995e-01 3.91677283e-02 9.15845096e-01
1.22956343e-01 8.34989667e-01 -3.02531645e-02 -7.61819124e-01
-1.49422228e+00 4.68920529e-01 4.50273082e-02 -2.53516603e-02
-4.83635455e-01 6.69343114e-01 3.51882353e-02 -4.37604815e-01
8.22863936e-01 -4.43277918e-02 -1.08958341e-01 3.98012370e-01
7.15956986e-01 -2.49950346e-02 4.39344086e-02 -7.23913133e-01
-2.36384213e-01 7.29311824e-01 3.31004530e-01 -7.20497847e-01
1.35683858e+00 -1.39600500e-01 -2.08087295e-01 1.28957283e+00
1.50324965e+00 2.32004851e-01 -3.72628361e-01 -2.97722518e-01
1.75455377e-01 -1.64735720e-01 8.32161903e-02 -6.98722243e-01
-5.34573376e-01 1.31168151e+00 6.20204389e-01 3.83678079e-01
1.03910220e+00 4.34466034e-01 9.30278957e-01 7.93089509e-01
3.52262765e-01 -1.00700903e+00 4.50268924e-01 5.58838725e-01
9.58742559e-01 -1.08440459e+00 -4.57178980e-01 6.54705688e-02
-5.65371752e-01 1.26922071e+00 8.28227669e-04 1.11512758e-01
6.01894677e-01 -3.22982147e-02 2.83093136e-02 -2.18791828e-01
-1.08923781e+00 -4.20374513e-01 8.16320300e-01 4.70064759e-01
9.65046227e-01 -1.05931833e-01 1.55320957e-01 7.22566664e-01
-3.11748356e-01 -5.33937514e-02 9.33976248e-02 7.79138684e-01
-4.09196168e-01 -1.32244563e+00 -2.74749160e-01 7.16963485e-02
-4.49669480e-01 -2.93259889e-01 -7.69038916e-01 5.02795696e-01
-1.49794489e-01 1.01156926e+00 2.11875126e-01 -5.29423237e-01
3.00185323e-01 7.00607240e-01 4.92502749e-01 -8.67374897e-01
-6.94441855e-01 4.61308360e-01 2.09047914e-01 -6.65447295e-01
-6.04131699e-01 -5.05127490e-01 -1.00855923e+00 1.89782277e-01
-1.68388367e-01 3.65749240e-01 5.79634607e-01 7.78873265e-01
3.33327234e-01 6.70671225e-01 4.47328657e-01 -1.16818321e+00
-4.38247204e-01 -1.10093260e+00 -6.39048755e-01 5.47300816e-01
4.37865794e-01 -5.91743350e-01 -1.77601770e-01 1.68243647e-01] | [15.506377220153809, 5.0801920890808105] |
f786fdd9-bea2-4de3-bc26-981e1af54978 | budget-sensitive-reannotation-of-noisy | 2112.13320 | null | https://arxiv.org/abs/2112.13320v1 | https://arxiv.org/pdf/2112.13320v1.pdf | Budget Sensitive Reannotation of Noisy Relation Classification Data Using Label Hierarchy | Large crowd-sourced datasets are often noisy and relation classification (RC) datasets are no exception. Reannotating the entire dataset is one probable solution however it is not always viable due to time and budget constraints. This paper addresses the problem of efficient reannotation of a large noisy dataset for the RC. Our goal is to catch more annotation errors in the dataset while reannotating fewer instances. Existing work on RC dataset reannotation lacks the flexibility about how much data to reannotate. We introduce the concept of a reannotation budget to overcome this limitation. The immediate follow-up problem is: Given a specific reannotation budget, which subset of the data should we reannotate? To address this problem, we present two strategies to selectively reannotate RC datasets. Our strategies utilize the taxonomic hierarchy of relation labels. The intuition of our work is to rely on the graph distance between actual and predicted relation labels in the label hierarchy graph. We evaluate our reannotation strategies on the well-known TACRED dataset. We design our experiments to answer three specific research questions. First, does our strategy select novel candidates for reannotation? Second, for a given reannotation budget is our reannotation strategy more efficient at catching annotation errors? Third, what is the impact of data reannotation on RC model performance measurement? Experimental results show that our both reannotation strategies are novel and efficient. Our analysis indicates that the current reported performance of RC models on noisy TACRED data is inflated. | ['Amit Awekar', 'Ashish Anand', 'Akshay Parekh'] | 2021-12-26 | null | null | null | null | ['relation-classification'] | ['natural-language-processing'] | [ 1.80915907e-01 5.25672495e-01 -5.16981408e-02 -2.81749994e-01
-7.00441897e-01 -4.68777210e-01 1.84892461e-01 2.60386109e-01
-4.92021531e-01 9.92804825e-01 3.82766277e-01 -3.17640722e-01
-3.47704411e-01 -8.26502025e-01 -4.42924172e-01 -6.12807572e-01
3.25040311e-01 5.57359219e-01 1.53150201e-01 -3.41939896e-01
2.68531650e-01 3.69302213e-01 -1.74034095e+00 2.79329479e-01
8.05484176e-01 8.21965694e-01 4.74456027e-02 6.21841908e-01
-1.89117372e-01 1.18171740e+00 -1.07233250e+00 -6.51137650e-01
5.08833110e-01 -5.35667598e-01 -1.32580161e+00 2.43722677e-01
2.07465783e-01 2.53932446e-01 1.52402431e-01 1.01325703e+00
6.79544210e-01 2.46735930e-01 2.62705475e-01 -1.29350817e+00
-4.85581160e-01 1.27472830e+00 -4.42372531e-01 4.23292190e-01
2.85483748e-01 -5.28290756e-02 1.34758770e+00 -8.50852191e-01
1.01398146e+00 9.57347929e-01 8.54639053e-01 3.77667338e-01
-1.33928144e+00 -2.81239837e-01 -1.81134090e-01 -1.67262014e-02
-1.70729947e+00 -3.12297702e-01 6.23215854e-01 -3.40683490e-01
8.63394678e-01 6.78627789e-01 3.18010867e-01 7.23491371e-01
-2.39644557e-01 5.20469606e-01 1.08622468e+00 -5.69969535e-01
4.98824984e-01 1.07379898e-01 2.31837928e-01 2.77753592e-01
7.11348832e-01 -2.78348058e-01 -8.22706819e-01 -4.65242505e-01
2.65012473e-01 -2.48264372e-01 -4.03045416e-01 -1.98083401e-01
-1.03794074e+00 7.61932671e-01 5.71763992e-01 3.64745259e-01
-3.99560988e-01 8.33588243e-02 5.35880744e-01 3.33676189e-01
7.53195167e-01 1.20821118e+00 -7.75820851e-01 -3.08889031e-01
-6.68107808e-01 1.90280661e-01 8.65797818e-01 9.69425261e-01
8.83345544e-01 -3.11985135e-01 -7.23885670e-02 1.22749841e+00
-1.06357649e-01 1.04801238e-01 5.52868009e-01 -9.70825195e-01
5.14810920e-01 8.71769965e-01 1.29691228e-01 -9.67654288e-01
-6.18811488e-01 -2.88565993e-01 -4.63300228e-01 -2.67188966e-01
5.92361152e-01 -4.84825641e-01 -6.26828372e-01 1.51920724e+00
4.79961008e-01 -3.29284221e-01 1.83170378e-01 1.05847096e+00
8.93441141e-01 2.12916061e-01 2.16287836e-01 -3.04235786e-01
1.26296711e+00 -6.82065427e-01 -7.51289666e-01 -1.74305797e-01
1.34780729e+00 -9.18935895e-01 1.13287079e+00 1.76992953e-01
-4.81358886e-01 -1.20930642e-01 -6.86389089e-01 -2.48117447e-01
-4.42384571e-01 3.21155608e-01 7.42263138e-01 7.26998925e-01
-7.76011348e-01 6.81962490e-01 -3.19231361e-01 -7.51231968e-01
2.73969024e-01 1.95119888e-01 -5.74795485e-01 -5.03620468e-02
-1.12617612e+00 1.06560659e+00 4.80511278e-01 3.38386223e-02
-1.95893377e-01 -4.83202159e-01 -5.87232113e-01 -5.10317720e-02
9.61116135e-01 -4.11308557e-01 1.23866570e+00 -1.03759360e+00
-1.07282341e+00 8.36872399e-01 -1.29870608e-01 -4.39777762e-01
3.72473508e-01 -2.07310580e-02 -1.29763097e-01 -1.75238505e-01
2.50022948e-01 3.65505517e-01 1.91689461e-01 -1.37385440e+00
-6.38956487e-01 -3.51996720e-01 -5.74741438e-02 2.97534555e-01
-1.57849029e-01 -2.48443186e-02 6.45352304e-02 -6.50894225e-01
3.95688534e-01 -1.04778469e+00 -2.49937072e-01 -4.61912602e-01
-6.22607887e-01 -4.88319814e-01 5.04915178e-01 -2.21176192e-01
1.26483810e+00 -1.78998053e+00 6.81484342e-02 8.82774219e-02
4.86181945e-01 1.81520060e-01 -4.49853987e-02 5.08318067e-01
-1.23235054e-01 5.97081661e-01 -1.25665441e-01 -1.74894691e-01
-4.00351435e-01 4.00050074e-01 -1.93443745e-01 4.01632935e-01
-7.08998963e-02 7.32693732e-01 -1.12717628e+00 -4.23269957e-01
-2.66406208e-01 -9.60467532e-02 -3.67331177e-01 2.45653614e-01
-1.41184837e-01 -1.83587104e-01 -1.02983005e-01 8.30658078e-01
6.05146825e-01 -2.43508741e-01 4.33376193e-01 -2.21758023e-01
1.65612042e-01 4.44869369e-01 -1.13576055e+00 1.06960642e+00
-1.16578750e-01 5.99495173e-01 -3.06655139e-01 -7.42155552e-01
1.17814970e+00 5.23807146e-02 4.22950685e-01 -2.38740712e-01
3.49277794e-01 6.09613180e-01 1.95155680e-01 -6.03317022e-01
1.09209895e+00 -3.02686960e-01 -1.70856029e-01 6.11629248e-01
-7.93251172e-02 -1.42725050e-01 1.50167033e-01 4.50557947e-01
1.56514347e+00 -1.68381557e-01 6.03464365e-01 -3.32248569e-01
-5.13371676e-02 5.72044134e-01 8.05034697e-01 8.55824232e-01
-3.68675053e-01 8.07327509e-01 9.28461730e-01 -4.48176444e-01
-9.51487124e-01 -3.86284679e-01 9.37809609e-03 1.19478083e+00
1.41300812e-01 -5.96519470e-01 -5.64105511e-01 -1.15169990e+00
-5.70429042e-02 7.21694052e-01 -8.19141805e-01 -9.14566859e-04
-2.69003242e-01 -1.04633033e+00 7.25132465e-01 2.85809994e-01
2.46890351e-01 -1.04758072e+00 -6.42490208e-01 1.58549808e-02
-4.66810197e-01 -8.98297966e-01 -2.31311858e-01 4.27635133e-01
-6.32936895e-01 -1.24564719e+00 -3.08457404e-01 -3.30424100e-01
8.66600037e-01 6.63466513e-01 1.31318903e+00 3.65709871e-01
-5.01877218e-02 1.21334232e-01 -8.62771928e-01 -4.25083458e-01
-3.67519259e-01 4.92442191e-01 -5.35204709e-02 -2.34189451e-01
9.15728569e-01 -3.02000463e-01 -3.37100327e-01 4.86743152e-01
-9.44757462e-01 -2.60346681e-01 4.14112210e-01 8.61990392e-01
4.66625988e-01 -2.04222295e-02 1.02646935e+00 -1.58804679e+00
9.83139753e-01 -7.65061319e-01 -2.64229774e-01 2.89791763e-01
-8.53586137e-01 2.55061290e-03 4.17217821e-01 -3.82140845e-01
-8.63639295e-01 1.49097264e-01 2.80315075e-02 -1.14178263e-01
5.23402244e-02 7.32385039e-01 -7.95994923e-02 2.72775263e-01
1.29823518e+00 -5.09976745e-01 -1.66324452e-01 -4.29530054e-01
4.83857065e-01 8.72063935e-01 2.50871599e-01 -7.24297762e-01
3.15399051e-01 2.54448831e-01 -1.95738479e-01 -5.57497323e-01
-1.17913890e+00 -7.48948812e-01 -4.77691203e-01 -2.31449738e-01
4.88291651e-01 -6.50463521e-01 -6.23763144e-01 9.04227048e-03
-1.19837749e+00 -1.22363731e-01 -7.63405681e-01 9.67621431e-02
-1.89039752e-01 3.00314367e-01 -2.64676332e-01 -8.56693149e-01
-3.06719005e-01 -1.00062406e+00 8.83505762e-01 6.60255179e-02
-5.07715344e-01 -6.83871388e-01 2.61790920e-02 5.33900261e-01
2.10677132e-01 1.57014728e-01 5.91196954e-01 -1.16799533e+00
-3.23813945e-01 -2.50676185e-01 -4.23848420e-01 1.60790041e-01
2.63372719e-01 -8.96178558e-02 -9.12463367e-01 9.41166133e-02
-1.72984868e-01 -4.90982115e-01 8.07306707e-01 6.16439199e-03
9.30697680e-01 -2.66310036e-01 -2.66748965e-01 -2.93321516e-02
1.20735586e+00 -2.57912397e-01 5.79055309e-01 4.52627987e-01
8.13110173e-01 1.04547596e+00 1.07388484e+00 3.51260632e-01
4.90298837e-01 5.75022697e-01 4.06272084e-01 2.54453629e-01
-1.17042929e-01 -3.92287999e-01 -7.97887985e-03 7.16911554e-01
-1.21754840e-01 -5.81178784e-01 -1.08102596e+00 9.07504499e-01
-2.10002279e+00 -6.56422436e-01 -3.92004371e-01 2.12882304e+00
9.88430083e-01 -6.81701005e-02 5.42819239e-02 2.96331108e-01
7.39340961e-01 -5.66096120e-02 -2.73465604e-01 -3.23229253e-01
-3.00263464e-01 -1.71216056e-01 7.56990135e-01 4.16824251e-01
-7.51569271e-01 9.86079335e-01 6.02556419e+00 8.43694568e-01
-8.26422215e-01 2.33675763e-01 5.90307415e-01 -2.16247916e-01
-2.56172001e-01 4.03749257e-01 -7.94717252e-01 2.76943564e-01
9.51155007e-01 -1.89718172e-01 3.11421603e-01 1.17395389e+00
7.34731555e-02 -4.96632993e-01 -9.96465802e-01 8.56576324e-01
1.39760509e-01 -1.47064602e+00 -2.19282702e-01 5.19136228e-02
6.93200171e-01 4.98512089e-02 -6.15462124e-01 2.84328967e-01
8.21603954e-01 -9.48031008e-01 5.79007268e-01 3.48344862e-01
3.40130836e-01 -6.55847728e-01 1.33994746e+00 2.57430822e-01
-5.81088006e-01 -1.19867258e-01 -5.28642952e-01 -2.76354134e-01
7.88823962e-02 1.09007704e+00 -1.35079575e+00 5.31930566e-01
6.42240703e-01 3.00639451e-01 -9.17102218e-01 9.57448423e-01
-2.61257678e-01 6.41769290e-01 -3.54143620e-01 -2.79887766e-01
-2.63323635e-01 1.97110862e-01 4.93376195e-01 1.13057375e+00
2.86811143e-01 2.62825131e-01 1.35881096e-01 4.57247019e-01
-5.24601460e-01 3.39556426e-01 -6.50596321e-01 1.05993766e-02
1.04816878e+00 1.34182918e+00 -1.01712906e+00 -3.68562043e-01
2.43769307e-02 5.29205620e-01 7.98759282e-01 7.28648379e-02
-3.76845330e-01 -4.58557338e-01 4.11593914e-01 2.41091490e-01
5.12221195e-02 1.86584964e-01 -9.30647790e-01 -1.17269194e+00
-1.49328396e-01 -8.42455864e-01 6.39471531e-01 -1.00212121e+00
-1.42437577e+00 7.11910129e-01 -1.18050784e-01 -1.05595374e+00
9.76879150e-02 -2.44603425e-01 2.70286798e-02 6.50527835e-01
-9.67026293e-01 -9.19036388e-01 -3.96731824e-01 -9.97343808e-02
5.08206904e-01 5.94401993e-02 5.95469892e-01 1.62959218e-01
-6.36140049e-01 4.77525890e-01 -2.23715812e-01 6.08747639e-02
8.21226001e-01 -1.41244376e+00 -9.83626023e-02 1.07896698e+00
2.72359848e-01 7.89212346e-01 9.52798247e-01 -1.00097454e+00
-9.59053755e-01 -1.10435247e+00 1.24183619e+00 -7.46984959e-01
6.60265267e-01 -1.84399590e-01 -1.02419829e+00 7.65007377e-01
-2.33424067e-01 2.57597595e-01 7.78285086e-01 4.38420653e-01
-4.37373132e-01 6.32598437e-03 -1.38299000e+00 5.25127888e-01
1.11278093e+00 -2.63155103e-01 -5.37017822e-01 4.55143511e-01
9.32585239e-01 -4.70655352e-01 -8.76214206e-01 2.81116128e-01
4.33195978e-02 -7.51186073e-01 4.17650968e-01 -5.63591659e-01
4.86597359e-01 -4.50333476e-01 -4.07740265e-01 -1.47879696e+00
5.27599407e-03 -5.27433336e-01 1.33754686e-01 1.48003805e+00
7.61700273e-01 -6.14163637e-01 8.18992198e-01 9.79716837e-01
1.28873900e-01 -6.93105280e-01 -9.69455183e-01 -6.77170992e-01
-2.07530141e-01 -5.72057068e-01 7.18720376e-01 1.12622380e+00
3.77924293e-01 6.86065257e-01 -5.11054993e-01 5.12075499e-02
1.88879207e-01 1.13029545e-02 9.56679583e-01 -1.17003381e+00
-1.56904105e-02 1.25968009e-02 -3.34686607e-01 -5.72342157e-01
-1.01026453e-01 -8.44986022e-01 4.03936923e-01 -1.39964712e+00
1.37321934e-01 -9.03907716e-01 1.09324753e-01 8.24121654e-01
-4.09449488e-01 2.69255489e-01 2.71746308e-01 5.43085277e-01
-7.57588625e-01 2.31643483e-01 7.62814641e-01 3.08228523e-01
-3.54969323e-01 -2.19991192e-01 -1.17277038e+00 4.46183264e-01
8.53324890e-01 -8.80670249e-01 -4.28023756e-01 -2.17237249e-01
7.01748192e-01 -2.45067105e-01 1.70322116e-02 -4.04733628e-01
2.63492733e-01 -3.27035755e-01 -1.91382557e-01 -3.52109104e-01
-3.09454501e-02 -7.04872429e-01 2.33033732e-01 8.18919018e-02
-6.00912213e-01 4.64563537e-03 -2.24431649e-01 6.85683966e-01
6.05267920e-02 -5.80127895e-01 6.19983017e-01 -2.27926344e-01
-3.82166088e-01 -2.86287308e-01 -5.53881943e-01 2.43259445e-01
7.88535476e-01 8.59740153e-02 -6.85820341e-01 -2.01578617e-01
-6.20003939e-01 -4.72909249e-02 5.30833125e-01 4.54010159e-01
2.05261886e-01 -1.08189571e+00 -5.47760963e-01 -3.44870776e-01
3.55665296e-01 1.84223473e-01 -2.43288070e-01 7.24175155e-01
-4.52902466e-01 2.30821878e-01 2.44313970e-01 -3.41261715e-01
-1.36207533e+00 5.74062645e-01 2.51950920e-01 -2.63608903e-01
-2.52007455e-01 9.26855445e-01 -5.91188073e-01 -7.10039496e-01
-6.85219392e-02 -3.08686465e-01 -1.39181614e-01 3.50509852e-01
2.50117093e-01 6.12080872e-01 4.98423934e-01 -7.01600015e-01
-2.88834840e-01 -1.20868959e-01 -5.74961044e-02 9.96048599e-02
1.19916058e+00 -3.64999175e-01 -5.13573050e-01 4.32616323e-01
8.51609945e-01 3.04040879e-01 -5.46776831e-01 -3.31702590e-01
5.46102107e-01 -6.35123432e-01 -6.90729916e-02 -9.17800784e-01
-9.56230044e-01 6.22843541e-02 1.01748630e-01 6.40180647e-01
8.76753151e-01 2.05271736e-01 4.01594192e-01 5.08408368e-01
4.94983196e-01 -1.20350814e+00 -5.25540672e-02 4.81681556e-01
8.97130966e-01 -1.31916976e+00 3.39302510e-01 -8.22789192e-01
-8.74600828e-01 7.38355577e-01 9.58302736e-01 1.27491027e-01
4.51869667e-01 2.17769258e-02 2.71505773e-01 -6.28115892e-01
-8.62856925e-01 -4.95134711e-01 -9.83998403e-02 5.64772666e-01
5.10726094e-01 2.88631082e-01 -9.05313551e-01 7.57669926e-01
-5.37118793e-01 -2.91777682e-02 1.18280113e+00 9.83639061e-01
-6.01061523e-01 -8.48756909e-01 -5.20122528e-01 7.13744223e-01
-4.76702213e-01 4.51829322e-02 -8.75726521e-01 4.22215939e-01
3.52183670e-01 1.34070385e+00 -3.21977705e-01 -6.11563087e-01
4.46907252e-01 3.22630256e-02 -1.07098535e-01 -1.02182031e+00
-7.55367815e-01 -2.08962530e-01 6.87415838e-01 -2.56221563e-01
-5.37536740e-01 -6.54847264e-01 -1.18309009e+00 -4.10276473e-01
-9.28542852e-01 4.19398904e-01 5.33945620e-01 9.97447968e-01
6.95406675e-01 4.02077466e-01 3.24612379e-01 -4.51038331e-01
-4.34458166e-01 -1.18135285e+00 -4.24182236e-01 5.15891075e-01
-7.75197074e-02 -6.94294930e-01 -5.43981433e-01 -1.19439743e-01] | [9.442434310913086, 8.590822219848633] |
8c98189c-0874-442d-920a-1831f6e53091 | multi-modality-deep-network-for-jpeg | 2305.02760 | null | https://arxiv.org/abs/2305.02760v1 | https://arxiv.org/pdf/2305.02760v1.pdf | Multi-Modality Deep Network for JPEG Artifacts Reduction | In recent years, many convolutional neural network-based models are designed for JPEG artifacts reduction, and have achieved notable progress. However, few methods are suitable for extreme low-bitrate image compression artifacts reduction. The main challenge is that the highly compressed image loses too much information, resulting in reconstructing high-quality image difficultly. To address this issue, we propose a multimodal fusion learning method for text-guided JPEG artifacts reduction, in which the corresponding text description not only provides the potential prior information of the highly compressed image, but also serves as supplementary information to assist in image deblocking. We fuse image features and text semantic features from the global and local perspectives respectively, and design a contrastive loss built upon contrastive learning to produce visually pleasing results. Extensive experiments, including a user study, prove that our method can obtain better deblocking results compared to the state-of-the-art methods. | ['Liquan Shen', 'Bo Yan', 'Chenxi Ma', 'Qing Lin', 'Weimin Tan', 'Xuhao Jiang'] | 2023-05-04 | null | null | null | null | ['image-deblocking'] | ['computer-vision'] | [ 5.58439791e-01 -6.29886448e-01 -2.04873130e-01 -1.39413416e-01
-7.61802614e-01 9.29525029e-03 1.64084390e-01 1.82708632e-02
-3.64021468e-03 5.36925077e-01 6.56706512e-01 -3.88737693e-02
4.72680479e-02 -6.31506979e-01 -5.75527787e-01 -6.16521120e-01
2.89381623e-01 -3.10839951e-01 -2.92310156e-02 -2.62277037e-01
4.66090858e-01 7.73248896e-02 -1.36995661e+00 7.39381671e-01
1.13729846e+00 1.06857872e+00 5.92949092e-01 4.55517828e-01
1.24590307e-01 1.15891016e+00 -3.59870404e-01 -4.10781741e-01
2.21982524e-01 -6.37330353e-01 -6.00064695e-01 3.51607770e-01
5.91277063e-01 -8.97333145e-01 -8.37723613e-01 1.42940950e+00
6.07078731e-01 -1.06300130e-01 3.85513723e-01 -9.72168326e-01
-9.73534882e-01 4.73893613e-01 -8.52102160e-01 1.90842286e-01
2.91820973e-01 9.58780050e-02 7.24460542e-01 -8.47689629e-01
3.50002170e-01 1.22180033e+00 7.28011250e-01 5.04359901e-01
-1.06881869e+00 -6.65473342e-01 -1.86923757e-01 5.90421855e-01
-1.08123267e+00 -6.09841287e-01 1.13666403e+00 -7.27138743e-02
4.40602422e-01 4.06871259e-01 6.27533674e-01 1.14134383e+00
3.90599251e-01 1.09431851e+00 9.82181013e-01 -3.33277911e-01
3.27896699e-02 -2.06472740e-01 -3.50891620e-01 4.74006414e-01
2.20088631e-01 8.03562030e-02 -8.21710885e-01 1.38486862e-01
9.83112872e-01 2.57568300e-01 -7.93418288e-01 -3.72589350e-01
-1.21300340e+00 5.60104549e-01 3.37193906e-01 2.61145920e-01
-3.32160652e-01 2.72811592e-01 4.07015592e-01 3.62022281e-01
5.99215984e-01 1.56925544e-01 -1.42554536e-01 -1.19278751e-01
-1.30693841e+00 2.37246215e-01 3.77563298e-01 6.87997162e-01
5.30186057e-01 1.35062665e-01 -4.14914817e-01 1.11052775e+00
3.42480659e-01 4.02852386e-01 5.48549592e-01 -1.20154059e+00
7.78488338e-01 2.59916574e-01 2.64630049e-01 -1.37464440e+00
7.83919692e-02 -4.31662410e-01 -1.56047046e+00 3.41750443e-01
1.61595225e-01 3.48009676e-01 -8.66297483e-01 1.45990419e+00
-2.79812396e-01 4.38541993e-02 -1.37225524e-01 1.10827184e+00
6.48717880e-01 6.65618420e-01 -5.58874309e-02 -3.55272323e-01
1.23412883e+00 -1.11593413e+00 -1.19553542e+00 -2.47452319e-01
1.09157704e-01 -1.14692247e+00 1.06288707e+00 7.13895380e-01
-1.64861071e+00 -8.70541990e-01 -1.37147772e+00 -2.95211613e-01
4.72083241e-02 1.72215447e-01 3.84358317e-01 5.75930595e-01
-1.08314300e+00 8.16466212e-01 -6.79254234e-01 1.33526661e-02
7.32025623e-01 -7.94989429e-03 -5.87220155e-02 -5.25312304e-01
-1.25415266e+00 8.10952902e-01 2.51922905e-01 2.57948656e-02
-9.64663029e-01 -5.35550833e-01 -8.49954009e-01 3.26459855e-01
2.03964069e-01 -7.17281401e-01 1.13618493e+00 -9.96104181e-01
-1.33172691e+00 5.36283374e-01 1.95202453e-03 -2.74578601e-01
8.33333254e-01 -3.52835983e-01 -4.24022228e-01 3.17792207e-01
1.80464401e-03 6.21550798e-01 1.40834820e+00 -1.56589043e+00
-4.76818293e-01 -1.82751209e-01 -2.30886877e-01 3.79010916e-01
-8.38456094e-01 -8.64103958e-02 -8.40169430e-01 -1.23029208e+00
2.80022413e-01 -2.42923781e-01 -7.14454055e-02 4.78321761e-01
-2.41336450e-01 3.50757390e-01 1.05437446e+00 -1.30876708e+00
1.44400203e+00 -2.19206524e+00 2.86831588e-01 -2.48720065e-01
5.62288284e-01 3.69029224e-01 -3.03039581e-01 2.85091966e-01
-8.71552676e-02 1.43128127e-01 -4.62033272e-01 -5.98609447e-01
-2.24682793e-01 -1.39993116e-01 -3.24779391e-01 5.26997268e-01
-9.18730944e-02 8.42418551e-01 -8.78671288e-01 -5.57965100e-01
5.62963605e-01 6.62618101e-01 -5.02953410e-01 3.86037856e-01
-9.56273675e-02 4.59426194e-01 -2.17171028e-01 6.94014013e-01
1.23280060e+00 -4.66849953e-01 2.77628284e-02 -7.61182964e-01
1.20620817e-01 -1.71710044e-01 -8.72171521e-01 2.13880849e+00
-5.10782003e-01 8.41900647e-01 1.48649395e-01 -8.23016286e-01
4.88011420e-01 1.65138468e-01 5.48632443e-01 -1.04831886e+00
1.69678912e-01 7.43309334e-02 -5.14560163e-01 -9.05095339e-01
6.44137383e-01 -3.60949673e-02 4.72850829e-01 3.17342341e-01
-3.11626405e-01 -7.66712278e-02 -5.56119494e-02 2.28736043e-01
9.15283859e-01 3.51610482e-01 7.90185854e-02 6.59199208e-02
5.24340749e-01 -4.48646307e-01 3.10945421e-01 6.36774600e-01
-3.31982732e-01 1.18848324e+00 7.30391815e-02 -4.84636933e-01
-1.43287444e+00 -1.04047179e+00 -2.64987373e-03 6.73615515e-01
7.73787439e-01 -4.88004744e-01 -7.40110517e-01 -3.89074504e-01
-3.78031492e-01 4.53520060e-01 -3.26006740e-01 -4.22682494e-01
-5.57379067e-01 -7.04281569e-01 3.33972633e-01 4.91518408e-01
1.27021730e+00 -8.63830209e-01 -3.73597741e-01 1.87621847e-01
-1.00070882e+00 -9.57669973e-01 -8.17329466e-01 -2.45040894e-01
-9.61989701e-01 -8.49714279e-01 -1.19390535e+00 -9.69243050e-01
5.19727051e-01 7.19260037e-01 1.04273939e+00 4.15365517e-01
-2.76925743e-01 -3.54170091e-02 -4.76256013e-01 3.21500212e-01
-6.01590514e-01 -5.28029203e-01 -3.35071862e-01 1.42493069e-01
-7.86460936e-02 -5.54365933e-01 -1.01542497e+00 1.61782786e-01
-1.40145409e+00 5.07048607e-01 8.67569149e-01 1.07676971e+00
2.41476402e-01 5.63011885e-01 2.56925851e-01 -4.26932544e-01
8.02044272e-01 -2.04705164e-01 -1.16665184e-01 3.08442831e-01
-7.58463144e-01 -4.22893502e-02 7.59329498e-01 -3.83488059e-01
-1.24210501e+00 -3.58632416e-01 -3.23971398e-02 -6.64164364e-01
8.02897103e-03 5.78888178e-01 -3.07094455e-01 -2.20063895e-01
4.91345227e-01 7.18572080e-01 8.34647045e-02 -7.67940640e-01
1.73221678e-01 8.65808308e-01 6.38043284e-01 -2.50883251e-01
6.78383887e-01 4.32709903e-01 -5.20576052e-02 -5.50378025e-01
-6.67982876e-01 -1.91242635e-01 -3.25484484e-01 -3.04713666e-01
7.31245399e-01 -1.11770797e+00 -4.61148173e-01 8.52435946e-01
-1.22983372e+00 -1.24160923e-01 8.62418860e-02 2.35793740e-01
-6.88864648e-01 1.16684163e+00 -9.69518781e-01 -4.68061179e-01
-3.28928143e-01 -1.26307631e+00 8.95661533e-01 4.46354151e-02
1.49751723e-01 -7.96786487e-01 -1.67984918e-01 6.88604236e-01
8.61162603e-01 -4.74048033e-02 1.03902137e+00 2.01497748e-01
-6.97952271e-01 -6.32006451e-02 -6.95417404e-01 7.50218689e-01
3.27780485e-01 -5.60068369e-01 -8.25703681e-01 -6.77181125e-01
3.83920938e-01 -4.77697462e-01 1.23700440e+00 4.64420646e-01
1.70483196e+00 -4.28638905e-01 -1.50862575e-01 8.48359466e-01
1.46753156e+00 -5.56300059e-02 1.29789901e+00 3.89641374e-01
7.78195679e-01 1.34968609e-01 3.84967119e-01 5.79186380e-01
2.02175483e-01 6.16376638e-01 7.20728636e-01 -4.24444467e-01
-8.11743200e-01 -3.88135821e-01 3.11022639e-01 7.76019335e-01
-3.50611028e-03 -5.05979419e-01 -4.82255578e-01 4.53810483e-01
-1.84971261e+00 -9.39186811e-01 8.57295096e-02 2.22930050e+00
1.05213046e+00 -1.15535103e-01 -3.79942536e-01 2.49651000e-01
8.34184527e-01 5.08917511e-01 -4.85667497e-01 3.66757780e-01
-2.32179433e-01 -1.72005653e-01 3.46369714e-01 3.39216650e-01
-1.24375188e+00 6.08978748e-01 6.46890402e+00 1.25827491e+00
-9.39927459e-01 9.54903811e-02 9.94937837e-01 5.96307665e-02
-3.58249038e-01 -1.01310655e-01 5.49191982e-02 8.26950133e-01
4.96289492e-01 1.48306534e-01 6.57773972e-01 4.94117647e-01
5.17807603e-01 7.00201141e-03 -8.83403003e-01 1.61301696e+00
4.53578502e-01 -1.41071153e+00 4.73429412e-01 -1.34604752e-01
7.80986965e-01 -4.48102385e-01 4.61794645e-01 -1.19284056e-01
-8.21522400e-02 -1.07133949e+00 7.83118963e-01 5.57567954e-01
1.06586778e+00 -6.88659608e-01 7.76259124e-01 1.23183601e-01
-9.44342673e-01 -4.16154005e-02 -5.14251053e-01 2.17265829e-01
1.75859392e-01 6.45239294e-01 7.33922869e-02 7.04602659e-01
8.72880995e-01 1.25909328e+00 -5.78461409e-01 1.30599344e+00
-9.24500376e-02 5.15042067e-01 3.14535946e-01 4.35557783e-01
-1.13592274e-01 -5.21938875e-02 5.22050440e-01 1.01210237e+00
5.16676068e-01 -4.77755442e-02 -1.34880068e-02 1.00809312e+00
-3.70720237e-01 4.37095622e-03 -3.03428590e-01 4.87140007e-02
5.14666177e-02 1.07819426e+00 -3.67793262e-01 -4.85561579e-01
-5.54779053e-01 1.74942815e+00 -3.85496696e-03 5.65944731e-01
-7.92548656e-01 -3.63375336e-01 3.92541707e-01 7.79390335e-02
2.63510823e-01 -2.17289198e-02 -4.42396313e-01 -1.61476433e+00
1.75574064e-01 -1.19876063e+00 5.41695729e-02 -1.21777380e+00
-1.38723457e+00 3.69433612e-01 -3.47267717e-01 -1.76671278e+00
1.64815649e-01 -4.25260246e-01 -3.96159351e-01 7.97459602e-01
-1.77387202e+00 -1.28597105e+00 -7.89747834e-01 6.36781693e-01
9.17297006e-01 -1.67169258e-01 3.92725348e-01 6.52536154e-01
-2.94997305e-01 6.06746376e-01 3.55228066e-01 -4.09226753e-02
9.56844807e-01 -7.88662136e-01 8.80234018e-02 1.09438920e+00
-2.34223709e-01 2.41875947e-01 8.09228361e-01 -7.80474842e-01
-1.45406353e+00 -1.13342726e+00 5.22443712e-01 9.96889398e-02
2.15295911e-01 1.11075886e-01 -9.44723129e-01 2.26201907e-01
5.57385087e-01 -1.62037924e-01 3.75269532e-01 -4.76070285e-01
-4.66435015e-01 -1.44590884e-01 -1.22863293e+00 7.20021069e-01
8.48791778e-01 -4.65846479e-01 -1.86881319e-01 1.87326148e-01
7.09107518e-01 -3.52887392e-01 -6.14665151e-01 4.07032460e-01
5.41299820e-01 -1.31271005e+00 1.15086663e+00 -2.08956599e-01
1.22835040e+00 -3.15627545e-01 -3.32258642e-01 -1.25862253e+00
-6.34827137e-01 -5.59235215e-01 -3.93371344e-01 9.16293979e-01
-1.19574636e-01 4.34671715e-02 7.04356372e-01 2.74760783e-01
-7.69841895e-02 -4.97678190e-01 -7.09020138e-01 -4.69241321e-01
-5.88679761e-02 -2.56642699e-01 3.94613028e-01 9.22284186e-01
-4.19639014e-02 3.01551186e-02 -1.21116781e+00 2.86122542e-02
1.01285303e+00 7.76485652e-02 3.97215903e-01 -7.17830837e-01
-3.30148458e-01 -5.07416964e-01 -3.07388306e-01 -1.50080359e+00
-2.60028154e-01 -5.82240939e-01 7.33525679e-02 -1.57032466e+00
8.18303764e-01 -4.04633209e-02 -3.95992815e-01 3.18543106e-01
-4.06865776e-01 6.66328311e-01 1.74784958e-01 5.34802318e-01
-5.99958181e-01 8.11406970e-01 1.59490347e+00 -6.04425311e-01
5.69135360e-02 -3.25968713e-01 -8.61179471e-01 6.10040724e-01
6.53982162e-01 -8.03640932e-02 -3.20195854e-01 -8.51761460e-01
2.37951949e-01 3.38003069e-01 5.39769113e-01 -1.20189166e+00
3.02895188e-01 -1.57721415e-01 8.88009489e-01 -7.22062171e-01
3.62385541e-01 -8.85554671e-01 1.55269187e-02 5.00661254e-01
-5.76040030e-01 -2.14407027e-01 1.69757828e-02 7.16599405e-01
-5.06105423e-01 -2.02008069e-01 1.12205756e+00 -2.66935259e-01
-5.59077799e-01 4.15563434e-01 -2.57498205e-01 -3.45349014e-01
4.78543520e-01 -1.79098517e-01 -2.34839857e-01 -9.86565888e-01
-4.76535112e-01 -1.20363586e-01 6.92755938e-01 4.61078316e-01
1.15013957e+00 -1.45118809e+00 -1.01054549e+00 4.51725692e-01
3.21732722e-02 -4.10948217e-01 6.64923131e-01 6.42104268e-01
-7.50015259e-01 6.79612979e-02 -4.24905688e-01 -4.31899995e-01
-1.20600533e+00 7.76792765e-01 2.65058935e-01 -2.03993618e-01
-7.77295947e-01 5.17868102e-01 1.95790797e-01 3.81152153e-01
4.54441905e-01 -4.33023572e-02 -1.14535250e-01 -3.07734549e-01
9.10876691e-01 4.26058322e-01 -4.28221747e-02 -4.27360445e-01
1.04885556e-01 6.23073459e-01 -2.96185046e-01 1.21345595e-02
1.30866444e+00 -7.62263298e-01 -2.44009018e-01 -2.07993522e-01
1.23578477e+00 -1.17887542e-01 -1.40765584e+00 -3.37681651e-01
-6.16186678e-01 -1.18059719e+00 4.80526745e-01 -1.03202784e+00
-1.55621207e+00 9.81468022e-01 9.76961672e-01 1.49418145e-01
1.64196217e+00 -3.72763276e-01 1.18887305e+00 -5.44150509e-02
1.22080937e-01 -1.00689828e+00 6.55190647e-01 7.29556754e-02
1.15732741e+00 -1.56648731e+00 3.04346025e-01 -3.02888691e-01
-4.31798011e-01 1.29431820e+00 5.19189060e-01 -2.08301307e-03
2.77331620e-01 1.09795086e-01 2.42576804e-02 1.43169817e-02
-4.07750726e-01 2.36301467e-01 2.51374513e-01 5.65871060e-01
3.26244175e-01 -3.29892427e-01 -2.25658536e-01 3.50812018e-01
2.59598106e-01 2.31820449e-01 4.94855493e-01 7.31210351e-01
-5.25511980e-01 -1.06699467e+00 -5.14152050e-01 4.32154924e-01
-5.47087252e-01 -4.97674137e-01 4.62370627e-02 3.00758183e-01
1.82272539e-01 1.11285841e+00 -1.92008585e-01 -5.91727793e-01
-1.15844809e-01 -3.64150375e-01 6.33951306e-01 2.12300625e-02
-1.52242780e-01 4.24703866e-01 -2.91969538e-01 -6.92148864e-01
-4.77188617e-01 -2.46640518e-01 -6.66694462e-01 -5.74600160e-01
-2.62852073e-01 -3.31155956e-01 4.17320758e-01 7.19277143e-01
2.20513582e-01 6.57247007e-01 7.40605175e-01 -1.15219879e+00
-6.41079664e-01 -1.03548002e+00 -6.58393502e-01 8.66438925e-01
7.09101915e-01 -3.11231554e-01 -5.49570680e-01 5.01463592e-01] | [11.287613868713379, -1.8955105543136597] |
07629e08-0f49-4562-b0aa-da20a0731a3c | generative-diffusion-models-on-graphs-methods | 2302.02591 | null | https://arxiv.org/abs/2302.02591v2 | https://arxiv.org/pdf/2302.02591v2.pdf | Generative Diffusion Models on Graphs: Methods and Applications | Diffusion models, as a novel generative paradigm, have achieved remarkable success in various image generation tasks such as image inpainting, image-to-text translation, and video generation. Graph generation is a crucial computational task on graphs with numerous real-world applications. It aims to learn the distribution of given graphs and then generate new graphs. Given the great success of diffusion models in image generation, increasing efforts have been made to leverage these techniques to advance graph generation in recent years. In this paper, we first provide a comprehensive overview of generative diffusion models on graphs, In particular, we review representative algorithms for three variants of graph diffusion models, i.e., Score Matching with Langevin Dynamics (SMLD), Denoising Diffusion Probabilistic Model (DDPM), and Score-based Generative Model (SGM). Then, we summarize the major applications of generative diffusion models on graphs with a specific focus on molecule and protein modeling. Finally, we discuss promising directions in generative diffusion models on graph-structured data. | ['Chengyi Liu', 'Qing Li', 'Jiliang Tang', 'Hui Liu', 'Hang Li', 'Jiatong Li', 'Yunqing Liu', 'Wenqi Fan'] | 2023-02-06 | null | null | null | null | ['video-generation', 'image-inpainting'] | ['computer-vision', 'computer-vision'] | [ 4.50546175e-01 3.50595713e-01 1.55196831e-01 -6.10571839e-02
-6.18545234e-01 -4.71419692e-01 7.96795547e-01 -1.94367602e-01
1.74496949e-01 9.64599848e-01 3.79926354e-01 -1.63212582e-01
-5.50912868e-04 -1.10062861e+00 -5.45061171e-01 -1.08036220e+00
6.37660921e-02 6.74026012e-01 4.22984622e-02 -5.07734157e-02
1.71194077e-01 4.47700053e-01 -8.63900781e-01 1.21751092e-01
9.80354488e-01 3.11455786e-01 3.84701252e-01 1.00278521e+00
-3.02168429e-01 9.99131083e-01 -7.72432387e-01 -7.17366934e-01
1.33998534e-02 -1.14902532e+00 -4.42955941e-01 4.64109451e-01
6.64362311e-02 -1.62761688e-01 -9.01388586e-01 1.35011363e+00
7.73608625e-01 1.38203070e-01 1.06497967e+00 -1.40391004e+00
-1.42057085e+00 6.08495235e-01 -7.88442254e-01 1.62367858e-02
4.34829414e-01 2.10038275e-01 6.53007865e-01 -7.94621587e-01
1.39680707e+00 1.34108102e+00 3.00770283e-01 8.76338422e-01
-1.43434691e+00 -2.47973874e-01 -1.80154461e-02 1.96568877e-01
-1.29974484e+00 -1.92956239e-01 1.14934695e+00 -5.70189655e-01
7.05899715e-01 2.04274029e-01 9.67121661e-01 1.17173839e+00
5.91681898e-01 9.85571325e-01 1.07617879e+00 -3.50258023e-01
2.57265419e-01 -4.73542154e-01 -5.22082627e-01 1.01491892e+00
2.56541967e-01 -1.07955873e-01 -7.21802950e-01 -4.75661844e-01
1.14535058e+00 4.36790660e-02 -1.07852839e-01 -5.90262949e-01
-1.06315982e+00 1.19165897e+00 1.86430320e-01 3.97185907e-02
-6.85745716e-01 4.85343248e-01 -1.65072680e-02 1.75596282e-01
8.45707178e-01 -9.91600454e-02 5.02810776e-01 1.20930314e-01
-1.00769985e+00 5.97110391e-01 7.52778053e-01 1.23905969e+00
6.43378019e-01 4.28362459e-01 -4.94260967e-01 7.80263126e-01
5.31068385e-01 6.54563308e-01 2.68316507e-01 -8.50404382e-01
1.92376494e-01 2.21394330e-01 -3.23386312e-01 -1.14867234e+00
3.76249589e-02 -3.74758244e-03 -1.34846771e+00 -1.36459293e-02
8.30794722e-02 -2.07964465e-01 -1.26453173e+00 1.49762428e+00
3.75343204e-01 1.48338079e-01 -1.59979314e-01 6.77547812e-01
1.06922102e+00 8.53903353e-01 -3.06744426e-02 -3.60582620e-01
8.30462754e-01 -8.19969475e-01 -8.29242647e-01 1.76313482e-02
2.29637429e-01 -1.03247583e+00 4.03304815e-01 2.41328523e-01
-1.27820659e+00 -3.78355742e-01 -6.54389620e-01 -4.02163947e-03
-4.99402313e-03 -2.45847329e-01 9.38118935e-01 5.93607485e-01
-1.45538902e+00 7.30453193e-01 -9.43346500e-01 -4.77382809e-01
5.37610769e-01 1.42955437e-01 -5.41484877e-02 -3.52008432e-01
-8.95181835e-01 2.30166301e-01 -1.09235033e-01 -2.29296476e-01
-1.20296896e+00 -2.89861619e-01 -7.65193880e-01 -3.98969024e-01
1.21109905e-02 -1.48464549e+00 8.85978103e-01 -6.06791496e-01
-1.59165883e+00 1.01438534e+00 -3.11442196e-01 -5.22852480e-01
6.03237987e-01 2.80857533e-01 -1.68399647e-01 4.04565372e-02
-1.23846298e-03 9.24603701e-01 1.10540724e+00 -9.99917626e-01
1.33100143e-02 -3.56818259e-01 -3.41251701e-01 2.37719223e-01
1.12429544e-01 -1.51372984e-01 -4.92854893e-01 -1.05167520e+00
3.50795761e-02 -1.12200570e+00 -6.63557410e-01 -8.86339545e-02
-7.81112373e-01 -2.08824232e-01 7.14683712e-01 -6.30490780e-01
1.38603735e+00 -1.61269617e+00 6.88935995e-01 1.88258007e-01
7.76177347e-01 5.42716496e-02 -3.76436859e-01 8.97345424e-01
3.74997444e-02 7.54501224e-02 -4.27639723e-01 -3.31974030e-01
-2.63235480e-01 2.05616042e-01 -2.33935565e-01 2.62489855e-01
1.88395187e-01 1.48078108e+00 -1.05650520e+00 -5.92297852e-01
4.09797393e-02 7.11866200e-01 -6.50297940e-01 2.33206823e-01
-5.17041147e-01 7.35992134e-01 -6.54756606e-01 5.25894701e-01
5.97321630e-01 -4.76001710e-01 2.92130709e-01 1.50420263e-01
2.39904210e-01 -3.39642465e-01 -9.11284328e-01 1.80085456e+00
5.84406964e-02 4.80049580e-01 -1.48483515e-01 -5.96728027e-01
1.08623159e+00 7.10315257e-02 6.91624939e-01 -2.72088021e-01
-2.02993378e-01 7.29790032e-02 1.10115611e-03 -1.47999048e-01
5.37104905e-01 -2.29709893e-01 2.45408490e-01 6.76325560e-01
1.13228470e-01 -6.65299058e-01 5.73705494e-01 8.78240049e-01
1.27604914e+00 1.50379062e-01 7.14688897e-02 8.28257725e-02
1.42129704e-01 -7.13095665e-02 6.36702403e-02 5.88953078e-01
4.79825065e-02 9.64989007e-01 6.49501264e-01 -3.05980116e-01
-1.17527401e+00 -1.18527496e+00 6.55022860e-01 5.01740158e-01
-8.18789974e-02 -6.81452870e-01 -1.10129094e+00 -5.76990068e-01
-1.36615306e-01 4.22970146e-01 -6.13818288e-01 -3.01025122e-01
-4.27499741e-01 -1.40506399e+00 3.02459121e-01 2.83278197e-01
3.49865705e-01 -1.15185428e+00 3.94107401e-01 5.03899813e-01
-1.25676945e-01 -7.14454651e-01 -8.77016187e-01 -4.51978981e-01
-1.16042972e+00 -8.41181576e-01 -1.40338254e+00 -8.09356213e-01
1.06038380e+00 5.25314689e-01 1.30680978e+00 7.98121095e-02
-5.06492078e-01 6.13759339e-01 -1.19722515e-01 -3.00781935e-01
-9.22624767e-01 -5.38583547e-02 -3.28995287e-01 8.14613104e-02
6.10322915e-02 -5.70646644e-01 -8.20606530e-01 -1.42302485e-02
-1.21887684e+00 3.31721574e-01 5.46541154e-01 8.81419599e-01
1.25676966e+00 -1.76869109e-01 4.13284600e-01 -1.22166026e+00
1.36584997e+00 -4.57031161e-01 -5.74065089e-01 1.89975172e-01
-7.51685023e-01 2.43810222e-01 2.98703521e-01 -4.46172684e-01
-1.01667452e+00 8.65646452e-02 -2.77089983e-01 -4.83456850e-01
3.00380081e-01 5.74382007e-01 -1.69570465e-02 -2.37786308e-01
5.72152734e-01 5.95044255e-01 3.58330011e-01 -1.29980624e-01
8.47451806e-01 4.01501097e-02 6.74979985e-02 -4.45728272e-01
7.82152295e-01 6.60600245e-01 3.16113651e-01 -9.51135874e-01
-4.57589418e-01 -1.53957501e-01 -4.53618079e-01 -3.76618028e-01
1.02769864e+00 -5.43422818e-01 -1.84540898e-01 7.70222545e-01
-1.32072675e+00 -4.59954083e-01 -4.36821431e-01 7.90085346e-02
-9.57410336e-01 6.10716045e-01 -8.72611701e-01 -6.24360919e-01
-4.49296176e-01 -1.26467717e+00 1.28346586e+00 3.35463047e-01
-3.90063450e-02 -1.56137681e+00 4.91548538e-01 1.45837888e-01
4.44000483e-01 6.22885466e-01 7.55496502e-01 -3.61089483e-02
-1.17452896e+00 -2.73764487e-02 -7.37354308e-02 1.03382401e-01
1.94065839e-01 2.51130223e-01 -4.08762455e-01 -2.48250544e-01
-1.74580947e-01 -4.79609072e-02 9.81371641e-01 9.67297733e-01
8.74442101e-01 -6.49916902e-02 -5.68435967e-01 7.47565150e-01
1.18269527e+00 2.62349397e-01 8.66204798e-01 -2.55953461e-01
9.28712666e-01 1.93498656e-01 1.45013660e-01 3.17820609e-01
2.99835593e-01 4.30802584e-01 6.94978535e-02 -2.65389651e-01
-8.84635448e-01 -8.05202723e-01 3.54314417e-01 1.23941374e+00
-4.14859533e-01 -7.54792690e-01 -5.67609727e-01 2.42687151e-01
-1.91060114e+00 -1.21879530e+00 -5.38227797e-01 1.91489661e+00
5.32517552e-01 -1.55125275e-01 2.58454859e-01 -3.68246108e-01
9.72276270e-01 4.96544629e-01 -7.79212177e-01 -8.85980427e-02
-5.10406256e-01 2.92115480e-01 3.80865246e-01 5.89523971e-01
-6.82135344e-01 1.21816218e+00 7.20346737e+00 9.97122049e-01
-6.86908662e-01 1.81467250e-01 1.08228230e+00 2.22070128e-01
-7.63630271e-01 8.02403614e-02 -6.89634264e-01 4.13910210e-01
6.98268354e-01 -5.89652836e-01 6.09632969e-01 5.77031732e-01
2.10396484e-01 4.89134118e-02 -5.30988038e-01 1.19358170e+00
3.58338594e-01 -1.77487218e+00 5.92384398e-01 4.65057105e-01
1.35796583e+00 -6.33161366e-02 2.42260322e-01 -2.25393638e-01
8.16570520e-01 -9.19773936e-01 4.22063559e-01 8.35525215e-01
8.41222823e-01 -7.57697463e-01 2.21995935e-01 2.34701872e-01
-1.11102724e+00 6.72754407e-01 -6.82951093e-01 4.08856720e-01
8.10499370e-01 1.04932356e+00 -7.25302279e-01 4.44790781e-01
-4.02879864e-02 9.54090655e-01 -2.90458649e-01 9.98228431e-01
-5.52652001e-01 5.16459048e-01 9.84940454e-02 -1.07861593e-01
2.43593194e-02 -8.53007078e-01 7.51267076e-01 8.56719077e-01
6.57192767e-01 2.89143808e-02 1.19734973e-01 1.27454209e+00
-2.11513430e-01 1.84019096e-02 -9.26155984e-01 -5.43995440e-01
1.19460281e-02 1.19150996e+00 -9.71827924e-01 -4.36357975e-01
-2.29319364e-01 1.47253287e+00 2.79029459e-01 6.21752143e-01
-8.86449695e-01 -2.27393508e-01 4.94870156e-01 3.50186288e-01
-4.38344590e-02 -6.97337866e-01 1.96923688e-01 -1.15211701e+00
-4.52725828e-01 -7.22364962e-01 2.88055748e-01 -9.92814302e-01
-1.31654990e+00 7.07267404e-01 -1.61115885e-01 -8.56500089e-01
-3.96439731e-01 -3.38182479e-01 -5.53349197e-01 1.00344872e+00
-1.15304291e+00 -1.11730599e+00 -2.65696913e-01 5.58254004e-01
5.79096317e-01 -1.55898824e-01 4.89562541e-01 1.34797409e-01
-2.60048151e-01 3.37467454e-02 3.41236413e-01 -1.19913496e-01
6.42284036e-01 -1.22737873e+00 1.10123098e+00 7.72221267e-01
5.45870483e-01 5.85301995e-01 5.60606420e-01 -1.33595181e+00
-1.63346791e+00 -1.31606328e+00 7.54591525e-01 -4.99919593e-01
5.50587595e-01 -2.58012682e-01 -6.94911182e-01 4.78984535e-01
3.22360456e-01 -1.92987591e-01 4.88531381e-01 -6.42971218e-01
2.68681526e-01 2.90042907e-01 -9.55837905e-01 8.01538289e-01
1.59028506e+00 -3.46869826e-01 3.90393198e-01 9.26857293e-01
5.04807949e-01 -5.84952235e-01 -7.27088869e-01 -2.01985344e-01
1.57996178e-01 -1.02375269e+00 1.06290793e+00 -6.39910698e-01
4.56882209e-01 -2.78909534e-01 2.65530378e-01 -1.64208090e+00
-6.47103012e-01 -1.46852911e+00 -4.41049099e-01 1.05437040e+00
1.95075467e-01 -3.85556638e-01 1.21526742e+00 4.06906664e-01
-5.83858676e-02 -7.01823533e-01 -4.91133511e-01 -6.16185009e-01
-3.31077278e-02 -4.42301817e-02 5.71014047e-01 5.39202631e-01
-5.48864126e-01 4.16210771e-01 -5.97272635e-01 -4.61343437e-01
9.65382397e-01 4.97086272e-02 9.65713203e-01 -9.44151163e-01
-4.65597928e-01 -4.96859938e-01 -5.09855986e-01 -1.43146348e+00
-7.46263415e-02 -1.21577609e+00 -3.04240942e-01 -2.20699167e+00
5.84070861e-01 -5.99830374e-02 2.82428980e-01 -2.48399854e-01
-4.43117976e-01 3.16624910e-01 1.40781894e-01 4.38227713e-01
-4.83423412e-01 7.04305172e-01 2.06104207e+00 -2.87925899e-01
-1.19261965e-01 1.03798717e-01 -6.85611665e-01 4.35623884e-01
8.16407144e-01 -4.52249706e-01 -7.92202055e-01 -4.25429404e-01
4.81163770e-01 3.19988281e-01 1.15685530e-01 -5.74028552e-01
2.00735450e-01 -3.42140079e-01 2.74079561e-01 -4.77120578e-01
1.95468187e-01 -6.47139996e-02 7.98059642e-01 5.79999864e-01
-7.85909519e-02 4.77687925e-01 -4.01607752e-01 9.06770587e-01
-2.80386180e-01 -9.90240425e-02 8.18276346e-01 -4.69736069e-01
-5.41589797e-01 9.58895147e-01 -3.75976712e-01 5.92545569e-02
1.08118284e+00 -2.85119563e-01 -1.87923372e-01 -9.82199728e-01
-9.99178886e-01 -2.93634593e-01 5.85600853e-01 3.01621407e-01
1.05973887e+00 -1.60454106e+00 -9.56688404e-01 1.07882641e-01
-1.14107974e-01 -4.53493707e-02 4.22425151e-01 7.68286586e-01
-7.06017911e-01 1.66699842e-01 1.49408832e-01 -4.77531821e-01
-1.37194192e+00 6.12843454e-01 4.83276844e-02 -6.16232872e-01
-4.70447987e-01 1.10102654e+00 5.83229661e-01 6.21962249e-02
-2.40877509e-01 -2.49003712e-02 1.99057668e-01 -3.09200436e-01
3.96270901e-01 4.22295958e-01 -2.18625501e-01 -5.76705635e-01
7.23743141e-02 4.84471858e-01 -6.07297532e-02 -1.35537982e-01
1.27819526e+00 -1.17921531e-01 -2.63520420e-01 -1.65290087e-02
7.70690739e-01 -2.35830061e-02 -1.36323786e+00 4.91624288e-02
-3.69575828e-01 -2.58635938e-01 -1.12879395e-01 -4.09738213e-01
-1.42440403e+00 9.89151657e-01 2.41823256e-01 2.98223644e-01
8.04622710e-01 1.79896504e-01 1.10707104e+00 -2.69030213e-01
5.71931660e-01 -8.39676201e-01 5.44158280e-01 2.87779570e-01
1.03527534e+00 -8.29710960e-01 1.28482968e-01 -5.45043886e-01
-7.21637249e-01 8.66449833e-01 1.42632470e-01 -2.07273051e-01
6.18341684e-01 1.91773281e-01 -2.28340819e-01 -4.56594288e-01
-7.61150956e-01 -1.49150565e-01 4.42734450e-01 1.15889442e+00
8.17638040e-01 1.77038908e-01 -3.59763920e-01 -3.12594250e-02
1.35247856e-02 8.84536467e-03 5.77408314e-01 7.96189427e-01
-3.68668884e-01 -1.57939386e+00 -2.21802175e-01 5.20407200e-01
-1.52408868e-01 -3.46060812e-01 -8.23847175e-01 2.81582117e-01
-3.31628770e-01 7.35760868e-01 -2.89906234e-01 -1.54239967e-01
7.78688341e-02 -1.20566815e-01 9.45519388e-01 -7.08945036e-01
-7.18769953e-02 4.06031668e-01 -2.39356428e-01 -3.95021051e-01
-5.63987195e-01 -6.71799839e-01 -1.01531994e+00 -5.86204469e-01
-2.71601498e-01 1.37434840e-01 6.10037148e-01 2.34303743e-01
8.29684377e-01 6.11797869e-01 2.34239936e-01 -7.14641511e-01
-1.75415859e-01 -7.62256265e-01 -7.80849993e-01 4.98331785e-01
-3.13196063e-01 -4.10508245e-01 1.77993625e-02 5.16725957e-01] | [11.261378288269043, -0.09948194772005081] |
65d7d7e2-7b40-4af1-8d4a-a7555c132643 | empirical-study-on-airline-delay-analysis-and | 2002.10254 | null | https://arxiv.org/abs/2002.10254v1 | https://arxiv.org/pdf/2002.10254v1.pdf | Empirical Study on Airline Delay Analysis and Prediction | The Big Data analytics are a logical analysis of very large scale datasets. The data analysis enhances an organization and improve the decision making process. In this article, we present Airline Delay Analysis and Prediction to analyze airline datasets with the combination of weather dataset. In this research work, we consider various attributes to analyze flight delay, for example, day-wise, airline-wise, cloud cover, temperature, etc. Moreover, we present rigorous experiments on various machine learning model to predict correctly the delay of a flight, namely, logistic regression with L2 regularization, Gaussian Naive Bayes, K-Nearest Neighbors, Decision Tree classifier and Random forest model. The accuracy of the Random Forest model is 82% with a delay threshold of 15 minutes of flight delay. The analysis is carried out using dataset from 1987 to 2008, the training is conducted with dataset from 2000 to 2007 and validated prediction result using 2008 data. Moreover, we have got recall 99% in the Random Forest model. | ['Ripon Patgiri', 'Sajid Hussain', 'Aditya Nongmeikapam'] | 2020-02-17 | null | null | null | null | ['l2-regularization'] | ['methodology'] | [-3.65129769e-01 -6.49764657e-01 -2.64153630e-01 -8.39598835e-01
5.35093620e-02 -4.62209165e-01 3.96262780e-02 5.02318501e-01
-4.23403710e-01 8.81534815e-01 1.68229550e-01 -5.56511402e-01
-7.52610981e-01 -1.28515434e+00 -5.56675978e-02 -5.22455990e-01
-1.23086996e-01 5.01347721e-01 1.29177049e-02 -5.04577607e-02
1.69161633e-01 7.94840634e-01 -1.49051690e+00 4.03997630e-01
9.65247929e-01 1.22678792e+00 -3.12203974e-01 7.76258826e-01
1.62911024e-02 1.04258120e+00 -5.15737116e-01 1.44536212e-01
5.46591997e-01 1.23279110e-01 -5.12290359e-01 -3.45447123e-01
-1.28366947e-01 -1.70457989e-01 2.33971953e-01 4.01285231e-01
3.31189841e-01 4.27482575e-01 6.13986313e-01 -1.65782452e+00
-8.76445100e-02 -1.19959107e-02 -2.85393685e-01 5.23257792e-01
1.69638366e-01 -3.15272748e-01 3.23759645e-01 -3.53203326e-01
1.63847759e-01 9.77276504e-01 7.16710269e-01 -1.39390543e-01
-6.12284482e-01 -9.61526930e-01 4.22351658e-02 3.84273559e-01
-1.39855146e+00 3.46540809e-01 1.65821895e-01 -8.86514366e-01
1.08101940e+00 4.21458453e-01 6.91085577e-01 -1.82422772e-02
6.97127044e-01 4.35371995e-02 1.51021206e+00 -4.46419328e-01
1.50668129e-01 4.77801472e-01 8.69178355e-01 4.04751539e-01
3.10260564e-01 7.12747097e-01 -7.79856071e-02 -6.46265745e-02
-2.43043490e-02 7.55731106e-01 2.88939297e-01 4.93298382e-01
-8.06851089e-01 6.47784233e-01 4.67362911e-01 1.86072066e-01
-5.96085846e-01 -5.16304851e-01 2.30203807e-01 7.45419323e-01
7.92391539e-01 1.73168182e-01 -9.96572912e-01 4.44078408e-02
-1.16258526e+00 4.22815770e-01 9.86896515e-01 8.66439819e-01
8.29849005e-01 -2.33835988e-02 1.06150858e-01 3.75049919e-01
7.84061849e-02 8.83533478e-01 3.57216507e-01 -4.10954356e-01
3.76655400e-01 7.05811143e-01 2.89875418e-01 -1.28799284e+00
-1.10139608e+00 -4.26680267e-01 -1.01392853e+00 4.64142933e-02
2.03091949e-02 -7.29623079e-01 -7.88660586e-01 9.48550284e-01
3.54207009e-01 4.44087535e-02 9.70158949e-02 8.59354734e-01
6.52522922e-01 9.99774814e-01 1.74021572e-01 -6.57247782e-01
9.68012869e-01 -1.12000132e+00 -8.99072766e-01 3.01248670e-01
7.56341875e-01 -8.86683345e-01 7.51415789e-01 1.00883055e+00
-4.46998179e-01 -8.68046403e-01 -6.23584151e-01 6.36792243e-01
-1.14689481e+00 2.06126899e-01 9.22094226e-01 6.60915554e-01
-3.65615815e-01 5.21765053e-01 -5.13911128e-01 -3.73837709e-01
-1.41495228e-01 2.58132309e-01 -3.17731857e-01 -4.18454617e-01
-1.15479279e+00 7.20935404e-01 4.84434843e-01 6.50646687e-02
-3.84176582e-01 -8.68572176e-01 -2.70526528e-01 -1.76425174e-01
8.01275894e-02 -7.97499597e-01 6.80455565e-01 -5.91969192e-01
-5.12319744e-01 5.33026397e-01 -7.30547607e-02 -5.11415422e-01
-6.27346933e-02 -4.88739938e-01 -1.40810156e+00 -3.92173946e-01
-1.14968508e-01 -2.46854555e-02 3.84631157e-01 -6.40262485e-01
-1.20988250e+00 -7.57206738e-01 -1.45352498e-01 -3.89696449e-01
8.83058831e-02 4.29315776e-01 4.49832499e-01 -3.47649366e-01
-7.85390139e-02 -1.20547223e+00 -5.95037639e-01 -9.11066294e-01
-2.07200140e-01 -3.38880555e-03 6.39993966e-01 -6.88406765e-01
2.09994221e+00 -2.19290376e+00 -5.58791816e-01 7.57857025e-01
-2.35355034e-01 -5.52749969e-02 5.12314916e-01 6.58709586e-01
-1.32218361e-01 3.15783143e-01 3.34676690e-02 5.80910087e-01
-4.23058897e-01 1.93712547e-01 -3.76518518e-01 2.32355863e-01
-1.99689835e-01 1.29238799e-01 -2.90382862e-01 -5.06390154e-01
2.69823968e-01 4.26608436e-02 -3.38261575e-01 5.13222218e-01
1.24682531e-01 2.30117336e-01 -3.67859066e-01 9.08942521e-01
8.80607486e-01 7.61709362e-02 -5.57574749e-01 -2.46404484e-01
-6.01542830e-01 -4.62551087e-01 -1.20822048e+00 1.02381313e+00
-7.03649759e-01 3.12633574e-01 -2.60530978e-01 -5.78785956e-01
1.47894382e+00 1.95509464e-01 5.61820269e-01 -3.49483818e-01
-1.00876823e-01 1.04462676e-01 9.32214409e-02 -9.92617846e-01
3.09829742e-01 -2.27338314e-01 -3.61700542e-02 4.37116235e-01
-5.28653264e-01 -1.56555593e-01 1.79103076e-01 -1.43778995e-01
8.04743290e-01 -3.24280143e-01 2.15848997e-01 -3.59121934e-02
4.18538630e-01 5.44119418e-01 8.12997103e-01 1.18670121e-01
-1.89454868e-01 2.44070336e-01 3.14862490e-01 -1.10543787e+00
-6.82016313e-01 -6.75327659e-01 -2.60810226e-01 1.06186521e+00
-2.58418947e-01 -5.08714318e-01 -2.20941037e-01 -4.96304423e-01
3.43817025e-01 8.52420032e-01 -2.46789694e-01 -1.87413935e-02
-2.78285414e-01 -1.02074647e+00 1.37224793e-01 2.07991242e-01
7.00488269e-01 -6.98781848e-01 -5.24423838e-01 -3.82603668e-02
8.16468149e-02 -8.71868014e-01 2.31567040e-01 3.12265933e-01
-1.05723429e+00 -1.00159562e+00 2.20552847e-01 -5.54577291e-01
5.63918352e-01 1.00271799e-01 1.20029068e+00 1.18698962e-01
-6.33653581e-01 -1.04502350e-01 -5.34095526e-01 -7.89826751e-01
4.86713618e-01 1.05588622e-01 1.74993217e-01 -3.17443937e-01
8.90639544e-01 -5.32042980e-01 -3.70553583e-01 5.78506112e-01
-6.48136675e-01 -2.42477417e-01 4.90728766e-01 4.78780061e-01
7.10695982e-01 8.72731805e-01 5.67636907e-01 -1.13143110e+00
6.54758573e-01 -7.42668450e-01 -5.97286761e-01 2.22139463e-01
-1.36638319e+00 -3.40582728e-01 7.93033361e-01 3.50890577e-01
-7.69322276e-01 -6.65153563e-02 -7.59130046e-02 -2.24326700e-01
-6.98960662e-01 4.97211516e-01 1.09991677e-01 1.20758064e-01
5.47299087e-01 -1.91299126e-01 -2.73349702e-01 -5.61723173e-01
1.78905614e-02 9.49536026e-01 -7.05870762e-02 -1.22207686e-01
7.66749501e-01 5.46156131e-02 2.72247732e-01 -4.75996614e-01
-8.35627079e-01 -7.94724822e-01 -1.15083992e+00 -4.33730334e-01
8.68814468e-01 -8.15606773e-01 -7.11137891e-01 1.74058095e-01
-5.96272707e-01 9.87487435e-02 9.83093027e-03 1.06929803e+00
-2.15181038e-01 -4.26781267e-01 -2.48249725e-01 -8.43566358e-01
-5.02547681e-01 -5.67978084e-01 3.67395848e-01 3.71742517e-01
-1.28096312e-01 -1.09158504e+00 2.86782742e-01 3.76260430e-01
5.79217851e-01 7.85126984e-01 1.19777668e+00 -9.58458066e-01
-3.93472880e-01 -7.47313380e-01 -1.56887397e-01 3.25542539e-01
6.02567308e-02 7.10082948e-01 -7.86134362e-01 -2.07458377e-01
-2.85500973e-01 -4.49663475e-02 6.08294785e-01 6.07670426e-01
1.44979894e+00 -3.64174128e-01 -4.72857863e-01 7.63234079e-01
1.62969232e+00 5.73572934e-01 4.23331797e-01 4.13627058e-01
2.65310138e-01 8.94136310e-01 1.77811277e+00 7.24346876e-01
3.58549058e-01 7.80237764e-02 2.87017405e-01 -2.90678352e-01
4.69717026e-01 -1.98224843e-01 -1.13058329e-01 8.34521592e-01
-4.11107630e-01 -1.42095923e-01 -1.18128359e+00 6.20869957e-02
-1.68059433e+00 -9.96623874e-01 -6.68957531e-01 2.41124749e+00
9.87555832e-02 2.56679356e-01 3.92360002e-01 4.58664417e-01
3.68179858e-01 1.04376458e-01 4.62144092e-02 -9.28870738e-01
1.47452597e-02 -4.10476141e-02 8.86367440e-01 5.65388799e-01
-1.13036704e+00 8.60115230e-01 6.48503256e+00 4.96832550e-01
-1.22448075e+00 2.03609932e-02 5.29249966e-01 -3.38223815e-01
-2.06479102e-01 1.67575121e-01 -1.10227013e+00 5.80191493e-01
1.61352193e+00 -2.14497179e-01 4.76828247e-01 8.72737765e-01
7.58832753e-01 -2.45926946e-01 -5.82954228e-01 6.81747735e-01
-1.98926628e-01 -9.37176228e-01 -3.37014198e-01 6.01709820e-02
6.41770244e-01 -1.12336501e-01 -3.74665618e-01 5.20827830e-01
2.62847990e-01 -1.23470092e+00 -1.42824233e-01 1.03268540e+00
3.67900044e-01 -1.15605462e+00 1.11603832e+00 7.51757026e-01
-1.05840337e+00 -5.81988871e-01 -6.58225656e-01 -6.73782945e-01
1.62652314e-01 1.30044603e+00 -8.54070067e-01 7.48594344e-01
1.02522516e+00 8.96060526e-01 -4.96335030e-01 1.10836148e+00
-3.00324112e-02 9.14091229e-01 -4.17977691e-01 -1.08305819e-03
1.59805149e-01 -4.52508718e-01 1.04718087e-02 1.05212677e+00
3.65423143e-01 4.26929623e-01 5.16030848e-01 8.37839395e-02
7.08219051e-01 5.80455422e-01 -8.03147852e-01 2.18455166e-01
4.91943210e-01 1.39494574e+00 -2.92310834e-01 -9.99571234e-02
-3.15161139e-01 6.41096979e-02 -3.91601264e-01 2.66016901e-01
-7.94952154e-01 -8.53978753e-01 4.26977396e-01 4.23168898e-01
-2.80400306e-01 -3.74486148e-01 -5.13476312e-01 -5.56486189e-01
-2.16121957e-01 -3.88090014e-01 6.50435090e-01 -1.01025486e+00
-1.03254926e+00 7.94949293e-01 1.32689118e-01 -1.51553857e+00
-2.06085593e-01 -5.57699859e-01 -9.78935421e-01 9.97735322e-01
-1.70016372e+00 -1.11552501e+00 -8.24764013e-01 9.06804144e-01
2.47866541e-01 -6.31173790e-01 1.08274877e+00 8.42345834e-01
-6.92353427e-01 1.52241081e-01 2.45905846e-01 -2.24883988e-01
9.07447636e-01 -1.04394305e+00 -2.25354269e-01 5.47140598e-01
-4.30959553e-01 4.04136688e-01 4.80274737e-01 -8.52866411e-01
-8.54516327e-01 -1.36113262e+00 1.20584130e+00 -2.99450636e-01
2.94837505e-01 1.32140905e-01 -4.20804560e-01 4.98109072e-01
-1.74925163e-01 2.19785079e-01 1.42439640e+00 5.98695159e-01
8.55068043e-02 -9.10555780e-01 -1.47180676e+00 -3.10516119e-01
2.62932062e-01 -9.75341871e-02 -1.77770734e-01 9.63899434e-01
1.06713486e+00 7.07787788e-03 -1.61822760e+00 5.07591844e-01
6.93676174e-01 -1.08863497e+00 9.90256011e-01 -1.26917195e+00
2.96140343e-01 -4.24870908e-01 -3.45347881e-01 -1.34071529e+00
-5.79983711e-01 6.49518743e-02 3.88938636e-01 1.30019426e+00
6.66193783e-01 -6.57924414e-01 5.05594313e-01 8.47822785e-01
-6.02354221e-02 -9.64442551e-01 -4.68811810e-01 -8.49240899e-01
-1.17725104e-01 -5.71149170e-01 1.09480059e+00 9.29189146e-01
-4.13065553e-01 1.51709005e-01 -5.43864548e-01 4.42513794e-01
2.18300909e-01 4.76865202e-01 7.76350796e-01 -1.57407618e+00
1.92048013e-01 4.46099669e-01 -4.47105497e-01 -2.43577093e-01
5.12889475e-02 -5.29938459e-01 -4.32982057e-01 -1.60779393e+00
-4.92333978e-01 -1.09778666e+00 -5.57408869e-01 3.63371253e-01
2.22708762e-01 -3.89399409e-01 8.47418699e-03 3.62955004e-01
-1.19502343e-01 -1.88098460e-01 1.10660541e+00 6.44011274e-02
-3.84643078e-01 1.01074839e+00 -2.31545120e-01 6.12094223e-01
1.04937243e+00 -7.27281570e-01 -4.64885265e-01 -1.89490929e-01
4.79831606e-01 2.28493899e-01 -1.57389045e-01 -1.16007924e+00
3.86239380e-01 -8.13840449e-01 6.21967912e-01 -1.14047515e+00
3.59292445e-03 -1.51354575e+00 4.52782184e-01 7.22065210e-01
-2.75690854e-01 6.94876730e-01 6.81842342e-02 3.49771321e-01
-6.04717135e-01 -8.94225910e-02 2.86486953e-01 3.72934006e-02
-8.15433800e-01 4.86532599e-01 -2.54689723e-01 -7.50248060e-02
1.43490708e+00 -7.86762610e-02 -2.86530733e-01 -2.27640923e-02
-7.97089219e-01 5.32627463e-01 2.68327165e-02 3.29282254e-01
4.18089360e-01 -1.27506053e+00 -5.87087274e-01 5.56363523e-01
9.34565961e-02 -1.89506277e-01 2.08744228e-01 1.08087730e+00
-8.84723723e-01 8.29066038e-01 -5.34495294e-01 -2.25610182e-01
-1.59635222e+00 9.84740257e-01 -3.12864929e-02 -3.33196521e-01
7.75019079e-02 6.09518230e-01 -7.21706569e-01 -5.46031892e-01
3.36228535e-02 -8.19162905e-01 -5.18506408e-01 3.21008950e-01
5.05488575e-01 9.75278258e-01 2.82114238e-01 -1.36205032e-01
-6.63992107e-01 1.05931163e+00 1.98423564e-01 2.91302502e-01
1.22869134e+00 -4.67409529e-02 -3.10294867e-01 6.99677885e-01
9.34432685e-01 1.84262142e-01 -2.64389426e-01 1.44277409e-01
-1.55596793e-01 -9.88368213e-01 2.36307949e-01 -1.17964768e+00
-1.04741466e+00 1.16825783e+00 9.04957116e-01 5.55409610e-01
1.48341274e+00 -6.71787858e-01 9.32086766e-01 5.50649285e-01
4.32303876e-01 -1.21842170e+00 -9.51427102e-01 4.04263020e-01
3.91952038e-01 -1.63177347e+00 2.61497140e-01 -3.31049353e-01
-6.55939996e-01 1.22906768e+00 6.63457096e-01 -1.36931807e-01
1.61270773e+00 3.31675857e-01 -1.11514525e-02 1.87816806e-02
-1.17073834e+00 -1.08153291e-01 6.38539530e-03 5.35411298e-01
5.59897423e-01 5.27272940e-01 -3.99947435e-01 1.00582981e+00
-6.47395670e-01 7.04308510e-01 2.45765418e-01 8.55820775e-01
-1.08127987e+00 -7.84504056e-01 -5.28176904e-01 8.47701788e-01
-2.65933245e-01 -1.61883801e-01 -1.40634879e-01 7.22027779e-01
7.34145880e-01 1.34391296e+00 2.46524170e-01 -8.96680117e-01
6.09120786e-01 1.38236403e-01 -2.50136167e-01 -3.46175760e-01
-9.97537315e-01 -4.67249751e-01 2.40154654e-01 -4.23423022e-01
-1.87849090e-01 -3.12155008e-01 -1.40480828e+00 -9.30783510e-01
-1.90048620e-01 5.69456637e-01 9.36104715e-01 7.07675695e-01
2.08010420e-01 5.11933029e-01 1.33535647e+00 2.50349641e-01
-2.07921956e-02 -7.18416154e-01 -9.99805570e-01 3.79828401e-02
3.76202434e-01 -4.75787431e-01 -5.75806737e-01 8.56252164e-02] | [8.408199310302734, 4.763503551483154] |
c1c39fc2-4f21-4038-ab35-d5eb6a2dce08 | a-transformer-based-approach-to-video-frame | 2303.09293 | null | https://arxiv.org/abs/2303.09293v2 | https://arxiv.org/pdf/2303.09293v2.pdf | A transformer-based approach to video frame-level prediction in Affective Behaviour Analysis In-the-wild | In recent years, transformer architecture has been a dominating paradigm in many applications, including affective computing. In this report, we propose our transformer-based model to handle Emotion Classification Task in the 5th Affective Behavior Analysis In-the-wild Competition. By leveraging the attentive model and the synthetic dataset, we attain a score of 0.4775 on the validation set of Aff-Wild2, the dataset provided by the organizer. | ['Hyung-Jeong Yang', 'Sudarshan Pant', 'Ngoc-Huynh Ho', 'Dang-Khanh Nguyen'] | 2023-03-16 | null | null | null | null | ['emotion-classification', 'emotion-classification'] | ['computer-vision', 'natural-language-processing'] | [-2.20734850e-01 -6.66272268e-02 2.55438667e-02 -5.50192416e-01
-3.83581132e-01 -3.43251556e-01 2.04380080e-01 3.45678926e-02
-5.40730357e-01 5.88789761e-01 1.26489952e-01 4.60176200e-01
3.77016097e-01 -3.89027715e-01 -1.60411701e-01 -4.22129691e-01
-2.06217721e-01 4.51187938e-01 -1.69584140e-01 -5.63535750e-01
-3.25155288e-01 4.30254042e-02 -1.48207331e+00 7.57696867e-01
7.10096300e-01 1.75401437e+00 -5.60945749e-01 3.38544041e-01
1.85710654e-01 1.00812030e+00 -6.01407647e-01 -9.86099422e-01
-3.08902085e-01 -2.69942850e-01 -8.13142240e-01 -9.13776308e-02
-2.21391380e-01 1.04886882e-01 -1.64445296e-01 8.74857426e-01
6.45501018e-01 2.93453366e-01 5.36221623e-01 -1.51455486e+00
-3.33450496e-01 7.34501719e-01 -3.12630117e-01 1.46379814e-01
4.86998320e-01 -8.04951936e-02 9.83607650e-01 -1.17301261e+00
5.29178679e-01 9.05526221e-01 6.80226743e-01 6.84329331e-01
-8.76815617e-01 -6.38139784e-01 2.18556494e-01 7.09102869e-01
-1.38696194e+00 -6.73736751e-01 1.19012260e+00 7.06031173e-02
1.19019389e+00 3.21854234e-01 1.06657898e+00 1.58731890e+00
1.27836391e-01 1.25328994e+00 1.12033927e+00 -2.24992111e-01
5.41859865e-01 2.12371185e-01 1.61435261e-01 4.76427048e-01
-5.23858726e-01 -2.75819898e-01 -6.32029533e-01 -4.68260869e-02
-2.06647709e-01 -2.61027724e-01 -3.31990868e-02 -1.54919669e-01
-5.74642777e-01 7.56358922e-01 5.04244208e-01 4.53116745e-01
-7.52758086e-01 9.53092650e-02 9.45780635e-01 5.49051881e-01
1.00445020e+00 5.90942919e-01 -4.88950342e-01 -7.32030332e-01
-4.77576017e-01 1.00151807e-01 7.02552557e-01 6.02104485e-01
1.18762717e-01 1.98233545e-01 -1.73734531e-01 1.10593295e+00
-2.59874929e-02 3.69322538e-01 6.61116660e-01 -7.04888225e-01
3.91079113e-02 6.78878844e-01 -6.22593015e-02 -8.76435161e-01
-6.21222079e-01 -3.98755938e-01 -1.00556183e+00 -2.37252340e-01
-1.65384218e-01 -3.82346123e-01 -6.45782530e-01 2.08844638e+00
2.22754747e-01 -6.17043972e-02 1.83927596e-01 8.96861970e-01
9.01333213e-01 6.17298424e-01 4.92495716e-01 -3.72896880e-01
1.34040058e+00 -1.08760619e+00 -9.61358070e-01 -3.71009439e-01
5.91445446e-01 -4.09936100e-01 1.15728867e+00 7.71726668e-01
-9.03547585e-01 -2.29216889e-01 -1.10066903e+00 3.57991874e-01
-4.67855752e-01 1.11615337e-01 8.97622705e-01 4.14343864e-01
-1.14871991e+00 1.22723751e-01 -6.87402964e-01 -5.91774225e-01
3.75186920e-01 2.85632402e-01 -5.38228393e-01 2.53653973e-01
-1.51268685e+00 9.95459557e-01 5.61536513e-02 1.94193035e-01
-7.93690324e-01 -3.18458408e-01 -5.61074674e-01 1.69456467e-01
9.30654109e-02 -3.86699706e-01 1.41782010e+00 -1.29204631e+00
-1.67687345e+00 1.16026795e+00 7.47690201e-02 -5.22212029e-01
3.66550148e-01 -4.99462724e-01 -9.52202618e-01 2.28065196e-02
-3.26975465e-01 4.22389179e-01 6.26526535e-01 -9.59874392e-01
-5.13550714e-02 -5.64467728e-01 -1.43783158e-02 9.48831141e-02
-6.81363106e-01 8.46641287e-02 -3.89578462e-01 -4.45826709e-01
-3.70232046e-01 -1.09168875e+00 -3.04058403e-01 -5.33590555e-01
-3.48020732e-01 -4.16787207e-01 6.12874985e-01 -1.55431211e-01
1.20230663e+00 -2.55294919e+00 2.31973305e-01 6.68060854e-02
1.52392328e-01 2.30404243e-01 -9.06048343e-02 2.13923767e-01
-4.71851081e-01 -2.53330022e-01 9.40025449e-02 -7.35846460e-01
2.04369307e-01 -1.87745243e-01 -3.82792115e-01 2.05821529e-01
1.11354105e-01 9.35860515e-01 -8.19196463e-01 -4.78688478e-01
1.09445892e-01 3.41995537e-01 -4.45917249e-01 6.85276985e-01
-2.12119911e-02 -2.83382386e-02 -5.39911985e-01 7.31560290e-01
3.38779896e-01 -9.05549377e-02 2.83379704e-01 -3.57767820e-01
2.69459069e-01 -1.35379255e-01 -3.08957517e-01 1.81640065e+00
-6.00780964e-01 4.89917785e-01 2.43038610e-02 -8.24322760e-01
1.01212823e+00 5.94825566e-01 7.16926992e-01 -1.13274133e+00
6.19339287e-01 -9.04839486e-02 -4.22581173e-02 -4.89980400e-01
5.33405840e-01 -3.68796170e-01 -7.71850407e-01 2.83075809e-01
3.10507864e-01 -3.41870487e-02 3.18293236e-02 1.92142949e-01
1.31096983e+00 -1.83032379e-02 2.15343684e-01 -2.13347197e-01
3.64868850e-01 -1.14707448e-01 5.74868858e-01 1.32931858e-01
-7.34019637e-01 2.68179804e-01 4.45316315e-01 -7.83093750e-01
-4.88060117e-01 -8.78052771e-01 5.07263280e-02 1.33976460e+00
-1.58582181e-01 -8.03106189e-01 -9.54921186e-01 -8.03822815e-01
-5.55029333e-01 8.10090125e-01 -1.15223825e+00 -7.63190985e-01
-4.88122180e-03 -9.29768384e-01 7.03604102e-01 4.98029888e-01
5.04620135e-01 -1.45496261e+00 -8.80963326e-01 2.13720411e-01
-5.09338617e-01 -1.30807948e+00 1.09686721e-02 4.53863233e-01
-2.43140385e-01 -6.56212807e-01 -2.11519793e-01 -4.34142560e-01
8.80252495e-02 -4.36130792e-01 1.54379189e+00 -1.51661232e-01
3.46799307e-02 1.75001740e-01 -6.81728125e-01 -5.40493846e-01
1.37401342e-01 2.43816674e-01 1.65645614e-01 4.18289185e-01
6.29591227e-01 -8.03389668e-01 -5.90903699e-01 1.97902530e-01
-5.21800637e-01 -1.25035658e-01 2.50086248e-01 5.72553277e-01
4.12039787e-01 -2.58283913e-01 8.76325727e-01 -8.31034184e-01
8.69455278e-01 -5.50805390e-01 -3.07118576e-02 1.58807606e-01
-3.80909085e-01 -2.78892606e-01 8.68773401e-01 -3.71591538e-01
-9.10648406e-01 1.23792142e-02 -5.55899620e-01 -7.92370021e-01
1.68823898e-02 4.90221560e-01 -4.30995613e-01 2.60216445e-01
3.73699456e-01 -7.87771028e-03 -4.31599081e-01 -6.71907142e-02
1.79935947e-01 6.24082804e-01 3.83150488e-01 -5.37595093e-01
1.97862387e-02 8.97246897e-02 -3.17902535e-01 -4.12860036e-01
-1.17822909e+00 -1.77037399e-02 5.81016019e-02 -5.91917515e-01
8.87947321e-01 -9.78129268e-01 -1.09580100e+00 4.94689852e-01
-9.52219307e-01 -3.51831317e-01 -3.28155309e-01 2.26848438e-01
-5.69269657e-01 -5.21271944e-01 -6.98885262e-01 -9.41645801e-01
-8.49233985e-01 -1.01206017e+00 1.00469422e+00 1.90708727e-01
-8.42925787e-01 -8.05562496e-01 5.14371932e-01 6.59892187e-02
7.36515641e-01 5.32874107e-01 7.87472010e-01 -8.95488918e-01
5.63584805e-01 -3.53829920e-01 2.44140357e-01 2.81834036e-01
-2.62331963e-01 -4.05206233e-02 -1.52869976e+00 1.71299398e-01
6.67123348e-02 -1.21200085e+00 6.86670959e-01 -1.59648985e-01
1.62855422e+00 1.34321049e-01 8.53020698e-03 5.26298761e-01
1.05212951e+00 3.15728039e-01 7.83527315e-01 9.11527351e-02
2.78465003e-01 5.74047506e-01 7.86323726e-01 7.11972535e-01
5.15832245e-01 7.46665597e-01 5.05922377e-01 -1.32429779e-01
5.31843126e-01 -2.02634260e-01 3.72858882e-01 1.09504783e+00
-3.22058946e-02 -4.32366699e-01 -7.27563143e-01 4.33599800e-01
-1.99499178e+00 -9.20309544e-01 2.98804015e-01 1.56704390e+00
6.64971173e-01 1.94930345e-01 1.24096088e-01 1.52439579e-01
3.39738727e-01 3.53499383e-01 -4.30966705e-01 -1.07921600e+00
-6.51042759e-02 4.34876412e-01 -5.70048273e-01 1.96722120e-01
-1.24025047e+00 1.17888606e+00 6.62511110e+00 7.41840422e-01
-1.36418355e+00 1.25513285e-01 1.08851910e+00 -3.93587619e-01
-8.57740790e-02 -7.03225136e-01 -1.27928719e-01 5.43065786e-01
1.35869026e+00 -3.56252074e-01 3.19200695e-01 1.04001772e+00
-5.57551794e-02 9.09520760e-02 -1.13149869e+00 1.13840032e+00
3.25143278e-01 -4.77019757e-01 -5.68266138e-02 -2.99249202e-01
3.74336660e-01 -5.29867830e-03 2.82401919e-01 1.06766236e+00
6.11103103e-02 -1.23972917e+00 6.82764947e-01 3.75104785e-01
9.29964542e-01 -1.02971566e+00 1.04277241e+00 -4.51894961e-02
-9.59209204e-01 2.13242471e-01 -7.62986913e-02 -2.10020021e-01
-4.59696725e-03 6.49802923e-01 -4.51352865e-01 3.52187544e-01
9.90880609e-01 7.39906669e-01 -4.97159332e-01 4.24099892e-01
-8.74766856e-02 7.56554008e-01 -1.78742215e-01 -3.31054032e-01
1.57069311e-01 -1.31186381e-01 2.24648774e-01 1.30216670e+00
1.27036884e-01 2.40862250e-01 -8.88418928e-02 2.61620760e-01
-6.15709662e-01 3.83882910e-01 -6.64729118e-01 -2.12133572e-01
2.15259016e-01 1.83246458e+00 -4.08004850e-01 -2.94183284e-01
3.88283865e-03 1.17731118e+00 8.14786971e-01 7.65676573e-02
-1.19484341e+00 -3.15465152e-01 6.15774453e-01 -5.42345107e-01
3.97017319e-03 5.36341667e-01 5.00951819e-02 -1.22575212e+00
-9.20031667e-02 -1.09111285e+00 4.30039942e-01 -1.15284979e+00
-1.28693986e+00 1.51490414e+00 -2.92531490e-01 -1.00800633e+00
-4.46774721e-01 -5.21270752e-01 -6.43998384e-01 3.74291271e-01
-7.93735564e-01 -1.02451980e+00 -4.30239707e-01 6.38739824e-01
3.56639832e-01 -1.94820955e-01 1.38999164e+00 4.34044898e-01
-8.85343492e-01 8.88478100e-01 -2.32724592e-01 1.20493062e-01
8.01002860e-01 -1.19588256e+00 1.11176573e-01 3.28869045e-01
1.04753666e-01 2.11261898e-01 8.57738256e-01 2.15213224e-01
-1.26373279e+00 -1.01489413e+00 8.13347042e-01 -5.60708880e-01
6.60268843e-01 -7.11513102e-01 -6.46176815e-01 6.39455914e-01
7.14284778e-01 5.96007407e-02 1.01183844e+00 5.30821145e-01
-3.66485208e-01 -4.21291739e-01 -1.22109509e+00 5.80156565e-01
7.67785490e-01 -4.98395115e-01 -4.97225642e-01 7.35968873e-02
6.13866568e-01 -2.44814336e-01 -1.01002979e+00 6.02480412e-01
7.82189250e-01 -9.89593506e-01 5.71803510e-01 -8.47055733e-01
9.44603980e-01 3.32954913e-01 -4.13692802e-01 -1.88309467e+00
-2.05234528e-01 -4.72623050e-01 -2.08673432e-01 1.27868342e+00
3.21153849e-01 -1.52639523e-01 8.08928251e-01 6.22631073e-01
-1.07854269e-01 -1.29251826e+00 -1.14361537e+00 -2.40013465e-01
-2.59685844e-01 -7.00408280e-01 6.02915645e-01 1.05645323e+00
5.69481492e-01 9.81226861e-01 -4.02948409e-01 -7.90375352e-01
2.52752721e-01 2.18241423e-01 5.54684639e-01 -1.02644718e+00
-1.44099653e-01 -3.25463474e-01 -4.43946511e-01 -2.71625429e-01
6.03080690e-01 -7.05988228e-01 7.35415220e-02 -9.46349859e-01
3.59962374e-01 -1.02427714e-01 -8.59938204e-01 6.28607512e-01
-1.37900025e-01 6.45434380e-01 2.04451457e-01 -5.16153455e-01
-1.30289662e+00 1.26358628e+00 7.33549058e-01 -2.67562181e-01
1.47322848e-01 -2.97550619e-01 -8.59708309e-01 8.71780753e-01
1.04729092e+00 -3.05855930e-01 -4.31282312e-01 -1.96025297e-01
4.94017124e-01 -4.56592776e-02 -4.38416824e-02 -9.93646264e-01
-1.95011035e-01 7.74434954e-02 4.52025890e-01 -1.43261477e-01
9.86612678e-01 -9.97596085e-01 1.48500398e-01 1.65689707e-01
-4.99000221e-01 4.57395256e-01 5.63793063e-01 1.60379946e-01
-4.20073450e-01 3.56125802e-01 6.03161693e-01 2.50888407e-01
-7.41601884e-01 3.65176380e-01 -4.65451986e-01 3.82235557e-01
1.17730510e+00 4.52073514e-01 -4.26082104e-01 -5.39509475e-01
-9.03019965e-01 2.67616838e-01 3.03515226e-01 5.60331464e-01
7.07727790e-01 -1.46268308e+00 -6.55396760e-01 -4.87161726e-02
6.53315187e-01 -7.87050724e-01 3.61278117e-01 8.40900362e-01
4.93251048e-02 1.37703419e-01 -5.36644161e-01 -3.29135060e-01
-1.18511283e+00 5.42329550e-01 6.08844221e-01 -5.90848744e-01
-1.53646559e-01 8.21674585e-01 1.06892586e-01 -3.81985694e-01
-1.60241332e-02 1.22617118e-01 -4.23649520e-01 1.46036252e-01
6.18174672e-01 3.05857100e-02 3.36151510e-01 -5.46158433e-01
-8.07517171e-01 6.08135425e-02 -9.84512568e-02 -2.77588248e-01
1.47265470e+00 1.58508137e-01 -1.55991763e-01 9.08355713e-01
1.27666950e+00 -8.69755819e-02 -4.85875666e-01 1.82601716e-02
-1.69982493e-01 5.31110503e-02 1.48280561e-01 -1.09301555e+00
-1.36856031e+00 8.47482324e-01 7.37677991e-01 2.82608509e-01
1.57665730e+00 -7.62744024e-02 6.68508828e-01 5.93374908e-01
4.90557730e-01 -1.30734503e+00 8.34202319e-02 7.68822253e-01
1.11899352e+00 -1.11955655e+00 -4.39749777e-01 -1.94022194e-01
-1.20690322e+00 5.71556687e-01 1.04527152e+00 -2.04727635e-01
6.54233754e-01 4.58326727e-01 3.31456512e-01 -3.96927953e-01
-1.64892912e+00 -4.09076139e-02 -5.09213246e-02 4.29076314e-01
7.63188183e-01 3.93168122e-01 -2.14415863e-01 1.22315621e+00
-2.44515687e-01 1.96727246e-01 9.02202725e-02 6.70006156e-01
-2.43067537e-02 -8.97059441e-01 1.83347017e-01 3.55585307e-01
-6.29354239e-01 -1.25647247e-01 -9.52388704e-01 4.01512325e-01
-2.15191439e-01 1.00798881e+00 5.62778749e-02 -1.06172872e+00
8.01348448e-01 3.18701953e-01 4.21904504e-01 -5.81612922e-02
-1.17238653e+00 -2.78944284e-01 5.69841325e-01 -9.92481053e-01
-3.65866512e-01 -3.78894299e-01 -1.12534928e+00 -3.78742814e-01
2.51737207e-01 6.53249562e-01 3.29477042e-01 8.17469597e-01
5.46439290e-01 8.04379880e-01 9.47553158e-01 -5.06679773e-01
-1.32864058e-01 -1.27060890e+00 -6.99777544e-01 6.28211617e-01
-1.36793137e-01 -7.25444198e-01 -3.22715223e-01 -3.52439880e-01] | [13.568730354309082, 2.3003458976745605] |
1cdc0d5b-4a35-4cbc-8d07-87fbf6e1a2cb | the-2021-image-similarity-dataset-and | 2106.09672 | null | https://arxiv.org/abs/2106.09672v4 | https://arxiv.org/pdf/2106.09672v4.pdf | The 2021 Image Similarity Dataset and Challenge | This paper introduces a new benchmark for large-scale image similarity detection. This benchmark is used for the Image Similarity Challenge at NeurIPS'21 (ISC2021). The goal is to determine whether a query image is a modified copy of any image in a reference corpus of size 1~million. The benchmark features a variety of image transformations such as automated transformations, hand-crafted image edits and machine-learning based manipulations. This mimics real-life cases appearing in social media, for example for integrity-related problems dealing with misinformation and objectionable content. The strength of the image manipulations, and therefore the difficulty of the benchmark, is calibrated according to the performance of a set of baseline approaches. Both the query and reference set contain a majority of "distractor" images that do not match, which corresponds to a real-life needle-in-haystack setting, and the evaluation metric reflects that. We expect the DISC21 benchmark to promote image copy detection as an important and challenging computer vision task and refresh the state of the art. Code and data are available at https://github.com/facebookresearch/isc2021 | ['Cristian Canton Ferrer', 'Ondřej Chum', 'Ismail Elezi', 'Laura Leal-Taixé', 'Maxim Maximov', 'Tomas Jenicek', 'Filip Radenovic', 'Lowik Chanussot', 'Zoë Papakipos', 'Ed Pizzi', 'Giorgos Tolias', 'Matthijs Douze'] | 2021-06-17 | null | null | null | null | ['image-similarity-detection'] | ['computer-vision'] | [ 6.29857719e-01 -1.01838671e-01 -3.91847491e-02 -2.64169097e-01
-6.68687880e-01 -6.68167949e-01 9.07691061e-01 1.71267703e-01
-8.65981698e-01 3.76405269e-01 1.94656625e-01 -1.86542064e-01
1.15234844e-01 -3.12173873e-01 -1.09704745e+00 -4.66397285e-01
-9.51689258e-02 2.95443892e-01 3.96937877e-01 -4.41634178e-01
5.80489814e-01 3.36454570e-01 -1.74801302e+00 3.82588655e-01
6.72367871e-01 8.56182098e-01 5.31471193e-01 7.22700238e-01
3.14703524e-01 5.69594920e-01 -7.75282919e-01 -9.34276700e-01
6.55108690e-01 -3.18236798e-01 -8.59269321e-01 9.89010483e-02
1.10109067e+00 -1.96237013e-01 -5.90222657e-01 1.82183206e+00
6.54343784e-01 -6.20134845e-02 5.12749493e-01 -1.53449249e+00
-6.55186653e-01 2.56134182e-01 -6.76628649e-01 9.08877194e-01
3.57916266e-01 2.84952998e-01 7.74051011e-01 -7.11257994e-01
1.21881938e+00 1.39709198e+00 3.88176441e-01 2.84102499e-01
-9.78047311e-01 -5.97269297e-01 -2.13651970e-01 7.03797281e-01
-1.20653212e+00 -5.03812432e-01 2.98863024e-01 -4.33191389e-01
5.15734315e-01 4.16224390e-01 3.05659562e-01 1.24617088e+00
2.77928382e-01 5.71182013e-01 8.59960675e-01 -4.01716322e-01
8.60603154e-02 3.64891797e-01 -4.81837802e-02 5.73840737e-01
2.95997262e-01 2.00648353e-01 -2.47286767e-01 -1.13593899e-01
4.75442499e-01 1.78205855e-02 -5.77563822e-01 -4.04564589e-01
-1.60898638e+00 6.45001054e-01 6.59791589e-01 4.30050045e-01
-6.00629710e-02 -1.37956277e-01 5.61132550e-01 8.15868258e-01
6.58189580e-02 7.22881079e-01 1.88541755e-01 -1.32346377e-01
-7.97206223e-01 5.29352903e-01 6.14261746e-01 1.27388275e+00
5.78719974e-01 -6.29040599e-01 -4.23356503e-01 8.38120520e-01
-3.00940603e-01 4.83700275e-01 8.52918386e-01 -1.17193329e+00
5.80052495e-01 4.28064466e-01 6.88712820e-02 -1.23143566e+00
-1.54206054e-02 -5.18286824e-01 -7.77652383e-01 3.23918283e-01
5.74170530e-01 3.59773546e-01 -7.43456125e-01 1.73019922e+00
2.79358238e-01 2.95542270e-01 -1.65358365e-01 9.32010949e-01
7.26534247e-01 -5.03000803e-03 -5.80223322e-01 6.39246255e-02
1.64453840e+00 -1.18171799e+00 -4.80881572e-01 -1.15384065e-01
7.27503419e-01 -1.00174296e+00 1.18412328e+00 2.47071728e-01
-1.15713227e+00 -4.39283311e-01 -1.12179983e+00 -1.41730785e-01
-5.32023311e-01 -1.82986826e-01 -1.11276932e-01 4.03977513e-01
-1.09664392e+00 6.45266593e-01 -1.72869042e-01 -6.99221909e-01
5.18070579e-01 -8.53324458e-02 -6.61881864e-01 -3.17299098e-01
-1.07721400e+00 1.20968044e+00 2.40958422e-01 -4.40486252e-01
-7.60290742e-01 -7.71852076e-01 -4.78646487e-01 -3.31166774e-01
4.36172009e-01 -5.51105559e-01 1.29166412e+00 -1.14787018e+00
-5.47826350e-01 1.73134935e+00 1.50024101e-01 -6.75597489e-01
1.05163312e+00 4.25793082e-02 -4.71904397e-01 3.80545706e-01
3.82286847e-01 7.91931689e-01 1.01560414e+00 -1.07829595e+00
-6.81356430e-01 -4.48471040e-01 1.80658177e-01 1.54438809e-01
-1.82419166e-01 3.42270732e-01 -9.37916696e-01 -8.11732113e-01
-3.80372733e-01 -1.28523278e+00 -6.38410673e-02 3.89020562e-01
-4.62185800e-01 1.94553718e-01 6.76092386e-01 -5.04288614e-01
8.31320047e-01 -2.18393564e+00 8.38645361e-03 2.67046750e-01
4.48556781e-01 4.16021019e-01 -7.07254171e-01 3.31013173e-01
-3.22336704e-01 9.37757045e-02 -1.66133747e-01 -2.39995718e-01
-1.23069197e-01 -6.65022284e-02 -2.78978318e-01 7.05356836e-01
-4.98953499e-02 9.49565589e-01 -8.91053081e-01 -3.92382681e-01
-1.97954457e-02 2.61561759e-02 -2.76396364e-01 1.69812590e-01
1.80170778e-02 2.56139427e-01 7.21792877e-02 4.41681743e-01
7.43960798e-01 -4.28001851e-01 -2.21410215e-01 -4.53269958e-01
1.96582720e-01 -1.68545529e-01 -1.05163527e+00 1.75211167e+00
1.38190642e-01 7.46662736e-01 -1.33519903e-01 -9.02145207e-01
4.55330282e-01 -1.65475309e-01 9.59509090e-02 -1.20461428e+00
2.71329492e-01 7.20557198e-02 2.29715094e-01 -6.65005803e-01
5.75178862e-01 5.22621870e-01 2.46535316e-01 5.09071887e-01
-2.86925789e-02 -9.39136185e-03 6.72745764e-01 6.04833663e-01
1.57554901e+00 -4.77368087e-01 3.34737659e-01 -3.47627968e-01
3.81136358e-01 -1.26719683e-01 2.86699086e-01 1.14953291e+00
-6.01569891e-01 9.85109448e-01 5.15667558e-01 -1.95809439e-01
-1.38977170e+00 -9.86463010e-01 -2.43605897e-01 1.02877223e+00
6.25962913e-01 -3.88716012e-01 -9.57843423e-01 -4.94364947e-01
1.35019705e-01 4.70189840e-01 -7.21396267e-01 -4.06035542e-01
-2.58683771e-01 -4.47327673e-01 6.03695035e-01 -9.01210085e-02
5.51876605e-01 -1.09648287e+00 -5.60601890e-01 -2.75419742e-01
-4.73497123e-01 -1.37666011e+00 -1.00183654e+00 -4.80201036e-01
-4.89187032e-01 -1.45920289e+00 -9.42102671e-01 -8.54341805e-01
7.87193060e-01 7.19198883e-01 1.38757694e+00 4.17115271e-01
-8.49659979e-01 4.32123810e-01 -3.49872261e-01 -3.88743341e-01
-7.11671412e-01 -2.61947185e-01 7.65713304e-02 8.49570781e-02
9.19213817e-02 -4.90152359e-01 -9.51444745e-01 5.94624102e-01
-1.27207911e+00 -7.29861408e-02 6.33771360e-01 6.97549701e-01
5.83980858e-01 -8.25405121e-02 6.89257607e-02 -8.21372271e-01
8.06867540e-01 -3.73611838e-01 -5.19208848e-01 4.80536789e-01
-5.69149435e-01 -2.43883231e-03 2.12682948e-01 -6.77915573e-01
-4.01664525e-01 -3.65202755e-01 2.11319059e-01 -7.24773586e-01
-7.16083348e-02 6.52066097e-02 3.10585722e-02 -3.80869120e-01
9.07081604e-01 3.54101956e-01 2.37247854e-01 -3.45728546e-01
3.27923834e-01 5.82891464e-01 1.16224074e+00 -1.28220886e-01
9.22051072e-01 5.68191528e-01 -2.01155633e-01 -6.01185381e-01
-4.92079109e-01 -6.72641754e-01 -4.21258926e-01 -1.48354858e-01
3.84803295e-01 -8.15129995e-01 -4.96441841e-01 5.57286441e-01
-1.12217069e+00 -5.99482730e-02 -1.91384628e-01 2.70737499e-01
-6.47111118e-01 7.78341889e-01 -3.14477921e-01 -1.24571942e-01
-5.18034816e-01 -1.25433707e+00 8.84424865e-01 1.07779883e-01
-1.03758685e-01 -6.12705469e-01 1.27751678e-02 4.92669612e-01
4.41061169e-01 1.14599101e-01 8.34913254e-01 -7.09288955e-01
-5.53945184e-01 -3.53855044e-01 -4.25939351e-01 3.34991455e-01
-8.06593597e-02 -1.79622352e-01 -6.95436597e-01 -6.77082419e-01
5.50081395e-03 -3.58571589e-01 8.22824478e-01 -1.42421901e-01
1.19417441e+00 -3.87345910e-01 -2.55870968e-01 5.28272688e-01
1.18283057e+00 -1.83487441e-02 1.02905285e+00 6.19114578e-01
3.14681262e-01 6.13462150e-01 6.17264032e-01 2.52230436e-01
2.20805749e-01 1.00844717e+00 7.20776498e-01 -7.31512159e-02
-2.91404694e-01 2.99004521e-02 1.26485348e-01 3.93585473e-01
7.97534883e-02 -2.05758244e-01 -7.46464729e-01 4.72378790e-01
-1.81141806e+00 -1.26458013e+00 1.93020143e-02 2.45457506e+00
8.64423096e-01 2.10949823e-01 2.42796898e-01 -9.08081755e-02
1.11752045e+00 2.67712753e-02 -7.87559867e-01 7.31883049e-02
-3.29574019e-01 -3.24244678e-01 6.28554165e-01 -5.62865883e-02
-1.07246852e+00 5.25371253e-01 5.67954683e+00 1.04097927e+00
-9.63071406e-01 2.08853453e-01 6.45085394e-01 -2.92725563e-01
-8.30325112e-02 -4.42992032e-01 -5.32690883e-01 9.56662118e-01
7.64769316e-01 -4.48481023e-01 6.30427063e-01 5.49395740e-01
-3.62266526e-02 -2.05913216e-01 -1.21618164e+00 1.40790761e+00
4.28510547e-01 -1.44902968e+00 1.27798140e-01 1.07337333e-01
6.79408252e-01 5.96109927e-01 3.59526843e-01 1.01766540e-02
-6.73620552e-02 -7.89925516e-01 7.86134005e-01 5.28810501e-01
8.23098779e-01 -4.22673404e-01 7.88089395e-01 1.95569262e-01
-7.00612903e-01 8.21448937e-02 -5.08060575e-01 5.19846082e-01
-1.58991367e-01 4.61606801e-01 -6.22895420e-01 1.27653271e-01
1.05912709e+00 6.84868634e-01 -1.13116312e+00 1.53421021e+00
3.10337067e-01 4.86031957e-02 -1.60427079e-01 2.70408750e-01
1.57332066e-02 -1.64133608e-01 8.75293612e-01 1.22091460e+00
2.18221962e-01 -4.40789193e-01 -3.43534142e-01 7.99937665e-01
-6.00508451e-01 1.30961224e-01 -7.89959610e-01 -6.19626641e-02
5.46328962e-01 1.24491930e+00 -5.85772455e-01 -3.38711739e-01
-4.26672846e-01 1.30493474e+00 2.93898225e-01 2.18162999e-01
-9.06423986e-01 -5.49100816e-01 6.35238647e-01 1.81280166e-01
4.16613102e-01 3.88456732e-01 2.84472615e-01 -1.20596766e+00
4.01408970e-01 -1.34082568e+00 2.70623654e-01 -8.76351297e-01
-1.43892145e+00 6.87974811e-01 -1.47796467e-01 -1.46712005e+00
-1.72784645e-02 -4.12879050e-01 -8.19473803e-01 3.36513788e-01
-1.38757503e+00 -6.76509976e-01 -6.35012805e-01 7.03977168e-01
2.72327453e-01 -4.71600205e-01 3.82039845e-01 4.96443063e-01
-3.17387700e-01 1.13754904e+00 4.78348047e-01 1.66184440e-01
1.11475480e+00 -1.00669754e+00 8.05585861e-01 9.03533399e-01
2.05601305e-01 5.17556190e-01 1.06949818e+00 -4.06098634e-01
-1.30467296e+00 -1.21566141e+00 6.01799011e-01 -4.28863794e-01
8.30609024e-01 -3.12849730e-01 -9.65991199e-01 4.04312909e-01
2.92046607e-01 3.13967645e-01 1.84761733e-01 -5.59984207e-01
-6.72773361e-01 -6.69698268e-02 -1.23458147e+00 8.22981000e-01
1.43803465e+00 -6.15408480e-01 -5.19004881e-01 6.60928071e-01
5.26204109e-01 -4.43648458e-01 -7.63750792e-01 3.06462407e-01
5.46051562e-01 -1.03938997e+00 1.24831331e+00 -6.34066880e-01
5.74689269e-01 -2.18173087e-01 -1.09167211e-01 -1.19443023e+00
-8.16892833e-02 -6.74599349e-01 1.82720885e-01 9.56789315e-01
8.48272368e-02 -4.39503998e-01 5.33166587e-01 3.25415432e-01
1.17224537e-01 -4.69322324e-01 -9.85458612e-01 -1.12576914e+00
-1.51986063e-01 9.69364867e-02 4.15201455e-01 9.63077903e-01
-2.69843280e-01 -1.61624455e-03 -1.84737295e-01 -1.17914438e-01
9.53201532e-01 2.66100932e-02 8.82395029e-01 -8.12707663e-01
-2.65967339e-01 -6.32446527e-01 -9.61915195e-01 -7.66600013e-01
-1.29968747e-01 -1.09422195e+00 -1.84769019e-01 -8.64589751e-01
5.41985810e-01 -3.92837524e-02 -3.14415842e-01 5.42237796e-02
-2.65284419e-01 6.11653268e-01 4.51200783e-01 4.93929327e-01
-1.02090037e+00 2.84365028e-01 1.33232510e+00 -4.82709438e-01
2.95882553e-01 5.89126423e-02 -7.18555331e-01 4.50257421e-01
7.64394045e-01 -4.77738440e-01 -3.42551231e-01 -2.41914824e-01
2.26250559e-01 -3.00719857e-01 6.04399860e-01 -1.04757142e+00
3.31424087e-01 1.20323487e-01 -5.09139858e-02 -3.77514035e-01
6.03792220e-02 -6.64158702e-01 1.85816929e-01 6.49560690e-01
-6.38290584e-01 4.33022171e-01 -5.32421730e-02 7.73847222e-01
-1.65358886e-01 -4.93408531e-01 1.02035081e+00 -3.15134585e-01
-8.77542138e-01 3.40094537e-01 4.24728170e-02 4.77343887e-01
1.11187541e+00 -3.18015724e-01 -9.24707413e-01 -5.43640137e-01
-2.58377939e-01 1.08341530e-01 7.63766348e-01 7.83644438e-01
6.93751693e-01 -1.31593752e+00 -8.56017530e-01 6.87796399e-02
7.84606755e-01 -6.27866209e-01 7.59557858e-02 1.07472849e+00
-6.03418469e-01 5.87279312e-02 -6.24368489e-01 -6.84270620e-01
-1.77469134e+00 7.28930771e-01 4.19373214e-01 3.34084034e-05
-7.00755537e-01 7.79977798e-01 3.51721734e-01 -2.03128412e-01
5.01624703e-01 -3.75479832e-02 1.62319727e-02 -1.35752901e-01
9.42989469e-01 2.98951924e-01 2.80037612e-01 -6.82986140e-01
-1.91734597e-01 2.91999370e-01 -6.08398855e-01 2.14407265e-01
9.05885398e-01 -3.55688721e-01 -1.60214588e-01 1.21693954e-01
1.61267006e+00 -4.28877652e-01 -7.45019972e-01 -7.37984896e-01
1.60206631e-02 -7.37377524e-01 -1.52491316e-01 -7.57643878e-01
-1.07368088e+00 5.54140449e-01 1.17329538e+00 -5.38564241e-03
9.33779418e-01 1.71040535e-01 6.51310503e-01 6.72344267e-01
3.99378151e-01 -1.08201623e+00 3.17778260e-01 3.15204114e-01
1.39972603e+00 -1.51021123e+00 1.11929839e-02 -1.80127576e-01
-6.87326789e-01 7.37927616e-01 5.01012862e-01 -1.59288019e-01
3.40459049e-01 1.53758690e-01 7.02373160e-04 -7.37935081e-02
-6.65105462e-01 -1.91460118e-01 3.06876302e-01 5.73283374e-01
-1.44211680e-01 -2.71765321e-01 -1.91815645e-01 1.10126063e-01
-3.82288396e-01 -1.75597057e-01 5.50827682e-01 8.43892276e-01
-3.98852855e-01 -8.47179532e-01 -3.40681195e-01 7.40186691e-01
-3.61273587e-01 -1.88470811e-01 -3.56402606e-01 4.65900481e-01
5.07832915e-02 7.66346335e-01 2.45969489e-01 -4.03502405e-01
4.52273637e-01 -5.19637942e-01 6.20001137e-01 -2.40140811e-01
-6.00698531e-01 -3.14555168e-01 -1.45246923e-01 -1.08914924e+00
-3.67535472e-01 -5.22579014e-01 -8.28340650e-01 -5.66102982e-01
-3.40639986e-02 -2.41603300e-01 6.14852011e-01 6.51928782e-01
6.78015411e-01 4.05123308e-02 4.91860032e-01 -7.72403300e-01
-6.70752227e-01 -8.35590124e-01 -5.19824982e-01 1.35778558e+00
3.68395507e-01 -6.49855018e-01 -4.03797656e-01 2.04225212e-01] | [10.649794578552246, 0.8357535004615784] |
3053d10e-3093-4bd4-b7f9-9058f39d5f1d | an-inexact-newton-krylov-algorithm-for | 1408.6299 | null | http://arxiv.org/abs/1408.6299v3 | http://arxiv.org/pdf/1408.6299v3.pdf | An inexact Newton-Krylov algorithm for constrained diffeomorphic image registration | We propose numerical algorithms for solving large deformation diffeomorphic
image registration problems. We formulate the nonrigid image registration
problem as a problem of optimal control. This leads to an infinite-dimensional
partial differential equation (PDE) constrained optimization problem.
The PDE constraint consists, in its simplest form, of a hyperbolic transport
equation for the evolution of the image intensity. The control variable is the
velocity field. Tikhonov regularization on the control ensures well-posedness.
We consider standard smoothness regularization based on $H^1$- or
$H^2$-seminorms. We augment this regularization scheme with a constraint on the
divergence of the velocity field rendering the deformation incompressible and
thus ensuring that the determinant of the deformation gradient is equal to one,
up to the numerical error.
We use a Fourier pseudospectral discretization in space and a Chebyshev
pseudospectral discretization in time. We use a preconditioned, globalized,
matrix-free, inexact Newton-Krylov method for numerical optimization. A
parameter continuation is designed to estimate an optimal regularization
parameter. Regularity is ensured by controlling the geometric properties of the
deformation field. Overall, we arrive at a black-box solver. We study spectral
properties of the Hessian, grid convergence, numerical accuracy, computational
efficiency, and deformation regularity of our scheme. We compare the designed
Newton-Krylov methods with a globalized preconditioned gradient descent. We
study the influence of a varying number of unknowns in time.
The reported results demonstrate excellent numerical accuracy, guaranteed
local deformation regularity, and computational efficiency with an optional
control on local mass conservation. The Newton-Krylov methods clearly
outperform the Picard method if high accuracy of the inversion is required. | ['George Biros', 'Andreas Mang'] | 2014-08-27 | null | null | null | null | ['constrained-diffeomorphic-image-registration'] | ['computer-vision'] | [ 3.22522298e-02 9.58906114e-02 2.78723061e-01 -3.83714177e-02
-6.13237321e-01 -1.99487045e-01 2.53611743e-01 7.09644184e-02
-7.16375470e-01 8.51788819e-01 -1.44187072e-02 1.66063830e-01
-3.49301428e-01 -6.61565602e-01 -5.82510293e-01 -9.93588805e-01
-1.39631450e-01 4.34967041e-01 -6.39733335e-04 -3.38405460e-01
2.78160751e-01 6.88821971e-01 -1.11275697e+00 -5.38839579e-01
1.06504369e+00 9.44278657e-01 -2.00745985e-01 6.30236804e-01
2.82337874e-01 4.22330856e-01 1.28488898e-01 1.84365705e-01
4.64695930e-01 -5.07545650e-01 -1.03851998e+00 3.19737792e-01
1.51518598e-01 -3.47119868e-01 1.63286641e-01 1.14881265e+00
5.77623785e-01 5.32693684e-01 7.89760768e-01 -8.46566796e-01
-3.42328131e-01 -9.46640819e-02 -4.78659809e-01 -1.13945007e-02
1.38038829e-01 1.12780454e-02 4.52538401e-01 -1.12232435e+00
9.39075947e-01 6.38773739e-01 1.04239857e+00 4.62585896e-01
-1.42169476e+00 9.42133218e-02 -5.87710738e-01 -3.71122688e-01
-1.59361088e+00 -1.57215014e-01 6.40683651e-01 -9.62939143e-01
4.44123745e-01 5.94568908e-01 5.82334936e-01 6.08980283e-02
4.85626400e-01 -1.89066187e-01 1.00673115e+00 -4.92585778e-01
3.88017207e-01 9.22659412e-02 2.62111157e-01 8.33628774e-01
1.64046422e-01 1.37395382e-01 1.51628582e-02 -6.04163706e-01
1.12740004e+00 -5.49123660e-02 -5.36928654e-01 -3.37448865e-01
-9.21166420e-01 1.01877105e+00 1.95988819e-01 7.12310731e-01
-5.73197067e-01 4.99263359e-03 1.34345636e-01 4.42600586e-02
8.43640268e-01 2.79037565e-01 -1.20379642e-01 1.60040557e-01
-7.54117012e-01 2.64082164e-01 8.10158968e-01 4.28572595e-01
9.20267820e-01 1.36403069e-01 -1.38581276e-01 5.43954372e-01
3.12137693e-01 5.24173677e-01 3.69632959e-01 -1.25627136e+00
-2.00102732e-01 4.44879293e-01 3.11401725e-01 -9.48366761e-01
-2.93940783e-01 -2.69411951e-01 -1.00007808e+00 5.99767864e-01
5.82597256e-01 -2.48148590e-01 -4.80594456e-01 1.50265884e+00
8.35486472e-01 2.24324599e-01 -2.25469738e-01 1.19149184e+00
6.94608688e-02 5.49269676e-01 1.98388975e-02 -7.83718109e-01
1.03547180e+00 -3.82794559e-01 -7.80126989e-01 4.34053779e-01
8.11543882e-01 -8.70146990e-01 7.93052912e-01 2.53206235e-03
-1.36994791e+00 -2.75264382e-01 -7.45726943e-01 2.02613920e-02
1.00775652e-01 1.88561291e-01 -6.47540987e-02 4.86120999e-01
-1.01238561e+00 1.20382404e+00 -9.84714627e-01 1.17050558e-01
-1.23456948e-01 4.79605436e-01 -4.33200389e-01 4.46140975e-01
-8.72463405e-01 9.23516452e-01 -1.61820903e-01 4.52519089e-01
-1.90671250e-01 -1.08732855e+00 -7.37074137e-01 -2.20561326e-01
-1.89904615e-01 -6.82953596e-01 5.22528529e-01 -1.01048386e+00
-1.70041990e+00 1.19704270e+00 -5.19830696e-02 -2.47356575e-03
9.43410099e-01 2.24373013e-01 2.20212147e-01 1.53001130e-01
1.01781331e-01 -1.56324714e-01 6.84994280e-01 -1.13499391e+00
2.88112104e-01 -3.87039661e-01 -4.67129499e-01 2.89939612e-01
-1.75761372e-01 -8.47484693e-02 7.69628538e-03 -7.24755228e-01
3.74724180e-01 -1.21412802e+00 -5.03156543e-01 2.09664837e-01
-4.98559922e-02 4.20997918e-01 5.18535852e-01 -1.06409860e+00
9.75675821e-01 -2.23638535e+00 4.87189412e-01 4.05541301e-01
1.09913595e-01 -5.03251217e-02 2.91592389e-01 2.90623158e-01
-6.55326545e-02 -2.10269675e-01 -9.09362555e-01 -2.08816037e-01
-3.19137871e-01 5.83460219e-02 7.69225284e-02 1.27309632e+00
-8.41147900e-02 4.80241567e-01 -5.71181774e-01 -4.21012938e-01
-3.85735966e-02 9.39080536e-01 -8.27805877e-01 1.51219770e-01
1.95017029e-02 1.02425599e+00 -6.81003332e-01 -3.71436588e-02
8.94841969e-01 -8.27626586e-02 6.71539828e-02 -3.40448052e-01
-8.03237975e-01 -3.05538386e-01 -1.52413094e+00 1.43486094e+00
-3.90082181e-01 7.37048462e-02 8.99152458e-01 -1.23316395e+00
7.21081614e-01 4.31274503e-01 8.92358601e-01 -2.08984554e-01
2.88400888e-01 7.43903935e-01 -2.56161749e-01 -4.66829985e-01
2.19338909e-01 -5.37379920e-01 4.27156180e-01 2.18031436e-01
-3.50526124e-01 -3.30050766e-01 3.21032642e-03 -9.93466079e-02
6.30983055e-01 8.25413987e-02 4.34555449e-02 -1.21018100e+00
8.87614787e-01 2.88332909e-01 4.13064092e-01 3.96488905e-02
1.37561545e-01 4.94259238e-01 5.10354698e-01 -4.28368747e-01
-1.03293324e+00 -3.88268828e-01 -7.81732380e-01 4.90483224e-01
1.45806596e-01 1.82226405e-01 -9.82485592e-01 1.38061553e-01
-4.47279811e-02 2.41390973e-01 -6.10898316e-01 5.19896075e-02
-1.01245272e+00 -1.06079829e+00 -3.94537225e-02 -1.03303147e-02
5.45084178e-01 -7.21970618e-01 -5.68612576e-01 2.44695097e-01
9.16534960e-02 -7.09995568e-01 -6.30667567e-01 -3.04344922e-01
-1.12182951e+00 -9.38225389e-01 -9.63517070e-01 -8.95949841e-01
8.96500468e-01 -5.32440603e-01 6.54870689e-01 3.31939340e-01
-4.12576228e-01 5.51508605e-01 3.58684629e-01 4.81640697e-01
-5.59585810e-01 -3.96447092e-01 5.08647449e-02 2.81549603e-01
-6.25237823e-01 -5.55123806e-01 -7.90615678e-01 3.10861439e-01
-8.99770021e-01 -3.31994206e-01 -7.83105940e-02 7.52011120e-01
8.58045042e-01 -2.30577558e-01 7.34252334e-02 -7.53753662e-01
4.63070869e-01 -1.51920050e-01 -9.28179622e-01 -2.04713598e-01
-5.97133219e-01 1.67563051e-01 5.20964503e-01 -5.50664365e-01
-8.77399385e-01 3.62884402e-01 -1.78995699e-01 -4.07813966e-01
4.02850330e-01 4.82337356e-01 2.30044529e-01 -9.21620905e-01
7.38020480e-01 2.82344997e-01 3.36705297e-01 -5.44000089e-01
8.51141065e-02 1.64733514e-01 3.20215970e-01 -7.08234429e-01
5.89392126e-01 7.65951872e-01 5.41627467e-01 -1.22066772e+00
-1.31983295e-01 -5.15183032e-01 -4.62462902e-01 -2.82283455e-01
1.07023299e+00 -3.19873095e-01 -9.89078283e-01 6.23978496e-01
-1.05105245e+00 -5.25402546e-01 -6.54676259e-01 6.19138122e-01
-8.33807051e-01 5.65926135e-01 -9.24488366e-01 -7.84984171e-01
-4.43173558e-01 -1.16327381e+00 6.40785158e-01 -2.72801161e-01
-1.69885810e-02 -1.40037596e+00 5.30889750e-01 3.51422243e-02
6.87653840e-01 7.93502212e-01 6.62658930e-01 1.34581968e-01
-3.47817600e-01 -9.74977911e-02 7.72743002e-02 3.17136943e-01
-1.03819422e-01 -1.83621019e-01 -4.80303347e-01 -3.24144334e-01
9.42322135e-01 2.05820739e-01 4.10696149e-01 6.08735442e-01
5.77739418e-01 -6.70253634e-01 -2.10197806e-01 8.03977132e-01
1.84205282e+00 -3.26182008e-01 2.13149935e-01 -3.79977860e-02
7.28068292e-01 6.97934866e-01 3.14481527e-01 6.07238293e-01
-8.81166086e-02 7.43559778e-01 2.01042727e-01 -9.17727202e-02
-2.78708506e-02 3.22193831e-01 9.14856419e-02 7.78484404e-01
-5.75846612e-01 4.95912731e-01 -9.54067588e-01 4.77363825e-01
-1.64940476e+00 -4.91521418e-01 -7.56481826e-01 2.60952306e+00
1.09404171e+00 -3.37255418e-01 -3.95708643e-02 1.14521034e-01
6.11771107e-01 -2.72418320e-01 -1.19271480e-01 -3.05057019e-01
1.97845444e-01 1.75149202e-01 7.34680355e-01 1.39372563e+00
-7.42509425e-01 2.54628181e-01 5.46972561e+00 4.84387994e-01
-1.41132426e+00 3.65368336e-01 3.23270291e-01 3.91720533e-01
-3.57675344e-01 1.57792233e-02 -5.05101800e-01 5.99207222e-01
7.02794731e-01 -4.66388762e-02 3.91718298e-01 6.04659081e-01
4.78934199e-01 -1.60579637e-01 -5.26275635e-01 6.17547810e-01
-1.63103297e-01 -1.36692965e+00 -5.04884303e-01 2.89487928e-01
8.40354800e-01 -1.14126422e-01 -1.10456273e-01 -4.55922663e-01
-3.94761413e-01 -7.70521879e-01 4.32690680e-01 5.71371377e-01
9.02780056e-01 -4.33950454e-01 4.14298415e-01 3.18209380e-01
-1.16078329e+00 3.98443937e-01 9.26024839e-03 -5.25336377e-02
4.76099610e-01 7.57408738e-01 -1.06701657e-01 2.93740928e-01
1.53915033e-01 3.82044077e-01 2.06589907e-01 7.82021582e-01
2.85749823e-01 2.01865062e-01 -5.63253164e-01 3.15000921e-01
1.61989376e-01 -1.28131020e+00 8.46648753e-01 8.14651430e-01
3.48160446e-01 8.64857316e-01 2.10639313e-01 9.63870049e-01
2.55205333e-01 5.49791813e-01 -2.26834267e-01 4.87737417e-01
-2.41940439e-01 1.08298051e+00 -7.38596380e-01 -1.61512773e-02
-1.28803268e-01 8.51512492e-01 1.07581578e-01 4.33349699e-01
-2.91744977e-01 7.30072707e-02 5.30223668e-01 6.22108638e-01
1.34089831e-02 -4.53588247e-01 -3.95732462e-01 -1.13534498e+00
9.79978144e-02 -2.80752093e-01 3.75314265e-01 -2.38919720e-01
-9.55069125e-01 5.84844828e-01 -1.73580982e-02 -8.82458925e-01
-2.97948807e-01 -5.26241541e-01 -5.90997279e-01 1.27865684e+00
-1.37273741e+00 -8.88309598e-01 1.04463554e-03 8.38822901e-01
-1.57833144e-01 4.60985690e-01 7.62159169e-01 4.26157475e-01
-2.93888271e-01 -5.81501424e-02 2.03661814e-01 2.80810650e-02
3.39138985e-01 -9.28443849e-01 -5.72698534e-01 7.30560660e-01
-9.24575388e-01 5.61384678e-01 9.55999017e-01 -7.98748493e-01
-1.45750523e+00 -6.87501431e-01 1.12543905e+00 5.76974973e-02
6.74671590e-01 9.12233721e-03 -1.28303027e+00 7.02036500e-01
-5.02995290e-02 6.40630126e-01 2.35030398e-01 -5.51376462e-01
4.31627721e-01 1.13257162e-01 -1.51351190e+00 1.53449371e-01
2.86510676e-01 -3.10518175e-01 -2.32923240e-03 5.86698890e-01
2.26731896e-01 -6.49718106e-01 -1.31182075e+00 4.31755304e-01
2.13931426e-01 -6.53750420e-01 8.93325865e-01 -3.26076210e-01
1.27966121e-01 -2.58940011e-01 2.23117664e-01 -7.83687055e-01
-1.92942590e-01 -1.12345481e+00 4.18190509e-01 7.97588229e-01
2.69153640e-02 -8.18435192e-01 5.93982399e-01 1.12077832e+00
-1.73554927e-01 -8.75162721e-01 -1.12193286e+00 -6.31183147e-01
4.49335694e-01 1.35022536e-01 -4.05908346e-01 1.30021954e+00
2.59836167e-01 -2.69412041e-01 -2.44138986e-01 2.73017794e-01
9.71317589e-01 -1.29789308e-01 1.06181517e-01 -1.15171504e+00
-2.51608223e-01 -1.53977305e-01 -2.19595462e-01 -4.76445287e-01
1.52188867e-01 -6.83283746e-01 8.84691477e-02 -1.04731536e+00
-2.30816543e-01 -6.19605243e-01 2.83776909e-01 -2.23142281e-02
-2.94735419e-05 4.25358891e-01 -3.24128747e-01 4.25565660e-01
4.40721959e-01 3.95778924e-01 1.38595366e+00 3.56394053e-01
-5.96264422e-01 -3.05334311e-02 2.60653049e-01 6.92665279e-01
5.83860993e-01 -3.33739340e-01 -1.27576485e-01 -2.98222274e-01
3.04145247e-01 4.09354389e-01 3.65669221e-01 -6.43309653e-01
3.02828878e-01 -2.44325846e-02 -2.44805470e-01 2.21246257e-01
2.76297331e-01 -5.79249859e-01 5.37501395e-01 7.81951725e-01
-1.80016771e-01 -1.14723764e-01 3.54601294e-02 1.16794907e-01
-9.07573402e-02 -4.93475288e-01 1.45107901e+00 -6.74531981e-02
-6.04725815e-02 4.50522095e-01 -4.53632504e-01 1.63988858e-01
7.21573889e-01 -8.81873593e-02 3.17805439e-01 -4.19401564e-02
-1.01721442e+00 -2.86269665e-01 6.61480367e-01 -6.03138745e-01
3.30075532e-01 -1.34881234e+00 -8.18014324e-01 4.83011395e-01
-5.49821079e-01 -9.31970328e-02 4.48911786e-01 1.39000368e+00
-1.05294526e+00 -1.64079428e-01 -1.72972083e-01 -4.79740411e-01
-9.69513178e-01 3.45243126e-01 8.65446150e-01 -2.96128333e-01
-4.97785866e-01 6.12335265e-01 -1.32597357e-01 -2.28972301e-01
-1.60575226e-01 -1.07920833e-01 -5.83181120e-02 -5.57792149e-02
1.82334512e-01 8.47151577e-01 1.79277107e-01 -8.55279863e-01
-3.12845618e-01 1.23548043e+00 6.40445292e-01 -4.01166290e-01
1.22883558e+00 -8.08361843e-02 -8.21570039e-01 4.56553549e-02
1.57279325e+00 2.83083379e-01 -1.25754774e+00 9.24353227e-02
-3.26726228e-01 -1.47361144e-01 4.64978993e-01 -7.96496645e-02
-1.04828227e+00 2.40680307e-01 4.48112637e-01 2.87049621e-01
8.63758326e-01 -4.33907330e-01 6.08298123e-01 -2.86103725e-01
-1.26111284e-01 -9.76462483e-01 -4.38340068e-01 3.74208868e-01
9.97312129e-01 -1.03865623e+00 2.85696566e-01 -7.90766120e-01
-9.38267931e-02 1.13835037e+00 1.55505165e-01 -6.75711334e-01
1.03311598e+00 3.96626472e-01 4.50747237e-02 -9.77298990e-02
2.15013802e-01 7.90852830e-02 4.69285965e-01 -1.62297711e-01
7.03227878e-01 -2.87660837e-01 -1.36232960e+00 -6.52932376e-02
1.35180488e-01 1.47371054e-01 2.81618893e-01 7.89238334e-01
-1.86788887e-01 -1.05605745e+00 -5.20765305e-01 -1.37463361e-01
-5.23263216e-01 1.12370171e-01 2.31321380e-01 6.78018272e-01
9.45636854e-02 4.46805984e-01 1.30824903e-02 6.37175202e-01
2.91022748e-01 8.25653374e-02 5.72033525e-01 -3.18705589e-01
-6.34927750e-01 2.58185685e-01 -2.15544298e-01 -5.45528173e-01
-5.93147576e-01 -7.50695229e-01 -1.55722070e+00 -3.09039861e-01
-1.20150961e-01 5.78469574e-01 8.53541791e-01 8.98112953e-01
9.65287536e-02 1.65439565e-02 5.42866111e-01 -1.09200919e+00
-5.86402297e-01 -5.99159777e-01 -8.26966405e-01 5.02914190e-01
4.30862546e-01 -3.60765427e-01 -7.42120743e-01 1.72043011e-01] | [6.513278961181641, 3.3870885372161865] |
50821bff-f7df-4b76-9ac9-de706a1c2699 | hybrid-sequence-to-sequence-model-for-video | 2010.05069 | null | https://arxiv.org/abs/2010.05069v2 | https://arxiv.org/pdf/2010.05069v2.pdf | Hybrid-S2S: Video Object Segmentation with Recurrent Networks and Correspondence Matching | One-shot Video Object Segmentation~(VOS) is the task of pixel-wise tracking an object of interest within a video sequence, where the segmentation mask of the first frame is given at inference time. In recent years, Recurrent Neural Networks~(RNNs) have been widely used for VOS tasks, but they often suffer from limitations such as drift and error propagation. In this work, we study an RNN-based architecture and address some of these issues by proposing a hybrid sequence-to-sequence architecture named HS2S, utilizing a dual mask propagation strategy that allows incorporating the information obtained from correspondence matching. Our experiments show that augmenting the RNN with correspondence matching is a highly effective solution to reduce the drift problem. The additional information helps the model to predict more accurate masks and makes it robust against error propagation. We evaluate our HS2S model on the DAVIS2017 dataset as well as Youtube-VOS. On the latter, we achieve an improvement of 11.2pp in the overall segmentation accuracy over RNN-based state-of-the-art methods in VOS. We analyze our model's behavior in challenging cases such as occlusion and long sequences and show that our hybrid architecture significantly enhances the segmentation quality in these difficult scenarios. | ['Andreas Dengel', 'Joern Hees', 'Federico Raue', 'Stanislav Frolov', 'Fatemeh Azimi'] | 2020-10-10 | null | null | null | null | ['one-shot-visual-object-segmentation'] | ['computer-vision'] | [ 3.51156414e-01 -2.63760298e-01 -2.48689160e-01 -3.25608075e-01
-6.78036273e-01 -2.89704651e-01 4.64031696e-02 -4.79158223e-01
-6.00587964e-01 3.51831764e-01 -1.38597786e-01 -1.29403815e-01
5.55042446e-01 -3.76192719e-01 -9.30879414e-01 -7.02943027e-01
2.25429624e-01 2.68725544e-01 1.09527552e+00 -1.93258852e-01
4.90254238e-02 3.04302871e-01 -1.54805052e+00 2.38433301e-01
8.73754144e-01 1.14783490e+00 6.09727323e-01 6.29049957e-01
-4.71233934e-01 1.00345826e+00 -7.86098182e-01 -2.93839961e-01
5.56403995e-01 -5.02843082e-01 -7.67501950e-01 1.59783855e-01
5.62571526e-01 -3.95349532e-01 -5.21317959e-01 1.12788832e+00
4.22701210e-01 1.94417015e-01 2.18294099e-01 -1.17670739e+00
-1.93725780e-01 7.50760674e-01 -8.76525044e-01 4.76094991e-01
-3.24191675e-02 4.57034945e-01 8.56478810e-01 -7.41023421e-01
9.79486465e-01 1.21698523e+00 7.95127213e-01 1.02673054e+00
-1.03148150e+00 -6.42912388e-01 5.63316703e-01 4.37973469e-01
-1.31523633e+00 -5.02902985e-01 6.94442153e-01 -4.07495588e-01
7.17501760e-01 8.34827125e-02 8.02494228e-01 1.07005107e+00
7.82583281e-02 1.68471515e+00 5.90692401e-01 -1.89524099e-01
9.58284363e-02 -2.45529383e-01 2.25797653e-01 5.43316603e-01
-1.24581240e-01 -2.51219928e-01 -5.07088244e-01 5.46253443e-01
8.03431571e-01 1.11925304e-01 -5.44092417e-01 -2.57279634e-01
-1.14942586e+00 5.13205469e-01 6.40262902e-01 3.15321088e-01
-4.43854868e-01 4.66339916e-01 5.27100742e-01 -2.46960483e-02
5.90519369e-01 1.36889607e-01 -4.07678425e-01 -2.21289113e-01
-1.51478159e+00 2.41742045e-01 4.76624608e-01 1.10786736e+00
4.24047470e-01 4.22627419e-01 -4.11428005e-01 8.91770720e-01
3.38982940e-01 3.77559453e-01 4.87246066e-01 -1.14385426e+00
6.31823540e-01 4.50825840e-01 8.89667124e-02 -7.02509701e-01
-2.66231686e-01 -5.98947525e-01 -7.45716870e-01 7.92841837e-02
4.59396362e-01 -2.94268399e-01 -1.54882658e+00 1.79647708e+00
2.44440392e-01 7.72587657e-01 -8.38313550e-02 1.08176422e+00
1.10267329e+00 8.43573868e-01 1.27277285e-01 -2.19133362e-01
1.02371013e+00 -1.48728931e+00 -1.07842433e+00 -3.73338521e-01
3.94606799e-01 -4.97718394e-01 7.36825466e-01 8.10678825e-02
-1.26463807e+00 -6.11771226e-01 -1.03321075e+00 5.16396537e-02
-6.11533178e-03 -5.35527840e-02 1.40288219e-01 4.44463462e-01
-1.25709021e+00 7.91046500e-01 -1.11443615e+00 -2.20123172e-01
7.77970672e-01 3.53494018e-01 1.89967111e-01 1.65202320e-02
-1.07656181e+00 5.35834908e-01 3.26019287e-01 4.19174612e-01
-8.70984316e-01 -7.12157011e-01 -7.14887202e-01 7.60168135e-02
8.48650575e-01 -6.13440037e-01 1.45956039e+00 -1.05650210e+00
-1.46403837e+00 5.33305585e-01 -5.40751159e-01 -8.68705392e-01
8.95441234e-01 -3.05774778e-01 -1.79960206e-01 2.02921987e-01
1.23106286e-01 1.21928632e+00 8.92097771e-01 -1.19654107e+00
-8.71588111e-01 -7.01608509e-02 -1.97628409e-01 1.36231810e-01
5.83362021e-03 1.39519542e-01 -1.14858365e+00 -8.14196110e-01
1.72231525e-01 -9.96166825e-01 -5.45721710e-01 3.78812067e-02
-4.90877956e-01 -1.99224964e-01 1.00344622e+00 -9.27368283e-01
1.39664745e+00 -2.04815960e+00 3.59586835e-01 -1.70905337e-01
2.52788782e-01 7.52532542e-01 -2.31247440e-01 -1.89572185e-01
2.28970900e-01 2.31785253e-01 -5.63144982e-01 -7.91955769e-01
-3.75582069e-01 2.39077970e-01 8.96902056e-04 2.18657717e-01
2.84117758e-01 1.26162767e+00 -6.15355134e-01 -6.75407946e-01
1.07691504e-01 4.61549580e-01 -3.07432741e-01 3.82878870e-01
-6.81773305e-01 4.78242636e-01 -1.39877141e-01 7.21156418e-01
4.88001913e-01 -3.69276255e-01 -7.65085891e-02 4.74249944e-02
-2.71242503e-02 -2.83804606e-03 -1.14052117e+00 1.98790038e+00
5.07489666e-02 1.02648449e+00 4.13687527e-02 -7.32665598e-01
6.79880917e-01 2.87190437e-01 5.86582482e-01 -6.75326586e-01
2.74216563e-01 6.91583902e-02 3.43751274e-02 -5.47879100e-01
5.81532896e-01 2.65680879e-01 4.14665908e-01 3.00661679e-02
-2.98863044e-03 2.87265062e-01 3.92871201e-01 2.46564686e-01
1.03922856e+00 3.85939300e-01 -2.74829566e-01 -3.09987441e-02
4.57094461e-01 -3.42853367e-02 1.16703856e+00 8.35659981e-01
-5.56064963e-01 9.72767532e-01 4.09348488e-01 -3.67706954e-01
-9.90471184e-01 -6.55116558e-01 2.46788800e-01 8.51146162e-01
4.24924225e-01 -3.42364639e-01 -1.14739048e+00 -7.75910378e-01
-3.57031137e-01 5.68694592e-01 -4.66248214e-01 2.54125029e-01
-9.44299281e-01 -6.63274229e-01 5.30826211e-01 8.85161579e-01
8.08034956e-01 -1.23371232e+00 -7.62267232e-01 4.42249626e-01
-4.65455234e-01 -1.51221204e+00 -5.88555634e-01 -1.49110630e-01
-1.00270307e+00 -9.39356506e-01 -1.10384965e+00 -8.54416490e-01
4.49403524e-01 4.35442954e-01 1.10350311e+00 1.47974178e-01
-1.62143707e-01 6.55531287e-02 -2.88945347e-01 -2.97855139e-01
-3.94895345e-01 3.30113918e-01 -2.57735550e-01 1.30822718e-01
1.35657459e-01 -2.61908859e-01 -8.47079635e-01 4.67223078e-01
-9.01690602e-01 2.32519373e-01 3.50909859e-01 6.17136180e-01
7.14287519e-01 -3.75337422e-01 4.86993909e-01 -1.03902733e+00
1.56906307e-01 -3.27387899e-01 -6.95312202e-01 3.06901872e-01
-3.42382282e-01 -9.20974240e-02 3.15865457e-01 -3.83970231e-01
-9.63579655e-01 3.07100862e-01 -4.27435458e-01 -9.49817061e-01
-1.12888878e-02 1.19196236e-01 -8.27352330e-02 -2.20898882e-01
2.77684957e-01 2.11395979e-01 8.10259115e-03 -6.35116279e-01
1.36217222e-01 5.10813653e-01 7.60653019e-01 -5.66290841e-02
5.81135750e-01 3.72511685e-01 -1.93030402e-01 -6.72458649e-01
-7.23996043e-01 -7.01433539e-01 -5.48693538e-01 -5.28747320e-01
1.08903837e+00 -1.11416185e+00 -7.41425991e-01 7.94188678e-01
-1.37048841e+00 -5.49278200e-01 -2.30905235e-01 1.77728787e-01
-4.54226345e-01 3.55462819e-01 -8.50974619e-01 -7.47904778e-01
-4.93354797e-01 -1.52760899e+00 8.07876945e-01 4.42850679e-01
8.98035169e-02 -6.97596729e-01 -2.66005605e-01 3.19732994e-01
4.40201759e-01 2.96817392e-01 3.19571733e-01 -5.96777141e-01
-1.00987136e+00 1.02782317e-01 -2.40774542e-01 4.54421133e-01
-1.50640756e-01 4.29714099e-02 -8.50344598e-01 -1.76674917e-01
1.37967216e-02 1.30598307e-01 1.24652219e+00 6.92445159e-01
1.33172286e+00 4.78418097e-02 -2.91914880e-01 7.96363771e-01
1.43496728e+00 4.99414831e-01 8.65739942e-01 3.27690214e-01
1.16964984e+00 3.09821308e-01 7.58790791e-01 1.19395092e-01
3.18101585e-01 7.47059047e-01 5.57885170e-01 -1.38775274e-01
-4.13824260e-01 -9.12955031e-03 3.05843443e-01 8.18284631e-01
-7.90674239e-02 -6.31741226e-01 -8.65146041e-01 7.57770538e-01
-2.26721239e+00 -8.98066282e-01 -2.25188881e-01 1.85276282e+00
6.52039528e-01 3.13449949e-01 2.84480542e-01 -8.18877518e-02
1.10772324e+00 5.41590810e-01 -8.66184652e-01 -3.63835543e-02
-2.47244388e-01 -1.91297576e-01 6.70152783e-01 3.73717576e-01
-1.15598989e+00 1.48167360e+00 6.08738852e+00 7.98868060e-01
-1.08987355e+00 1.18874073e-01 8.10452044e-01 -1.95083082e-01
-1.84572756e-03 -3.89014632e-01 -9.90987241e-01 6.55684292e-01
8.82333279e-01 3.50850254e-01 2.56136924e-01 6.21028900e-01
3.20843041e-01 -9.79286432e-02 -8.89423251e-01 1.15603197e+00
1.74415231e-01 -1.69561791e+00 -1.13957293e-01 -3.53929788e-01
8.27789724e-01 4.98421669e-01 -1.28723621e-01 1.51016250e-01
6.01668246e-02 -7.08470404e-01 9.54656959e-01 4.15006846e-01
6.44386232e-01 -6.08299017e-01 8.70356023e-01 1.78792700e-01
-1.29515207e+00 1.78045139e-01 -1.43807158e-01 3.03009331e-01
6.04444027e-01 2.67417729e-01 -6.52408600e-01 3.44428450e-01
8.30671668e-01 1.03219283e+00 -5.47764659e-01 1.43920028e+00
-1.88464105e-01 7.26832926e-01 -3.09515178e-01 6.85161054e-02
5.57334363e-01 -8.62111822e-02 7.06685960e-01 1.31675255e+00
-2.82109622e-03 4.78543527e-02 2.45077744e-01 8.81816149e-01
-4.38727647e-01 -1.27032727e-01 -2.16870964e-01 1.30334124e-01
3.05009186e-01 9.70030665e-01 -1.18021262e+00 -5.59377015e-01
-3.42837781e-01 1.12469757e+00 1.99589245e-02 5.63730717e-01
-1.01478219e+00 -2.76911318e-01 8.00207615e-01 -6.07701344e-03
8.28465700e-01 -1.41545430e-01 -2.04264805e-01 -1.14179265e+00
2.43300766e-01 -6.96077228e-01 8.87308642e-02 -6.27994537e-01
-7.20347106e-01 8.61629367e-01 -2.85120457e-01 -1.15760660e+00
-2.73569673e-01 -1.72200993e-01 -4.66803223e-01 5.28109789e-01
-1.67672014e+00 -8.92618954e-01 -2.54988611e-01 3.85445505e-01
1.18853831e+00 3.53520713e-03 1.19266033e-01 5.28032541e-01
-9.94313538e-01 5.78160167e-01 3.24472189e-02 4.79366243e-01
4.68047976e-01 -1.05191422e+00 1.11139953e+00 1.24884140e+00
2.48423323e-01 1.61354423e-01 6.88173056e-01 -9.35840607e-01
-1.09259295e+00 -1.29749274e+00 5.42383969e-01 -1.75613850e-01
3.44124079e-01 -3.85032833e-01 -1.12581193e+00 7.54356205e-01
1.44409567e-01 3.07919830e-01 8.35481733e-02 -4.60817337e-01
-3.52258086e-02 -1.23641463e-02 -9.65386212e-01 7.11872339e-01
1.41667163e+00 -1.99448019e-01 -2.95123607e-01 3.12503986e-02
1.18544722e+00 -8.67686152e-01 -5.93422234e-01 4.84796911e-01
3.73075426e-01 -1.00079131e+00 8.63505304e-01 -4.84345973e-01
4.87211496e-01 -5.36103487e-01 1.30988985e-01 -9.99938726e-01
-3.21410187e-02 -8.34580839e-01 -2.48182580e-01 1.21830440e+00
2.87643850e-01 -1.65473580e-01 1.16367781e+00 6.60800457e-01
-2.89556175e-01 -9.34110761e-01 -9.78642941e-01 -7.72986174e-01
-3.28229249e-01 -5.44591963e-01 4.36744988e-01 5.53919315e-01
-7.38395154e-01 1.93759888e-01 -5.94646275e-01 7.66500607e-02
5.25418282e-01 -1.49517938e-01 6.99811816e-01 -9.90192592e-01
-3.80113423e-02 -4.06716108e-01 -5.28905213e-01 -1.63396823e+00
2.27234870e-01 -4.34082448e-01 4.42699939e-01 -1.62320745e+00
1.56074902e-02 -2.25606725e-01 -2.84176677e-01 3.21139902e-01
-3.85857999e-01 4.43427801e-01 7.76645839e-01 3.83368790e-01
-1.13160908e+00 4.83055741e-01 1.15491486e+00 -1.97449505e-01
-4.48537588e-01 2.39913642e-01 -1.80336773e-01 8.74417782e-01
5.78503191e-01 -6.04036689e-01 -2.60394663e-01 -7.18330204e-01
-1.88530564e-01 1.39220253e-01 1.50594667e-01 -1.15062225e+00
5.51471114e-01 9.75552276e-02 2.37834290e-01 -9.34436500e-01
3.02093238e-01 -8.34888339e-01 7.44728297e-02 5.98089755e-01
-2.18178406e-01 3.88134643e-02 3.14694375e-01 7.77963459e-01
-2.05908313e-01 -2.59570569e-01 7.60584056e-01 -1.44543320e-01
-1.17727566e+00 4.84467268e-01 -2.46351197e-01 2.49868855e-01
1.01970327e+00 -3.58742058e-01 -1.42306434e-02 -3.40148032e-01
-6.15907669e-01 5.84711552e-01 3.79506648e-01 5.75461566e-01
7.83546209e-01 -9.20458436e-01 -6.75901532e-01 -5.60637414e-02
-2.08786115e-01 4.42656100e-01 2.87999809e-01 8.05734694e-01
-7.22564936e-01 2.55587727e-01 -4.69966568e-02 -1.10933971e+00
-1.32267320e+00 3.47317308e-01 4.16271299e-01 -8.86803791e-02
-1.01939368e+00 1.21931970e+00 2.39791125e-01 1.84173808e-01
7.26470590e-01 -4.92056489e-01 -3.99066329e-01 2.69840732e-02
6.20918334e-01 4.36415464e-01 -1.70580044e-01 -7.89055288e-01
-3.80616397e-01 7.44199157e-01 -3.69286418e-01 -6.92423582e-02
1.33649063e+00 -3.03916752e-01 2.25900441e-01 5.48207760e-01
9.35311496e-01 -5.06911278e-01 -1.88217270e+00 -5.08876085e-01
1.78757086e-01 -5.55050850e-01 -6.65684715e-02 -6.88534141e-01
-1.87643051e+00 7.89061785e-01 5.45644939e-01 -1.14335269e-01
1.05484724e+00 -1.72333598e-01 1.34878147e+00 4.21823524e-02
1.09142743e-01 -1.00769460e+00 -8.86174440e-02 4.19853508e-01
4.40671742e-01 -1.35885477e+00 -2.68548816e-01 -5.11966228e-01
-8.63895297e-01 9.42590177e-01 7.03721881e-01 -1.74058184e-01
1.62410140e-01 1.03186883e-01 1.92803070e-01 -4.28303666e-02
-5.93103588e-01 -3.49754035e-01 2.79647082e-01 2.21041441e-01
1.11917518e-01 -2.47635797e-01 -1.15607552e-01 3.54379982e-01
3.01562190e-01 1.16639204e-01 5.31375110e-01 8.87178421e-01
-2.90079623e-01 -8.97085130e-01 -2.40400955e-01 3.80804658e-01
-6.01079464e-01 -4.86483425e-02 -1.61472842e-01 6.74783647e-01
2.91447937e-02 8.40587556e-01 1.40150741e-01 -4.10719186e-01
2.94954568e-01 9.94879529e-02 1.69253483e-01 -3.89549732e-01
-8.24116051e-01 2.96045572e-01 -1.37161016e-01 -9.28564191e-01
-7.69227862e-01 -7.95189857e-01 -1.48761761e+00 -2.74462700e-01
-4.57392663e-01 -9.67285410e-02 6.93633974e-01 1.11228383e+00
4.31961179e-01 1.11279523e+00 3.49822223e-01 -9.70925450e-01
-1.12682171e-01 -7.15423584e-01 -2.17791557e-01 2.94003934e-01
5.62530816e-01 -3.85099113e-01 -7.83156455e-02 5.87919280e-02] | [9.205306053161621, -0.08392012119293213] |
6f0343db-c4b2-498a-ad70-5c08605f53cc | compositional-processing-emerges-in-neural | 2105.08961 | null | https://arxiv.org/abs/2105.08961v1 | https://arxiv.org/pdf/2105.08961v1.pdf | Compositional Processing Emerges in Neural Networks Solving Math Problems | A longstanding question in cognitive science concerns the learning mechanisms underlying compositionality in human cognition. Humans can infer the structured relationships (e.g., grammatical rules) implicit in their sensory observations (e.g., auditory speech), and use this knowledge to guide the composition of simpler meanings into complex wholes. Recent progress in artificial neural networks has shown that when large models are trained on enough linguistic data, grammatical structure emerges in their representations. We extend this work to the domain of mathematical reasoning, where it is possible to formulate precise hypotheses about how meanings (e.g., the quantities corresponding to numerals) should be composed according to structured rules (e.g., order of operations). Our work shows that neural networks are not only able to infer something about the structured relationships implicit in their training data, but can also deploy this knowledge to guide the composition of individual meanings into composite wholes. | ['Jianfeng Gao', 'Paul Smolensky', 'Nebojsa Jojic', 'Eric Rosen', 'Hamid Palangi', 'Roland Fernandez', 'Jacob Russin'] | 2021-05-19 | null | null | null | null | ['mathematical-reasoning'] | ['natural-language-processing'] | [ 4.91037190e-01 4.04834837e-01 1.60863355e-01 -5.39579272e-01
2.60313898e-01 -1.01555645e+00 8.40530574e-01 6.19228482e-01
-3.03496808e-01 4.36632484e-01 5.00787497e-01 -8.09245825e-01
-1.73901886e-01 -1.25377238e+00 -8.32226455e-01 -4.44370389e-01
-1.50109187e-01 3.93829674e-01 -2.16362439e-02 -6.01975620e-01
3.89069766e-01 4.42026705e-01 -1.68860888e+00 5.86677670e-01
8.99835348e-01 8.75933349e-01 4.07687396e-01 5.66420197e-01
-5.86444855e-01 1.11242008e+00 -4.03155029e-01 -4.18730944e-01
-8.43123645e-02 -6.52896643e-01 -9.46081817e-01 -1.20119475e-01
3.02479416e-01 -2.90166855e-01 5.76828681e-02 1.22087598e+00
-3.24490458e-01 1.37209937e-01 9.24358308e-01 -6.81733012e-01
-1.35574329e+00 1.20414650e+00 2.36906007e-01 1.40926346e-01
3.86432260e-01 -5.68575040e-02 1.41576040e+00 -5.57542801e-01
3.61595720e-01 1.56676936e+00 4.67052788e-01 4.63415980e-01
-1.43009531e+00 -2.68218666e-01 4.33025479e-01 2.45185688e-01
-1.00470388e+00 -4.84150916e-01 5.62298894e-01 -5.15832663e-01
1.18122673e+00 1.74237370e-01 8.58050287e-01 4.68571126e-01
4.07800913e-01 5.39104283e-01 1.28300917e+00 -7.37598598e-01
3.06018442e-01 -1.45195900e-02 3.46817821e-01 7.31833577e-01
4.97236073e-01 2.84387976e-01 -4.88873512e-01 2.53603548e-01
9.49042141e-01 -2.63389409e-01 -6.37524650e-02 9.29298699e-02
-1.30611873e+00 8.01823497e-01 6.25942230e-01 6.92573369e-01
-3.26936841e-01 1.35383561e-01 2.72679120e-01 5.01778722e-01
-1.63530603e-01 8.28472972e-01 -5.62656641e-01 4.03465360e-01
-2.81432271e-01 1.43554032e-01 9.19324219e-01 7.17461646e-01
8.55285645e-01 6.59565628e-02 3.25197071e-01 6.35027289e-01
5.40736973e-01 5.80389380e-01 4.55126733e-01 -1.16635716e+00
2.50556767e-01 3.79603833e-01 -1.10536441e-02 -1.01880264e+00
-3.77531767e-01 -1.66336909e-01 -5.87272704e-01 7.86850229e-02
6.87396824e-01 -9.75785255e-02 -5.78045905e-01 2.19796443e+00
-8.75680000e-02 -3.26012582e-01 2.38571808e-01 7.53391087e-01
7.41957605e-01 6.49843216e-01 5.18509448e-01 -1.38450235e-01
1.47066629e+00 -7.32583478e-02 -4.74807650e-01 -5.44067204e-01
5.87354481e-01 -3.48739445e-01 1.07220256e+00 3.39838982e-01
-1.05356073e+00 -6.93558574e-01 -9.48694825e-01 -2.93400586e-01
-5.57031691e-01 -2.35263914e-01 1.34370768e+00 3.96805853e-01
-1.15587556e+00 7.92546988e-01 -6.69255137e-01 -5.25382996e-01
2.05551088e-01 3.59267861e-01 -1.96503654e-01 3.22800219e-01
-1.26638234e+00 1.12885928e+00 8.52126062e-01 2.37578005e-01
-5.63419938e-01 -2.95847505e-01 -9.69143271e-01 2.34571129e-01
3.96028399e-01 -7.93823600e-01 1.52152455e+00 -1.47894323e+00
-1.15150368e+00 9.77263927e-01 -2.09805846e-01 -4.20755595e-01
-4.97284859e-01 1.14044026e-01 -2.73634136e-01 1.02486104e-01
1.39737530e-02 8.65837097e-01 4.76624101e-01 -1.39389539e+00
-7.34059393e-01 -3.99898320e-01 6.71712518e-01 2.25205168e-01
-8.60065371e-02 2.96607427e-02 4.61142242e-01 -5.87899864e-01
6.12005293e-01 -7.31076777e-01 1.54030204e-01 5.47650568e-02
-2.10999072e-01 -5.62113404e-01 -1.58893228e-01 -6.49177313e-01
9.88975585e-01 -1.96907628e+00 3.87727499e-01 2.99994022e-01
3.42219502e-01 -2.85404712e-01 -4.53664549e-02 4.23036456e-01
-1.61612198e-01 2.79629678e-01 -2.51358151e-01 3.38614970e-01
5.78456700e-01 7.33060181e-01 -5.71777821e-01 1.74207836e-02
2.76012748e-01 9.93687987e-01 -8.98755252e-01 -2.20602095e-01
8.58414248e-02 1.50231719e-01 -6.39195323e-01 1.32561252e-01
-5.52960694e-01 2.10283220e-01 -3.81164014e-01 1.98178813e-01
6.39843717e-02 -3.81569684e-01 5.99213421e-01 -1.70135513e-01
-3.37412693e-02 1.18805110e+00 -9.07686830e-01 1.28799355e+00
-6.62679434e-01 6.75321102e-01 -1.62552148e-02 -1.39542615e+00
5.71310282e-01 1.90180779e-01 -3.14884365e-01 -7.42097914e-01
4.05408680e-01 2.12430462e-01 8.90372455e-01 -3.32284331e-01
8.17188025e-02 -8.87549579e-01 -1.79524213e-01 8.21990788e-01
-8.41938481e-02 -3.48072231e-01 4.38864052e-01 9.39713567e-02
6.29199088e-01 1.11559600e-01 5.79146862e-01 -4.76204097e-01
5.46492219e-01 -6.25014082e-02 3.05870771e-01 6.48922980e-01
2.59084731e-01 -1.29696667e-01 4.50587809e-01 -6.68703020e-01
-1.12218940e+00 -1.36971843e+00 -1.08549498e-01 1.30127645e+00
-1.11556761e-01 -2.28824034e-01 -6.40789986e-01 -3.92515510e-02
-4.14475463e-02 1.21366048e+00 -7.23445535e-01 -3.11376154e-01
-6.32333994e-01 -4.19036210e-01 2.10271612e-01 7.29561448e-01
2.70182341e-01 -1.74518359e+00 -9.05530334e-01 5.03322780e-01
5.10682948e-02 -1.03387702e+00 -2.49848049e-02 5.00032067e-01
-1.17949891e+00 -8.17161560e-01 1.71350971e-01 -9.60089386e-01
8.53701472e-01 -1.69972271e-01 1.39627504e+00 4.94953573e-01
1.44273445e-01 3.86711538e-01 -2.35921472e-01 -5.94058573e-01
-7.01864004e-01 -3.77908707e-01 -1.16354257e-01 -1.97381690e-01
3.91577691e-01 -8.11810434e-01 -1.20939687e-01 4.74608764e-02
-1.07708561e+00 1.99369222e-01 5.79768360e-01 5.21964252e-01
3.66272479e-01 2.22696453e-01 4.69944745e-01 -8.92916977e-01
8.70908976e-01 -3.88194472e-01 -4.46529061e-01 5.15706003e-01
-3.38066310e-01 5.72028279e-01 9.09943521e-01 -3.96245778e-01
-1.07555306e+00 -3.90963644e-01 -6.21452648e-03 4.92028117e-01
-5.92015088e-01 9.15996909e-01 -2.51183510e-01 3.39600593e-01
5.16162813e-01 5.69106042e-01 -7.28408545e-02 -1.79319531e-01
6.46009505e-01 2.41722748e-01 5.75311482e-01 -1.31026912e+00
5.96638918e-01 1.52954698e-01 5.03299460e-02 -7.35055029e-01
-1.07200634e+00 1.66216195e-01 -7.30579734e-01 1.14247009e-01
9.83356655e-01 -6.05016351e-01 -9.51131463e-01 1.85860306e-01
-1.19802737e+00 -3.72165173e-01 -3.89928579e-01 3.85268837e-01
-7.95689642e-01 1.58984348e-01 -8.23554575e-01 -8.08744669e-01
3.20263542e-02 -6.77634478e-01 7.54145741e-01 1.29958421e-01
-9.17101085e-01 -1.21095121e+00 -4.98536408e-01 -2.24604607e-02
2.35012844e-01 -1.44130141e-01 1.73899913e+00 -7.42646575e-01
-6.78640425e-01 3.24439079e-01 -1.35469034e-01 1.91561118e-01
3.30725163e-01 -2.09005140e-02 -5.42087853e-01 3.53269368e-01
1.83207259e-01 -2.81308830e-01 6.17017210e-01 2.73968607e-01
1.03871286e+00 -5.07755876e-01 6.51996583e-02 2.87669748e-01
1.12084508e+00 6.42575979e-01 6.19310021e-01 1.45166725e-01
2.90379703e-01 1.00583661e+00 3.32625471e-02 -1.16559856e-01
7.78474629e-01 2.85056293e-01 -1.73241424e-03 5.03507376e-01
-1.13068521e-01 -4.47400212e-01 2.65362769e-01 1.00265181e+00
-2.50276685e-01 6.93852156e-02 -7.99185157e-01 3.75873148e-01
-1.52802813e+00 -1.10925412e+00 -1.34660006e-02 2.00184107e+00
1.33643508e+00 3.64403099e-01 -2.66571343e-01 3.39740723e-01
4.39784676e-01 -7.41091594e-02 -4.29435819e-01 -8.80328953e-01
-1.05857931e-01 5.00574708e-01 -1.12510912e-01 6.30077302e-01
-6.89294636e-01 1.10078335e+00 6.88444233e+00 3.68681282e-01
-9.05873418e-01 -2.72867233e-01 3.73087645e-01 5.53611338e-01
-7.53551483e-01 8.41969401e-02 -5.06858230e-01 1.39622882e-01
1.03653395e+00 -1.24041602e-01 1.04278922e+00 1.77757576e-01
7.46562481e-02 -3.52229089e-01 -1.64020717e+00 5.35694957e-01
-8.19081292e-02 -1.30015993e+00 4.55704898e-01 -1.31078601e-01
3.54256511e-01 -6.05389535e-01 -2.39217907e-01 1.46088928e-01
6.91370785e-01 -1.22587764e+00 1.19380867e+00 5.33929944e-01
3.52075011e-01 -4.46480960e-01 3.37863833e-01 5.51451743e-01
-1.02834868e+00 -1.86849803e-01 -5.43331385e-01 -8.52445662e-01
4.76152338e-02 3.75117183e-01 -5.85901856e-01 1.03926085e-01
3.17137837e-01 6.28689468e-01 -4.23727214e-01 5.12365103e-01
-1.05924213e+00 5.12162030e-01 -3.92970324e-01 -5.14439940e-01
6.49403185e-02 -4.56181973e-01 5.67729436e-02 7.48198330e-01
2.38447383e-01 6.24774694e-01 -1.59888536e-01 1.38254488e+00
8.14887136e-02 9.84804556e-02 -5.97014129e-01 -5.02209604e-01
4.90510941e-01 7.68000782e-01 -9.76950526e-01 -6.23810232e-01
-3.10574323e-01 3.97550344e-01 4.49809521e-01 3.57135653e-01
-2.88485110e-01 -3.88119593e-02 6.24007106e-01 -5.09131476e-02
3.31997395e-01 -6.36167049e-01 -7.08088100e-01 -1.03650284e+00
-1.27922907e-01 -9.51536953e-01 1.51919931e-01 -1.21668291e+00
-1.32940328e+00 3.24102730e-01 2.20490500e-01 -4.22321349e-01
-4.75295693e-01 -1.03286171e+00 -6.49296820e-01 1.06224155e+00
-1.02397323e+00 -8.75615656e-01 3.80361736e-01 5.15307784e-01
3.74335498e-01 1.08022295e-01 1.12495875e+00 -3.54758978e-01
5.10358512e-02 1.87406186e-02 -4.26255852e-01 4.60394979e-01
-9.40675437e-02 -1.44237673e+00 3.47038925e-01 6.63016021e-01
7.60287106e-01 1.40946412e+00 7.10310578e-01 -4.88858759e-01
-1.40562403e+00 -2.81881511e-01 1.33213556e+00 -5.96760750e-01
1.09985912e+00 -4.71377343e-01 -8.97460878e-01 8.49063396e-01
2.81104773e-01 -3.52167964e-01 8.89872670e-01 4.33568984e-01
-8.57173860e-01 -1.20860286e-01 -8.27507079e-01 7.43758976e-01
1.15649796e+00 -8.86091173e-01 -1.62475276e+00 -2.30834130e-02
6.51342988e-01 -1.93401463e-02 -5.82339227e-01 2.43019816e-02
8.15578103e-01 -8.99156570e-01 8.10828507e-01 -9.95790839e-01
7.21253574e-01 -5.03138721e-01 -3.86705577e-01 -1.43796062e+00
-3.56028736e-01 -6.60468191e-02 1.73279028e-02 7.97301173e-01
7.44507313e-01 -9.48454261e-01 1.84686989e-01 8.51842463e-01
-1.87930107e-01 -2.71273583e-01 -5.82805395e-01 -5.32970846e-01
3.56696337e-01 -6.12379193e-01 6.79627895e-01 8.66472006e-01
2.03301519e-01 6.48030043e-01 3.16563040e-01 9.94546935e-02
5.08500397e-01 5.70086002e-01 1.69373199e-01 -1.56748927e+00
-4.83233064e-01 -5.90218365e-01 -2.86004931e-01 -1.04657567e+00
4.84408081e-01 -1.23891854e+00 9.94014665e-02 -1.72908914e+00
1.53604159e-02 -4.24753278e-01 -2.67379373e-01 6.79600358e-01
-1.00017115e-01 -1.69259518e-01 5.53551555e-01 -1.23355567e-01
-2.55600154e-01 1.04966424e-01 1.30930936e+00 -1.17535405e-01
1.51072294e-01 -3.38971019e-01 -1.27228868e+00 1.05031323e+00
9.00557995e-01 -1.89521790e-01 -4.81798679e-01 -5.75471699e-01
8.75298679e-01 -2.34631170e-02 5.32228231e-01 -7.13605523e-01
2.02169746e-01 -3.68952960e-01 4.84161139e-01 1.51741933e-02
-1.84529141e-01 -9.93180096e-01 3.11858915e-02 4.83718961e-01
-7.15478003e-01 2.23275185e-01 3.53147268e-01 2.07419708e-01
-1.23518467e-01 -3.55439276e-01 2.45349884e-01 -4.27623719e-01
-9.52398241e-01 -3.76317799e-01 -6.64334655e-01 1.22684188e-01
3.70997131e-01 -1.69221118e-01 -2.36472011e-01 -3.19265187e-01
-1.14327288e+00 -1.31638050e-01 1.63354009e-01 3.29492122e-01
5.14639378e-01 -1.35761857e+00 -4.00282115e-01 1.62203789e-01
-5.27480803e-02 -2.68010169e-01 -3.29671949e-01 5.41901529e-01
-3.80092233e-01 5.23354769e-01 -3.55714381e-01 -2.26591781e-01
-8.51872563e-01 6.39176071e-01 3.92957509e-01 2.46331096e-01
-4.41043437e-01 8.49850297e-01 6.57401502e-01 -4.12701488e-01
-9.66005325e-02 -1.08708096e+00 -2.91069984e-01 3.43027897e-02
6.13502562e-01 -3.79410297e-01 -2.59238482e-01 -5.75521231e-01
-2.42182031e-01 7.89733469e-01 1.91720769e-01 -4.27088328e-03
1.26490736e+00 -1.11263610e-01 -6.99051380e-01 8.88966262e-01
7.97074258e-01 -1.61016341e-02 -8.10196579e-01 -1.90001026e-01
1.97754279e-01 -9.25240293e-02 -3.97598892e-01 -9.52908874e-01
-5.50350785e-01 1.05858052e+00 -2.68913150e-01 6.37721181e-01
1.15573895e+00 4.49087054e-01 4.27856266e-01 8.20041478e-01
6.32980287e-01 -8.78313303e-01 -2.64328033e-01 8.84132624e-01
9.30142879e-01 -9.25499558e-01 -2.09557176e-01 -5.39682567e-01
-2.76061624e-01 1.49426222e+00 2.02259064e-01 -2.73944736e-01
5.23366034e-01 1.59196362e-01 -1.62815452e-01 -3.17289174e-01
-9.75019693e-01 -3.76186073e-01 4.49472666e-01 5.83770573e-01
8.42501581e-01 4.75897253e-01 -4.56967384e-01 4.94051486e-01
-7.49297857e-01 -2.85109311e-01 4.82313484e-01 6.48634553e-01
-7.96735823e-01 -1.12967014e+00 -4.73105848e-01 5.67420900e-01
-2.58203804e-01 -5.46402872e-01 -6.35577559e-01 6.93128824e-01
5.33894718e-01 1.03538013e+00 2.27623612e-01 1.04643991e-02
1.53339297e-01 3.91716242e-01 1.01100087e+00 -9.74640548e-01
-4.72397178e-01 -1.79347038e-01 1.87468320e-01 -2.60712445e-01
-7.41800725e-01 -7.90697157e-01 -1.87522542e+00 -2.64326096e-01
2.89547533e-01 2.22592145e-01 3.57106924e-01 1.31837726e+00
-1.79938808e-01 6.38605297e-01 -6.29786998e-02 -4.77846444e-01
-5.87965131e-01 -6.96718454e-01 -5.67002177e-01 5.22609353e-01
2.56482154e-01 -4.19894665e-01 -4.18448269e-01 4.86128509e-01] | [9.463369369506836, 7.3119306564331055] |
4494babc-bf86-4f0a-ae0a-3a512d4d0de3 | designing-bert-for-convolutional-networks | 2301.03580 | null | https://arxiv.org/abs/2301.03580v2 | https://arxiv.org/pdf/2301.03580v2.pdf | Designing BERT for Convolutional Networks: Sparse and Hierarchical Masked Modeling | We identify and overcome two key obstacles in extending the success of BERT-style pre-training, or the masked image modeling, to convolutional networks (convnets): (i) convolution operation cannot handle irregular, random-masked input images; (ii) the single-scale nature of BERT pre-training is inconsistent with convnet's hierarchical structure. For (i), we treat unmasked pixels as sparse voxels of 3D point clouds and use sparse convolution to encode. This is the first use of sparse convolution for 2D masked modeling. For (ii), we develop a hierarchical decoder to reconstruct images from multi-scale encoded features. Our method called Sparse masKed modeling (SparK) is general: it can be used directly on any convolutional model without backbone modifications. We validate it on both classical (ResNet) and modern (ConvNeXt) models: on three downstream tasks, it surpasses both state-of-the-art contrastive learning and transformer-based masked modeling by similarly large margins (around +1.0%). Improvements on object detection and instance segmentation are more substantial (up to +3.5%), verifying the strong transferability of features learned. We also find its favorable scaling behavior by observing more gains on larger models. All this evidence reveals a promising future of generative pre-training on convnets. Codes and models are released at https://github.com/keyu-tian/SparK. | ['Zehuan Yuan', 'LiWei Wang', 'Chen Lin', 'Qishuai Diao', 'Yi Jiang', 'Keyu Tian'] | 2023-01-09 | null | null | null | null | ['self-supervised-image-classification', '2d-object-detection'] | ['computer-vision', 'computer-vision'] | [ 5.40522456e-01 5.47299683e-01 1.45413317e-02 -2.89255321e-01
-8.61036062e-01 -6.84585989e-01 7.92255461e-01 -4.93570328e-01
-3.37083608e-01 5.11717021e-01 7.88835213e-02 -6.40891790e-01
6.87835217e-02 -8.57185721e-01 -1.25314510e+00 -6.29744351e-01
-6.09074272e-02 4.29050952e-01 5.37313521e-01 -1.29380867e-01
1.41570708e-02 7.69155085e-01 -1.55693650e+00 5.83445430e-01
5.17242014e-01 9.80948627e-01 3.87438506e-01 6.95015192e-01
-5.34650609e-02 7.98067629e-01 -2.89125413e-01 -3.43204886e-01
5.32278895e-01 -1.62084118e-01 -7.59976387e-01 1.71168208e-01
6.91293061e-01 -3.82862747e-01 -4.91814852e-01 7.67328680e-01
2.33552590e-01 -3.01483601e-01 5.13545990e-01 -1.16914690e+00
-7.34019756e-01 7.87359595e-01 -5.84442735e-01 2.45597377e-01
-2.36105368e-01 4.28783149e-01 8.75461876e-01 -1.09735823e+00
4.91001964e-01 1.13447821e+00 1.05366504e+00 3.83873075e-01
-1.43732667e+00 -5.94975531e-01 2.08838195e-01 -3.09926718e-01
-1.36879694e+00 -5.00514388e-01 2.86099792e-01 -5.32162786e-01
1.30467832e+00 3.14553320e-01 6.66695118e-01 9.63343263e-01
9.51282127e-05 6.69177353e-01 1.16451120e+00 -3.21800202e-01
-1.62628442e-02 4.55229916e-02 -1.32364124e-01 7.53465354e-01
1.98706657e-01 4.27489430e-01 -3.30451936e-01 1.72059551e-01
1.05529022e+00 -1.01586588e-01 -2.27652583e-02 -4.25853655e-02
-1.25302327e+00 8.06908607e-01 7.24368870e-01 2.45122105e-01
-3.04576784e-01 6.08674765e-01 -5.08893095e-03 1.56304836e-01
5.84574461e-01 3.47489953e-01 -6.06883109e-01 8.81519318e-02
-1.31351173e+00 2.90270299e-01 5.39178193e-01 1.07534933e+00
1.14369869e+00 3.89153898e-01 -5.60188182e-02 4.28835869e-01
2.51208663e-01 3.74584645e-01 2.52846092e-01 -9.98652518e-01
2.22470373e-01 4.49658692e-01 -2.72530377e-01 -5.10643959e-01
-4.29974198e-01 -7.94788122e-01 -7.38876879e-01 4.91419733e-01
4.66754287e-01 -1.12040602e-01 -1.51317334e+00 1.54596567e+00
1.63690634e-02 4.42060113e-01 -1.03944667e-01 6.63180470e-01
9.93140280e-01 6.52194262e-01 5.94734140e-02 4.08877432e-01
1.30683982e+00 -9.00927663e-01 5.54889441e-02 -5.71419716e-01
5.18917203e-01 -7.97624648e-01 8.87628555e-01 3.37416977e-01
-1.30545032e+00 -6.87947631e-01 -9.82033968e-01 -3.07285637e-01
-2.87344992e-01 8.09730440e-02 9.48053777e-01 7.48500764e-01
-1.49191761e+00 7.46731460e-01 -9.74200368e-01 -1.03183992e-01
8.18996966e-01 6.13088191e-01 -2.17319086e-01 -1.47157893e-01
-7.56751537e-01 7.89463818e-01 1.49644434e-01 2.22743824e-02
-1.23573732e+00 -1.10889792e+00 -8.27749550e-01 6.34378940e-02
6.25211746e-02 -8.00575435e-01 1.30724812e+00 -1.02779233e+00
-1.12394440e+00 1.17774558e+00 -2.76100755e-01 -6.49082541e-01
5.48077524e-01 -6.57963604e-02 -3.80596444e-02 1.43620670e-01
1.62881255e-01 1.26763511e+00 9.38136280e-01 -1.55061579e+00
-5.13578355e-01 -6.92632571e-02 1.52101871e-02 2.80164368e-02
5.52413613e-02 -7.84195364e-02 -4.58786279e-01 -6.95216656e-01
4.38236117e-01 -8.49634230e-01 -3.78521234e-01 -6.98790699e-02
-5.65709651e-01 1.44081697e-01 7.55310595e-01 -5.85654318e-01
4.76752967e-01 -2.16400647e+00 -1.33470953e-01 1.24790296e-01
5.06756365e-01 9.31184590e-02 -2.46833622e-01 1.97351143e-01
-3.21094275e-01 3.30684572e-01 -5.12444615e-01 -7.95930684e-01
-1.38935447e-02 2.81833917e-01 -5.45559645e-01 4.30563182e-01
6.82254910e-01 1.34864283e+00 -3.95531505e-01 -2.58089483e-01
3.04266036e-01 5.97013652e-01 -6.99470282e-01 -9.08368453e-02
-2.32397661e-01 5.02169490e-01 2.88651101e-02 6.35918438e-01
9.20824409e-01 -5.35432756e-01 -1.16037302e-01 -4.24321949e-01
-4.06823933e-01 5.22604167e-01 -1.00072801e+00 1.56744933e+00
-3.47148418e-01 7.54724562e-01 1.40860021e-01 -1.00921094e+00
7.37610161e-01 1.45873562e-01 4.29185033e-01 -6.46151543e-01
-6.08175956e-02 2.04856083e-01 -1.29696401e-02 -1.16161756e-01
6.05679631e-01 -2.31149971e-01 2.07054377e-01 2.76659369e-01
3.92142206e-01 -3.23670626e-01 3.37050259e-02 3.51638079e-01
1.20549035e+00 3.81802768e-01 -2.38393128e-01 -4.53419328e-01
-1.45493329e-01 2.66589463e-01 2.58356273e-01 8.79449844e-01
3.28173488e-01 1.08852959e+00 5.26593864e-01 -2.57693797e-01
-1.26309299e+00 -1.10836267e+00 -3.39259475e-01 1.01040673e+00
-7.34060928e-02 -2.84127295e-01 -6.91575110e-01 -3.77957761e-01
1.06952868e-01 5.81702590e-01 -8.11082661e-01 -2.00959016e-02
-6.52196765e-01 -9.36022282e-01 8.37531328e-01 7.96455741e-01
4.41116482e-01 -9.12525773e-01 -4.30085778e-01 1.36182189e-01
1.59190163e-01 -1.36359191e+00 -4.56636846e-02 7.19986498e-01
-9.92795289e-01 -8.29090536e-01 -6.21369660e-01 -9.51435387e-01
7.55419016e-01 3.99498403e-01 1.38909376e+00 3.54373366e-01
-1.97881609e-01 1.10190503e-01 -2.34672621e-01 -3.20063025e-01
-4.84537363e-01 2.57911533e-01 -3.34399581e-01 -2.91029811e-01
8.09069499e-02 -8.66334617e-01 -5.14664292e-01 3.93014997e-01
-9.32954311e-01 4.06322211e-01 9.15299594e-01 6.09195471e-01
6.07803106e-01 -9.92760211e-02 3.21905464e-01 -9.52604175e-01
-1.62105728e-02 -4.36721087e-01 -5.36959112e-01 -2.11779296e-01
-5.51371813e-01 -2.70008072e-02 2.92411059e-01 -3.42554808e-01
-8.52337182e-01 1.87526017e-01 -5.16671538e-01 -5.04865289e-01
-4.03593034e-01 2.47806653e-01 6.87607154e-02 -3.38908255e-01
7.42291749e-01 1.70801818e-01 5.93893193e-02 -5.52533150e-01
4.70320851e-01 2.54176378e-01 6.69946551e-01 -5.76008916e-01
1.10561371e+00 7.89341569e-01 -4.35015894e-02 -6.02338135e-01
-8.00309300e-01 -6.67868406e-02 -6.75423920e-01 2.18486652e-01
9.87838864e-01 -1.34744453e+00 -4.10718918e-01 4.29178953e-01
-1.17910647e+00 -9.63084877e-01 -5.25295258e-01 2.18982831e-01
-5.10479152e-01 4.31051813e-02 -7.80445158e-01 -4.29742038e-01
-4.96611483e-02 -1.30154693e+00 1.29381168e+00 -2.09923182e-02
-3.05554662e-02 -7.73373902e-01 -4.19014245e-01 2.90518701e-01
5.52632809e-01 1.60540223e-01 7.11976647e-01 -4.34949011e-01
-9.34339881e-01 6.46431595e-02 -6.62457943e-01 3.59716415e-01
-9.08237994e-02 1.75908864e-01 -1.38046432e+00 -1.56481743e-01
-1.45194083e-02 -2.61795193e-01 1.26196563e+00 4.97189611e-01
1.26116455e+00 -1.09257743e-01 -3.43255818e-01 1.01633894e+00
1.38116693e+00 -2.05402240e-01 9.16136086e-01 1.93687707e-01
8.48388076e-01 1.93863183e-01 -4.61475439e-02 1.53015912e-01
3.43041927e-01 4.26461637e-01 8.15310001e-01 -5.89661896e-01
-5.93033016e-01 -3.08299631e-01 2.52380967e-01 4.60351229e-01
-1.35936558e-01 3.20210978e-02 -9.28452432e-01 5.82421064e-01
-1.55862784e+00 -7.83397436e-01 -4.47473764e-01 1.87676859e+00
7.80192256e-01 4.49562132e-01 2.83168945e-02 -6.60635624e-03
4.79279488e-01 9.76732597e-02 -3.78672928e-01 -1.25668526e-01
-3.77758324e-01 6.75988853e-01 9.10775304e-01 5.60711443e-01
-1.10566986e+00 1.13020217e+00 6.67136860e+00 8.71431708e-01
-1.16321266e+00 3.66923362e-01 9.94042397e-01 1.31989561e-03
-4.73028153e-01 6.84233382e-02 -1.02325678e+00 3.81017029e-01
8.89748275e-01 3.76020342e-01 3.51168841e-01 7.90217280e-01
-1.26832053e-01 -2.39346400e-02 -1.13572872e+00 6.99140966e-01
-1.88982174e-01 -1.93438971e+00 -1.11283809e-01 2.89759755e-01
9.51526999e-01 7.00179517e-01 2.69944578e-01 4.33692157e-01
5.97896516e-01 -1.49740946e+00 1.12145996e+00 1.48408651e-01
1.04056668e+00 -3.95872086e-01 4.53753620e-01 2.85614163e-01
-1.07679284e+00 2.10142180e-01 -5.08749306e-01 -1.25448823e-01
8.34437832e-02 7.57935405e-01 -8.35960567e-01 3.28541309e-01
7.47357309e-01 7.11641133e-01 -7.37683773e-01 8.64450932e-01
-3.18958133e-01 9.38625336e-01 -5.75673342e-01 4.74777013e-01
5.23423076e-01 1.83616415e-01 3.18795472e-01 1.35259306e+00
2.26189375e-01 -1.04163051e-01 -1.62309930e-02 1.29095054e+00
-1.45937845e-01 -5.57866395e-01 -5.60841084e-01 2.15902373e-01
2.87384242e-01 1.29785860e+00 -9.71352696e-01 -2.90073484e-01
-5.03671229e-01 8.11452568e-01 1.84973449e-01 3.77547026e-01
-9.85549629e-01 1.11626618e-01 4.97517824e-01 4.11659896e-01
8.40055287e-01 -3.69185030e-01 -8.48325312e-01 -7.83183753e-01
-1.26904964e-01 -6.96182430e-01 -4.79781702e-02 -8.18869650e-01
-1.20372045e+00 7.16993630e-01 3.07679176e-02 -1.07932627e+00
-8.59430507e-02 -8.03392947e-01 -5.93873918e-01 9.74160850e-01
-1.66242909e+00 -1.47229099e+00 -4.23571944e-01 5.90994895e-01
4.20013905e-01 4.48912084e-02 6.51098371e-01 3.57355595e-01
-2.02551574e-01 4.70703930e-01 -1.91280499e-01 1.67621076e-01
2.03954965e-01 -1.19410276e+00 9.82429683e-01 9.68596816e-01
3.40990454e-01 4.53026265e-01 4.26981300e-01 -4.90655780e-01
-1.34044576e+00 -1.31682205e+00 5.67152679e-01 -6.49762273e-01
5.93926847e-01 -5.66862047e-01 -7.80346870e-01 1.12243283e+00
1.98704302e-01 2.26189733e-01 3.14730138e-01 -6.90832809e-02
-4.64001745e-01 1.53333798e-01 -1.06090772e+00 4.42769706e-01
1.25257409e+00 -2.33544782e-01 -2.07389414e-01 4.06790316e-01
1.10948074e+00 -6.02750301e-01 -6.60005510e-01 5.14116585e-01
2.20480919e-01 -1.15720296e+00 1.04418039e+00 -4.16850597e-01
6.45644963e-01 -1.72353119e-01 -3.74279916e-01 -1.03702939e+00
-4.93905336e-01 -6.25620782e-01 3.67675275e-02 9.45884764e-01
6.38416648e-01 -7.48548746e-01 1.15149713e+00 5.13691425e-01
-6.16586685e-01 -8.49211454e-01 -8.63542914e-01 -6.96096301e-01
4.24708307e-01 -7.95071483e-01 5.76364815e-01 8.25972497e-01
-7.75845230e-01 1.10221066e-01 -8.36638454e-03 4.06487316e-01
5.13405681e-01 -9.78953391e-02 7.06798315e-01 -1.03703415e+00
-5.67788303e-01 -5.84207833e-01 -3.12147319e-01 -1.37814701e+00
-6.44611195e-02 -1.08024251e+00 -1.99992377e-02 -1.61813760e+00
-1.31040625e-02 -8.91374707e-01 -2.62023527e-02 9.23268139e-01
1.56026825e-01 9.38005805e-01 2.18277723e-01 3.02975535e-01
-9.79344100e-02 2.41811663e-01 1.14457309e+00 6.29107207e-02
8.34737644e-02 -5.90306297e-02 -1.00478256e+00 7.36324072e-01
9.13660049e-01 -5.40801942e-01 -1.63399935e-01 -8.44158888e-01
2.33925864e-01 -3.20489675e-01 9.86900806e-01 -1.19396293e+00
-8.58865380e-02 1.29749432e-01 5.48757553e-01 -6.13629103e-01
4.68178868e-01 -7.21607924e-01 3.04476619e-01 2.86303103e-01
-3.60268392e-02 -8.60533852e-04 6.61881268e-01 1.22894488e-01
5.54319732e-02 -2.18136549e-01 8.09960365e-01 -3.89821827e-01
-6.94786906e-01 3.31315517e-01 -2.50194669e-01 -4.77841543e-03
6.94442153e-01 -5.42144239e-01 -5.40561676e-01 -1.74075231e-01
-8.53303969e-01 -1.01672828e-01 5.23157895e-01 1.26876041e-01
5.41121185e-01 -9.16308284e-01 -8.53013396e-01 4.35734987e-01
-3.66982520e-01 5.76484144e-01 1.46023467e-01 1.02161324e+00
-6.80667877e-01 1.91805407e-01 -1.06476350e-02 -8.67868423e-01
-9.99050498e-01 2.77651727e-01 3.47443521e-01 -1.07315682e-01
-8.41787815e-01 1.23256886e+00 5.14863908e-01 -4.73855704e-01
5.78716695e-02 -4.78134364e-01 4.41718638e-01 -2.53134906e-01
2.71455348e-01 -3.41483019e-02 4.49651062e-01 -4.80653197e-01
-3.19313109e-01 4.04373616e-01 -2.28681955e-02 -1.27664059e-01
1.68652093e+00 2.17187718e-01 -1.93677351e-01 1.15937956e-01
1.26016641e+00 5.73599571e-03 -1.71471632e+00 -1.41162053e-01
-3.78831536e-01 -2.18503818e-01 1.57973796e-01 -7.47917533e-01
-1.39417708e+00 1.03005576e+00 3.77464980e-01 2.27097422e-01
8.70660782e-01 4.34371740e-01 5.24013042e-01 8.61291662e-02
1.52932376e-01 -3.93579543e-01 -1.62894484e-02 4.96460885e-01
7.84719884e-01 -9.63773906e-01 1.45544246e-01 -6.26050115e-01
-5.29782057e-01 8.86019945e-01 5.50649464e-01 -4.59952235e-01
7.69771516e-01 8.29975426e-01 -1.48109137e-03 -3.77168953e-01
-6.15958691e-01 -3.51820379e-01 2.04798505e-01 6.46689415e-01
3.59998554e-01 -9.72949043e-02 3.38571370e-01 4.11070377e-01
-5.18187940e-01 -1.10397361e-01 4.41526592e-01 7.88393021e-01
-3.71321172e-01 -8.28750193e-01 -3.30025852e-01 5.30733943e-01
-3.26786667e-01 -5.40560961e-01 -1.31123051e-01 9.30046320e-01
3.53948891e-01 7.08045602e-01 4.30002391e-01 -4.55995232e-01
3.56037021e-02 -7.29385614e-02 5.58704853e-01 -7.89795458e-01
-6.98713422e-01 1.67139143e-01 -1.05451467e-02 -5.32635748e-01
-3.90474349e-01 -5.52759171e-01 -1.30501485e+00 -3.71358216e-01
-1.49962351e-01 -3.13087553e-01 7.08491445e-01 9.31682825e-01
4.98104274e-01 7.56375194e-01 3.42380345e-01 -1.53439510e+00
-4.12680119e-01 -8.65453959e-01 -3.80178630e-01 1.00846946e-01
2.61116892e-01 -5.78809083e-01 -4.82688695e-01 2.35461086e-01] | [9.768131256103516, 0.29606303572654724] |
5f2fd55b-7474-466b-9860-5fa6d2b4d8a8 | reddit-a-gold-mine-for-personality-prediction | null | null | https://aclanthology.org/W18-1112 | https://aclanthology.org/W18-1112.pdf | Reddit: A Gold Mine for Personality Prediction | Automated personality prediction from social media is gaining increasing attention in natural language processing and social sciences communities. However, due to high labeling costs and privacy issues, the few publicly available datasets are of limited size and low topic diversity. We address this problem by introducing a large-scale dataset derived from Reddit, a source so far overlooked for personality prediction. The dataset is labeled with Myers-Briggs Type Indicators (MBTI) and comes with a rich set of features for more than 9k users. We carry out a preliminary feature analysis, revealing marked differences between the MBTI dimensions and poles. Furthermore, we use the dataset to train and evaluate benchmark personality prediction models, achieving macro F1-scores between 67{\%} and 82{\%} on the individual dimensions and 82{\%} accuracy for exact or one-off accurate type prediction. These results are encouraging and comparable with the reliability of standardized tests. | ['Jan {\\v{S}}najder', "Matej Gjurkovi{\\'c}"] | 2018-06-01 | null | null | null | ws-2018-6 | ['type-prediction'] | ['computer-code'] | [-2.42290720e-01 4.39186692e-01 -2.07303748e-01 -5.48689723e-01
-4.61564004e-01 -4.27362233e-01 5.23701906e-01 7.02233553e-01
-4.68660951e-01 1.00129735e+00 3.39519769e-01 3.33150417e-01
-3.79881442e-01 -5.85886061e-01 5.11765806e-03 -3.96268249e-01
-2.14278847e-01 6.98570788e-01 -5.34781720e-04 2.54857000e-02
3.24536324e-01 -3.59363593e-02 -1.69470048e+00 3.09535354e-01
1.08215570e+00 1.09842002e+00 -7.85502121e-02 3.59680891e-01
7.50079006e-02 6.54645979e-01 -6.20805681e-01 -1.12324286e+00
-5.93628548e-02 -1.54994726e-01 -8.78994405e-01 2.88154427e-02
5.19035101e-01 -9.62799340e-02 1.23450514e-02 9.23813403e-01
3.78835768e-01 -3.60171013e-02 7.80111790e-01 -1.05876064e+00
-7.50214934e-01 5.38083136e-01 -6.30865037e-01 -7.32081011e-02
5.86759031e-01 -2.78130136e-02 1.36988986e+00 -5.99218488e-01
6.02692068e-01 9.33902323e-01 7.83835292e-01 4.66195285e-01
-1.42028773e+00 -5.38738370e-01 -1.92741960e-01 4.02863510e-02
-1.25998211e+00 -3.91290158e-01 5.83389342e-01 -7.98768222e-01
6.30645275e-01 2.01338917e-01 6.03973567e-01 1.20615804e+00
-5.97361811e-02 4.36281234e-01 1.81987083e+00 -1.36043906e-01
1.58391714e-01 8.48273218e-01 4.19582456e-01 7.66048551e-01
1.48033351e-01 -2.63306171e-01 -5.98865747e-01 -4.50266331e-01
5.53547263e-01 -3.82086545e-01 2.94345260e-01 1.29764140e-01
-1.18017435e+00 6.37023509e-01 4.15152311e-02 2.42292151e-01
-3.94054234e-01 -5.76467991e-01 4.62096006e-01 4.35489506e-01
6.82717562e-01 7.13918030e-01 -4.32686299e-01 -5.24685860e-01
-8.00049722e-01 4.72016364e-01 1.00902188e+00 8.08802187e-01
5.58973670e-01 -2.11192355e-01 -2.54952729e-01 1.29876757e+00
-1.74137816e-01 4.13689911e-01 5.61504006e-01 -1.03993225e+00
2.32315049e-01 7.41715550e-01 1.56710252e-01 -1.41903198e+00
-7.61274040e-01 -4.63086069e-01 -9.51796293e-01 -1.41693816e-01
7.55976737e-01 -7.86221847e-02 4.86986004e-02 1.68591213e+00
2.12248608e-01 -3.79279554e-01 -1.37383893e-01 6.67753577e-01
8.95583272e-01 1.60893798e-01 1.37766019e-01 -3.68473560e-01
1.55404651e+00 -3.48858476e-01 -2.31341556e-01 -4.42425199e-02
4.49402839e-01 -6.05289519e-01 1.24017978e+00 6.83983624e-01
-1.17690480e+00 -6.37369454e-01 -5.96633971e-01 1.58659965e-01
-2.07571447e-01 1.19384728e-01 8.24213564e-01 1.03822649e+00
-1.03346741e+00 7.75764406e-01 -1.64124280e-01 -4.54529732e-01
5.16161501e-01 4.92998242e-01 -5.77121079e-01 2.71201372e-01
-8.53160739e-01 6.49334431e-01 4.79370505e-02 -4.86394614e-01
-2.04654068e-01 -5.94575942e-01 -4.01562005e-01 1.22960836e-01
2.26276860e-01 -3.83576214e-01 8.50483418e-01 -8.82867575e-01
-1.35579038e+00 1.20249176e+00 1.03245147e-01 -3.80852669e-01
4.50639278e-01 1.65451959e-01 -6.03634953e-01 9.89189968e-02
2.80031115e-01 4.96522784e-01 5.99234819e-01 -8.59939337e-01
-8.15346181e-01 -6.08125448e-01 -2.31775552e-01 -2.16708228e-01
-1.00656438e+00 3.07963431e-01 6.92816265e-03 -5.48573494e-01
-9.99738425e-02 -9.15704966e-01 1.33889109e-01 -5.65560699e-01
-3.07112426e-01 -4.70368087e-01 9.18070450e-02 -1.04898822e+00
1.21884394e+00 -1.88856900e+00 7.75321573e-02 2.15166077e-01
5.57140827e-01 1.40172362e-01 7.04090148e-02 2.85551846e-01
1.75522745e-01 1.69930995e-01 1.53526008e-01 -5.91879308e-01
4.04118329e-01 -1.44798666e-01 -6.19284920e-02 2.24655718e-01
7.01177493e-02 6.59949958e-01 -8.98539066e-01 -4.67511058e-01
-2.41324365e-01 7.59704039e-02 -6.95159733e-01 4.17828262e-02
9.30072516e-02 4.66693342e-01 -2.80003995e-01 6.81358874e-01
3.32037568e-01 -6.07233718e-02 2.64410406e-01 2.54956394e-01
-1.55855045e-01 3.21354449e-01 -7.32179165e-01 9.63604212e-01
-2.13993667e-03 5.12976408e-01 -1.48343667e-01 -5.96331060e-01
1.41633821e+00 7.90452436e-02 7.08007812e-01 -6.95976138e-01
1.71486631e-01 1.69669494e-01 1.70653671e-01 -3.80632967e-01
8.38143587e-01 -1.86387464e-01 -5.07140577e-01 4.53943163e-01
9.68742445e-02 2.34308720e-01 2.73168564e-01 1.58062339e-01
1.02825713e+00 -1.30301923e-01 2.12809712e-01 -5.41613042e-01
3.44136059e-01 -1.35894954e-01 8.59014094e-01 5.37134349e-01
-3.42433840e-01 3.86449188e-01 1.14023840e+00 -3.24955523e-01
-1.12329268e+00 -6.83301151e-01 -2.65480280e-01 1.14521027e+00
-5.56848943e-01 -7.45494962e-01 -7.45301127e-01 -6.28267169e-01
2.55700439e-01 3.99438620e-01 -6.01333618e-01 1.50564965e-02
1.19959705e-01 -7.53796935e-01 8.14801276e-01 3.39265436e-01
4.23681915e-01 -1.00731182e+00 -4.42917794e-02 -2.60702521e-02
-1.61278337e-01 -1.14492106e+00 -7.95700960e-03 -2.65102804e-01
-5.59940815e-01 -8.27573121e-01 -6.24907374e-01 -6.03802085e-01
2.89375693e-01 -2.42012873e-01 1.37893498e+00 1.91425756e-02
6.24154769e-02 1.30615339e-01 -2.78861880e-01 -3.90531033e-01
-3.31798702e-01 3.66646826e-01 5.59893191e-01 9.26662609e-02
6.35395467e-01 -7.24336028e-01 -2.24103272e-01 4.29621339e-01
-2.27132156e-01 -1.31423891e-01 5.77116787e-01 5.96914113e-01
5.95120937e-02 2.44147554e-01 1.20861673e+00 -9.22279954e-01
7.87512541e-01 -5.22455812e-01 -2.96406001e-01 -2.14085579e-01
-9.49529231e-01 -4.66335893e-01 5.42810440e-01 -2.69457459e-01
-8.61368835e-01 -3.46171618e-01 -6.03858717e-02 1.26001775e-01
-5.32511234e-01 4.26241279e-01 5.40203936e-02 1.36779264e-01
4.76311058e-01 2.83316299e-02 6.11281991e-02 -6.08554304e-01
-2.85962760e-01 1.04100204e+00 4.62886870e-01 -6.55630648e-01
4.39266235e-01 3.86318401e-03 -7.02262446e-02 -1.04879570e+00
-9.95437920e-01 -2.64766663e-01 -8.59035134e-01 -2.74539322e-01
5.61751246e-01 -7.22089171e-01 -1.28186619e+00 4.99076843e-01
-4.73981380e-01 -2.16921300e-01 -7.20515708e-03 3.31790417e-01
-5.11877179e-01 5.14746189e-01 -6.84318721e-01 -8.69552374e-01
-3.27707440e-01 -6.59681618e-01 6.63453460e-01 8.11750591e-02
-8.84922922e-01 -9.36941624e-01 -4.10559364e-02 8.62220645e-01
1.84390292e-01 2.53074914e-01 1.12018192e+00 -7.92056441e-01
1.65861443e-01 -4.64590162e-01 -3.29811513e-01 3.38092357e-01
-1.65181711e-01 -1.05775863e-01 -1.00631618e+00 -9.58787128e-02
-2.16640770e-01 -5.71832061e-01 6.10738099e-01 5.56605458e-02
1.12526345e+00 -3.25950325e-01 -2.74305195e-02 1.21027753e-01
9.42077637e-01 -3.48147243e-01 5.34929395e-01 3.70485902e-01
4.71889853e-01 1.06049228e+00 4.29229289e-01 8.12170267e-01
7.90419877e-01 7.18574226e-01 -1.70349285e-01 2.92813778e-01
2.87171483e-01 -1.75939754e-01 3.73599827e-01 7.54515052e-01
-4.25159365e-01 2.16228411e-01 -1.09808671e+00 4.14148808e-01
-1.64456773e+00 -1.04258835e+00 -2.70152181e-01 2.48373485e+00
8.65894914e-01 3.79383445e-01 1.02726138e+00 2.91776955e-01
5.17231762e-01 -1.14699580e-01 -1.34815201e-01 -5.30818582e-01
-2.51318187e-01 -6.94711357e-02 1.88383758e-01 5.59502207e-02
-9.14035857e-01 8.17383111e-01 6.26043367e+00 5.03764212e-01
-7.47536659e-01 -1.85423389e-01 8.21153939e-01 -1.50543481e-01
7.62273222e-02 -2.61837631e-01 -8.84685993e-01 5.83157718e-01
1.19458687e+00 -2.31268987e-01 4.06909764e-01 7.80685008e-01
7.27105886e-02 -1.64205581e-01 -8.99942875e-01 1.00093651e+00
5.96819185e-02 -7.61824012e-01 -4.98394847e-01 4.79371935e-01
7.19613552e-01 -2.04598799e-01 2.19026029e-01 6.25075579e-01
2.39718765e-01 -1.01472938e+00 5.59831262e-01 6.08202517e-01
8.42431068e-01 -9.20815527e-01 6.40165687e-01 4.99255151e-01
-5.65804541e-01 -4.02147442e-01 -6.68836117e-01 -5.85484922e-01
-3.01633239e-01 8.80094349e-01 -8.39419067e-01 1.90701157e-01
8.10767412e-01 7.26556361e-01 -9.43335056e-01 7.98109829e-01
9.26663950e-02 6.68201447e-01 2.96511631e-02 -3.04562032e-01
-1.69313580e-01 -3.70947301e-01 2.87622154e-01 9.90930557e-01
2.38793045e-01 -3.56248207e-02 1.81154117e-01 7.52940357e-01
-1.53702855e-01 4.72943097e-01 -3.93564582e-01 -4.68748301e-01
3.19024026e-01 1.71021736e+00 -6.21646762e-01 -3.02066892e-01
-5.66414893e-01 7.59103417e-01 7.50176728e-01 -1.35475948e-01
-5.76498687e-01 -1.81944624e-01 6.06554627e-01 2.67113984e-01
-1.51134089e-01 -2.74016619e-01 -6.12472415e-01 -1.16028094e+00
-6.80809543e-02 -8.53786647e-01 4.06392753e-01 -4.80072111e-01
-1.80387580e+00 5.02339423e-01 -1.88651368e-01 -9.37456250e-01
-1.48020208e-01 -6.20747864e-01 -2.44625583e-01 8.63208532e-01
-1.00199270e+00 -1.14182401e+00 -3.67863089e-01 2.79494941e-01
-7.48430789e-02 -5.66578209e-01 1.02643406e+00 2.90278554e-01
-7.18868375e-01 7.62883186e-01 1.86688080e-01 1.15389757e-01
9.66797352e-01 -1.52000356e+00 1.25861540e-01 9.55557078e-03
-3.50894332e-01 6.07083023e-01 6.00581288e-01 -7.55392194e-01
-1.03475165e+00 -8.31294119e-01 1.48261642e+00 -7.82679558e-01
8.13038409e-01 -4.86520469e-01 -8.57345402e-01 5.64539909e-01
6.78422228e-02 -7.46372104e-01 1.20295632e+00 8.73843253e-01
-3.14841479e-01 -5.14278524e-02 -1.43199098e+00 4.75552499e-01
1.01656878e+00 -4.21646684e-01 -3.68925899e-01 2.77434409e-01
-6.02843007e-03 1.49789631e-01 -1.60495007e+00 1.06122851e-01
9.71616328e-01 -1.31469226e+00 7.72093892e-01 -5.46179891e-01
9.30743992e-01 5.25451899e-01 -6.36906922e-02 -1.32977343e+00
-8.00702870e-01 -4.71764028e-01 1.53059527e-01 1.76903200e+00
3.23818237e-01 -6.55569196e-01 9.62609291e-01 1.12667572e+00
8.21080953e-02 -7.33307362e-01 -5.51193953e-01 -6.03696942e-01
2.68159181e-01 -2.02733114e-01 5.27337790e-01 1.07700992e+00
6.75920010e-01 5.97591519e-01 -5.85486293e-01 -2.92153955e-01
7.57447600e-01 1.56558394e-01 8.65469873e-01 -1.88503861e+00
-4.44565952e-01 -5.97867608e-01 -5.33392191e-01 -2.69235283e-01
4.05744523e-01 -9.15041983e-01 -5.80929875e-01 -1.10855985e+00
5.68620205e-01 -6.20044351e-01 -1.43726364e-01 4.09635365e-01
-1.05648950e-01 4.46957737e-01 -3.14808004e-02 2.50878751e-01
-4.89264607e-01 3.21960539e-01 8.87635946e-01 1.09973289e-01
-1.50496885e-01 8.27015005e-03 -1.02873445e+00 7.88072228e-01
9.77971375e-01 -9.95787755e-02 -9.37803686e-02 -9.81917698e-03
4.50410306e-01 6.54170811e-02 2.62503237e-01 -1.17093706e+00
-2.10838884e-01 -2.53108084e-01 6.24282002e-01 -1.90278471e-01
4.57969546e-01 -3.33715975e-01 -1.12290576e-01 2.83682793e-02
-5.17829716e-01 -8.01216885e-02 -1.38632074e-01 2.41111815e-01
1.00533806e-01 -2.64039598e-02 6.82781458e-01 -1.89206004e-01
-4.72325236e-01 2.59649694e-01 -3.89224946e-01 -7.19801560e-02
8.27010036e-01 -8.93919915e-02 -4.56608444e-01 -4.50494677e-01
-7.09776282e-01 -9.13737640e-02 7.26342916e-01 4.11636919e-01
1.68726176e-01 -1.13313651e+00 -7.97512174e-01 1.95033073e-01
2.75562972e-01 -7.43319571e-01 2.58493304e-01 1.06732416e+00
1.00571953e-01 3.73967737e-01 -6.03935003e-01 -1.80421531e-01
-1.44435859e+00 4.77337241e-01 -1.87287748e-01 -1.44172892e-01
-4.37782317e-01 6.37888074e-01 -1.32701010e-01 -4.73280579e-01
1.14779444e-02 2.20168993e-01 -5.70264876e-01 4.29754078e-01
5.56099057e-01 7.67062306e-01 3.24422270e-02 -9.41587031e-01
-3.93568501e-02 -2.60397699e-02 -9.62075964e-02 -5.78355268e-02
1.45453060e+00 -1.74752176e-01 -4.87802386e-01 6.97654009e-01
8.85665834e-01 3.59929472e-01 -7.34774768e-01 -1.89806223e-01
2.93479055e-01 -6.10607803e-01 -3.17883253e-01 -8.04762602e-01
-5.04729271e-01 6.30812466e-01 1.80633426e-01 7.89368153e-01
6.40177310e-01 -2.43985411e-02 6.07452571e-01 2.41512030e-01
5.64729154e-01 -1.44250977e+00 1.33289635e-01 6.72894359e-01
6.93541765e-01 -1.16954350e+00 7.42469877e-02 -3.63518655e-01
-9.06304419e-01 6.75004244e-01 6.50673211e-01 -8.92468169e-02
3.45947385e-01 -1.25670835e-01 -1.82995871e-01 -1.87292874e-01
-9.12598133e-01 -1.97891705e-02 3.79751712e-01 6.07470572e-01
7.25829124e-01 4.85326141e-01 -5.71941555e-01 1.36574721e+00
-8.56567204e-01 -4.93790172e-02 5.43461978e-01 3.78537267e-01
-5.19195318e-01 -1.14514160e+00 -2.56556928e-01 1.12512338e+00
-6.54573202e-01 1.65972680e-01 -8.50971758e-01 5.43064237e-01
1.87448133e-02 9.84970987e-01 -5.72605757e-04 -7.98719108e-01
1.20403938e-01 2.52463907e-01 4.94456053e-01 -5.32697499e-01
-7.27160156e-01 -2.37564370e-01 5.84131777e-01 -1.22875839e-01
-6.50553182e-02 -1.19034076e+00 -8.15149546e-01 -8.60349536e-01
8.43234882e-02 1.44782349e-01 3.76525640e-01 8.13330114e-01
4.09314662e-01 5.28333597e-02 7.36509562e-01 -6.67617321e-01
-5.65903366e-01 -1.24506676e+00 -1.14761138e+00 6.93787277e-01
-2.10954338e-01 -6.78916812e-01 -8.76043141e-02 -7.36893043e-02] | [9.442756652832031, 10.279058456420898] |
06ce4e9a-f86f-4bcf-b4e3-95935f065141 | a-framework-for-real-time-object-detection | 2303.09190 | null | https://arxiv.org/abs/2303.09190v2 | https://arxiv.org/pdf/2303.09190v2.pdf | Resolution Enhancement Processing on Low Quality Images Using Swin Transformer Based on Interval Dense Connection Strategy | The Transformer-based method has demonstrated remarkable performance for image super-resolution in comparison to the method based on the convolutional neural networks (CNNs). However, using the self-attention mechanism like SwinIR (Image Restoration Using Swin Transformer) to extract feature information from images needs a significant amount of computational resources, which limits its application on low computing power platforms. To improve the model feature reuse, this research work proposes the Interval Dense Connection Strategy, which connects different blocks according to the newly designed algorithm. We apply this strategy to SwinIR and present a new model, which named SwinOIR (Object Image Restoration Using Swin Transformer). For image super-resolution, an ablation study is conducted to demonstrate the positive effect of the Interval Dense Connection Strategy on the model performance. Furthermore, we evaluate our model on various popular benchmark datasets, and compare it with other state-of-the-art (SOTA) lightweight models. For example, SwinOIR obtains a PSNR of 26.62 dB for x4 upscaling image super-resolution on Urban100 dataset, which is 0.15 dB higher than the SOTA model SwinIR. For real-life application, this work applies the lastest version of You Only Look Once (YOLOv8) model and the proposed model to perform object detection and real-life image super-resolution on low-quality images. This implementation code is publicly available at https://github.com/Rubbbbbbbbby/SwinOIR. | ['Chun-Tse Chien', 'Wei-Han Chen', 'Yu-Shian Lin', 'Jen-Shiun Chiang', 'Chih-Chia Chen', 'Rui-Yang Ju'] | 2023-03-16 | null | null | null | null | ['image-super-resolution', 'real-time-object-detection', 'image-cropping'] | ['computer-vision', 'computer-vision', 'computer-vision'] | [ 1.95369184e-01 -2.86074698e-01 -3.42976525e-02 3.82858291e-02
-4.33389664e-01 1.80557221e-02 2.09392205e-01 -6.05565429e-01
-4.11208928e-01 7.33820260e-01 1.27456769e-01 -2.11972877e-01
-1.41438410e-01 -9.72985744e-01 -8.47313285e-01 -6.94454432e-01
1.15223631e-01 -3.01991582e-01 5.73172390e-01 -5.39214611e-01
2.39163056e-01 5.65420270e-01 -1.70147765e+00 5.88507295e-01
9.52958226e-01 1.04938614e+00 7.15667486e-01 5.34276247e-01
3.61745665e-03 7.77836859e-01 -4.43099678e-01 -3.12039405e-01
4.63558018e-01 -2.97701895e-01 -6.72842920e-01 -2.69275248e-01
6.23419642e-01 -6.68852687e-01 -5.57176828e-01 1.23965085e+00
6.59835219e-01 5.12409881e-02 7.14894608e-02 -8.70337486e-01
-1.20056617e+00 7.42696762e-01 -9.53156292e-01 9.03858602e-01
4.10009502e-03 8.30325335e-02 6.42793179e-01 -9.98343587e-01
4.12885219e-01 1.22088242e+00 6.70381486e-01 5.94181299e-01
-1.08315349e+00 -1.09592330e+00 -1.24548294e-01 8.01241875e-01
-1.51493335e+00 -3.07675272e-01 5.16090274e-01 7.50046745e-02
9.33803856e-01 3.74488056e-01 4.34962898e-01 8.66714358e-01
2.15037927e-01 5.20050287e-01 1.23194969e+00 -1.10844202e-01
-9.86031443e-02 6.29597008e-02 2.02706173e-01 5.80586374e-01
3.49905699e-01 2.91177601e-01 -6.48000002e-01 4.13137823e-01
1.35385740e+00 1.27263710e-01 -5.03313482e-01 3.40975016e-01
-1.20549917e+00 5.08508623e-01 1.04169214e+00 4.58245605e-01
-3.43191624e-01 1.14238828e-01 2.37881020e-01 2.47000962e-01
4.95074511e-01 2.02971801e-01 -2.92833507e-01 2.37287879e-01
-9.17151093e-01 -4.22009639e-02 -4.11758646e-02 8.33764434e-01
6.83325589e-01 3.25322807e-01 -1.95577309e-01 9.44899142e-01
-1.12869926e-01 4.06931370e-01 7.05725372e-01 -9.92384613e-01
2.59437948e-01 3.81345242e-01 9.83935446e-02 -9.29212391e-01
-1.94942668e-01 -8.62441897e-01 -1.38974690e+00 5.91287792e-01
1.60998747e-01 1.56730294e-01 -8.49797428e-01 1.38397610e+00
1.12065047e-01 4.32791889e-01 2.46841282e-01 1.35871851e+00
1.37576401e+00 1.00123167e+00 -9.17774886e-02 -1.04930006e-01
1.59611309e+00 -1.17017233e+00 -7.33303308e-01 6.31707860e-03
-5.78571931e-02 -7.53719807e-01 1.27497458e+00 6.18339360e-01
-1.31226790e+00 -1.07834852e+00 -1.25370359e+00 -3.12136889e-01
-1.80997178e-01 2.74579167e-01 4.97839242e-01 3.98780882e-01
-1.27062619e+00 7.72119939e-01 -6.16938531e-01 -2.74480313e-01
5.79346001e-01 1.63718715e-01 -4.64951754e-01 -1.72141850e-01
-1.22339773e+00 7.94775784e-01 3.35172802e-01 1.61976472e-01
-7.19480217e-01 -9.86872137e-01 -5.28885067e-01 2.82144040e-01
2.48924300e-01 -5.48258305e-01 9.14260268e-01 -9.83706653e-01
-1.47968578e+00 5.00592947e-01 -1.39479876e-01 -6.58897936e-01
3.37687433e-01 -3.79784107e-01 -4.34919000e-01 2.80803323e-01
-9.76590216e-02 6.56127810e-01 8.55453610e-01 -1.26548719e+00
-8.17334831e-01 -3.04711819e-01 2.97999054e-01 9.36161950e-02
-5.53890526e-01 2.83111483e-01 -5.55777729e-01 -6.98054552e-01
3.68268229e-02 -5.30835927e-01 -8.74122903e-02 6.49330616e-02
-1.61160097e-01 -1.62292446e-03 9.50485528e-01 -9.32510376e-01
1.22808409e+00 -2.15399170e+00 -1.14896663e-01 -2.86685437e-01
2.81998307e-01 6.20667100e-01 -2.86707819e-01 -5.23661524e-02
-3.59015077e-01 1.91667870e-01 -2.52858326e-02 -9.77360308e-02
-5.85980892e-01 -4.22441252e-02 -2.03977212e-01 2.37515748e-01
1.40613943e-01 8.16971302e-01 -5.58261573e-01 -4.12377000e-01
3.40673625e-01 9.58089590e-01 -4.82187390e-01 7.25153536e-02
3.01152319e-01 3.81941617e-01 -8.01578537e-02 6.00991607e-01
1.16412890e+00 -5.24871707e-01 -2.02371418e-01 -9.13535655e-01
-3.46059412e-01 -2.53684700e-01 -1.05965543e+00 1.48747170e+00
-7.43266642e-01 7.15758264e-01 -1.39756529e-02 -6.84835613e-01
9.36802745e-01 9.59425867e-02 2.28971660e-01 -1.22635841e+00
6.89576790e-02 3.80406119e-02 -9.34840813e-02 -5.27835965e-01
5.54978251e-01 1.90241948e-01 5.85137546e-01 -5.90860436e-04
-1.42428592e-01 4.83459979e-01 2.00987116e-01 2.15095967e-01
9.75670993e-01 3.90430391e-02 1.12704836e-01 -2.71279812e-01
6.58465445e-01 -4.67131734e-02 4.33247030e-01 5.67502439e-01
9.42620914e-03 8.39410305e-01 1.87971760e-02 -5.68704009e-01
-1.21839082e+00 -1.16276073e+00 -3.66132587e-01 8.04858327e-01
5.39663255e-01 -1.81452394e-01 -7.45741904e-01 -1.87508598e-01
-4.53883111e-01 4.03578967e-01 -6.07914448e-01 2.09207758e-02
-6.87368155e-01 -1.14579749e+00 3.45625341e-01 6.56603634e-01
1.44696510e+00 -1.09572172e+00 -8.01225305e-01 2.71174032e-02
-3.69822353e-01 -1.20415437e+00 -4.26813334e-01 -3.77575785e-01
-8.04545581e-01 -9.47899282e-01 -8.63001108e-01 -6.97053492e-01
4.50157523e-01 7.17796564e-01 1.01660383e+00 2.79464662e-01
-4.59647596e-01 -9.54509899e-02 -4.98839349e-01 6.77028373e-02
-1.12633958e-01 -1.56755686e-01 -1.89125076e-01 8.03533718e-02
-2.96897236e-02 -7.38346815e-01 -1.02483308e+00 4.02848929e-01
-1.10142970e+00 4.91165072e-01 8.32482994e-01 9.06709611e-01
5.94006360e-01 4.24934447e-01 6.35807395e-01 -5.73064566e-01
5.05511701e-01 -2.58807421e-01 -6.83233023e-01 6.15508482e-02
-6.94722652e-01 -7.57487491e-02 7.40465403e-01 -5.52207112e-01
-1.28199673e+00 -2.90970743e-01 -1.66494265e-01 -6.70054555e-01
1.58032209e-01 2.17223376e-01 -8.65584165e-02 -1.68895826e-01
6.56394839e-01 4.79885757e-01 1.79210845e-02 -7.44934559e-01
1.53283030e-01 6.93020999e-01 8.15208554e-01 3.11886519e-02
8.22034359e-01 6.36806488e-01 -2.13575978e-02 -6.24293864e-01
-7.38448083e-01 1.20902676e-02 -2.23810181e-01 -6.68741688e-02
8.77185345e-01 -1.13161588e+00 -8.51355374e-01 4.76013273e-01
-1.08235669e+00 -2.86522627e-01 -9.67901200e-02 4.35038179e-01
-2.64111042e-01 2.99457759e-01 -8.50514054e-01 -4.03187722e-01
-8.79305542e-01 -1.21417081e+00 7.60279119e-01 5.62333703e-01
5.08325458e-01 -4.34507906e-01 -5.40652215e-01 5.77423275e-01
1.00124228e+00 -6.72414852e-03 5.54558337e-01 -1.36289224e-02
-8.14403713e-01 1.95177421e-01 -1.04448831e+00 6.56249464e-01
3.50988209e-02 -2.42353573e-01 -8.43987644e-01 -3.96547675e-01
2.07790360e-02 1.58191502e-01 1.08030534e+00 2.92256564e-01
1.45645809e+00 -3.59049410e-01 -4.76430021e-02 9.19888020e-01
1.83935392e+00 1.10885546e-01 1.27173841e+00 6.75066829e-01
6.36325359e-01 9.73419175e-02 7.23328531e-01 2.96231568e-01
2.56333739e-01 8.51920247e-01 7.66188443e-01 -5.16515136e-01
-7.86602557e-01 8.04446787e-02 3.85891795e-01 5.47464192e-01
-5.91093957e-01 1.05974175e-01 -5.16759276e-01 3.79616827e-01
-1.60347211e+00 -1.13929760e+00 -2.57440716e-01 2.02615261e+00
8.18720162e-01 -4.16508093e-02 -2.22254619e-01 1.74691260e-01
8.44335556e-01 5.49066104e-02 -4.88345742e-01 -1.66859359e-01
-2.35034376e-01 4.64044780e-01 5.70859849e-01 4.05517817e-01
-9.08881783e-01 8.08745861e-01 5.27653265e+00 1.32521760e+00
-1.45664001e+00 5.01552463e-01 7.88774431e-01 -2.13004723e-01
1.41745433e-01 -3.36363703e-01 -8.75310481e-01 5.09786963e-01
8.22722435e-01 -2.06291541e-01 7.09933162e-01 6.52751088e-01
4.92574275e-01 -5.02309538e-02 -5.51274598e-01 1.17648232e+00
1.36561736e-01 -1.61063004e+00 2.22656697e-01 -1.96932793e-01
5.72824955e-01 2.05459725e-02 2.90566266e-01 1.06239244e-01
-5.48072197e-02 -1.11970723e+00 4.92579490e-01 3.50812316e-01
1.16491616e+00 -8.65559459e-01 9.04827118e-01 2.44398247e-02
-1.32006526e+00 -3.74857903e-01 -7.31091082e-01 2.01644331e-01
-1.58078507e-01 6.66909873e-01 -4.87205565e-01 7.54164159e-01
1.40777409e+00 7.30326176e-01 -7.98580647e-01 1.04775727e+00
-1.45433610e-02 5.07226288e-01 2.03607697e-03 4.71836418e-01
-7.27463141e-02 -5.24642505e-02 6.10940576e-01 1.09050524e+00
5.65981746e-01 1.30833760e-01 -4.30944532e-01 1.00887942e+00
-4.91480008e-02 3.92082408e-02 -1.28410682e-01 6.31187737e-01
3.63007665e-01 1.61014581e+00 -4.79984015e-01 -5.58250189e-01
-4.11832154e-01 1.12195563e+00 -2.73106247e-03 3.18522900e-01
-1.11938024e+00 -3.77183288e-01 5.71775913e-01 4.58592087e-01
4.62903172e-01 6.92757145e-02 -2.96221077e-01 -1.15242541e+00
6.13078028e-02 -1.00168991e+00 1.77845597e-01 -1.35108447e+00
-1.01670969e+00 1.09123373e+00 5.02664633e-02 -1.49307466e+00
3.35349321e-01 -4.69650865e-01 -4.51576769e-01 9.22528744e-01
-1.90349913e+00 -1.13384283e+00 -9.57032025e-01 7.60762632e-01
8.07116628e-01 -1.35687321e-01 4.72118050e-01 7.14391768e-01
-7.47908354e-01 6.33516192e-01 1.12314127e-01 1.55977653e-02
5.58088958e-01 -8.12112212e-01 2.27412909e-01 1.06023490e+00
-2.03427672e-01 4.43666309e-01 7.28658020e-01 -3.93242627e-01
-1.33144081e+00 -1.39585900e+00 2.93468446e-01 5.59727699e-02
4.00613934e-01 -5.29255085e-02 -1.09487188e+00 4.81854558e-01
4.76701349e-01 4.42212999e-01 5.79486527e-02 -3.42622161e-01
-5.14294803e-01 -6.40586913e-01 -1.43753135e+00 6.10237241e-01
1.18527877e+00 5.05346283e-02 -1.67639077e-01 2.14729402e-02
1.00498819e+00 -3.46978515e-01 -1.02227986e+00 5.81441283e-01
4.49166477e-01 -1.31760848e+00 1.32462335e+00 2.49103196e-02
8.44689369e-01 -6.29720986e-01 -2.02901512e-01 -1.02395022e+00
-5.72150171e-01 -2.28238881e-01 -6.51080236e-02 1.08828497e+00
1.27222389e-01 -7.06479669e-01 4.79603678e-01 6.79738969e-02
-1.34283192e-02 -8.37358415e-01 -9.24049497e-01 -8.60578597e-01
-2.06298202e-01 8.99634734e-02 7.70843029e-01 6.78280115e-01
-5.06430149e-01 1.94944203e-01 -4.65757519e-01 4.41537768e-01
8.05538595e-01 7.10585713e-02 5.90410948e-01 -8.35946143e-01
-2.78842419e-01 -2.67800301e-01 -3.98046136e-01 -8.83003116e-01
-3.71127516e-01 -6.74102545e-01 -4.58143026e-01 -1.61494005e+00
4.96527404e-01 -2.45903328e-01 -4.96306837e-01 5.72605252e-01
-1.28775492e-01 1.03393328e+00 4.09403443e-01 3.03280681e-01
-4.50444221e-01 4.62802529e-01 1.51224351e+00 -1.54988647e-01
-2.65006647e-02 -1.50221378e-01 -9.69740570e-01 5.43291450e-01
9.28093195e-01 -6.63628578e-02 -3.11329991e-01 -7.20267296e-01
-3.49151678e-02 1.03713818e-01 7.29916453e-01 -1.34632468e+00
1.43031985e-01 4.98739295e-02 6.16893828e-01 -6.28183007e-01
4.26468819e-01 -7.18598008e-01 3.84095281e-01 5.67688167e-01
-1.37179464e-01 1.87040552e-01 4.08661336e-01 1.53763771e-01
-2.15234160e-01 -8.27615783e-02 1.15446866e+00 -1.77325815e-01
-9.51338351e-01 3.27221423e-01 2.60732900e-02 -4.88301098e-01
8.93847048e-01 -3.08179766e-01 -7.99463868e-01 -1.96817577e-01
-5.79595864e-01 -2.15556219e-01 3.69980752e-01 4.59338725e-01
9.78449285e-01 -1.25897515e+00 -1.07067847e+00 2.11657837e-01
-3.14843595e-01 -2.07218193e-02 8.08864772e-01 7.05397844e-01
-6.13001943e-01 1.46830752e-01 -6.63642704e-01 -4.34404135e-01
-1.54941881e+00 7.74222612e-01 4.37069416e-01 -3.32503766e-01
-1.04379201e+00 6.02426529e-01 3.44905585e-01 2.17852831e-01
-1.10280856e-01 -3.90911475e-02 -5.25827706e-01 -3.37621927e-01
1.03614402e+00 5.66227138e-01 1.28943220e-01 -4.72054869e-01
-1.96492106e-01 6.84340358e-01 -2.55690545e-01 1.62376344e-01
1.47577214e+00 -3.87575954e-01 -2.87452847e-01 -1.81052789e-01
1.00505543e+00 -6.13102987e-02 -1.17267895e+00 -4.41875935e-01
-5.68488955e-01 -8.25431943e-01 3.87712896e-01 -9.03373301e-01
-1.63466549e+00 7.63582230e-01 1.17972279e+00 2.14666780e-03
1.75836480e+00 -1.57302469e-01 9.36801553e-01 2.41667163e-02
4.35493320e-01 -5.28296173e-01 1.95207864e-01 2.11954251e-01
1.20885408e+00 -1.22465980e+00 1.89265028e-01 -4.67481405e-01
-3.82085383e-01 9.06140029e-01 9.46568727e-01 -3.93509179e-01
4.95235771e-01 3.99240285e-01 -5.06049916e-02 1.41869798e-01
-7.37931848e-01 -2.49852747e-01 -1.07823545e-02 5.86023152e-01
2.28916541e-01 -1.52347952e-01 -4.21945304e-01 5.83567321e-01
-2.63455063e-01 2.99899727e-01 7.63345659e-01 3.25441301e-01
-3.80055398e-01 -7.37452030e-01 -6.19183838e-01 3.07999581e-01
-6.70215189e-01 -4.89217043e-01 4.41154271e-01 6.60001695e-01
2.34536454e-01 9.35441852e-01 2.20664725e-01 -4.76195902e-01
2.65864998e-01 -7.17729211e-01 4.27307725e-01 -1.60140246e-01
-5.44155955e-01 -1.34341940e-01 -2.68861502e-01 -7.36566484e-01
-4.73518908e-01 -1.01344071e-01 -1.18677783e+00 -5.56446373e-01
-2.01492459e-01 -1.56402439e-01 6.86535597e-01 3.80834818e-01
5.83010375e-01 9.43176746e-01 5.99173784e-01 -1.00062144e+00
-2.81353027e-01 -1.05207324e+00 -3.94467026e-01 2.84003019e-01
2.44009599e-01 -5.76390088e-01 -2.10140601e-01 4.78958078e-02] | [11.048661231994629, -1.954351544380188] |
60bd3bff-fabb-4755-97bc-65ebdd7a299d | mina-multilevel-knowledge-guided-attention | 1905.11333 | null | https://arxiv.org/abs/1905.11333v3 | https://arxiv.org/pdf/1905.11333v3.pdf | MINA: Multilevel Knowledge-Guided Attention for Modeling Electrocardiography Signals | Electrocardiography (ECG) signals are commonly used to diagnose various cardiac abnormalities. Recently, deep learning models showed initial success on modeling ECG data, however they are mostly black-box, thus lack interpretability needed for clinical usage. In this work, we propose MultIlevel kNowledge-guided Attention networks (MINA) that predict heart diseases from ECG signals with intuitive explanation aligned with medical knowledge. By extracting multilevel (beat-, rhythm- and frequency-level) domain knowledge features separately, MINA combines the medical knowledge and ECG data via a multilevel attention model, making the learned models highly interpretable. Our experiments showed MINA achieved PR-AUC 0.9436 (outperforming the best baseline by 5.51%) in real world ECG dataset. Finally, MINA also demonstrated robust performance and strong interpretability against signal distortion and noise contamination. | ['Hongyan Li', 'Cao Xiao', 'Tengfei Ma', 'Shenda Hong', 'Jimeng Sun'] | 2019-05-27 | null | null | null | null | ['electrocardiography-ecg'] | ['methodology'] | [ 1.40444905e-01 4.20420527e-01 -1.44103676e-01 -6.11226499e-01
-8.48473251e-01 -2.99894720e-01 -2.37201110e-01 4.69572812e-01
1.09419711e-01 7.75588453e-01 3.12510282e-01 -6.35547876e-01
-5.34995317e-01 -4.77685630e-01 -5.94770074e-01 -3.44783068e-01
-4.79570717e-01 3.87885153e-01 -6.27488256e-01 1.17832147e-01
3.00792214e-02 1.86302990e-01 -6.83045626e-01 6.38986647e-01
1.16784453e+00 1.02648866e+00 -3.77094775e-01 1.09715772e+00
3.92817467e-01 8.96614194e-01 -7.06343234e-01 -3.47910225e-01
-1.36975706e-01 -9.09811378e-01 -8.79933238e-01 -1.86577946e-01
8.10922608e-02 -3.07696432e-01 -3.02144021e-01 5.73472977e-01
9.46622074e-01 -4.85661447e-01 5.01236379e-01 -1.02149320e+00
-9.66403723e-01 9.24176097e-01 -4.39794421e-01 6.79571271e-01
8.72232467e-02 2.67361760e-01 9.73494053e-01 -8.53629768e-01
1.74990356e-01 7.29331613e-01 1.25942612e+00 3.30374956e-01
-1.15926814e+00 -4.59714830e-01 -4.62848097e-02 3.70086610e-01
-1.24314177e+00 -4.31813002e-02 8.47086906e-01 -2.65985548e-01
1.04847932e+00 3.99992406e-01 8.89583647e-01 1.05566645e+00
4.57269907e-01 5.66405952e-01 9.71597135e-01 -2.39480302e-01
-1.15252584e-01 -5.55768143e-03 5.92141032e-01 6.10417724e-01
3.95815194e-01 -2.30071262e-01 -4.79201555e-01 -1.48284376e-01
9.09228325e-01 1.22094661e-01 -3.21530163e-01 2.08720222e-01
-1.39981139e+00 5.36400318e-01 6.35681748e-01 2.73410469e-01
-8.43935609e-01 1.51563004e-01 5.53283989e-01 3.02122682e-01
1.85424149e-01 7.56260157e-01 -8.30340862e-01 -6.62958175e-02
-5.74574411e-01 3.18935178e-02 4.75378007e-01 4.84245628e-01
2.73351166e-02 4.54964668e-01 -5.67883909e-01 5.30190885e-01
3.08487922e-01 5.34852803e-01 5.55920482e-01 -8.74824405e-01
2.53797591e-01 5.41846037e-01 -1.66112155e-01 -1.13487411e+00
-8.55007410e-01 -1.11108565e+00 -1.39701271e+00 -2.06229955e-01
1.86330050e-01 -4.77665454e-01 -9.86665070e-01 1.51565373e+00
-1.44255042e-01 5.64634442e-01 2.09453091e-01 1.01773393e+00
1.10812986e+00 3.34721893e-01 3.97436470e-01 -2.85500526e-01
1.58806443e+00 -5.28205156e-01 -9.78121042e-01 -5.75767979e-02
4.58101600e-01 -1.93710644e-02 1.00650585e+00 5.53600967e-01
-1.25175571e+00 -8.45325232e-01 -1.12649333e+00 6.13603322e-03
6.06417432e-02 9.53010991e-02 4.91048932e-01 7.10966766e-01
-8.42897832e-01 7.52507865e-01 -7.61615694e-01 -1.94943454e-02
8.99505258e-01 5.06455541e-01 -7.91501850e-02 4.52167183e-01
-1.51775730e+00 7.22732663e-01 2.99339831e-01 2.69954622e-01
-8.50708485e-01 -9.69006062e-01 -6.45843208e-01 3.36031109e-01
-9.38998982e-02 -1.22834718e+00 9.94700193e-01 -9.30020034e-01
-1.41241837e+00 7.51555264e-01 1.41210062e-02 -9.76536989e-01
3.71845186e-01 -5.56048214e-01 -6.00268900e-01 2.62454540e-01
-1.13751128e-01 4.10657018e-01 6.57498598e-01 -1.08965743e+00
-1.85590193e-01 -5.18879354e-01 2.86731571e-02 1.76875591e-02
-2.14573249e-01 -3.51805389e-01 -1.14260614e-01 -8.21409345e-01
1.52730688e-01 -6.44239008e-01 -2.95436054e-01 -4.05419618e-01
-5.47143579e-01 7.96021819e-02 4.18053657e-01 -1.12427974e+00
1.65495205e+00 -2.03705382e+00 5.03759272e-02 3.27766329e-01
7.42939472e-01 5.40134668e-01 1.53912365e-01 -6.75580930e-03
-5.23509562e-01 5.32866240e-01 -3.50568295e-01 7.57828131e-02
-1.52253270e-01 1.84351444e-01 -1.83520779e-01 1.88921586e-01
5.22909343e-01 1.59372056e+00 -8.83941293e-01 -4.89843041e-01
2.72667617e-01 5.84581375e-01 -6.32294893e-01 1.30631149e-01
1.70071036e-01 7.82296658e-01 -4.01195079e-01 6.20828331e-01
2.57648200e-01 -7.20236659e-01 4.62020606e-01 -4.34084713e-01
6.42523825e-01 3.56554426e-02 -7.92862833e-01 1.48391438e+00
-3.14089745e-01 7.46178627e-01 -4.75501239e-01 -1.18591952e+00
7.65162408e-01 7.70140350e-01 6.45452678e-01 -5.67722559e-01
2.21431911e-01 -6.01071976e-02 5.73290229e-01 -9.46240127e-01
-3.79249841e-01 -3.16441476e-01 -9.29608010e-03 3.31829190e-01
3.56838405e-02 3.63219589e-01 -5.78896880e-01 6.18707165e-02
1.01190960e+00 -1.58586472e-01 5.71909845e-01 -3.57245862e-01
5.15130401e-01 -3.44105303e-01 7.05355048e-01 1.09492135e+00
-3.40401232e-01 8.97971272e-01 5.83711803e-01 -1.02269351e+00
-7.93482602e-01 -1.01636708e+00 -2.77436614e-01 5.38212240e-01
-1.54439136e-01 -5.18608391e-01 -8.28543425e-01 -8.14641595e-01
-2.89268773e-02 5.40198088e-01 -8.15505087e-01 -3.81839216e-01
-5.25826454e-01 -1.03904593e+00 1.07961226e+00 1.22380030e+00
4.35579419e-01 -1.20465040e+00 -8.29999864e-01 5.25590897e-01
-4.61105675e-01 -8.87859046e-01 -1.60243601e-01 1.90385342e-01
-1.15112412e+00 -1.07636237e+00 -6.06839597e-01 -4.31638509e-01
3.39394033e-01 -4.85389799e-01 1.49473214e+00 2.79275656e-01
-6.92360401e-01 2.24772915e-01 -1.54567540e-01 -9.85234797e-01
-1.11405529e-01 -4.06467468e-02 1.52096832e-02 2.51922458e-01
1.96358532e-01 -7.43061721e-01 -9.17500496e-01 -1.54952466e-01
-5.25495112e-01 1.12832841e-02 7.35439956e-01 1.19175994e+00
5.78474402e-01 -2.58676559e-01 1.33860922e+00 -9.52337384e-01
8.34134102e-01 -5.38213730e-01 1.76156729e-01 1.08550012e-01
-8.74445856e-01 -1.99137166e-01 6.89986646e-01 -1.51279330e-01
-6.53370500e-01 -2.54419446e-01 -1.95835829e-01 -4.82203841e-01
-3.54948610e-01 8.01293075e-01 -5.36609888e-02 2.93350458e-01
8.01462233e-01 9.58710983e-02 -1.56938553e-01 -2.93368936e-01
-5.21010011e-02 7.46804655e-01 7.58997321e-01 -4.05219316e-01
2.49320105e-01 1.33495077e-01 1.47719949e-01 -5.49785197e-01
-1.08065236e+00 -5.39399423e-02 -6.57988667e-01 -1.41377281e-02
1.15867651e+00 -8.35259855e-01 -8.06698263e-01 2.70616025e-01
-1.14345527e+00 -6.07099198e-02 -3.51439893e-01 4.57916111e-01
-3.30134183e-01 4.18896794e-01 -6.64847076e-01 -8.88720334e-01
-9.25669789e-01 -6.64735794e-01 8.46532285e-01 -6.94881473e-03
-7.37775981e-01 -1.19445980e+00 -2.75257170e-01 5.10693610e-01
3.98542732e-01 9.27602232e-01 1.21598136e+00 -7.19719648e-01
-2.66543776e-01 -7.99087062e-02 -2.62595981e-01 3.01700145e-01
1.89438269e-01 -3.27318519e-01 -1.06067133e+00 8.20113495e-02
-1.44584011e-02 -1.21635772e-01 5.21008849e-01 8.38515699e-01
1.72817028e+00 -2.52152890e-01 1.45021724e-02 7.22694695e-01
1.07920671e+00 3.00438464e-01 6.98864818e-01 -6.97733387e-02
8.77946377e-01 -5.75850438e-03 7.23845437e-02 5.29880106e-01
5.43266177e-01 3.01478207e-01 2.13342875e-01 -7.04729557e-01
-2.86291521e-02 4.02973741e-02 -4.79670148e-03 8.93582344e-01
-5.58927894e-01 -1.06758803e-01 -1.28293848e+00 5.72344542e-01
-1.93870676e+00 -8.98057461e-01 -6.13484919e-01 1.87319481e+00
8.04499388e-01 2.69743562e-01 1.23184003e-01 4.57266361e-01
3.58328462e-01 -3.14741731e-01 -8.69540215e-01 -6.07616305e-01
-2.82529145e-01 5.14126599e-01 -2.59427540e-03 3.39022368e-01
-1.09977186e+00 1.96174741e-01 6.86570883e+00 9.92832407e-02
-9.74813461e-01 7.58890510e-02 1.23586392e+00 1.74028426e-01
-1.32254988e-01 -4.99432147e-01 -1.88084096e-02 3.94938648e-01
1.30625033e+00 -1.19173534e-01 -1.06130376e-01 4.13854450e-01
2.85177976e-01 6.30158842e-01 -1.22429395e+00 1.17127454e+00
-1.68104377e-02 -1.29185510e+00 1.36228740e-01 -1.71233371e-01
6.37491465e-01 -1.69938415e-01 9.82993767e-02 3.98893028e-01
-3.26272726e-01 -1.58929586e+00 2.11959347e-01 7.82227755e-01
8.68315041e-01 -5.95521748e-01 1.29008031e+00 1.00385241e-01
-7.30453789e-01 -2.07501128e-01 2.27699816e-01 -2.20432013e-01
-1.34737734e-02 5.46798050e-01 -8.08236361e-01 8.81366432e-01
8.03382218e-01 9.79397774e-01 -5.35784483e-01 7.85398602e-01
-1.44430071e-01 1.20156610e+00 7.98691213e-02 4.56236541e-01
5.80093004e-02 2.93266386e-01 4.22144711e-01 1.41880262e+00
9.98566002e-02 5.66091716e-01 1.95708916e-01 8.07323754e-01
-9.93637964e-02 1.15452297e-01 -3.49240780e-01 8.43188241e-02
-5.97209148e-02 9.69224095e-01 -5.74519992e-01 -7.11657524e-01
-1.08090714e-01 8.11345994e-01 -1.77369658e-02 3.70706409e-01
-1.34792745e+00 -5.88408053e-01 3.83995414e-01 1.29244432e-01
-2.16842428e-01 4.75537032e-01 -1.18342996e+00 -1.10394287e+00
3.01851612e-02 -1.17114079e+00 6.97316229e-01 -7.85756171e-01
-1.39360332e+00 7.39704728e-01 -3.29746395e-01 -1.03648424e+00
-5.42787947e-02 -3.60580891e-01 -6.01365030e-01 9.64304745e-01
-1.26588523e+00 -1.07546687e+00 -4.87645775e-01 4.85323548e-01
5.24222314e-01 8.58189762e-02 1.22123647e+00 4.43219364e-01
-5.01283109e-01 8.92228723e-01 -4.23321098e-01 4.60530192e-01
4.46130693e-01 -1.70534754e+00 3.17810178e-01 4.05172646e-01
2.67426312e-01 7.65027583e-01 4.04143512e-01 -4.43253726e-01
-1.08002591e+00 -1.22670150e+00 9.96521354e-01 -6.90285742e-01
2.73171067e-01 1.82217300e-01 -1.28314066e+00 6.55528188e-01
4.31007385e-01 9.63461921e-02 1.14198649e+00 2.88121313e-01
-9.77029204e-02 -2.96567231e-01 -9.99457598e-01 2.87304521e-01
8.03666055e-01 -4.91348803e-01 -9.67895687e-01 1.56135038e-01
5.02743781e-01 -4.80122864e-01 -1.47127616e+00 8.74809980e-01
8.41455877e-01 -7.35108197e-01 1.08285165e+00 -1.13959587e+00
6.26491070e-01 -2.53426641e-01 3.62988859e-01 -1.22436821e+00
-7.63240159e-01 -6.64691806e-01 -5.45503497e-01 7.90915430e-01
5.45125127e-01 -4.99550104e-01 4.24538195e-01 4.47173446e-01
-2.15077773e-01 -1.08197367e+00 -4.77928877e-01 -2.21972883e-01
-1.35664016e-01 -4.07364875e-01 5.18381894e-01 1.31232452e+00
3.51484269e-01 6.15238190e-01 -6.93751872e-01 4.18186128e-01
5.57162404e-01 7.70991296e-02 1.95309356e-01 -1.36462128e+00
-5.67530453e-01 -3.95644993e-01 -5.95278382e-01 -3.28894794e-01
-2.31015012e-01 -1.13906658e+00 -3.54294538e-01 -1.67042673e+00
2.41854832e-01 -6.10185228e-02 -1.32841802e+00 7.21814692e-01
-8.00459564e-01 5.85227907e-01 5.35439029e-02 -2.01367587e-01
-4.90594208e-01 1.20363347e-01 1.05742514e+00 -1.32432148e-01
-5.13813198e-01 -7.69884661e-02 -1.15110993e+00 8.15290213e-01
1.15771723e+00 -3.43632281e-01 -5.17051280e-01 -5.42824864e-01
2.38711890e-02 5.03465891e-01 5.85263193e-01 -1.13527501e+00
-2.01223671e-01 2.99223721e-01 9.29171562e-01 -3.01341951e-01
-2.91452464e-02 -5.71523070e-01 3.35107684e-01 7.63250649e-01
-5.86459637e-01 1.84520051e-01 4.17164087e-01 5.86756706e-01
-2.39145234e-01 4.43289399e-01 6.03302360e-01 3.91886719e-02
-2.03780338e-01 3.76006484e-01 -3.40781868e-01 2.83196837e-01
8.15107107e-01 -1.84347630e-01 2.27307886e-01 -4.11157757e-01
-1.42018104e+00 2.45362744e-01 -4.85837787e-01 3.76783133e-01
7.27225065e-01 -1.20295441e+00 -1.12774551e+00 2.39878044e-01
5.92602380e-02 -5.25484495e-02 6.92528129e-01 1.11835361e+00
-6.55758321e-01 4.24733251e-01 -2.59597778e-01 -8.12193751e-01
-1.23140597e+00 3.39815527e-01 5.68067729e-01 -3.25485647e-01
-1.05385971e+00 6.16808891e-01 5.81474863e-02 -7.04814121e-02
3.03173423e-01 -6.96375608e-01 -2.33809620e-01 -2.85020888e-01
5.04783571e-01 3.31156611e-01 3.58580798e-02 -1.23159945e-01
-5.52079976e-01 6.53487980e-01 2.82727778e-01 2.12139443e-01
1.35895514e+00 -2.55742893e-02 9.47100446e-02 5.52837491e-01
7.04855323e-01 -5.02579272e-01 -7.24871814e-01 2.10129861e-02
5.77001907e-02 -4.44101505e-02 8.66825879e-02 -1.44985938e+00
-1.14379787e+00 1.38999724e+00 8.25373769e-01 1.77827507e-01
1.41910923e+00 -3.92812401e-01 7.46209681e-01 3.15627933e-01
-5.03726229e-02 -6.75000966e-01 1.27351480e-02 1.10033214e-01
7.47198462e-01 -1.16716337e+00 -3.77925076e-02 -8.86432156e-02
-9.47081447e-01 1.09305835e+00 5.00605226e-01 -5.65042906e-02
7.56709039e-01 6.57454506e-02 3.97863954e-01 -3.34707469e-01
-6.68379843e-01 2.47498259e-01 5.38096964e-01 6.96221530e-01
6.12954915e-01 1.94286570e-01 -1.71373621e-01 1.44945836e+00
1.50750160e-01 3.16960990e-01 3.56563598e-01 4.46588546e-01
-9.51832458e-02 -6.52416468e-01 -1.80392876e-01 6.55730247e-01
-1.18185008e+00 -3.67355883e-01 -2.95636714e-01 4.49103057e-01
3.63497198e-01 8.88501048e-01 -3.09325844e-01 -3.24282527e-01
3.34179997e-01 4.50935721e-01 3.10002774e-01 -4.16272610e-01
-1.04954851e+00 7.97791630e-02 4.74176854e-02 -3.04536134e-01
-2.15597332e-01 -2.71239102e-01 -1.45617735e+00 2.04548597e-01
8.72098804e-02 1.73395321e-01 1.74314931e-01 8.18041086e-01
9.97035325e-01 1.28454614e+00 4.34928894e-01 -2.21830070e-01
-3.03847849e-01 -1.07064033e+00 -3.19725156e-01 5.96666753e-01
6.58607543e-01 -1.09904818e-01 9.72902179e-02 5.77080965e-01] | [14.339581489562988, 3.267864465713501] |
3a394b25-ee9b-4504-b108-540b8325a6c5 | tafim-targeted-adversarial-attacks-against | 2112.09151 | null | https://arxiv.org/abs/2112.09151v2 | https://arxiv.org/pdf/2112.09151v2.pdf | TAFIM: Targeted Adversarial Attacks against Facial Image Manipulations | Face manipulation methods can be misused to affect an individual's privacy or to spread disinformation. To this end, we introduce a novel data-driven approach that produces image-specific perturbations which are embedded in the original images. The key idea is that these protected images prevent face manipulation by causing the manipulation model to produce a predefined manipulation target (uniformly colored output image in our case) instead of the actual manipulation. In addition, we propose to leverage differentiable compression approximation, hence making generated perturbations robust to common image compression. In order to prevent against multiple manipulation methods simultaneously, we further propose a novel attention-based fusion of manipulation-specific perturbations. Compared to traditional adversarial attacks that optimize noise patterns for each image individually, our generalized model only needs a single forward pass, thus running orders of magnitude faster and allowing for easy integration in image processing stacks, even on resource-constrained devices like smartphones. | ['Matthias Niessner', 'Lev Markhasin', 'Shivangi Aneja'] | 2021-12-16 | null | null | null | null | ['detecting-image-manipulation'] | ['computer-vision'] | [ 8.46931875e-01 1.79470584e-01 1.23161800e-01 -6.89767599e-02
-6.02951586e-01 -1.04016745e+00 5.47821283e-01 -1.64386630e-01
-2.42448032e-01 3.47386986e-01 -1.06239552e-02 -3.28257561e-01
2.35125482e-01 -7.81122565e-01 -1.03668082e+00 -5.94735086e-01
1.15214035e-01 -1.36350974e-01 -8.07077810e-02 -6.36324510e-02
2.64487088e-01 7.41932273e-01 -1.54923379e+00 2.81302303e-01
5.98668694e-01 8.83298934e-01 -2.94431090e-01 9.90857780e-01
2.60518372e-01 5.43297648e-01 -1.03764319e+00 -8.81424308e-01
7.38911748e-01 -3.36371452e-01 -3.72003824e-01 1.37116805e-01
9.60309446e-01 -8.35031807e-01 -5.47072470e-01 1.32156503e+00
5.61639130e-01 -2.87007131e-02 3.37368071e-01 -1.44811237e+00
-7.70338476e-01 3.87277573e-01 -6.64623499e-01 -1.74821001e-02
4.66341287e-01 7.87345886e-01 2.89046556e-01 -3.32056552e-01
4.89349812e-01 1.27776074e+00 3.93011659e-01 1.02523839e+00
-1.37739110e+00 -9.04042602e-01 1.76316462e-02 -9.85129848e-02
-1.34853971e+00 -9.01552141e-01 6.29646540e-01 -1.81397319e-01
4.71736580e-01 7.20434189e-01 2.84287333e-01 1.33678031e+00
-6.70016557e-02 4.35969979e-01 1.03517449e+00 -2.40927890e-01
1.66749164e-01 1.08056076e-01 -3.13083261e-01 4.58566308e-01
4.85268563e-01 3.28591615e-02 -3.67136955e-01 -6.09293342e-01
5.59553623e-01 9.56090391e-02 -6.54568374e-01 -1.52422637e-01
-7.30577886e-01 7.46203482e-01 2.19057336e-01 -1.15459464e-01
-1.80469975e-01 5.04358232e-01 2.55049825e-01 2.12113261e-01
2.00047150e-01 4.37438011e-01 -1.77677095e-01 9.51808244e-02
-9.42397416e-01 5.82405031e-01 6.55742705e-01 8.32046986e-01
7.17574656e-01 -2.02088431e-02 -3.92038107e-01 2.59464443e-01
5.29945642e-02 6.27236426e-01 2.10588261e-01 -9.91525531e-01
5.84101140e-01 2.46563166e-01 1.53840005e-01 -1.50701809e+00
9.93735865e-02 6.63562194e-02 -7.29249716e-01 5.05896688e-01
5.32213688e-01 -2.47968897e-01 -9.74284768e-01 1.93192804e+00
3.35046887e-01 4.58584428e-01 -2.63093829e-01 6.77684069e-01
1.47772446e-01 3.23570162e-01 -6.71564043e-02 4.16578958e-03
1.39269781e+00 -5.78320205e-01 -6.34727061e-01 -1.34568691e-01
2.33504787e-01 -5.59290230e-01 1.15751541e+00 4.74824339e-01
-1.17504263e+00 -3.39014590e-01 -1.28615808e+00 -1.41231731e-01
-5.42623818e-01 -3.56378406e-01 3.47731858e-01 1.38087070e+00
-1.04318452e+00 5.57283819e-01 -6.37614906e-01 2.94869542e-02
8.98173988e-01 6.26231909e-01 -5.08994997e-01 3.90815847e-02
-8.67968321e-01 6.23467267e-01 -2.02451721e-02 -1.47477120e-01
-9.62035954e-01 -1.00405812e+00 -8.05287182e-01 2.25361168e-01
4.30741996e-01 -5.03759205e-01 1.01407719e+00 -1.01451182e+00
-1.64174044e+00 9.28834856e-01 -1.24780782e-01 -7.08516896e-01
7.62111008e-01 -2.31173843e-01 -2.66829908e-01 4.74670917e-01
-3.79476368e-01 6.80620193e-01 1.79094183e+00 -1.43797052e+00
-5.53472266e-02 -4.82674956e-01 2.89028972e-01 -8.28001350e-02
-8.31231534e-01 2.29918644e-01 -5.37220776e-01 -8.59159112e-01
-5.70772052e-01 -9.88847077e-01 -2.44079217e-01 2.68174827e-01
-7.09985495e-01 6.02415383e-01 1.25478148e+00 -7.05065072e-01
1.27674913e+00 -2.35613418e+00 1.42836496e-01 4.27295059e-01
4.31228876e-01 7.52091825e-01 -3.61502916e-01 2.20452845e-01
-1.30356714e-01 6.58714294e-01 -4.36047226e-01 -6.55167818e-01
-4.18259650e-02 8.45962167e-02 -7.10898578e-01 6.09331071e-01
3.28234524e-01 9.47947323e-01 -5.60846984e-01 -2.33587530e-02
2.36792758e-01 6.45374060e-01 -8.26393485e-01 1.38026148e-01
-5.01478851e-01 5.93099177e-01 -2.80164599e-01 4.96422648e-01
1.39532280e+00 3.41686755e-01 1.36372089e-01 -2.08901111e-02
3.86239111e-01 -2.62255240e-02 -1.02819133e+00 1.18409348e+00
-3.84634048e-01 4.59729344e-01 4.94092077e-01 -5.46418548e-01
5.25629938e-01 1.57353118e-01 7.40946531e-02 -1.75859600e-01
3.56973916e-01 -1.39240921e-01 -2.00927526e-01 -3.93773437e-01
4.53617066e-01 2.81715631e-01 -9.87300947e-02 6.42956555e-01
-4.12762493e-01 -1.16972141e-01 -2.60455698e-01 1.63309455e-01
1.32173669e+00 -1.85401753e-01 4.74278890e-02 2.38050669e-01
5.25411248e-01 -4.53822374e-01 3.71060640e-01 9.33637202e-01
-2.14619875e-01 7.29696393e-01 6.62691414e-01 -1.62418097e-01
-1.02470934e+00 -8.16459477e-01 1.69492468e-01 7.92127013e-01
-2.64928862e-02 -5.18920064e-01 -1.31954908e+00 -1.04422653e+00
4.33478594e-01 5.13434410e-01 -5.91000974e-01 -3.87136936e-01
-7.52804935e-01 -3.94130677e-01 1.14050210e+00 1.50439277e-01
5.93066812e-01 -8.13056290e-01 -6.25711322e-01 -2.92402834e-01
1.71449766e-01 -1.12911868e+00 -9.09682035e-01 -5.35071373e-01
-4.47020173e-01 -1.06800771e+00 -5.43435931e-01 -2.49362826e-01
8.88420582e-01 3.51035386e-01 6.98130548e-01 4.69890982e-01
-4.76214647e-01 5.33417106e-01 -1.75516471e-01 -4.62267280e-01
-5.77021062e-01 -1.13709316e-01 1.25759915e-01 4.56376284e-01
2.58325636e-01 -7.11452842e-01 -6.67331457e-01 4.48685745e-03
-1.43923295e+00 -5.17524660e-01 2.07829550e-01 5.67738652e-01
1.53985947e-01 2.57271320e-01 1.21917956e-01 -8.35826993e-01
7.70459712e-01 -4.11639482e-01 -8.05696845e-01 1.15317814e-01
-1.29690811e-01 -1.82451040e-01 9.24332321e-01 -8.26660097e-01
-7.33985662e-01 1.43287435e-01 8.78131911e-02 -8.97295952e-01
-3.35873574e-01 -2.30913207e-01 -5.66499174e-01 -5.82185328e-01
6.44442260e-01 2.11462289e-01 3.13637376e-01 -4.01520610e-01
6.61691546e-01 7.10727036e-01 7.85118163e-01 -4.28903431e-01
1.32419908e+00 6.52846754e-01 4.56515513e-02 -6.79899275e-01
-1.68574542e-01 1.80079266e-01 -2.75510252e-01 -8.12929496e-02
6.35773480e-01 -5.29379725e-01 -1.27865803e+00 7.77370989e-01
-1.17075658e+00 -2.33428851e-01 6.35264218e-02 -1.41134679e-01
-4.03742820e-01 7.17146158e-01 -4.36288506e-01 -8.04663718e-01
-2.84590989e-01 -1.37893498e+00 1.13026953e+00 1.57080054e-01
-1.09035030e-01 -5.29080570e-01 -3.77004802e-01 3.93435389e-01
6.52441919e-01 6.19532883e-01 4.79757488e-01 -4.94798005e-01
-8.94375324e-01 -5.01708925e-01 -1.02209844e-01 5.52209616e-01
1.80829585e-01 9.05431807e-02 -1.06213808e+00 -5.87793887e-01
3.53642672e-01 -2.99517781e-01 6.10705256e-01 -4.53800000e-02
1.72773194e+00 -8.44758928e-01 -2.66982734e-01 9.38940048e-01
1.30682492e+00 8.55957717e-02 1.07848382e+00 1.64900482e-01
6.81405246e-01 5.51240206e-01 1.90671057e-01 5.59035659e-01
-2.17452645e-01 8.58362556e-01 5.92098832e-01 8.95685703e-02
3.17729414e-02 -2.77126491e-01 5.96169889e-01 -1.82715759e-01
2.63891965e-01 -4.31588352e-01 -5.37364304e-01 1.95916355e-01
-1.49507439e+00 -1.13598967e+00 1.78948849e-01 2.54896164e+00
7.76230037e-01 1.37400553e-02 7.98827559e-02 3.31954025e-02
7.68542111e-01 3.02516967e-01 -6.06579721e-01 -5.01676679e-01
4.84203883e-02 4.99422520e-01 9.67872679e-01 5.50263107e-01
-1.15021217e+00 9.45423663e-01 6.69682074e+00 9.03183997e-01
-1.12279618e+00 6.67242035e-02 6.49127483e-01 -5.28677285e-01
-3.21735501e-01 -8.47213566e-02 -7.71217406e-01 7.96446264e-01
8.42826843e-01 -1.24901779e-01 9.41218853e-01 6.55490935e-01
2.30197445e-01 4.66499962e-02 -9.55709994e-01 9.96730268e-01
2.75252253e-01 -1.32753289e+00 3.35588276e-01 4.44107145e-01
3.93422276e-01 -6.28662467e-01 6.79365993e-01 -1.79809481e-01
3.31000507e-01 -1.21069551e+00 4.66041058e-01 3.13198328e-01
9.93117273e-01 -1.01466560e+00 1.16049528e-01 1.44308701e-01
-6.21852696e-01 -1.66457459e-01 -3.32648844e-01 6.87634051e-02
1.17225304e-01 4.24976796e-01 -6.78942680e-01 2.04162598e-01
5.27123690e-01 3.46788205e-03 -5.24121583e-01 4.17116761e-01
-3.36041600e-01 5.61539292e-01 -5.48109531e-01 3.83091301e-01
-4.35899533e-02 2.94140354e-02 8.09761465e-01 1.03130698e+00
2.50922263e-01 -4.77233483e-03 -1.25957966e-01 8.14824462e-01
-5.94147444e-01 -2.41882741e-01 -1.01306081e+00 -9.88955647e-02
7.22474813e-01 1.13180304e+00 -2.92848825e-01 -2.01870516e-01
-2.09205002e-01 1.55418563e+00 2.50741988e-01 3.30336571e-01
-1.13756800e+00 -4.61867183e-01 1.17690992e+00 1.07916385e-01
4.98577744e-01 -1.48248985e-01 -1.88283697e-01 -1.13566589e+00
3.71991158e-01 -1.51988852e+00 9.56597030e-02 -4.04491603e-01
-1.07819605e+00 3.30436587e-01 -4.67253923e-02 -9.76114154e-01
-2.29176637e-02 -4.69599038e-01 -6.02591455e-01 9.40480649e-01
-1.22911704e+00 -1.10816765e+00 -1.79843128e-01 8.30074072e-01
2.81993765e-02 -1.17175199e-01 7.74605870e-01 8.43350962e-02
-6.94830954e-01 1.27543855e+00 -1.25659734e-01 2.24663287e-01
7.53086329e-01 -8.08126330e-01 8.21189821e-01 1.16231716e+00
-1.36654302e-01 1.10135388e+00 6.35642588e-01 -7.55122364e-01
-1.63479793e+00 -1.06398451e+00 3.40169936e-01 -5.36155760e-01
3.87034237e-01 -7.49615610e-01 -9.10945356e-01 6.60323679e-01
2.22134769e-01 4.22463976e-02 6.49809837e-01 -6.35282338e-01
-8.20454359e-01 -8.81481767e-02 -1.82277453e+00 1.05524373e+00
1.05226767e+00 -7.11362064e-01 -6.43775165e-02 3.24359864e-01
7.82482445e-01 -4.60777044e-01 -5.27643919e-01 2.24682674e-01
4.45249707e-01 -9.55063522e-01 1.11349332e+00 -6.82089329e-01
3.03855449e-01 -5.40072262e-01 -1.66201934e-01 -9.30524230e-01
-2.95996014e-02 -1.41537881e+00 -5.57125390e-01 1.39107919e+00
1.13007529e-02 -7.30366886e-01 8.95300984e-01 9.23550010e-01
4.79207397e-01 -2.97863901e-01 -8.73284161e-01 -7.22450852e-01
-8.56626332e-02 -5.00759065e-01 1.08339596e+00 7.85057843e-01
-2.59510070e-01 -5.62521994e-01 -7.78432667e-01 6.59257233e-01
7.02351809e-01 -4.60503459e-01 1.23829579e+00 -3.86211097e-01
-5.50083041e-01 -3.27684313e-01 -6.93360984e-01 -8.27476561e-01
1.73248440e-01 -5.89815021e-01 -1.99913949e-01 -4.37950820e-01
2.61833775e-03 -1.70769006e-01 5.38304448e-02 5.75820267e-01
-4.03483301e-01 7.64823318e-01 5.00336349e-01 2.02342235e-02
5.14899679e-02 2.01157913e-01 9.18338895e-01 -1.77733287e-01
-3.65880243e-02 -4.87257205e-02 -1.10615218e+00 5.51059484e-01
8.62632275e-01 -5.40494084e-01 -5.19392014e-01 -4.57024783e-01
-4.64093871e-02 -3.25930923e-01 6.27080321e-01 -1.12831390e+00
6.95131198e-02 -1.28374785e-01 2.37797365e-01 4.51055449e-03
3.91883761e-01 -9.04962301e-01 8.46368596e-02 3.61000180e-01
-2.76401311e-01 -1.27436757e-01 3.28850746e-01 6.16870105e-01
1.49518177e-01 -1.20871395e-01 9.63698328e-01 -7.67111108e-02
-1.18656699e-02 4.56607491e-01 -3.61568987e-01 -3.33618760e-01
1.37863779e+00 -1.28451794e-01 -4.57897276e-01 -6.58984244e-01
-5.18419802e-01 -6.33430555e-02 1.08547699e+00 3.70385975e-01
6.10332668e-01 -9.78025556e-01 -5.80681860e-01 5.52465737e-01
-2.17564091e-01 -2.33688623e-01 2.61786550e-01 2.86739200e-01
-5.55917323e-01 -1.48774385e-01 -4.72901426e-02 -1.21547677e-01
-1.69302440e+00 1.14722216e+00 2.96436638e-01 4.09633666e-02
-4.18752909e-01 8.12032580e-01 -6.00109762e-03 -1.10966779e-01
2.82221943e-01 8.05768359e-04 2.05570340e-01 -2.67524689e-01
1.14745593e+00 2.35801086e-01 -1.45053476e-04 -5.63755393e-01
-2.23173544e-01 4.94405389e-01 -2.97742963e-01 -2.25887761e-01
7.71087289e-01 -1.43089131e-01 -2.67421573e-01 -4.92216676e-01
1.41908348e+00 6.27765954e-01 -1.43564868e+00 2.01406419e-01
-3.70024055e-01 -1.06918406e+00 -2.38568693e-01 -4.66361582e-01
-1.21003938e+00 6.65335953e-01 5.34818888e-01 2.72862285e-01
1.42359090e+00 -4.17572618e-01 1.01236820e+00 1.64581209e-01
2.20738426e-01 -5.52234888e-01 -1.29060566e-01 -4.28775735e-02
8.93321216e-01 -8.38549614e-01 1.08762152e-01 -6.72696769e-01
-4.46775287e-01 8.67091894e-01 4.52915639e-01 -7.97144175e-02
4.83365536e-01 5.80823123e-01 2.83586476e-02 1.74548268e-01
-4.55716133e-01 2.53452778e-01 2.92623695e-02 1.07394123e+00
-2.28577584e-01 -1.46618769e-01 -2.72786375e-02 2.36851573e-01
-1.88867658e-01 -3.02095735e-03 6.02587998e-01 1.16112721e+00
-4.38111499e-02 -1.55991948e+00 -8.04272771e-01 3.98348629e-01
-8.11944485e-01 -9.58012491e-02 -5.85406005e-01 3.18301976e-01
2.75964051e-01 1.12354040e+00 1.61878280e-02 -4.79210109e-01
1.50521711e-01 -1.06577324e-02 3.38304579e-01 -3.98003638e-01
-8.30749333e-01 -3.80309820e-01 -2.96117187e-01 -8.96731436e-01
-6.07290380e-02 -6.19670212e-01 -6.31320894e-01 -8.07976782e-01
-3.45674977e-02 -3.98856878e-01 5.60089111e-01 6.05687380e-01
8.40251207e-01 1.18807286e-01 1.00990927e+00 -1.09829414e+00
-6.73192859e-01 -5.95033824e-01 -9.31808800e-02 7.64212489e-01
5.43698907e-01 -4.14098799e-01 -5.90014696e-01 1.80268764e-01] | [12.707466125488281, 0.9327552914619446] |
a06a73bf-79bd-4691-a9f7-61e9c6eafe37 | video-reflection-removal-through-spatio | null | null | http://openaccess.thecvf.com/content_iccv_2017/html/Nandoriya_Video_Reflection_Removal_ICCV_2017_paper.html | http://openaccess.thecvf.com/content_ICCV_2017/papers/Nandoriya_Video_Reflection_Removal_ICCV_2017_paper.pdf | Video Reflection Removal Through Spatio-Temporal Optimization | Reflections can obstruct content during video capture and hence their removal is desirable. Current removal techniques are designed for still images, extracting only one reflection (foreground) and one background layer from the input. When extended to videos, unpleasant artifacts such as temporal flickering and incomplete separation are generated. We present a technique for video reflection removal by jointly solving for motion and separation. The novelty of our work is in our optimization formulation as well as the motion initialization strategy. We present a novel spatio-temporal optimization that takes n frames as input and directly estimates 2n frames as output, n for each layer. We aim to fully utilize spatio-temporal information in our objective terms. Our motion initialization is based on iterative frame-to-frame alignment instead of the direct alignment used by current approaches. We compare against advanced video extensions of the state of the art, and we significantly reduce temporal flickering and improve separation. In addition, we reduce image blur and recover moving objects more accurately. We validate our approach through subjective and objective evaluations on real and controlled data.
| ['Mohamed Hefeeda', 'Ajay Nandoriya', 'Wojciech Matusik', 'Mohamed Elgharib', 'Changil Kim'] | 2017-10-01 | null | null | null | iccv-2017-10 | ['reflection-removal'] | ['computer-vision'] | [ 7.29397058e-01 -2.79551029e-01 2.72250414e-01 4.19419780e-02
-6.49702907e-01 -6.56722665e-01 3.99401665e-01 -4.62412655e-01
-3.61489207e-01 5.66280186e-01 4.13024306e-01 -4.68209460e-02
7.60214627e-02 -6.21780008e-02 -7.16708302e-01 -7.47374058e-01
-5.37074283e-02 -3.51704031e-01 5.51896811e-01 1.97068498e-01
3.80851984e-01 4.43002820e-01 -1.52138996e+00 4.61415052e-01
7.78564572e-01 6.54361188e-01 3.05755466e-01 1.20640373e+00
3.14097911e-01 1.30274200e+00 -6.48306608e-01 1.39661267e-01
4.39489990e-01 -5.89729726e-01 -6.48134470e-01 6.63177729e-01
1.00009465e+00 -7.85698831e-01 -4.71491605e-01 8.23079407e-01
3.95772636e-01 4.67544764e-01 2.60765851e-01 -1.09410036e+00
-2.28231743e-01 -3.47910449e-02 -7.81779349e-01 3.75189602e-01
5.32163978e-01 1.95975229e-01 3.00823361e-01 -9.00765002e-01
5.69934607e-01 1.02055025e+00 6.58489645e-01 6.39685750e-01
-1.25602973e+00 -2.52427846e-01 3.18266481e-01 3.32953811e-01
-1.34517610e+00 -1.11665189e+00 7.99709797e-01 -3.40459347e-01
9.35568869e-01 5.93838632e-01 6.23348355e-01 9.03820217e-01
-3.39455046e-02 9.19650793e-01 7.86838293e-01 -5.79468668e-01
7.42266551e-02 -1.34263694e-01 -1.06694236e-01 4.57604915e-01
2.16721058e-01 7.11677000e-02 -6.33301854e-01 -2.84698922e-02
9.48466599e-01 -2.37961970e-02 -9.44446206e-01 -2.50007033e-01
-1.16274631e+00 4.59477641e-02 -1.67947665e-01 1.95647374e-01
-3.68561983e-01 4.00793940e-01 6.09137630e-03 -6.10232614e-02
5.84475398e-01 1.25049129e-01 -8.43049884e-02 -2.96117395e-01
-1.57730544e+00 2.08621889e-01 8.00497293e-01 1.04719794e+00
5.09002805e-01 2.51030594e-01 -3.51226240e-01 7.10482597e-01
1.57089695e-01 5.81093848e-01 1.08008310e-01 -1.63812196e+00
3.19904059e-01 -2.88262013e-02 5.64644098e-01 -1.09100080e+00
-7.69031420e-02 -1.31173447e-01 -5.75113535e-01 4.18711662e-01
4.34176952e-01 -1.54426306e-01 -9.06109750e-01 1.39407229e+00
2.90786654e-01 8.07933509e-01 -1.13277487e-01 1.09511793e+00
4.47829038e-01 7.06011295e-01 -3.56663048e-01 -7.35782266e-01
9.39847767e-01 -1.37229085e+00 -1.05589926e+00 -3.73287320e-01
2.64005274e-01 -1.05898821e+00 5.64489126e-01 7.72734880e-01
-1.60012293e+00 -4.04380351e-01 -9.99018610e-01 -1.25129715e-01
3.71883929e-01 1.93317324e-01 1.30111635e-01 7.47973204e-01
-1.20468485e+00 5.31291664e-01 -9.75235641e-01 -8.39475393e-02
2.23929256e-01 4.20417428e-01 -1.78184509e-01 -2.77619481e-01
-5.77710092e-01 8.80670726e-01 2.08754186e-02 3.36805254e-01
-7.64772356e-01 -5.13500690e-01 -8.53120148e-01 -7.65921474e-02
8.06555092e-01 -6.57640040e-01 1.37833619e+00 -1.36480033e+00
-1.66936183e+00 5.91802359e-01 -6.65258527e-01 -2.05691889e-01
5.82502007e-01 -5.92954397e-01 -2.75795013e-01 5.32822907e-01
-3.23268652e-01 3.85154188e-01 1.20812023e+00 -1.61957443e+00
-5.81836343e-01 2.20842391e-01 1.20876864e-01 4.01154906e-01
-3.11613262e-01 2.13072166e-01 -1.08001757e+00 -7.09501565e-01
-3.73769295e-03 -9.28012013e-01 -3.02058876e-01 9.16640535e-02
-5.93835823e-02 6.22301936e-01 1.07232499e+00 -1.00920427e+00
1.51994169e+00 -2.14837480e+00 2.42451981e-01 -1.27064183e-01
4.35783207e-01 5.24940193e-01 -1.81335002e-01 1.05478667e-01
-2.76065376e-02 -2.45113730e-01 -3.11998934e-01 -7.22217381e-01
-3.80603731e-01 1.66772544e-01 -1.09212726e-01 8.51672232e-01
1.31046921e-01 5.46329260e-01 -9.84945238e-01 -4.21674520e-01
6.74031675e-01 7.46571839e-01 -8.00467491e-01 3.04167151e-01
3.53926532e-02 4.27763611e-01 8.01787972e-02 5.53920209e-01
1.01274836e+00 -1.03923507e-01 2.28063002e-01 -5.26736021e-01
-4.09928441e-01 1.53659835e-01 -1.41841590e+00 1.37041044e+00
-3.81105572e-01 1.08174062e+00 4.82284606e-01 -6.29017651e-01
2.55334944e-01 3.75891715e-01 6.71606719e-01 -5.63765645e-01
-2.48237737e-02 -8.69413279e-03 -2.08161920e-01 -7.91474462e-01
8.59980941e-01 1.50283426e-01 4.47033197e-01 2.21636683e-01
-2.92039901e-01 2.37350106e-01 3.14032346e-01 1.98894098e-01
1.23566949e+00 4.99195248e-01 2.27477536e-01 -1.86888725e-01
6.28497005e-01 -1.91268563e-01 6.16048813e-01 7.35458136e-01
-1.81350455e-01 1.23008752e+00 1.51996955e-01 -1.28312767e-01
-1.02126503e+00 -9.06675696e-01 3.47223490e-01 6.82137847e-01
5.45515776e-01 -4.57444519e-01 -8.84863853e-01 -3.54146123e-01
-4.45786864e-01 6.09777749e-01 -2.33781129e-01 1.39361009e-01
-1.11193120e+00 -5.77012658e-01 3.00776422e-01 4.11150813e-01
2.61340082e-01 -6.80528164e-01 -9.55722392e-01 2.96384692e-01
-6.14729702e-01 -1.64325976e+00 -8.59104753e-01 -2.95509636e-01
-9.36255753e-01 -1.15230715e+00 -8.57228279e-01 -3.58312577e-01
8.53634596e-01 1.02750993e+00 1.12347960e+00 1.86556369e-01
-4.57372904e-01 7.28280485e-01 -2.12554336e-01 6.08187243e-02
-2.84294158e-01 -5.73934317e-01 -9.34021026e-02 2.77130187e-01
-8.27323943e-02 -3.88578326e-01 -9.07648146e-01 4.06121671e-01
-1.23499942e+00 4.23418552e-01 3.33374441e-01 4.83766019e-01
1.19682118e-01 2.21655414e-01 -3.51779908e-01 -6.55529797e-01
6.66876743e-03 -1.38273284e-01 -6.65938258e-01 2.80691106e-02
-9.32209119e-02 -2.00566098e-01 4.86233234e-01 -6.12012923e-01
-1.49082518e+00 8.24897215e-02 3.74310315e-01 -9.10733759e-01
-1.76293731e-01 -2.12545633e-01 -1.20631456e-01 -1.54864162e-01
3.70294929e-01 6.31885901e-02 -2.01936468e-01 -4.64344233e-01
1.87586769e-01 4.71176773e-01 7.02023804e-01 -2.04794660e-01
7.73289323e-01 1.00136495e+00 -1.49358556e-01 -1.31295681e+00
-6.82234168e-01 -7.90836513e-01 -6.48559928e-01 -6.30156100e-01
7.84448087e-01 -9.06089485e-01 -7.15674937e-01 6.46940708e-01
-1.42711377e+00 -4.58202153e-01 -3.20984907e-02 7.19798267e-01
-4.54977006e-01 8.13618124e-01 -6.20881617e-01 -1.17484891e+00
-7.93624669e-02 -1.08353543e+00 1.19104207e+00 1.18543983e-01
-1.58301830e-01 -9.73706782e-01 4.62464876e-02 4.60988730e-01
4.88870651e-01 7.31713995e-02 2.77392995e-02 2.96816587e-01
-1.12178147e+00 2.94667333e-01 -3.07070702e-01 4.50455099e-01
2.48566389e-01 1.44841209e-01 -1.10142171e+00 -4.47271436e-01
3.33428174e-01 3.94183785e-01 9.21894729e-01 6.27190828e-01
8.17804813e-01 -4.58331078e-01 -2.41199583e-01 8.18998396e-01
1.63789201e+00 3.42485905e-01 9.79300022e-01 3.85375828e-01
9.92760122e-01 5.63834190e-01 6.81173980e-01 3.83356035e-01
1.45088891e-02 1.00643849e+00 3.62779319e-01 -3.56674612e-01
-4.26778764e-01 3.95423859e-01 7.74045646e-01 5.58610618e-01
-5.03414214e-01 -6.68209255e-01 -6.34213269e-01 6.55538678e-01
-2.02395582e+00 -1.28951478e+00 -4.72587913e-01 2.46651530e+00
5.69276989e-01 -2.39762813e-01 7.62853548e-02 2.06936538e-01
6.35443568e-01 2.15311736e-01 -1.77157179e-01 9.48385894e-02
-1.31810695e-01 1.31754586e-02 7.90253818e-01 9.87647176e-01
-1.07995057e+00 6.14654481e-01 6.48363447e+00 3.84446263e-01
-9.73792434e-01 9.86909717e-02 3.45095247e-01 -8.62655640e-01
-1.22786365e-01 1.35273576e-01 -5.89988947e-01 3.95066679e-01
6.64314032e-01 2.19208822e-01 6.19184017e-01 1.93185970e-01
7.66697228e-01 -4.96616453e-01 -1.05019665e+00 1.27902377e+00
6.57898664e-01 -1.12269795e+00 -3.55258137e-01 1.14790723e-01
7.77887762e-01 -3.28325778e-01 -1.56934470e-01 -4.24493492e-01
-2.04619020e-01 -8.62495244e-01 9.46360290e-01 7.03845620e-01
5.00041664e-01 -5.15545070e-01 3.76130760e-01 1.06886551e-02
-1.17964518e+00 5.78540675e-02 -8.33237618e-02 -1.02147631e-01
6.81520522e-01 6.77497625e-01 -3.58438462e-01 5.41619182e-01
6.82418883e-01 8.41090143e-01 -3.28287035e-01 1.24773133e+00
-1.71044357e-02 4.60284382e-01 -3.85152251e-01 6.14107907e-01
-2.34063081e-02 -4.13232565e-01 9.31682944e-01 1.50298524e+00
2.94861525e-01 3.32988113e-01 1.67427376e-01 5.75373650e-01
3.35057884e-01 -2.65672415e-01 -4.43972647e-01 2.58192450e-01
8.15709457e-02 1.05740905e+00 -7.36771286e-01 -4.56908226e-01
-6.97395146e-01 1.50830877e+00 -1.75521940e-01 7.29307592e-01
-1.09154582e+00 -1.34503812e-01 7.63405025e-01 2.07667366e-01
5.05975783e-01 -4.93899673e-01 2.87328679e-02 -1.52910578e+00
2.92887539e-01 -9.15574610e-01 3.00024915e-02 -8.33371043e-01
-6.05264902e-01 3.37746412e-01 1.26044601e-01 -1.47590661e+00
-1.19076364e-01 -4.94178087e-01 -4.52317268e-01 6.79001749e-01
-1.63554490e+00 -7.76337266e-01 -5.66444516e-01 5.88927865e-01
8.84863496e-01 4.23248857e-01 2.13909611e-01 7.84535408e-01
-6.25996947e-01 2.15683535e-01 1.47262558e-01 -1.50259435e-01
8.35295618e-01 -9.35479760e-01 2.11441129e-01 1.54019916e+00
-2.91028339e-02 5.41700780e-01 9.84264731e-01 -4.77192014e-01
-1.53707862e+00 -9.01315689e-01 6.88537657e-01 -3.77191722e-01
3.35840732e-01 -2.84907252e-01 -9.20876026e-01 6.94854140e-01
3.32679063e-01 -6.83262721e-02 4.17458206e-01 -4.53368217e-01
6.79409131e-02 1.93628762e-02 -7.46100783e-01 7.93088198e-01
1.14109504e+00 -2.84053892e-01 -2.45300591e-01 1.19372591e-01
3.52869511e-01 -6.42631650e-01 -3.55157822e-01 1.29884332e-01
7.28051305e-01 -1.33492589e+00 1.22403562e+00 3.90917882e-02
3.36704910e-01 -7.83079684e-01 -8.78818259e-02 -8.39895427e-01
-2.73268014e-01 -1.13832414e+00 -5.72608113e-01 1.14773571e+00
-1.58614099e-01 -1.62756532e-01 6.74053371e-01 6.88619018e-01
-1.24745548e-01 -1.02209985e-01 -6.04763746e-01 -7.43621945e-01
-8.18917096e-01 -4.39422399e-01 -2.23722681e-01 7.56721079e-01
-2.74334550e-01 1.88201070e-01 -1.02772582e+00 4.77046341e-01
1.09620488e+00 2.22175140e-02 9.71555233e-01 -6.09023452e-01
-5.65945268e-01 -1.70759544e-01 -2.31798425e-01 -1.47651196e+00
-2.96027094e-01 -6.77745938e-02 3.96984577e-01 -1.38128948e+00
1.86007053e-01 2.26444960e-01 6.76192576e-03 8.57518762e-02
-3.14130455e-01 5.65557480e-01 5.37307501e-01 1.30200267e-01
-7.89130211e-01 2.69762605e-01 9.47354317e-01 1.11947231e-01
-4.74021941e-01 -2.15192869e-01 -4.20381099e-01 9.48699474e-01
4.85370427e-01 -3.64958435e-01 -4.69635069e-01 -8.38813126e-01
-6.27362952e-02 2.49091178e-01 6.50640786e-01 -1.07593393e+00
2.74665475e-01 -2.23606586e-01 4.21617657e-01 -5.28639138e-01
6.17624700e-01 -1.06804180e+00 4.28936213e-01 2.42600515e-01
7.58260787e-02 -1.08489238e-01 4.06287462e-01 6.18419707e-01
-4.93547656e-02 -3.22178036e-01 8.37816179e-01 -8.20022821e-02
-6.55541480e-01 -4.36037369e-02 -6.53722227e-01 -1.28579393e-01
9.40322518e-01 -5.64567864e-01 -4.02598470e-01 -6.39033377e-01
-6.44652128e-01 9.57554281e-02 8.41332436e-01 2.33525902e-01
7.84733832e-01 -1.02167618e+00 -6.78760886e-01 1.10325329e-02
-4.16071683e-01 -1.09440379e-01 5.28956115e-01 1.29325151e+00
-8.95897150e-01 1.72811463e-01 4.55209874e-02 -7.08558261e-01
-1.85352516e+00 5.46260238e-01 4.21162933e-01 -4.87727113e-03
-8.14726353e-01 5.59825957e-01 5.58388829e-01 5.41932046e-01
2.70035297e-01 -3.71152580e-01 -1.48592889e-01 -1.56334072e-01
9.30627465e-01 8.60749722e-01 -1.38586774e-01 -6.71208143e-01
-4.39889878e-01 7.35044539e-01 3.99127677e-02 -3.99025202e-01
1.21233034e+00 -6.68095171e-01 -6.03442602e-02 2.86303520e-01
1.14716566e+00 4.99447644e-01 -1.69130206e+00 -6.69808760e-02
-2.72732049e-01 -1.04753327e+00 6.45543113e-02 -3.36760700e-01
-1.03626966e+00 6.21876836e-01 5.12120962e-01 1.03916079e-01
1.49865723e+00 -4.10214990e-01 8.92857969e-01 3.63369919e-02
1.52058033e-02 -9.93299663e-01 1.29564255e-01 3.65419537e-01
5.27805984e-01 -9.46475863e-01 3.66863728e-01 -7.39740252e-01
-3.74020308e-01 1.10560727e+00 3.88815761e-01 -9.30917636e-02
1.47319540e-01 6.99440420e-01 1.59900576e-01 1.09101675e-01
-8.04306507e-01 -3.26022059e-01 3.45674634e-01 5.78013420e-01
5.07198930e-01 -6.64255977e-01 -9.09534022e-02 -1.20629899e-01
4.65923965e-01 7.63610825e-02 9.18592930e-01 1.25143337e+00
-3.70126635e-01 -9.05396521e-01 -8.67771566e-01 -1.16073705e-01
-6.54534638e-01 -2.13805154e-01 -8.61534923e-02 4.99470055e-01
-3.18461247e-02 1.23312902e+00 9.03213695e-02 4.66403589e-02
2.66072959e-01 -3.09808999e-01 8.73795629e-01 -3.57598037e-01
-4.04430360e-01 7.26657271e-01 2.44574815e-01 -9.85019863e-01
-1.06021714e+00 -6.99607790e-01 -9.57182944e-01 -2.03057185e-01
-4.20658946e-01 -1.32433563e-01 3.95330727e-01 7.47060180e-01
2.64603764e-01 9.03416574e-01 4.92906272e-01 -1.52392745e+00
1.47629827e-01 -5.61744034e-01 -2.81684190e-01 4.91173238e-01
8.88072729e-01 -3.35725576e-01 -6.43179297e-01 7.57078350e-01] | [10.681740760803223, -1.7394400835037231] |
17e48713-662a-4660-8b32-88cba93c52d7 | toward-risk-based-optimistic-exploration-for | 2303.01768 | null | https://arxiv.org/abs/2303.01768v1 | https://arxiv.org/pdf/2303.01768v1.pdf | Toward Risk-based Optimistic Exploration for Cooperative Multi-Agent Reinforcement Learning | The multi-agent setting is intricate and unpredictable since the behaviors of multiple agents influence one another. To address this environmental uncertainty, distributional reinforcement learning algorithms that incorporate uncertainty via distributional output have been integrated with multi-agent reinforcement learning (MARL) methods, achieving state-of-the-art performance. However, distributional MARL algorithms still rely on the traditional $\epsilon$-greedy, which does not take cooperative strategy into account. In this paper, we present a risk-based exploration that leads to collaboratively optimistic behavior by shifting the sampling region of distribution. Initially, we take expectations from the upper quantiles of state-action values for exploration, which are optimistic actions, and gradually shift the sampling region of quantiles to the full distribution for exploitation. By ensuring that each agent is exposed to the same level of risk, we can force them to take cooperatively optimistic actions. Our method shows remarkable performance in multi-agent settings requiring cooperative exploration based on quantile regression appropriately controlling the level of risk. | ['Se-Young Yun', 'Minchan Jeong', 'Joonkee Kim', 'Jihwan Oh'] | 2023-03-03 | null | null | null | null | ['distributional-reinforcement-learning'] | ['methodology'] | [-3.50258142e-01 3.87633473e-01 -2.67695099e-01 -1.67184114e-01
-8.85115921e-01 -4.55893070e-01 4.62579250e-01 5.87182701e-01
-7.12622166e-01 1.31511176e+00 1.54492050e-01 -1.94294125e-01
-5.84716678e-01 -1.19781148e+00 -5.93299925e-01 -1.00794029e+00
-3.66052359e-01 7.28376389e-01 8.93245339e-02 -4.00496840e-01
3.28077883e-01 1.65550455e-01 -1.25919962e+00 -1.76212788e-01
1.03677499e+00 8.05622935e-01 -2.76501682e-02 5.45452118e-01
-3.91317979e-02 8.09806406e-01 -8.76811206e-01 -7.49194100e-02
3.13155800e-01 -2.30650738e-01 -2.31898323e-01 -2.89704412e-01
-6.46050692e-01 -5.11325121e-01 3.65826249e-01 1.30018961e+00
4.69800204e-01 3.32251996e-01 6.82104707e-01 -1.47183275e+00
-4.75044906e-01 1.18833113e+00 -1.02765524e+00 -9.30845737e-02
1.35777339e-01 1.35760635e-01 8.26661885e-01 -1.58718184e-01
3.36957902e-01 1.48938823e+00 2.23275691e-01 3.04050326e-01
-1.29344237e+00 -6.77528679e-01 5.79061508e-01 -2.04825953e-01
-1.05889320e+00 1.84060186e-01 7.17347324e-01 -2.50765651e-01
5.10438919e-01 1.28862809e-03 6.64988637e-01 7.04648972e-01
5.43457806e-01 5.73932827e-01 1.31153035e+00 -3.86245131e-01
1.05416656e+00 1.65648222e-01 -5.00426114e-01 1.81104988e-01
4.47073191e-01 3.62118334e-01 -4.47554052e-01 -5.90218008e-01
5.34714758e-01 -5.14461473e-02 1.97165042e-01 -5.33670247e-01
-7.65619934e-01 1.17792344e+00 2.88477987e-01 -1.55960679e-01
-8.84788156e-01 4.91631627e-01 2.34858751e-01 5.36047578e-01
4.44639593e-01 6.63647711e-01 -2.21266881e-01 -2.42097542e-01
-5.76917768e-01 5.68525791e-01 7.13463366e-01 4.76598978e-01
7.23328352e-01 2.40346357e-01 -2.18547955e-01 4.24217343e-01
6.00681126e-01 5.64138114e-01 1.84099585e-01 -1.20180035e+00
3.74600887e-01 4.30187792e-01 7.53830850e-01 -6.60695255e-01
-2.95888484e-01 -3.99713248e-01 -4.77115303e-01 1.03583944e+00
3.55375409e-01 -8.83847296e-01 -5.94365478e-01 1.92856860e+00
6.23609066e-01 -3.07869852e-01 3.33905160e-01 6.68294668e-01
-3.84289265e-01 5.78585386e-01 2.81966746e-01 -5.51790655e-01
8.16428006e-01 -4.18691725e-01 -6.02782726e-01 1.31265238e-01
3.00309747e-01 -1.68053150e-01 8.56145918e-01 6.22427285e-01
-1.11712933e+00 -7.36480171e-04 -9.71220136e-01 9.46098566e-01
-2.81508923e-01 -4.63125587e-01 2.97087759e-01 6.72612011e-01
-7.84167588e-01 6.60014451e-01 -1.02667391e+00 2.05859020e-01
3.42623621e-01 2.16373548e-01 2.56726176e-01 4.84641016e-01
-1.28229165e+00 7.96714842e-01 3.67663473e-01 -5.46587892e-02
-1.26249862e+00 -6.92992151e-01 -5.32241821e-01 1.35352477e-01
9.84493732e-01 -3.51465911e-01 1.18451667e+00 -8.07802677e-01
-2.00367689e+00 -1.61079660e-01 6.70056224e-01 -6.31513894e-01
9.50671971e-01 -2.11797640e-01 1.69696704e-01 2.89930236e-02
9.88840088e-02 4.86841083e-01 8.17167401e-01 -1.38752258e+00
-6.36930764e-01 -3.14866126e-01 2.40885481e-01 5.26084661e-01
-3.84826601e-01 -1.47922665e-01 6.05570018e-01 -3.63923073e-01
-3.56541455e-01 -9.32715535e-01 -8.51456404e-01 -3.40684682e-01
-3.58205765e-01 -2.11409524e-01 2.28901953e-01 1.41015500e-01
1.06942737e+00 -1.98074365e+00 2.23041907e-01 4.59329695e-01
2.72510331e-02 -2.66280770e-01 1.11150630e-01 8.29969108e-01
5.24148166e-01 2.04210863e-01 -1.62184611e-01 -3.72019440e-01
4.60155427e-01 2.44628727e-01 -4.83029306e-01 4.39456642e-01
-4.13856097e-03 3.25741678e-01 -1.04597461e+00 -1.20907739e-01
1.26934303e-02 -9.40642506e-02 -6.45708859e-01 3.86883080e-01
-7.02898443e-01 2.64190555e-01 -9.48059440e-01 4.69016135e-01
5.97574472e-01 1.64880469e-01 3.20061386e-01 5.80144405e-01
-4.70541120e-01 -3.52821916e-01 -1.44100702e+00 1.04335439e+00
-4.30111557e-01 -2.08624035e-01 3.60360235e-01 -7.19385684e-01
1.05707514e+00 7.24453777e-02 8.06319952e-01 -2.43287653e-01
-5.98659460e-03 2.80394524e-01 -3.58460657e-02 -6.15310408e-02
5.06973386e-01 -3.36325377e-01 -4.08703923e-01 9.04555917e-01
-4.88951534e-01 -4.99708861e-01 1.17810719e-01 2.03166846e-02
8.45301092e-01 3.66510272e-01 3.84102672e-01 -5.39708316e-01
1.46877393e-01 -1.03920825e-01 7.39048481e-01 8.83278728e-01
-4.08868879e-01 -3.14534530e-02 1.12362003e+00 -1.90345794e-01
-8.10592353e-01 -1.23817730e+00 2.11960465e-01 1.31807661e+00
1.99881032e-01 3.09619345e-02 -5.23859203e-01 -5.27132988e-01
5.18216491e-01 1.08178258e+00 -1.00607812e+00 -1.73613280e-01
-3.05509061e-01 -9.71760869e-01 2.33252421e-01 3.69391650e-01
2.81181514e-01 -1.04629469e+00 -1.23671198e+00 5.54266453e-01
3.96275371e-01 -2.06546962e-01 -1.62211448e-01 4.89746749e-01
-3.78031462e-01 -8.25830936e-01 -7.61635780e-01 2.52316564e-01
5.17928064e-01 -4.09131467e-01 6.43282831e-01 -3.64190340e-01
2.18276918e-01 3.80867124e-01 -2.42647961e-01 -9.18534040e-01
-4.36132818e-01 2.84408126e-03 2.03504398e-01 -9.84237045e-02
-1.74221858e-01 -6.74006045e-01 -6.59372628e-01 1.93853214e-01
-9.53654706e-01 -4.96251166e-01 3.94559175e-01 6.89981401e-01
5.42052209e-01 1.50660187e-01 1.37870562e+00 -5.84376931e-01
1.06373131e+00 -7.02831745e-01 -1.05571926e+00 2.70350933e-01
-7.70700276e-01 3.68647158e-01 6.11890495e-01 -6.37907028e-01
-1.26063788e+00 -3.46353561e-01 2.65269905e-01 -2.94243813e-01
1.20763853e-01 5.50599754e-01 -3.40638054e-03 3.60959649e-01
5.44961929e-01 -6.80528358e-02 3.93500686e-01 -9.41658318e-02
4.62825894e-01 5.39054632e-01 3.15081663e-02 -1.01471889e+00
4.70761299e-01 2.99772650e-01 8.23538899e-02 -2.95189798e-01
-5.83103120e-01 2.14304134e-01 -6.19493239e-02 -5.19636333e-01
5.33466637e-01 -7.86724210e-01 -1.06071043e+00 3.79534423e-01
-4.90995675e-01 -6.40399754e-01 -6.20752811e-01 5.83068490e-01
-9.36515808e-01 2.57764403e-02 -2.68579543e-01 -1.45327806e+00
-1.91349089e-01 -1.14786184e+00 6.53517544e-01 5.27941823e-01
6.03931472e-02 -8.18586648e-01 5.14595687e-01 -1.42948344e-01
7.53479838e-01 6.18113637e-01 7.18596816e-01 -6.49610341e-01
-4.34549421e-01 1.25931635e-01 3.01726341e-01 -6.74536452e-02
-4.01549488e-02 4.25357185e-02 -4.18192565e-01 -4.56367612e-01
-1.54235601e-01 -7.54560292e-01 5.40295064e-01 5.22816360e-01
6.95141017e-01 -4.31732118e-01 -2.00586602e-01 7.83471242e-02
1.32046378e+00 4.23638731e-01 2.19947293e-01 8.41658235e-01
-1.80323362e-01 7.75315642e-01 9.07963216e-01 1.32887614e+00
5.59477866e-01 3.74557137e-01 1.05000651e+00 4.87357080e-01
7.82672763e-01 -3.03719997e-01 5.98202109e-01 -2.12865219e-01
8.05384293e-02 -2.68767238e-01 -6.55257165e-01 4.01230663e-01
-2.09456801e+00 -9.52167392e-01 5.72125912e-01 2.43103242e+00
1.13581705e+00 3.53256226e-01 4.51343477e-01 -3.60594124e-01
5.88041544e-01 1.78963333e-01 -9.94474232e-01 -5.95819592e-01
9.40199420e-02 -4.57282066e-01 5.62622249e-01 6.54007852e-01
-7.19802856e-01 6.17231965e-01 5.89219236e+00 6.52801216e-01
-8.22498500e-01 -2.04243481e-01 7.48335063e-01 -4.43131894e-01
-7.21970618e-01 -7.84085132e-03 -6.63078070e-01 5.69949329e-01
1.03480053e+00 -6.06113315e-01 4.57784712e-01 8.66570354e-01
4.89159584e-01 -6.46446586e-01 -7.01821625e-01 1.26268238e-01
-6.15857184e-01 -1.02300274e+00 -3.28069001e-01 2.81869560e-01
8.03589463e-01 -5.77071980e-02 9.31835696e-02 4.94388461e-01
1.20999551e+00 -8.78905833e-01 9.23493326e-01 6.24286592e-01
1.54025510e-01 -1.35237873e+00 6.82609320e-01 5.85549474e-01
-8.03986251e-01 -6.21061504e-01 -3.53387564e-01 -2.40343019e-01
3.99872869e-01 7.56326079e-01 -5.56361914e-01 3.96998495e-01
5.97732663e-01 -8.64477158e-02 5.10533899e-02 7.50663817e-01
-1.24733500e-01 3.26393425e-01 -6.09630048e-01 -4.70406473e-01
5.24706304e-01 -5.34280062e-01 4.84506100e-01 6.06599510e-01
3.95090938e-01 -3.72497998e-02 6.28041089e-01 9.52072740e-01
2.12636948e-01 4.78212759e-02 -5.08849084e-01 7.54569396e-02
7.57928073e-01 9.89704609e-01 -6.05648160e-01 -1.91469699e-01
9.43828151e-02 2.15494335e-01 4.67466891e-01 1.92277506e-01
-7.31685400e-01 -3.62576693e-01 6.30348563e-01 -3.32444191e-01
2.40390524e-01 -2.03687876e-01 -2.71551788e-01 -6.38197660e-01
-1.95222989e-01 -5.59404492e-01 5.73467791e-01 -3.33222479e-01
-1.33369434e+00 4.46309716e-01 3.58595252e-01 -9.32378650e-01
-7.89313018e-01 1.49591081e-02 -9.27395403e-01 6.32077634e-01
-1.66793776e+00 -7.16867685e-01 4.05368924e-01 2.63101101e-01
2.19004765e-01 -1.73036441e-01 5.43337643e-01 -5.33698320e-01
-4.71403182e-01 2.38064811e-01 5.25414646e-01 -4.37485337e-01
5.99606931e-01 -1.62964225e+00 -4.03491795e-01 3.13922852e-01
-5.83617210e-01 1.69461489e-01 1.03650808e+00 -8.83961260e-01
-1.25648427e+00 -9.14542556e-01 -3.69059034e-02 -9.56277270e-03
1.03917944e+00 -1.18517034e-01 -7.65980542e-01 3.17217350e-01
4.05495435e-01 -2.29963884e-01 6.04013920e-01 2.28744950e-02
4.22654347e-03 -1.76707029e-01 -1.38951790e+00 7.93118834e-01
3.19267720e-01 1.81512743e-01 -2.75909752e-01 7.87914246e-02
7.31423378e-01 -1.33673683e-01 -9.82672095e-01 2.86241323e-01
5.00581324e-01 -9.52345490e-01 5.46306729e-01 -4.60681289e-01
1.10899918e-01 -1.87838286e-01 -3.33744317e-01 -1.99491739e+00
7.55521134e-02 -8.66218567e-01 2.66891401e-02 1.06365502e+00
4.00254190e-01 -8.20698559e-01 7.36935198e-01 6.90602779e-01
1.07075050e-01 -9.35452938e-01 -1.25796807e+00 -7.11953342e-01
6.43164992e-01 1.32607762e-02 7.43383467e-01 4.17250633e-01
4.43839282e-01 -3.82141173e-01 -4.21394050e-01 3.18161428e-01
1.10877526e+00 3.61308932e-01 5.82045794e-01 -8.07258368e-01
-5.27924836e-01 -5.65944016e-01 2.56873697e-01 -3.04826468e-01
1.70303270e-01 -3.65030989e-02 2.49831855e-01 -1.16587961e+00
-1.62554737e-02 -6.04565799e-01 -3.85876447e-01 4.24077153e-01
-1.69239342e-01 -3.62396449e-01 2.94794917e-01 4.60246652e-02
-7.46189296e-01 1.10253882e+00 1.01426375e+00 -5.90262935e-02
-7.25858748e-01 2.45446786e-01 -7.39214778e-01 6.89872503e-01
9.77327824e-01 -4.00835872e-01 -6.54591322e-01 3.53308730e-02
4.87076223e-01 5.56874692e-01 -1.12586215e-01 -5.98623097e-01
1.80057928e-01 -1.10847723e+00 8.50984082e-02 -3.86874795e-01
1.26365587e-01 -5.89635193e-01 2.08445787e-01 7.38540769e-01
-7.25449502e-01 2.58348316e-01 2.37265006e-02 9.83862936e-01
3.71371321e-02 -2.08850965e-01 9.34373498e-01 -2.01152012e-01
-2.90196519e-02 7.30563104e-02 -8.10068250e-01 5.41069498e-03
1.48538566e+00 3.29736054e-01 -2.60726511e-01 -7.19258130e-01
-6.42641842e-01 9.11688805e-01 3.86320978e-01 3.52840163e-02
3.71306419e-01 -9.84730005e-01 -8.50593209e-01 -1.90292656e-01
-9.79073867e-02 1.62945911e-02 1.69176653e-01 6.02994442e-01
6.85909018e-02 -9.99470130e-02 -2.79482275e-01 -2.35448673e-01
-5.36336005e-01 5.34088969e-01 6.71047151e-01 -4.69682783e-01
-2.48429954e-01 3.32728624e-01 -1.69683099e-01 -3.78236353e-01
2.20286623e-01 -1.14597641e-01 -1.55337244e-01 3.54128569e-01
5.38471758e-01 5.67957819e-01 -3.33964050e-01 2.03459248e-01
-2.43181929e-01 1.06023975e-01 6.25765417e-04 -6.50314033e-01
1.58703041e+00 -3.44054610e-01 4.37295400e-02 3.97269636e-01
2.94256777e-01 2.10201770e-01 -1.84916449e+00 3.79594937e-02
9.54321325e-02 -3.66994023e-01 -2.05425620e-02 -9.78263497e-01
-8.17855537e-01 4.64548469e-01 1.35342091e-01 5.79684734e-01
8.08779240e-01 -1.60693526e-01 6.32253140e-02 3.95949274e-01
6.35046184e-01 -1.55945718e+00 2.82230049e-01 2.42032036e-01
8.53042483e-01 -1.00608087e+00 9.12904218e-02 3.40009063e-01
-1.06299913e+00 1.03104818e+00 7.63002992e-01 -3.07246208e-01
7.48152077e-01 5.45579791e-01 5.08111566e-02 9.63251069e-02
-1.15711510e+00 -2.14915052e-02 -5.16975284e-01 6.01441324e-01
-2.09597245e-01 3.55256677e-01 -2.14922652e-01 4.49606627e-01
7.84576535e-02 -1.78395584e-01 9.96834636e-01 1.15155244e+00
-9.05458868e-01 -1.06516862e+00 -7.00910270e-01 4.19488192e-01
-3.67676377e-01 2.67833918e-01 1.23769350e-01 7.25462258e-01
-2.13186651e-01 9.79445755e-01 4.01741378e-02 2.47305080e-01
1.78498909e-01 -1.70053810e-01 1.12885393e-01 -2.93154120e-01
-4.47311193e-01 2.78436482e-01 -4.10859175e-02 -3.79529864e-01
-7.78156519e-02 -9.35752571e-01 -1.47087824e+00 -2.01817885e-01
-1.82054982e-01 5.92244029e-01 6.37198865e-01 8.26120913e-01
3.11917394e-01 3.43696773e-01 1.07309997e+00 -7.97000527e-01
-1.61967003e+00 -8.44811976e-01 -9.08695579e-01 -2.71319866e-01
3.94282222e-01 -9.81159925e-01 -4.81753379e-01 -7.56815910e-01] | [4.187962055206299, 2.4530372619628906] |
a8eb3310-c1f3-41df-96cd-d627ea64d48c | iiitsurat-lt-edi-acl2022-hope-speech | null | null | https://aclanthology.org/2022.ltedi-1.13 | https://aclanthology.org/2022.ltedi-1.13.pdf | IIITSurat@LT-EDI-ACL2022: Hope Speech Detection using Machine Learning | This paper addresses the issue of Hope Speech detection using machine learning techniques. Designing a robust model that helps in predicting the target class with higher accuracy is a challenging task in machine learning, especially when the distribution of the class labels is highly imbalanced. This study uses and compares the experimental outcomes of the different oversampling techniques. Many models are implemented to classify the comments into Hope and Non-Hope speech, and it found that machine learning algorithms perform better than deep learning models. The English language dataset used in this research was developed by collecting YouTube comments and is part of the task “ACL-2022:Hope Speech Detection for Equality, Diversity, and Inclusion”. The proposed model achieved a weighted F1-score of 0.55 on the test dataset and secured the first rank among the participated teams. | ['Bharathi Raja Chakravarthi', 'Abhinav Kumar', 'Snehaan Bhawal', 'Pradeep Roy'] | null | null | null | null | ltedi-acl-2022-5 | ['hope-speech-detection'] | ['natural-language-processing'] | [-4.43180859e-01 5.71812829e-03 -8.10936511e-01 -3.79366785e-01
-1.31685877e+00 5.90144545e-02 3.51650625e-01 5.74383676e-01
-2.23057643e-01 6.37389243e-01 8.75698090e-01 -2.64244735e-01
3.05424854e-02 -6.04341507e-01 -1.77842200e-01 -4.26224023e-01
4.35023963e-01 6.84285760e-01 -2.16388166e-01 -3.37954223e-01
6.43463433e-01 1.10646725e-01 -1.45029712e+00 9.18349028e-01
7.31771111e-01 8.71127248e-01 -4.55415517e-01 8.67139220e-01
-2.03975514e-01 1.30226171e+00 -9.56944942e-01 -7.97804296e-01
6.17314093e-02 -4.50788289e-01 -8.78604352e-01 -2.17400283e-01
5.99207520e-01 -1.78661227e-01 -2.54681885e-01 9.11663771e-01
1.01491487e+00 -1.84312358e-01 6.28268182e-01 -1.55280805e+00
-3.53599668e-01 1.02967381e+00 -6.20850384e-01 4.14634794e-01
3.43877405e-01 -7.59470612e-02 1.26439881e+00 -9.10910964e-01
4.70159203e-01 1.13968444e+00 9.31770086e-01 4.51346874e-01
-8.22042167e-01 -1.10016775e+00 -6.80233538e-01 4.71220523e-01
-1.23350084e+00 -7.26481259e-01 7.40384638e-01 -6.05091453e-01
9.10657823e-01 3.84745270e-01 7.00609565e-01 1.25842226e+00
9.16672647e-02 1.00891471e+00 9.31038141e-01 -3.61297160e-01
1.92181900e-01 5.57142735e-01 4.66004729e-01 4.98317569e-01
-1.30239412e-01 -4.20632780e-01 -9.93113399e-01 -3.86206388e-01
-1.32105887e-01 -3.03101003e-01 5.51587949e-03 2.92217731e-01
-9.00752127e-01 1.42881405e+00 5.13504446e-03 3.09963286e-01
-1.61934674e-01 -5.63326061e-01 6.94321871e-01 5.15057266e-01
7.83511758e-01 2.99819767e-01 -2.16317952e-01 -7.49709189e-01
-1.23120809e+00 1.56277865e-01 8.96781027e-01 5.00530005e-01
-2.80169910e-03 -8.82474482e-02 -3.46759021e-01 1.12803602e+00
2.11216331e-01 8.08004215e-02 8.05738091e-01 -8.97830904e-01
6.47616982e-01 8.71923506e-01 -3.00836295e-01 -9.85669672e-01
-5.18085480e-01 -6.67331994e-01 -9.42265987e-01 2.04545707e-01
5.47622323e-01 -2.27011696e-01 -4.06310201e-01 1.46986282e+00
1.46592751e-01 -4.12158698e-01 2.39807248e-01 5.03608644e-01
1.23485959e+00 8.23076129e-01 -7.92951360e-02 -4.82234269e-01
1.08536541e+00 -1.05029237e+00 -8.61581504e-01 -1.39185144e-02
7.49681413e-01 -1.14750671e+00 1.14306903e+00 6.67146683e-01
-1.09643996e+00 -2.07381263e-01 -9.99104798e-01 -1.96946174e-01
-1.06922768e-01 2.12534145e-01 4.43356812e-01 1.11449671e+00
-9.44925308e-01 1.84002861e-01 -8.71230103e-03 -3.97199184e-01
6.22289062e-01 2.57406831e-01 -3.00719738e-01 1.55455694e-01
-1.03365242e+00 6.93130076e-01 -8.66834894e-02 -4.89395469e-01
-6.17508590e-01 -4.91212666e-01 -4.02289629e-01 3.68132517e-02
-3.23170900e-01 -1.53933078e-01 1.24616945e+00 -1.18295109e+00
-1.08132684e+00 1.27130711e+00 -9.98044759e-02 -7.24016130e-01
6.89620078e-01 7.87174404e-02 -2.88766742e-01 -1.93836659e-01
-9.47934692e-04 4.05292243e-01 6.86023116e-01 -5.85105002e-01
-5.52133977e-01 -4.78691429e-01 -9.46462601e-02 2.96344347e-02
-7.78321266e-01 3.42413664e-01 3.47291470e-01 -4.75595206e-01
-1.77504718e-01 -6.34690642e-01 5.13620019e-01 -4.07692194e-01
-5.16982913e-01 -6.31198227e-01 8.14133644e-01 -8.26642275e-01
1.45737326e+00 -2.17225409e+00 -3.37262988e-01 -2.19624847e-01
5.18896759e-01 2.41443604e-01 1.08926244e-01 5.72290599e-01
-2.27361441e-01 3.25959831e-01 5.25909603e-01 -4.46007669e-01
4.87135164e-02 -2.69940346e-01 -1.34171546e-01 5.59703350e-01
-2.67567158e-01 1.43786758e-01 -4.67234850e-01 -5.54812312e-01
-3.12021583e-01 3.27459484e-01 -7.83893228e-01 3.09056014e-01
2.61609644e-01 -9.55048054e-02 9.40071270e-02 5.74345291e-01
6.35981202e-01 -1.24785461e-01 4.42510936e-03 -1.00579292e-01
-8.98077190e-02 6.81953073e-01 -8.08802664e-01 7.83738852e-01
-3.42083901e-01 1.15822017e+00 -1.95711121e-01 -1.05217195e+00
1.28959119e+00 4.25226688e-01 4.32895511e-01 -7.70261765e-01
2.93528229e-01 2.45100498e-01 3.17833364e-01 -7.37883866e-01
5.96896708e-01 -1.55436650e-01 -5.80655597e-03 4.53170359e-01
-9.56135914e-02 7.65295997e-02 1.20160811e-01 3.51456672e-01
9.76208389e-01 -9.64092791e-01 3.71294856e-01 -7.31521621e-02
3.91617507e-01 -4.09897342e-02 8.24232399e-01 7.14829266e-01
-7.31135726e-01 6.76947117e-01 1.00480068e+00 -5.05803049e-01
-1.22034013e+00 -6.98808193e-01 -2.90668786e-01 1.17820787e+00
-4.46680933e-01 -5.06450295e-01 -7.24824786e-01 -6.53891146e-01
-3.14540148e-01 8.63101721e-01 -4.52543139e-01 -4.66159135e-01
1.18480772e-02 -9.07193065e-01 7.90995896e-01 1.98368847e-01
5.44904828e-01 -9.18915749e-01 -2.66916573e-01 -1.99415058e-01
-7.58896589e-01 -9.65719283e-01 -3.05608749e-01 -5.12579232e-02
-5.02264202e-01 -1.13781118e+00 -4.42615807e-01 -9.34941947e-01
1.00174487e-01 -2.80516781e-02 1.17215645e+00 1.10669903e-01
-1.55548379e-01 -1.86498046e-01 -5.17186344e-01 -5.42917192e-01
-7.38750041e-01 3.22715104e-01 3.20539740e-03 -3.10954582e-02
6.23075783e-01 -1.37092829e-01 -3.42322588e-01 1.51671782e-01
-4.62955743e-01 3.85330431e-02 2.78853327e-01 9.60051179e-01
-2.84071732e-02 2.72545338e-01 1.02431798e+00 -6.79639220e-01
1.13696003e+00 -6.05946124e-01 3.26263458e-02 -9.21148658e-02
-5.17126679e-01 -4.68977004e-01 3.94539863e-01 -2.92645752e-01
-7.24062383e-01 -2.28128001e-01 -7.40322530e-01 1.54610679e-01
-3.10852714e-02 3.05062681e-01 -7.28197694e-02 2.63383836e-01
5.66107571e-01 4.21723016e-02 3.32164466e-02 -4.28323299e-01
-2.36717686e-01 1.58268821e+00 3.86105478e-02 2.83150114e-02
4.75130929e-03 6.90098479e-02 -5.93121231e-01 -9.89358425e-01
-1.12228787e+00 -6.03484988e-01 -1.81706443e-01 -2.83844084e-01
6.96162224e-01 -9.77079511e-01 -6.22770727e-01 8.99326324e-01
-1.14898515e+00 -1.36745512e-01 -2.23012403e-01 4.59450752e-01
-3.66369009e-01 4.46957303e-04 -6.21130705e-01 -1.12912548e+00
-6.88425541e-01 -1.23606694e+00 8.11570048e-01 2.39753306e-01
-7.16528594e-01 -7.73928225e-01 2.36458242e-01 1.30565286e+00
3.32698792e-01 -4.08359058e-02 1.17736685e+00 -1.40282214e+00
2.31079757e-01 -4.77964818e-01 -9.77035835e-02 3.10709149e-01
-2.96448320e-01 2.89408296e-01 -1.13038647e+00 -5.69505990e-02
2.11750805e-01 -8.90055180e-01 7.48629391e-01 4.33852106e-01
1.04782045e+00 -6.40266716e-01 1.98411480e-01 -6.08165376e-02
9.32580948e-01 -1.45916343e-01 6.98156714e-01 5.05612612e-01
1.67324081e-01 6.87558174e-01 5.04928589e-01 8.42794478e-01
7.04285860e-01 5.82757711e-01 5.12385011e-01 1.34711936e-01
-1.71899453e-01 -2.91807026e-01 5.31349301e-01 1.20278108e+00
5.09182811e-01 -3.91087472e-01 -1.20908451e+00 5.79889655e-01
-1.57325900e+00 -1.26887798e+00 -5.96031129e-01 2.19198084e+00
9.20652449e-01 2.57267267e-01 7.90764391e-01 8.01481843e-01
7.07574368e-01 2.04763547e-01 -8.09694752e-02 -9.88299131e-01
-2.11838380e-01 -1.23348095e-01 -4.59192544e-02 5.16561568e-01
-1.16256356e+00 4.11899418e-01 6.72756243e+00 1.08291042e+00
-1.12134612e+00 2.67926514e-01 1.32801211e+00 -2.79524058e-01
-4.52568270e-02 -4.49370503e-01 -8.99899840e-01 5.33581495e-01
1.25563478e+00 -2.63397276e-01 -1.46948233e-01 7.20887005e-01
4.76042658e-01 -7.42649361e-02 -6.63827598e-01 1.10958540e+00
4.28789139e-01 -1.17364657e+00 -8.24423209e-02 1.62246625e-03
7.79846311e-01 1.48714021e-01 3.46545875e-01 6.09246254e-01
1.32652685e-01 -1.20195699e+00 7.06289887e-01 2.98886031e-01
5.19967377e-01 -1.00021815e+00 1.14929831e+00 8.66095424e-01
-2.72184908e-01 -6.04637742e-01 -2.55710781e-01 -3.49885672e-01
-2.51194417e-01 9.90724504e-01 -1.12966037e+00 -2.61028141e-01
7.10978150e-01 5.84645391e-01 -5.15633464e-01 1.34086490e+00
1.92571491e-01 9.24746752e-01 -2.70452537e-03 -4.29313153e-01
4.32614191e-03 2.69429833e-01 5.42551041e-01 1.19824278e+00
3.01293015e-01 -4.88790065e-01 1.53495476e-01 3.52163970e-01
-4.91216362e-01 6.63570404e-01 -4.71836209e-01 -2.28756323e-01
3.17545444e-01 1.15842760e+00 -3.31834763e-01 -3.11366767e-01
-1.41344130e-01 2.53457606e-01 2.24040210e-01 -3.29715133e-01
-7.49144256e-01 -2.82694817e-01 4.09358591e-01 4.50656205e-01
-2.63236016e-01 1.43963382e-01 -6.47736192e-01 -9.59631801e-01
-2.33862609e-01 -1.44864655e+00 5.50031602e-01 -4.88042504e-01
-1.40882623e+00 4.74551231e-01 -5.05169749e-01 -9.94503081e-01
-8.72154012e-02 -1.83704674e-01 -7.61550188e-01 5.14434516e-01
-1.16715813e+00 -7.82479942e-01 -4.44288850e-01 4.50124264e-01
1.08164787e+00 -7.21731901e-01 6.47288680e-01 6.62843108e-01
-6.43805265e-01 9.46092665e-01 1.23943374e-01 4.03521240e-01
7.91119814e-01 -1.10554659e+00 -1.40200138e-01 3.80736262e-01
1.10192887e-01 2.67957933e-02 8.15778494e-01 -3.48499238e-01
-8.72045994e-01 -7.00969100e-01 1.70062256e+00 -1.88250542e-01
6.07059419e-01 -2.22624287e-01 -5.41364372e-01 1.52540013e-01
5.52755415e-01 -4.01754737e-01 1.16833866e+00 1.81296408e-01
-3.94868076e-01 -1.25261173e-01 -1.51247537e+00 2.29999289e-01
4.53580886e-01 -5.94281733e-01 -3.35470229e-01 7.85031855e-01
3.84697765e-01 -2.27439478e-01 -7.24052966e-01 2.03062654e-01
6.50131106e-01 -1.46119189e+00 6.29030347e-01 -9.50507164e-01
8.99198949e-01 5.28865218e-01 -3.43533069e-01 -1.19818699e+00
2.58553401e-02 -3.57662678e-01 -5.07419780e-02 1.61441231e+00
6.97109997e-01 -2.83609360e-01 1.02170193e+00 3.12084496e-01
3.28654289e-01 -9.14445400e-01 -1.10507226e+00 -3.20485443e-01
3.71763319e-01 -3.81535053e-01 1.80375159e-01 1.17085683e+00
2.99333870e-01 7.75889695e-01 -4.56388682e-01 -4.50476527e-01
3.96320313e-01 -1.36914417e-01 7.80994594e-01 -1.29699779e+00
-1.50693078e-02 -6.77061379e-01 -6.41710758e-01 -4.10510629e-01
2.94324636e-01 -1.09312153e+00 -3.75928342e-01 -1.32327247e+00
7.16408968e-01 -2.89675325e-01 1.05085552e-01 5.44141531e-01
1.21858396e-01 3.58758003e-01 1.27244025e-01 1.95697978e-01
-5.77719688e-01 5.41535854e-01 6.89231813e-01 -5.99169016e-01
6.84043989e-02 5.67586541e-01 -9.85215724e-01 6.86446667e-01
1.17450893e+00 -6.12915754e-01 -2.16929495e-01 -9.24654901e-02
5.01281261e-01 1.19248420e-01 4.87674661e-02 -1.06752086e+00
1.71559036e-01 9.79650766e-02 1.20811462e-01 -7.47429967e-01
3.26062620e-01 -4.10964459e-01 -2.64998794e-01 4.46008474e-01
-9.02458072e-01 2.77566195e-01 -1.60497390e-02 4.13005911e-02
-3.79718482e-01 -4.80323851e-01 9.18753982e-01 1.41756028e-01
-5.15746176e-02 -7.63839558e-02 -4.98170674e-01 5.79870701e-01
8.68314564e-01 1.26939014e-01 -7.75363088e-01 -1.03328788e+00
-6.31284297e-01 -1.39970586e-01 1.49419263e-01 5.74182153e-01
6.64968550e-01 -1.28495204e+00 -1.13992035e+00 7.57102296e-02
3.98641871e-03 -7.07008600e-01 4.79977131e-02 1.14028108e+00
-5.11597395e-01 8.39206725e-02 -8.08304250e-02 -5.18162429e-01
-1.90515244e+00 -3.37041505e-02 3.94981652e-01 -4.50600058e-01
-2.76057988e-01 9.37994063e-01 -5.48181891e-01 -6.79394245e-01
5.85131884e-01 3.50576609e-01 -6.14261389e-01 4.97905523e-01
7.75956333e-01 6.91469491e-01 3.08747590e-01 -9.93027210e-01
-1.74328953e-01 1.35740088e-02 -1.26047611e-01 5.80916964e-02
1.40797925e+00 1.41982958e-01 -1.82090715e-01 6.23822987e-01
1.56092143e+00 2.88615912e-01 -3.28224778e-01 1.71611324e-01
-3.77442464e-02 -5.06584883e-01 2.77111053e-01 -8.66479993e-01
-1.09160662e+00 1.24250054e+00 8.96226108e-01 6.20035291e-01
8.12516868e-01 -2.16510803e-01 9.38576519e-01 6.85247332e-02
-3.69266629e-01 -1.29470539e+00 4.57491487e-01 6.42405689e-01
7.42074788e-01 -1.52407968e+00 -1.70277491e-01 -8.40667933e-02
-7.77979493e-01 1.21230876e+00 5.29999971e-01 2.17697114e-01
5.35053253e-01 4.52033192e-01 2.79888391e-01 -1.13284983e-01
-9.95282888e-01 2.58275330e-01 1.87205505e-02 5.97085416e-01
8.93000960e-01 3.26079540e-02 -6.16754353e-01 6.02338612e-01
-6.34172380e-01 4.42395955e-02 8.34059834e-01 2.92164654e-01
-7.06752181e-01 -8.57869327e-01 -3.10817748e-01 7.91696906e-01
-1.03304505e+00 -1.42227456e-01 -8.02721441e-01 4.49287504e-01
3.46369669e-02 1.33087575e+00 1.69249594e-01 -7.91529953e-01
1.71270803e-01 1.50397345e-01 -2.29833916e-01 -2.05237120e-01
-9.55649793e-01 -2.18598962e-01 5.08456469e-01 -1.31789476e-01
-3.26119274e-01 -6.86165810e-01 -9.49095666e-01 -7.74123788e-01
-3.14309061e-01 3.54351163e-01 7.42497981e-01 8.07724297e-01
2.98244685e-01 3.83420497e-01 8.48204434e-01 -3.09940398e-01
-8.49688768e-01 -9.88482773e-01 -5.07655144e-01 4.70018357e-01
4.30203289e-01 -3.06338876e-01 -6.58799350e-01 -3.82475078e-01] | [9.033310890197754, 10.684954643249512] |
116a7999-a32c-454a-9fa2-0f784c95975d | launcher-attitude-control-based-on | 2307.00372 | null | https://arxiv.org/abs/2307.00372v1 | https://arxiv.org/pdf/2307.00372v1.pdf | Launcher Attitude Control based on Incremental Nonlinear Dynamic Inversion: A Feasibility Study Towards Fast and Robust Design Approaches | The so-called ``New Space era'' has seen a disruptive change in the business models and manufacturing technologies of launch vehicle companies. However, limited consideration has been given to the benefits that innovation in control theory can bring; not only in terms of increasing the limits of performance but also reducing mission preparation or ``missionisation'' efforts. Moreover, there is a gap between the current state-of-practice that still relies on linear controls and other modern control techniques that could bring relevant improvements in launcher attitude control. Nonlinear Dynamic Inversion (NDI) is a technique that basically `cancels' the nonlinearities of a class of nonlinear systems, allowing for a single linear control law to be applied without the need for gain-scheduling across different operational points. Incremental NDI (INDI) is a variation of NDI that generates incremental commands and employs acceleration feedback to reduce model dependency, making it easier to design, and results in being more robust in closed-loop. While INDI has been applied successfully to several aerospace applications, its applicability to launch vehicles has not yet been adequately investigated. The objective of this paper is therefore to introduce and raise awareness of the INDI method among the launcher guidance, navigation, and control (GNC) community, showcasing its implementation on a representative launch ascent application scenario which highlights INDI's strengths and challenges. We present a new, practical approach for stability analysis of INDI for attitude control, and compare INDI with scheduled PD controllers with- and without angular acceleration estimates. Results show that, while INDI controllers are generally more sensitive to sensor noise and actuator delay than linear controllers, their potential benefits outweigh these limitations in terms of robustness and performance. | ['Samir Bennani', 'Paul Acquatella', 'Pedro Simplício'] | 2023-07-01 | null | null | null | null | ['robust-design'] | ['miscellaneous'] | [ 8.99742693e-02 1.00128897e-01 -1.96571976e-01 2.45861128e-01
3.10400967e-03 -9.44563329e-01 8.30168903e-01 3.48988809e-02
-2.31578767e-01 6.88158572e-01 -6.82570562e-02 -8.47656846e-01
-6.80020213e-01 -4.51828361e-01 -6.47486448e-01 -6.85005903e-01
-2.30893478e-01 2.72973865e-01 1.40361354e-01 -8.67034853e-01
7.59298578e-02 8.10169101e-01 -1.50232160e+00 -5.14755666e-01
5.74640810e-01 8.03738773e-01 -2.32925769e-02 4.40341920e-01
5.47432125e-01 4.69116300e-01 -5.31394839e-01 3.65538478e-01
4.73609328e-01 -4.09831226e-01 -1.95075542e-01 8.63949433e-02
-1.25385836e-01 -1.65502116e-01 -3.96783985e-02 6.68449044e-01
5.60030460e-01 2.47560576e-01 2.96152353e-01 -1.19377208e+00
1.92946002e-01 -2.39668816e-01 -1.52838066e-01 1.20920509e-01
2.44875491e-01 3.69726211e-01 4.63530838e-01 -3.97914797e-01
5.99342406e-01 9.30360138e-01 7.69310534e-01 2.52902508e-01
-1.25710094e+00 -4.56342787e-01 1.17632113e-02 -4.32940662e-01
-1.14504671e+00 -4.84005511e-01 3.61342847e-01 -7.60906994e-01
1.10452008e+00 7.60914087e-01 9.39920068e-01 3.64718467e-01
6.59909070e-01 1.92513466e-01 9.97523427e-01 -3.34162563e-01
1.85166657e-01 1.92117855e-01 -2.85101622e-01 2.79415727e-01
6.45863891e-01 6.50042355e-01 1.20429255e-01 5.97823076e-02
8.97368073e-01 -2.68377542e-01 -5.95819712e-01 -6.77174449e-01
-1.14459932e+00 6.41907334e-01 2.80289084e-01 4.59396988e-01
-4.44750726e-01 9.73184854e-02 4.05644268e-01 6.10553682e-01
1.75688252e-01 8.72852802e-01 -2.60090470e-01 -3.66500020e-01
-5.08879185e-01 7.09999204e-01 9.32097256e-01 7.54525304e-01
3.59688282e-01 7.75909185e-01 4.06058431e-01 3.98442656e-01
2.92038500e-01 6.52371585e-01 2.65876263e-01 -7.90233016e-01
1.49708867e-01 5.66927075e-01 4.54276830e-01 -1.03708565e+00
-7.14301109e-01 -9.03917611e-01 -3.91344786e-01 6.79624796e-01
1.33265436e-01 -6.36877239e-01 -6.05988145e-01 1.48023224e+00
3.30714941e-01 -4.23437804e-01 2.15637252e-01 1.15686035e+00
-1.40287772e-01 7.32759356e-01 -6.28171027e-01 -4.24177974e-01
9.11689401e-01 -2.78947711e-01 -7.68355548e-01 -6.32012606e-01
5.10245740e-01 -8.83963287e-01 6.39326155e-01 5.31721413e-01
-9.50335026e-01 -2.88778514e-01 -1.48121214e+00 5.95924377e-01
-1.67893350e-01 1.04031719e-01 4.04746532e-01 8.34803998e-01
-1.03876328e+00 5.05969405e-01 -7.57521749e-01 -4.60125983e-01
-4.67801094e-01 6.16652966e-01 -2.35762119e-01 3.25409800e-01
-1.18319583e+00 1.51213741e+00 3.25610459e-01 3.96330088e-01
-3.97074163e-01 -7.43843555e-01 -5.96943378e-01 -3.60740632e-01
7.57145643e-01 -5.65646529e-01 1.28966582e+00 -7.55479634e-01
-1.82290804e+00 4.42416631e-02 2.05443040e-01 -5.92106164e-01
2.72545189e-01 -2.51906067e-01 -5.21383107e-01 -1.70323223e-01
-2.66337365e-01 2.20693722e-01 8.19942892e-01 -1.15011644e+00
-7.87217915e-01 6.28757551e-02 2.17618763e-01 2.92155385e-01
1.97211310e-01 -1.02027118e-01 4.44538072e-02 -4.03768599e-01
2.68661797e-01 -1.52447140e+00 -3.40804517e-01 -5.08855879e-01
2.72866130e-01 2.05162466e-01 8.52764547e-01 -5.15981138e-01
1.32496023e+00 -1.96631193e+00 4.22423244e-01 3.90578926e-01
-3.59032184e-01 4.89003092e-01 4.93624598e-01 1.06352484e+00
-1.96951926e-01 -1.88184708e-01 1.38666123e-01 5.12122393e-01
1.42605882e-02 3.98065329e-01 -5.56272745e-01 6.04969084e-01
8.52181017e-02 4.43433404e-01 -7.35393047e-01 5.73357761e-01
4.80346292e-01 1.90935940e-01 -5.07851005e-01 -3.12151998e-01
5.43596596e-03 6.28966630e-01 -1.79069698e-01 4.53928411e-01
2.29178876e-01 5.31514406e-01 3.95691723e-01 -1.38598174e-01
-9.16210830e-01 2.70698488e-01 -1.35218775e+00 8.75278294e-01
-4.03272092e-01 7.11288393e-01 8.08347940e-01 -9.94381547e-01
1.02302611e+00 2.68401414e-01 4.49497670e-01 -8.32141578e-01
4.47959155e-01 6.98285758e-01 4.36496735e-01 -3.01451206e-01
5.79830766e-01 -6.90313518e-01 -1.40601709e-01 -6.28349483e-02
-2.33065486e-01 -8.17079425e-01 6.91031218e-02 -1.18174940e-01
6.08540595e-01 1.73435956e-01 2.44491041e-01 -6.69392467e-01
6.18970871e-01 4.10702199e-01 7.30709970e-01 9.63100269e-02
-2.91777775e-03 -1.89817533e-01 4.36734080e-01 1.42365089e-02
-9.78687882e-01 -4.16169196e-01 -6.13646097e-02 3.85209680e-01
1.80253103e-01 -1.38078585e-01 -3.29113841e-01 1.86522275e-01
4.34924036e-01 9.26391244e-01 -2.35302877e-02 -4.22310531e-01
-7.82351851e-01 -4.24685985e-01 5.22528291e-02 2.23032475e-01
-5.29484041e-02 1.56713292e-01 -9.83296096e-01 5.55868983e-01
1.90087721e-01 -6.83413684e-01 -5.83284795e-02 -1.93921532e-02
-1.06704068e+00 -1.10987401e+00 -3.99027377e-01 -3.46130371e-01
5.26883423e-01 4.75431114e-01 4.75932717e-01 -5.03892489e-02
-2.28653848e-03 6.83465362e-01 -2.69625515e-01 -7.66732872e-01
-4.63489950e-01 -3.15955549e-01 7.42667496e-01 -1.92572549e-01
-3.37173939e-01 -1.42128304e-01 -4.04442608e-01 7.62437761e-01
-7.96788871e-01 1.82875060e-02 6.09017968e-01 6.28273845e-01
1.09763123e-01 2.10552201e-01 7.33300805e-01 -3.02663475e-01
7.21561193e-01 -2.02555627e-01 -9.99761581e-01 -3.73975962e-01
-8.69877517e-01 -3.11125636e-01 4.73944575e-01 -1.49750128e-01
-9.13999438e-01 -1.87570870e-01 -1.12478718e-01 -1.57913253e-01
3.00441206e-01 8.62252712e-01 2.76729733e-01 -6.10531628e-01
5.07644773e-01 -1.85395449e-01 1.11027527e+00 -2.41919920e-01
2.21566409e-02 5.36387503e-01 4.81514663e-01 -4.37544227e-01
1.13902700e+00 4.02093083e-01 5.20006537e-01 -1.06807625e+00
-8.23375285e-02 -4.32702899e-01 -1.46758854e-01 -5.08834600e-01
3.25080484e-01 -9.40805316e-01 -6.40386403e-01 1.24434777e-01
-4.48797733e-01 -3.95346373e-01 -3.30071688e-01 8.89940381e-01
-5.47750115e-01 1.18011922e-01 -2.19556987e-01 -1.09510493e+00
3.61616053e-02 -1.26333952e+00 3.25591594e-01 2.11850047e-01
-3.67696971e-01 -9.75007713e-01 -1.09429292e-01 -4.35140543e-03
9.46178019e-01 7.30823398e-01 6.36695683e-01 -6.57156035e-02
-5.47648370e-01 -7.68784881e-01 5.21111250e-01 5.60435236e-01
1.92409679e-02 -1.03389792e-01 -3.01428139e-01 -1.12358475e+00
4.06002760e-01 1.46625802e-01 -5.20701408e-02 2.23964050e-01
-2.23443970e-01 -4.24929261e-01 -3.76047373e-01 1.45015076e-01
1.55637157e+00 7.77036190e-01 4.56892908e-01 8.78954113e-01
1.21136002e-01 8.58156502e-01 1.35130131e+00 9.91966650e-02
-3.80354971e-02 8.92221451e-01 6.10055089e-01 -8.22857469e-02
2.86195487e-01 9.33210105e-02 5.91167212e-01 6.94731057e-01
-4.69423085e-01 2.08125830e-01 -8.14690888e-01 4.69980091e-01
-1.64084613e+00 -6.52487457e-01 -3.38518262e-01 2.61272097e+00
2.64401197e-01 2.86751002e-01 9.04702693e-02 1.54255256e-01
4.85623538e-01 -2.15641394e-01 -1.59727663e-01 -7.58521616e-01
1.69087648e-01 -9.63263884e-02 1.10221016e+00 5.30426919e-01
-8.01493585e-01 2.39456236e-01 5.77374077e+00 3.84443671e-01
-1.49365199e+00 -2.31280133e-01 -1.76516607e-01 -1.70653090e-01
-1.40410393e-01 6.38772428e-01 -7.98180223e-01 7.89242312e-02
1.22734141e+00 -5.71265459e-01 3.53025347e-01 7.27402329e-01
6.65885806e-01 -3.75455856e-01 -7.57655621e-01 3.19379568e-01
5.58180250e-02 -1.21707475e+00 -4.40982640e-01 3.40379894e-01
5.93659341e-01 -2.51965284e-01 -1.08574271e-01 5.60765147e-01
-4.85093057e-01 -2.79441684e-01 1.05240595e+00 4.99763429e-01
4.77573842e-01 -1.04076827e+00 8.46538067e-01 6.39446318e-01
-8.93427730e-01 -5.44158399e-01 -1.29309297e-01 -6.70265734e-01
5.74721634e-01 2.12721795e-01 -8.68412018e-01 9.23097312e-01
3.03181916e-01 9.17891636e-02 -9.31018442e-02 1.06343853e+00
6.99085696e-03 4.24351901e-01 -4.55972433e-01 -2.73436844e-01
5.03548205e-01 -5.99686205e-01 1.03166580e+00 6.71131074e-01
4.31327254e-01 7.04120845e-02 2.73597687e-02 3.25329870e-01
1.15301299e+00 -1.87153280e-01 -7.26986408e-01 -1.96266040e-01
8.15910697e-02 1.00878680e+00 -2.84185171e-01 -8.94105062e-02
-3.35659653e-01 1.65026203e-01 -7.35131800e-01 2.14323908e-01
-8.27509940e-01 -6.40407860e-01 8.31800103e-01 6.95953369e-01
4.39922772e-02 -8.55061173e-01 -4.92671058e-02 -4.63357210e-01
-8.06615353e-02 -1.09278858e+00 -2.41786256e-01 -4.71820116e-01
-3.66996855e-01 2.07140595e-01 4.00985420e-01 -1.92365694e+00
-8.67813528e-01 -6.90805733e-01 -2.49518752e-01 1.02309656e+00
-1.17181349e+00 -6.79642498e-01 1.19681537e-01 -1.58225700e-01
2.90416211e-01 -6.85851127e-02 4.93895322e-01 3.59202653e-01
-4.27869737e-01 -1.36528656e-01 5.19504309e-01 -7.71868646e-01
6.58249736e-01 -7.20279157e-01 2.15260282e-01 9.89027798e-01
-9.24763680e-01 1.01879323e+00 1.27151263e+00 -7.84396052e-01
-2.24419975e+00 -8.29490721e-01 6.74590826e-01 -9.44785178e-02
1.00144279e+00 -1.21254466e-01 -6.05933547e-01 8.03440392e-01
1.61718696e-01 -7.16496885e-01 1.01226486e-01 -1.95050880e-01
7.75621712e-01 -2.98927128e-01 -7.72529244e-01 8.14417422e-01
3.77316803e-01 -1.75982550e-01 -5.97864687e-01 1.83034629e-01
5.37133157e-01 -8.56071293e-01 -1.00210869e+00 7.95165956e-01
4.88533080e-01 -7.73800075e-01 1.00805497e+00 -9.49169174e-02
-4.58006203e-01 -9.17523623e-01 2.50911981e-01 -1.47500205e+00
-3.45042408e-01 -1.07474613e+00 1.21396244e-01 9.81846690e-01
3.16178083e-01 -1.23510706e+00 2.64152199e-01 6.06028140e-01
-7.45068789e-01 -7.70292521e-01 -1.08318102e+00 -1.25909209e+00
-1.25819877e-01 -2.93915182e-01 1.67680681e-01 7.39980996e-01
2.64574856e-01 3.20107698e-01 -3.13001394e-01 4.84030098e-01
1.15604579e-01 -1.47069633e-01 1.17847812e+00 -1.11141992e+00
-4.06833082e-01 -2.17690513e-01 -5.12181580e-01 -7.66154230e-01
-5.18291116e-01 -3.45812440e-01 -3.60756996e-03 -1.57005799e+00
-8.80811512e-01 -3.55889142e-01 2.62124300e-01 1.34536028e-01
3.44496965e-01 2.39661187e-02 2.24052444e-01 4.27300990e-01
3.06007564e-01 2.97882110e-01 9.94668245e-01 1.32268593e-01
-4.36638713e-01 4.27103281e-01 -3.29315782e-01 5.55913270e-01
7.61025310e-01 1.77300405e-02 -6.05739295e-01 -2.23946616e-01
4.09907937e-01 2.61707217e-01 4.53116179e-01 -1.58326054e+00
2.11394504e-01 -1.54198736e-01 -1.55633658e-01 -5.45275271e-01
8.92142057e-02 -9.81271267e-01 8.61407340e-01 9.85400498e-01
5.78124464e-01 2.04573408e-01 7.52250850e-01 4.90843832e-01
-3.87388706e-01 -2.22546414e-01 7.63774037e-01 2.76627153e-01
-7.02649057e-01 -4.10097182e-01 -8.60445142e-01 -6.53240502e-01
1.35873258e+00 -5.50369799e-01 -2.73600131e-01 -3.99480164e-01
-5.88636994e-01 4.37015235e-01 6.82414174e-01 5.73500693e-01
7.37673640e-02 -1.00771427e+00 -1.67815179e-01 1.93431616e-01
-2.26028129e-01 -2.94564098e-01 3.43116879e-01 1.39418519e+00
-7.41980731e-01 1.07132661e+00 -1.91088438e-01 -6.24085724e-01
-1.02178490e+00 4.40029502e-01 4.34212565e-01 2.88074106e-01
-5.92383623e-01 3.53345424e-01 -1.47441253e-01 -3.34776849e-01
-2.92207599e-01 -4.58825290e-01 -1.22088417e-02 8.35794806e-02
1.79574668e-01 4.87800181e-01 3.65485042e-01 -5.92687309e-01
-2.17524454e-01 5.23573339e-01 2.25149795e-01 -2.64359325e-01
9.96702671e-01 -2.46141255e-01 -5.58441654e-02 5.66607356e-01
5.46646357e-01 7.95338377e-02 -1.29895604e+00 3.50939244e-01
-1.01872198e-02 -4.79850292e-01 1.60195068e-01 -5.96179068e-01
-5.66533923e-01 2.66216159e-01 5.78168333e-01 4.25733566e-01
1.06806600e+00 -6.96114123e-01 4.39659625e-01 2.29200229e-01
6.25250340e-01 -1.51548266e+00 -2.81133682e-01 7.67558217e-01
1.32032955e+00 -4.32025254e-01 2.51793772e-01 -4.73673016e-01
-5.56614935e-01 1.07262337e+00 3.51780891e-01 -2.21499607e-01
2.48396710e-01 3.41082811e-01 8.32238048e-02 -2.73248814e-02
-4.82146591e-01 -2.39258483e-01 1.49244323e-01 3.77748013e-01
4.31844741e-01 -1.04030356e-01 -1.00270593e+00 1.69703722e-01
9.49839279e-02 7.51049146e-02 6.86444461e-01 1.53013468e+00
-8.03210974e-01 -1.25003219e+00 -8.10153782e-01 2.24339128e-01
-1.15486369e-01 4.75869328e-01 -3.18514444e-02 1.52644849e+00
3.44925411e-02 8.15518200e-01 1.85001474e-02 -4.07161623e-01
1.17821574e+00 -7.15243816e-02 2.14293897e-01 -3.98650914e-01
-6.20335102e-01 4.79811877e-01 6.81868851e-01 -5.34936249e-01
-1.69003040e-01 -7.83633769e-01 -1.19984508e+00 -2.47305289e-01
-5.92581451e-01 5.08897722e-01 1.12998676e+00 6.79941177e-01
5.33018470e-01 9.08683598e-01 5.88237286e-01 -1.06899273e+00
-9.13515389e-01 -7.22834885e-01 -5.00334740e-01 -2.34904915e-01
2.59501845e-01 -1.15867925e+00 -4.73509997e-01 -4.99518126e-01] | [5.380246639251709, 2.283017635345459] |
e22be01f-5af1-40dd-b658-607b6db5a027 | cross-sectional-stock-price-prediction-using | 2002.06975 | null | https://arxiv.org/abs/2002.06975v1 | https://arxiv.org/pdf/2002.06975v1.pdf | Cross-sectional Stock Price Prediction using Deep Learning for Actual Investment Management | Stock price prediction has been an important research theme both academically and practically. Various methods to predict stock prices have been studied until now. The feature that explains the stock price by a cross-section analysis is called a "factor" in the field of finance. Many empirical studies in finance have identified which stocks having features in the cross-section relatively increase and which decrease in terms of price. Recently, stock price prediction methods using machine learning, especially deep learning, have been proposed since the relationship between these factors and stock prices is complex and non-linear. However, there are no practical examples for actual investment management. In this paper, therefore, we present a cross-sectional daily stock price prediction framework using deep learning for actual investment management. For example, we build a portfolio with information available at the time of market closing and invest at the time of market opening the next day. We perform empirical analysis in the Japanese stock market and confirm the profitability of our framework. | ['Kei Nakagawa', 'Masaya Abe'] | 2020-02-17 | null | null | null | null | ['stock-price-prediction'] | ['time-series'] | [-1.01674783e+00 -4.47635829e-01 -3.81547600e-01 -3.06048185e-01
-9.93956551e-02 -4.49011087e-01 3.97522330e-01 -1.82875484e-01
-2.46325925e-01 8.48403990e-01 2.15821251e-01 -5.30466378e-01
-1.04695901e-01 -1.39691627e+00 -5.97343862e-01 -5.77969193e-01
-1.20727159e-01 1.95326120e-01 -2.09143367e-02 -4.96334612e-01
8.05192530e-01 3.58713478e-01 -1.17576945e+00 -3.08820438e-02
3.85205239e-01 1.52585924e+00 1.12414714e-02 -5.80495000e-02
-6.09187841e-01 1.09851623e+00 -5.78248024e-01 -7.23516941e-01
9.57975745e-01 -2.62580127e-01 -1.91694871e-01 -4.23404612e-02
-3.17241460e-01 -8.12078536e-01 -3.28611165e-01 1.01625991e+00
1.36284977e-01 -3.74119133e-01 3.95685494e-01 -1.02844071e+00
-8.35683107e-01 1.16637993e+00 -7.93003201e-01 6.81598783e-01
-4.66620117e-01 -5.66033497e-02 1.53764629e+00 -8.89832497e-01
-8.70530382e-02 6.70862734e-01 5.76343119e-01 6.85454346e-03
-5.37885666e-01 -9.44639862e-01 5.47601655e-02 8.57894793e-02
-6.51528358e-01 2.92236865e-01 1.01763475e+00 -7.45697677e-01
6.45244300e-01 -6.23927191e-02 1.08191323e+00 3.43526751e-01
8.40239167e-01 8.26729000e-01 1.11000252e+00 -2.18259841e-01
1.50253028e-01 2.12413847e-01 -1.80877447e-02 -3.53103280e-02
6.68844163e-01 4.67429876e-01 -2.87627220e-01 1.11115657e-01
1.07120883e+00 4.11436200e-01 8.90027732e-02 6.81828782e-02
-1.05499196e+00 1.16948807e+00 3.86056751e-01 6.19607866e-01
-5.85936248e-01 7.17090145e-02 2.46950120e-01 7.77291954e-01
6.47486031e-01 5.94270527e-01 -8.05887640e-01 -1.77150398e-01
-1.03227663e+00 4.69778627e-01 1.00767243e+00 3.91779035e-01
4.82373744e-01 6.25106752e-01 1.75872818e-01 3.00575167e-01
2.74351686e-01 4.27951604e-01 1.19152617e+00 -5.90083599e-01
3.44189048e-01 6.89396143e-01 3.61237317e-01 -1.06245518e+00
-4.74004686e-01 -8.38566065e-01 -6.71024561e-01 4.73646730e-01
2.91820735e-01 -4.71332014e-01 -1.07355431e-01 9.82440472e-01
-2.95533210e-01 3.00105482e-01 1.27793714e-01 5.09909153e-01
3.10142457e-01 8.51471901e-01 -3.27718914e-01 -3.52892280e-01
1.16198790e+00 -7.28543162e-01 -8.46056640e-01 -1.15563355e-01
3.31590593e-01 -5.36727548e-01 4.21722174e-01 3.62027049e-01
-1.00527644e+00 -6.00877225e-01 -1.09110284e+00 5.21979868e-01
-5.18352389e-01 8.76933858e-02 7.81129360e-01 3.90532106e-01
-7.03036189e-01 1.13822317e+00 -6.31114304e-01 6.87860787e-01
1.44840598e-01 2.57767022e-01 2.18323395e-01 9.10019875e-01
-1.52847099e+00 1.01773357e+00 5.03919065e-01 7.28893653e-02
-1.92685753e-01 -7.19871342e-01 -3.72454315e-01 4.47660565e-01
1.45160124e-01 -1.71682894e-01 1.48671746e+00 -1.02775967e+00
-1.21623111e+00 4.72530961e-01 5.42500317e-01 -9.75185871e-01
5.01025915e-01 -2.76283324e-01 -6.53048694e-01 -4.07755852e-01
1.03434421e-01 -8.54454935e-02 6.84812129e-01 -5.30963063e-01
-1.08758092e+00 -2.62472838e-01 8.94907042e-02 -1.97494134e-01
-3.67160946e-01 -1.06861386e-02 3.41202021e-01 -1.13947892e+00
1.32280052e-01 -3.95438612e-01 -1.55486777e-01 -6.84242129e-01
7.34306499e-03 -2.83535302e-01 3.44990045e-01 -7.32807875e-01
1.45597982e+00 -1.76278603e+00 -5.13668060e-01 2.21708283e-01
-1.44541655e-02 -7.97283836e-03 4.61780250e-01 3.82552326e-01
-4.78063732e-01 2.61141449e-01 1.22700043e-01 2.91478515e-01
3.10421765e-01 -1.59539863e-01 -8.79428208e-01 3.07607263e-01
3.44217628e-01 1.18213940e+00 -4.00004625e-01 7.96745718e-02
4.52726223e-02 -2.40855098e-01 -2.67479569e-01 2.10785180e-01
-1.97538093e-01 -1.45999447e-01 -5.10318160e-01 7.11840212e-01
8.36408973e-01 -3.20447862e-01 -1.24475814e-01 3.13729271e-02
-8.80516469e-01 4.46914077e-01 -1.12902904e+00 5.86070657e-01
-9.79412943e-02 5.87311745e-01 -6.03417575e-01 -1.10286593e+00
1.37477374e+00 1.88721344e-01 6.31898284e-01 -9.22061026e-01
2.98703074e-01 7.32026100e-01 2.61649579e-01 -2.07888663e-01
6.85946107e-01 -7.42732406e-01 1.89250009e-03 7.65045404e-01
-5.14339983e-01 3.56945634e-01 3.65182817e-01 -5.00052691e-01
5.86534560e-01 -1.54077122e-03 4.01242882e-01 -3.54781151e-01
3.09290528e-01 -1.86231688e-01 7.84825563e-01 2.75069088e-01
-8.37600306e-02 3.59801859e-01 9.53510165e-01 -9.92906928e-01
-1.09887612e+00 -4.81402785e-01 -2.43038937e-01 6.91430807e-01
-3.12605143e-01 3.14889878e-01 -3.49805593e-01 -2.02865794e-01
7.26969242e-01 6.01448357e-01 -6.22349441e-01 2.01050669e-01
-5.02546549e-01 -9.81520772e-01 -4.60295416e-02 8.87476861e-01
7.08924830e-01 -1.47186792e+00 -6.58109367e-01 6.06215119e-01
5.15236259e-01 -6.64160848e-01 -1.60907283e-01 2.36067712e-01
-9.14538026e-01 -1.00317621e+00 -1.11731732e+00 -4.14906651e-01
1.74159169e-01 -9.36348736e-02 1.25935698e+00 1.18993767e-01
1.75899506e-01 -2.59943068e-01 -2.10296944e-01 -1.06224871e+00
-9.95841809e-03 1.08391523e-01 -1.12459352e-02 9.85023081e-02
7.04814911e-01 -2.47896910e-01 -6.66714847e-01 1.28379941e-01
-7.57022202e-01 -3.89647782e-01 7.84933567e-01 7.39732146e-01
2.88593262e-01 6.13503456e-01 1.07338405e+00 -6.88590705e-01
6.57430351e-01 -7.37313271e-01 -1.35932589e+00 7.13683814e-02
-9.33556974e-01 1.32025793e-01 3.11797589e-01 1.55955762e-01
-8.61230135e-01 -5.75549960e-01 -8.69712830e-02 -2.73172945e-01
4.70780522e-01 1.05044520e+00 1.70163900e-01 3.37706417e-01
-2.43186444e-01 4.89592761e-01 1.75675958e-01 -6.25915587e-01
-4.26879495e-01 3.56642723e-01 2.33354956e-01 2.20678514e-03
1.09962595e+00 1.83316395e-01 4.43469919e-02 -6.51278421e-02
-9.89005327e-01 -1.34603739e-01 -8.05426180e-01 -8.12173337e-02
4.74827260e-01 -1.10559165e+00 -8.02708566e-01 8.60733092e-01
-7.09397376e-01 4.83486727e-02 -3.83029640e-01 6.60490394e-01
-4.16698188e-01 -1.87343150e-01 -7.53377140e-01 -1.04904342e+00
-4.25755739e-01 -8.33278060e-01 3.91938508e-01 4.63332027e-01
1.19585104e-01 -1.37442482e+00 1.57031313e-01 -3.67941186e-02
5.24123967e-01 2.89275408e-01 7.91925848e-01 -8.68809640e-01
-7.51079798e-01 -5.02758801e-01 -2.42567301e-01 6.43030345e-01
3.81118417e-01 -8.15622229e-03 -5.50868988e-01 -6.49862140e-02
4.60285842e-01 1.77831188e-01 8.47110868e-01 7.16756582e-01
8.61788869e-01 -2.24696666e-01 1.99921653e-01 4.97198373e-01
1.58671796e+00 7.94745803e-01 6.35547996e-01 1.38152266e+00
1.71342522e-01 6.55775189e-01 7.43400216e-01 8.41198325e-01
3.17461997e-01 6.38232082e-02 1.95101693e-01 1.47613883e-01
8.46244752e-01 -2.96673894e-01 4.11907554e-01 9.01522756e-01
-2.80808330e-01 1.77320808e-01 -7.81860769e-01 3.34491134e-01
-1.73091316e+00 -1.32727897e+00 1.56126872e-01 1.79016209e+00
6.10977173e-01 7.49974787e-01 4.82911527e-01 3.31937164e-01
4.86697227e-01 3.58300090e-01 -6.58796430e-01 -2.06030328e-02
-3.16324621e-01 5.95289879e-02 9.48336244e-01 -4.90732007e-02
-1.20090055e+00 6.56488895e-01 6.52033424e+00 2.01631784e-01
-1.34926581e+00 -6.15446568e-01 1.10616255e+00 3.74886021e-02
-5.39847016e-01 -1.84827477e-01 -1.15896654e+00 9.11953509e-01
7.97721684e-01 -8.20372462e-01 -1.34991273e-01 1.19143164e+00
2.30354115e-01 1.74754575e-01 -8.07846010e-01 7.21410751e-01
-3.65479290e-01 -1.77126157e+00 -6.48168921e-02 3.35974306e-01
6.47933483e-01 -2.65222609e-01 5.21736801e-01 4.56603467e-01
2.90809758e-02 -7.77358532e-01 9.43382859e-01 8.76316786e-01
-3.59299183e-02 -1.17044199e+00 1.29754531e+00 3.34880680e-01
-1.12364507e+00 -5.03508866e-01 -6.79153681e-01 -6.28949225e-01
9.51755643e-02 6.92043662e-01 -2.69645363e-01 4.00184989e-01
7.63066232e-01 1.01686490e+00 -3.12272847e-01 9.77132082e-01
4.01066989e-02 4.75907832e-01 1.57731295e-01 -2.72283942e-01
4.90560144e-01 -7.38426924e-01 -1.47718817e-01 4.77063775e-01
8.66319060e-01 2.33610988e-01 -1.71401352e-01 1.04755175e+00
-1.04271099e-01 1.10422768e-01 -4.98800933e-01 -2.92792648e-01
1.95183352e-01 6.83950961e-01 -5.81126988e-01 -2.55582511e-01
-9.33911264e-01 1.76090449e-01 -3.00071537e-01 -1.33036658e-01
-4.65360582e-01 -3.62675190e-01 6.88906074e-01 5.12381315e-01
6.46351755e-01 -1.13095559e-01 -6.99923873e-01 -1.17877734e+00
1.95148960e-01 -5.38958848e-01 1.74584255e-01 -4.09668714e-01
-1.55672014e+00 1.60947904e-01 -1.74913496e-01 -1.54296410e+00
-4.66418117e-01 -1.16377032e+00 -1.06106734e+00 1.01700783e+00
-2.11832047e+00 -2.83677787e-01 3.55730295e-01 1.84163883e-01
5.47965884e-01 -1.01801825e+00 2.53280967e-01 5.36420532e-02
-5.07751882e-01 1.78564116e-01 6.04658782e-01 7.75241613e-01
-5.44286445e-02 -1.56158793e+00 8.50661397e-01 4.96685475e-01
3.97388861e-02 4.20313388e-01 4.39021111e-01 -9.29588437e-01
-9.72395658e-01 -6.84374511e-01 1.06980431e+00 -9.89805087e-02
1.37460232e+00 2.75876671e-01 -9.26377296e-01 8.55421960e-01
4.44087327e-01 -3.01776469e-01 7.28567243e-01 -1.89042076e-01
8.88663344e-03 -5.52346945e-01 -8.24233770e-01 2.25524038e-01
6.81026429e-02 -5.85848987e-02 -9.13886487e-01 4.96241730e-03
5.44177234e-01 5.07540777e-02 -8.93572748e-01 2.78876245e-01
6.37499273e-01 -1.47745764e+00 7.64990926e-01 -3.39470148e-01
5.13177574e-01 1.18146338e-01 4.93786037e-02 -1.29654646e+00
-6.43090904e-01 -3.67951840e-01 -4.90316451e-02 1.00767457e+00
4.25232470e-01 -9.81292009e-01 9.86351430e-01 8.01752090e-01
1.67068988e-01 -9.10393178e-01 -7.28775561e-01 -8.17024887e-01
5.83021820e-01 -8.03514570e-02 1.31781757e+00 8.55025828e-01
-1.14399351e-01 -1.28674731e-01 -4.77196723e-01 -1.65884972e-01
4.68130499e-01 8.15987170e-01 4.57297385e-01 -1.56582725e+00
-1.75626427e-01 -1.01979077e+00 -1.88117474e-01 -8.55480671e-01
2.14747712e-01 -4.82753336e-01 -6.45140111e-01 -1.15976965e+00
-5.06353788e-02 -1.65531740e-01 -9.84948277e-01 1.20601222e-01
-1.93274058e-02 -3.95322114e-01 1.92605510e-01 5.29031932e-01
2.57119745e-01 4.65385854e-01 1.29574239e+00 -1.35935292e-01
-8.13560709e-02 4.59661096e-01 -7.79389441e-01 8.32732499e-01
1.21777523e+00 -5.89316804e-03 1.07000284e-01 -1.22943401e-01
6.66855574e-01 1.08280145e-01 -8.78489017e-02 -6.96893752e-01
-5.19238934e-02 -4.99971598e-01 7.36559927e-01 -1.20537233e+00
-8.01632479e-02 -7.61005282e-01 9.54292417e-02 8.02255511e-01
-1.78262308e-01 7.95789123e-01 6.75363885e-03 1.60753265e-01
-7.43435502e-01 -5.91920018e-01 4.99245048e-01 -5.53626060e-01
-8.68142843e-01 5.36110759e-01 -2.44323298e-01 -3.04554164e-01
1.11800635e+00 -1.32436156e-01 -1.31697565e-01 -3.86557162e-01
-3.78809184e-01 1.97035730e-01 -1.67182256e-02 4.34380680e-01
5.61652601e-01 -1.65636539e+00 -8.29110384e-01 2.48151734e-01
-3.35502297e-01 -3.53769004e-01 -8.47989470e-02 4.08972055e-01
-8.25368583e-01 8.19175541e-01 -5.13792455e-01 1.71607807e-01
-2.92637348e-01 5.18603265e-01 5.76871216e-01 -5.31474054e-01
-5.47770679e-01 6.71901822e-01 3.17695551e-02 2.85880476e-01
-2.10666806e-02 -6.42456770e-01 -7.27014840e-01 8.97260606e-01
8.04273307e-01 3.00183684e-01 3.57390800e-03 -5.42423010e-01
1.23896241e-01 6.95940852e-01 -1.45760834e-01 1.72623858e-01
1.92982495e+00 6.80913776e-02 -3.01807672e-01 9.28923011e-01
9.98100698e-01 -2.70757109e-01 -1.20225060e+00 -2.19472408e-01
8.64036024e-01 -6.70893192e-01 6.18128255e-02 -3.36789608e-01
-1.82497513e+00 6.77312136e-01 2.87334025e-01 9.31527853e-01
8.42505038e-01 -4.00682420e-01 1.10542154e+00 4.07615393e-01
3.50462973e-01 -1.39458787e+00 1.25232220e-01 5.57355583e-01
9.55600679e-01 -1.42903662e+00 -4.11852524e-02 3.53544027e-01
-6.41489208e-01 1.40330637e+00 3.66006792e-01 -5.83758652e-01
1.41644251e+00 4.30978537e-01 3.68565798e-01 -1.46274984e-01
-7.57856786e-01 1.60063848e-01 1.09193981e-01 -3.16536687e-02
6.46300733e-01 1.51330397e-01 -3.62756819e-01 1.11782873e+00
-9.28460956e-01 -2.09535249e-02 7.18607426e-01 8.82070541e-01
-8.12293172e-01 -1.11036742e+00 -4.24389154e-01 7.43317068e-01
-1.05219269e+00 -1.69145197e-01 -3.85577343e-02 9.59588408e-01
-2.35214710e-01 3.47737312e-01 8.02207232e-01 -2.13715538e-01
4.39527124e-01 1.95777729e-01 -9.96945724e-02 -3.80443484e-01
-5.89621186e-01 2.33297840e-01 -4.55815136e-01 -1.09878415e-02
-2.98974514e-01 -9.45155978e-01 -1.22744083e+00 -5.75250268e-01
-2.10167408e-01 1.94984868e-01 4.13346052e-01 9.35494900e-01
-1.64659545e-01 5.18654406e-01 1.36500561e+00 -6.02674842e-01
-1.14051282e+00 -9.23031867e-01 -1.58339000e+00 3.83337103e-02
4.31540549e-01 -7.69668758e-01 -5.62627494e-01 -1.53686315e-01] | [4.435600280761719, 4.251047611236572] |
295bde75-249e-425c-84b5-0d35f92d9ef5 | bone-texture-analysis-for-prediction-of | 1902.04880 | null | http://arxiv.org/abs/1902.04880v1 | http://arxiv.org/pdf/1902.04880v1.pdf | Bone Texture Analysis for Prediction of Incident Radio-graphic Hip Osteoarthritis Using Machine Learning: Data from the Cohort Hip and Cohort Knee (CHECK) study | Our aim was to assess the ability of radiography-based bone texture
parameters in proximal femur and acetabulum to predict incident radiographic
hip osteoarthritis (rHOA) over a 10 years period. Pelvic radiographs from CHECK
(Cohort Hip and Cohort Knee) at baseline (987 hips) were analyzed for bone
texture using fractal signature analysis in proximal femur and acetabulum.
Elastic net (machine learning) was used to predict the incidence of rHOA
(Kellgren-Lawrence grade (KL) > 1 or total hip replacement (THR)), joint space
narrowing score (JSN, range 0-3), and osteophyte score (OST, range 0-3) after
10 years. Performance of prediction models was assessed using the area under
the receiver operating characteristic curve (ROC AUC). Of the 987 hips without
rHOA at baseline, 435 (44%) had rHOA at 10-year follow-up. Of the 667 hips with
JSN grade 0 at baseline, 471 (71%) had JSN grade > 0 at 10-year follow-up. Of
the 613 hips with OST grade 0 at baseline, 526 (86%) had OST grade > 0 at
10-year follow-up. AUCs for the models including age, gender, and body mass
index to predict incident rHOA, JSN, and OST were 0.59, 0.54, and 0.51,
respectively. The inclusion of bone texture parameters in the models improved
the prediction of incident rHOA (ROC AUC 0.66 and 0.71 when baseline KL was
also included in the model) and JSN (ROC AUC 0.62), but not incident OST (ROC
AUC 0.53). Bone texture analysis provides additional information for predicting
incident rHOA or THR over 10 years. | ['Willem Evert van Spil', 'Saeed Arbabi', 'Willem Paul Gielis', 'Rintje Agricola', 'Harrie Weinans', 'Vahid Arbabi', 'Jukka Hirvasniemi'] | 2019-02-13 | null | null | null | null | ['texture-classification'] | ['computer-vision'] | [-4.07263815e-01 1.69347033e-01 -7.24344432e-01 1.54862836e-01
-7.19170690e-01 1.00832008e-01 9.49444026e-02 4.44261283e-01
-4.25485402e-01 4.76999819e-01 6.06229961e-01 -3.42598557e-01
-7.27676868e-01 -1.22128069e+00 -6.71548486e-01 -9.08978656e-02
-1.08823550e+00 7.61624634e-01 6.05651200e-01 -1.70530334e-01
-7.39612430e-02 9.24794674e-02 -1.10783780e+00 2.00313911e-01
4.52320218e-01 8.35102499e-01 -7.75071755e-02 5.96821249e-01
6.74569786e-01 6.30858243e-01 -1.74459934e-01 -1.04290962e-01
1.74582675e-01 -3.90696168e-01 -4.57493424e-01 -2.00427175e-01
8.67690518e-02 -3.95618647e-01 -3.39112729e-01 4.01710659e-01
3.73314857e-01 -1.75273329e-01 1.12654984e+00 -4.58107293e-01
-6.56725824e-01 4.06618655e-01 -7.51004279e-01 7.54245445e-02
1.44634724e-01 1.92911014e-01 7.58059323e-01 -8.99971902e-01
8.03033769e-01 9.37800288e-01 9.13291931e-01 1.86764881e-01
-1.25029016e+00 -6.37053967e-01 -6.44639194e-01 -6.18515303e-04
-1.15450072e+00 -1.29455114e-02 1.35806441e-01 -8.02342474e-01
4.65960860e-01 1.59305215e-01 1.47300613e+00 3.80658269e-01
1.11139584e+00 6.55734092e-02 1.14474165e+00 -3.51457119e-01
4.50335413e-01 -6.91894054e-01 3.10504287e-01 7.69472182e-01
7.92323768e-01 3.47403109e-01 3.38836201e-03 -4.45247114e-01
1.13232255e+00 -4.17669192e-02 6.65217340e-02 -3.11043084e-01
-1.17851031e+00 8.16613257e-01 4.07612622e-01 1.47897363e-01
-3.34386140e-01 1.59001589e-01 5.01037896e-01 3.04028392e-01
-4.28002588e-02 2.08247781e-01 -1.41003951e-01 4.34096493e-02
-5.78338683e-01 3.45424831e-01 3.34250927e-01 -7.35713448e-03
5.28541990e-02 -1.66561514e-01 1.59424499e-01 1.43331766e+00
3.07773739e-01 8.26738954e-01 9.16250587e-01 -1.11551893e+00
2.02987026e-02 7.21240342e-01 -2.30843410e-01 -7.68926620e-01
-8.16454828e-01 -5.19129515e-01 -6.94658220e-01 3.74168873e-01
5.32406986e-01 2.85042942e-01 -1.14963901e+00 1.26860285e+00
-9.28967521e-02 -6.37916327e-01 -1.27278939e-01 1.02899420e+00
2.59671777e-01 1.17540009e-01 1.89538494e-01 -1.84729055e-01
1.91353834e+00 -3.87171060e-01 -1.10810310e-01 -3.05279493e-01
7.58567512e-01 -5.71164131e-01 1.00109434e+00 3.03820133e-01
-1.17018771e+00 -3.34733963e-01 -8.98905516e-01 2.29069680e-01
1.67223394e-01 5.89641333e-02 2.73263901e-01 7.13943303e-01
-6.82408392e-01 6.27509534e-01 -1.19045889e+00 -4.20239240e-01
3.92861962e-02 4.31858748e-01 -5.39998949e-01 -1.22492060e-01
-1.14796638e+00 1.05621290e+00 -1.32015795e-01 -5.50129175e-01
-2.35932156e-01 -7.30720282e-01 -5.70415616e-01 -7.58658722e-02
8.04460272e-02 -1.18449378e+00 8.11594546e-01 -4.37431425e-01
-1.18836582e+00 7.37292886e-01 1.28232956e-01 -2.61649191e-01
3.39652836e-01 -4.17735487e-01 -2.54933208e-01 2.97701687e-01
6.87912285e-01 -4.62100916e-02 -8.62726197e-02 -9.90565419e-01
-3.66400659e-01 -1.02141559e+00 -3.93879175e-01 3.83464769e-02
1.44215465e-01 -8.61667991e-02 -2.82173641e-02 -7.21425593e-01
5.69132626e-01 -1.30427432e+00 -4.82439071e-01 4.81049493e-02
-9.33791772e-02 7.03196079e-02 1.69170722e-01 -9.91089344e-01
1.16778481e+00 -1.93699729e+00 -3.24502230e-01 6.64188147e-01
4.07392949e-01 -2.34968901e-01 4.57048148e-01 4.18390892e-02
-1.82460546e-01 3.60012978e-01 -3.78606558e-01 5.99721730e-01
-5.76708972e-01 6.71333969e-02 7.08310366e-01 4.35819358e-01
-1.17632136e-01 8.24410141e-01 -8.19297552e-01 -5.37647784e-01
2.41895542e-01 2.19239309e-01 -7.16336071e-01 -5.81050932e-01
3.97184491e-01 -3.35234664e-02 -1.72165960e-01 8.29575360e-01
1.47897691e-01 -5.08225024e-01 3.42295825e-01 7.92825520e-02
-1.46779616e-03 -1.87940702e-01 -8.03455591e-01 9.83226895e-01
-3.55398059e-01 2.58581996e-01 -1.62839785e-01 -5.72641313e-01
1.24803019e+00 1.91601992e-01 1.00081158e+00 -1.03882396e+00
-1.40778154e-01 8.62321436e-01 7.20098972e-01 -1.43979117e-01
-1.66359141e-01 -6.13066971e-01 -2.57493526e-01 4.00522441e-01
-4.09703106e-01 3.22329223e-01 3.02449912e-01 3.97990011e-02
1.20935369e+00 -2.40565524e-01 5.60274243e-01 -5.47949553e-01
3.97707164e-01 1.72079638e-01 4.28816557e-01 6.85519159e-01
-6.95692003e-02 6.43413067e-01 7.85162687e-01 -7.06499934e-01
-1.36796224e+00 -1.61245620e+00 -6.58300579e-01 3.99333239e-01
-4.76138555e-02 -3.66784304e-01 -1.29078761e-01 9.29281861e-03
6.18142247e-01 8.37804675e-02 -4.48918492e-01 -4.16406721e-01
-9.04235005e-01 -1.18937004e+00 3.46820652e-01 6.32084966e-01
2.11050287e-01 -4.59024131e-01 -5.12361407e-01 4.90593553e-01
-1.32383257e-01 -2.91285366e-01 3.98208238e-02 8.22493806e-02
-1.55399799e+00 -1.32874775e+00 -1.21482384e+00 -5.61310291e-01
3.25983465e-01 -2.69083530e-01 7.66665459e-01 4.99494851e-01
-5.37864029e-01 3.49984676e-01 -3.83627079e-02 2.49707296e-01
-2.94563472e-01 -1.48984179e-01 2.97091037e-01 -5.78915179e-01
-1.95681036e-01 -5.10026276e-01 -1.16034102e+00 4.55993772e-01
-3.77161473e-01 -4.84823016e-03 9.01692510e-01 7.34833658e-01
7.73842394e-01 -2.64078140e-01 5.83070636e-01 -7.06888378e-01
9.40256789e-02 -7.08977818e-01 -4.30223867e-02 7.22952932e-02
-1.15594161e+00 -1.78586602e-01 1.57302454e-01 -4.24360514e-01
-6.47193611e-01 -4.34271276e-01 3.34242195e-01 1.89863786e-01
3.58354419e-01 9.48544085e-01 5.85638463e-01 3.04858953e-01
6.85917914e-01 -1.79980323e-01 5.25157690e-01 -5.68715930e-01
-3.90998930e-01 2.98363239e-01 4.94136244e-01 -8.53476346e-01
2.72921860e-01 5.08939147e-01 5.98081350e-01 -9.11721468e-01
-2.02455491e-01 -5.11955082e-01 -4.55107093e-01 -3.26289803e-01
9.04040456e-01 -9.45423663e-01 -5.95162570e-01 2.19290212e-01
-4.84226421e-02 -4.47056592e-01 -1.62831917e-01 1.17283547e+00
-6.54724777e-01 3.82487923e-01 -8.37110400e-01 -4.51879889e-01
-4.55770880e-01 -1.28365934e+00 7.31603861e-01 -1.38004437e-01
-8.68471205e-01 -7.12403893e-01 4.08113539e-01 4.70562726e-01
3.53671640e-01 8.40379477e-01 1.71243834e+00 -3.24367136e-01
-8.15196261e-02 -2.81414360e-01 -2.75694817e-01 -1.33346289e-01
1.94195986e-01 -6.97738305e-02 2.21143499e-01 1.36864051e-01
-1.35477230e-01 1.04372613e-01 5.75062752e-01 8.19021940e-01
5.50378621e-01 -6.79040281e-03 -3.88195306e-01 2.23718092e-01
1.68698418e+00 6.26986682e-01 7.11577117e-01 9.90089238e-01
1.52293444e-01 3.05154830e-01 3.47735435e-01 2.58529246e-01
1.99275851e-01 8.63289833e-01 8.06205496e-02 2.76714951e-01
-6.09864771e-01 -1.61497463e-02 4.99064237e-01 8.41882288e-01
-8.09280038e-01 4.92183805e-01 -1.43933058e+00 7.32470334e-01
-1.20088196e+00 -5.64410269e-01 -7.41344929e-01 2.22895980e+00
5.32803714e-01 5.96363604e-01 2.60011166e-01 1.06824979e-01
7.00387597e-01 -2.12597802e-01 -2.59486228e-01 -3.11050087e-01
1.22096799e-01 6.36371613e-01 6.58859968e-01 6.19911909e-01
-3.31198335e-01 2.25760695e-02 6.17986679e+00 7.32319877e-02
-8.94029498e-01 -1.97830766e-01 7.31618762e-01 -2.27329761e-01
9.67396200e-02 4.34821993e-01 -1.73334971e-01 6.05854988e-01
1.29970109e+00 -2.38795385e-01 -1.54145107e-01 9.28049862e-01
2.66796052e-01 -9.35730278e-01 -4.82471049e-01 9.59528834e-02
-4.09656316e-01 -1.23597133e+00 -1.51930287e-01 5.07865489e-01
4.39437658e-01 4.41143140e-02 -1.65392626e-02 3.43410879e-01
2.14843586e-01 -7.29052484e-01 4.47638333e-01 7.23246455e-01
1.12193072e+00 -3.63561720e-01 1.00818956e+00 -2.14343235e-01
-1.20736420e+00 -3.57864276e-02 3.80174927e-02 -5.23399830e-01
4.59267259e-01 9.23227251e-01 -8.19716156e-01 4.05952513e-01
9.10678983e-01 2.74277806e-01 -5.74792385e-01 1.00545287e+00
3.15489858e-01 6.95471704e-01 -5.95676064e-01 2.94481516e-01
-1.29205287e-01 -1.49313688e-01 4.49496925e-01 4.61624950e-01
2.92391658e-01 3.62412900e-01 3.42192613e-02 3.34540278e-01
2.40047500e-01 4.61450577e-01 -8.52908716e-02 4.02277738e-01
4.50877905e-01 2.64964819e-01 -9.89274085e-01 -1.88407838e-01
-7.74078310e-01 1.32941529e-01 -3.01327080e-01 5.50967045e-02
-2.64717698e-01 -6.70497492e-02 3.20708275e-01 1.18289971e+00
-4.83504504e-01 -1.71288732e-03 -1.00107419e+00 -6.71821713e-01
-6.69807717e-02 -6.83223963e-01 6.70387566e-01 -6.43335342e-01
-1.05392206e+00 -2.85748690e-01 4.89796177e-02 -7.64127433e-01
-1.81115165e-01 -6.60729110e-01 -3.11850309e-01 6.94198906e-01
-4.36010242e-01 -6.66825950e-01 8.27588215e-02 -5.78815266e-02
1.75664827e-01 -9.44009349e-02 7.17049122e-01 3.73097956e-02
-5.37220091e-02 4.11571935e-03 2.11394459e-01 4.75889385e-01
6.56884313e-01 -1.14245319e+00 -1.36689559e-01 -4.13176604e-02
-8.48624408e-01 8.67006660e-01 2.61438310e-01 -1.62421167e+00
-8.13819230e-01 -7.74395883e-01 8.68747652e-01 -8.17503110e-02
7.25653708e-01 5.90257466e-01 -5.39872706e-01 5.31773150e-01
-8.14583540e-01 -2.72898972e-01 1.08403218e+00 3.31298620e-01
-1.26729608e-01 1.06084555e-01 -8.38455260e-01 7.29753017e-01
6.22388065e-01 -1.39418855e-01 -7.59078503e-01 1.03787836e-02
2.84444839e-02 -3.21515054e-01 -2.02835608e+00 1.20331776e+00
1.56875849e+00 -1.07169569e+00 1.50921249e+00 -1.82209760e-01
7.67103732e-01 3.35667445e-03 -1.39401674e-01 -2.84793433e-02
-6.33875191e-01 3.71093780e-01 2.84281582e-01 1.80072486e-01
4.10939932e-01 -4.62113857e-01 9.22833562e-01 4.96800631e-01
-3.59396309e-01 -1.32862782e+00 -1.01384699e+00 -9.65372622e-01
5.84210098e-01 -5.77737018e-02 1.49203399e-02 4.76670831e-01
1.36911646e-01 -2.35123068e-01 2.39843085e-01 -2.09319577e-01
5.05383372e-01 -1.67835295e-01 4.21869069e-01 -1.35641122e+00
-3.48849952e-01 -5.83711147e-01 -7.80380845e-01 2.13546917e-01
-8.12960804e-01 -9.87163126e-01 -8.15856755e-01 -2.14245534e+00
6.76012337e-01 -5.35320044e-01 -1.57783240e-01 2.19308317e-01
3.40788215e-02 2.05856994e-01 8.73704776e-02 1.01920712e+00
3.36176097e-01 2.47192547e-01 1.14615595e+00 1.37027562e-01
-5.13144910e-01 2.47973889e-01 -8.73711184e-02 8.54886949e-01
8.63662660e-01 -4.77315992e-01 -1.40305400e-01 2.01067105e-01
1.50742993e-01 6.88963294e-01 7.94079661e-01 -1.39801860e+00
-4.71581250e-01 -2.23220065e-01 9.61471438e-01 -4.74464506e-01
1.79128364e-01 -3.60871255e-01 7.96312273e-01 1.59158909e+00
7.80169740e-02 1.17869809e-01 -2.01968312e-01 3.21180910e-01
1.19025253e-01 -1.21197812e-01 8.63310397e-01 -3.77013534e-01
-2.10371166e-01 -2.29114518e-01 -9.94602799e-01 3.32886763e-02
7.30354369e-01 -5.38905144e-01 -1.57355919e-01 -1.33554950e-01
-1.60676074e+00 -2.73146123e-01 7.13274658e-01 2.69926250e-01
4.70118940e-01 -1.42569482e+00 -4.54525441e-01 -2.32195735e-01
6.69460520e-02 -2.60828674e-01 4.93393354e-02 1.28926754e+00
-1.07735598e+00 3.33672196e-01 -5.34398139e-01 -6.74336672e-01
-9.97638464e-01 -2.04950068e-02 5.20458817e-01 -6.40682697e-01
-9.39596951e-01 2.53751427e-01 1.84582159e-01 9.19777900e-02
-2.57001549e-01 -1.68667853e-01 2.08270609e-01 -2.70899087e-01
1.15970992e-01 9.32128906e-01 -8.58205631e-02 -5.44044614e-01
-1.63191006e-01 9.51733768e-01 -1.99928414e-02 -3.38827193e-01
1.23634672e+00 -1.13658994e-01 -2.63525933e-01 7.69025922e-01
1.06795061e+00 9.90211666e-02 -4.66688961e-01 5.92644960e-02
2.63438582e-01 -2.78056636e-02 -1.59490794e-01 -7.87321508e-01
-6.92319751e-01 1.27656683e-01 1.06569898e+00 -4.99472082e-01
6.21222436e-01 4.64418262e-01 7.93768406e-01 -3.25880200e-02
6.25438273e-01 -9.09672558e-01 1.67873859e-01 7.80492648e-02
7.01956987e-01 -6.82069540e-01 4.08337831e-01 -5.25619805e-01
-2.33139202e-01 1.27956247e+00 3.15565705e-01 -6.51648521e-01
9.48795617e-01 -2.20526159e-01 -7.22406283e-02 -5.66501558e-01
-2.33727664e-01 2.79213965e-01 2.36333564e-01 3.94080371e-01
6.03116274e-01 4.73659575e-01 -1.18191504e+00 8.85083079e-01
-6.49653494e-01 2.64506102e-01 5.20135999e-01 1.13047874e+00
-9.42545652e-01 -9.80643809e-01 -7.55065322e-01 1.27572167e+00
-6.90668285e-01 3.95694017e-01 -1.30355343e-01 1.51498699e+00
8.04122686e-02 2.02685073e-01 9.84295607e-02 -8.47956352e-03
3.13756466e-01 3.23769510e-01 2.48766690e-01 -4.94537115e-01
-2.52732188e-01 2.77164787e-01 3.36217523e-01 -2.85457820e-01
-2.82474101e-01 -9.71460938e-01 -1.60267031e+00 -1.50418490e-01
-1.94077849e-01 -2.14317098e-01 1.33351579e-01 9.40073788e-01
-2.35197812e-01 4.37067181e-01 7.17776418e-02 -5.77571802e-02
-1.14593364e-01 -7.72853792e-01 -9.93441582e-01 1.94887891e-01
3.92744392e-02 -1.15672767e+00 -2.76937395e-01 1.92810595e-01] | [14.522894859313965, -1.7803139686584473] |
5c9c2ffc-9849-45fa-9106-21862d6d6c55 | fast-algorithms-for-directed-graph | 2306.09128 | null | https://arxiv.org/abs/2306.09128v1 | https://arxiv.org/pdf/2306.09128v1.pdf | Fast Algorithms for Directed Graph Partitioning Using Flows and Reweighted Eigenvalues | We consider a new semidefinite programming relaxation for directed edge expansion, which is obtained by adding triangle inequalities to the reweighted eigenvalue formulation. Applying the matrix multiplicative weight update method to this relaxation, we derive almost linear-time algorithms to achieve $O(\sqrt{\log{n}})$-approximation and Cheeger-type guarantee for directed edge expansion, as well as an improved cut-matching game for directed graphs. This provides a primal-dual flow-based framework to obtain the best known algorithms for directed graph partitioning. The same approach also works for vertex expansion and for hypergraphs, providing a simple and unified approach to achieve the best known results for different expansion problems and different algorithmic techniques. | ['Robert Wang', 'Kam Chuen Tung', 'Lap Chi Lau'] | 2023-06-15 | null | null | null | null | ['graph-partitioning'] | ['graphs'] | [ 1.80251658e-01 5.02943873e-01 -5.38448155e-01 -4.74138670e-02
-6.73723161e-01 -9.13318872e-01 -3.25885028e-01 2.62883544e-01
-2.68407613e-01 6.56879008e-01 -1.10016622e-01 -6.89072847e-01
-7.56450176e-01 -1.15631199e+00 -5.14394522e-01 -7.30555534e-01
-5.64809859e-01 1.01367104e+00 1.20105118e-01 -3.91586602e-01
1.03228331e-01 6.06297314e-01 -8.61390233e-01 -2.17448771e-01
6.80339396e-01 6.95009291e-01 -2.22739503e-01 8.51343989e-01
-1.59753740e-01 2.61177987e-01 8.14187378e-02 -7.88637936e-01
7.29778647e-01 -1.67655215e-01 -1.27229822e+00 5.34829795e-01
1.93707705e-01 -1.42265603e-01 -6.27124488e-01 1.01163054e+00
3.38611513e-01 1.54553548e-01 2.78945833e-01 -1.59426916e+00
-5.48172593e-01 1.07559884e+00 -1.57214522e+00 1.17679879e-01
7.40152180e-01 -4.09353048e-01 1.55114627e+00 -5.95571935e-01
6.76396132e-01 1.10627258e+00 4.93513554e-01 2.74252087e-01
-1.52678502e+00 -3.93171906e-01 4.80953097e-01 3.23951244e-02
-1.42430973e+00 1.73797220e-01 6.34918392e-01 -2.18993071e-02
8.12903404e-01 7.39770949e-01 7.93262005e-01 1.04602650e-01
-3.48561615e-01 8.64154875e-01 6.81177437e-01 -5.41450679e-01
-9.56154913e-02 -6.88666105e-02 5.66099703e-01 7.78863490e-01
9.54169333e-01 -5.13725765e-02 -2.53023934e-02 -5.03457427e-01
5.51714182e-01 -1.00093298e-01 -3.38714123e-01 -8.00261617e-01
-7.74798810e-01 1.02415466e+00 2.49520436e-01 5.37356213e-02
-9.05405134e-02 4.68239874e-01 4.57329273e-01 3.10038924e-01
3.78437161e-01 9.22251400e-03 -2.73306698e-01 1.41996682e-01
-7.92598784e-01 3.09751898e-01 1.23762453e+00 1.38016486e+00
8.41798007e-01 -5.29092476e-02 -6.96236640e-02 4.34568405e-01
2.64402479e-01 7.38728344e-01 -7.95600295e-01 -9.60748196e-01
7.58494139e-01 6.83635533e-01 6.44997507e-02 -1.37922776e+00
-6.52675688e-01 -5.81579626e-01 -1.01481593e+00 -9.61536989e-02
3.17407936e-01 -1.90102249e-01 -5.97174585e-01 1.85216951e+00
5.12996733e-01 -1.85256287e-01 -3.92585695e-01 7.33900964e-01
4.10915196e-01 7.86088467e-01 -5.15020370e-01 -7.41303444e-01
1.04631782e+00 -1.01044261e+00 -4.48122203e-01 -1.36461914e-01
9.69644547e-01 -4.06283826e-01 5.29134393e-01 3.72907251e-01
-1.46541488e+00 2.11946264e-01 -7.55406916e-01 6.22191057e-02
-1.29065737e-02 -5.31931341e-01 1.03594351e+00 1.05851090e+00
-1.31555212e+00 5.81626631e-02 -4.69347209e-01 -4.94551621e-02
3.41348946e-02 7.70843804e-01 -4.31332737e-01 -3.72226179e-01
-6.94242060e-01 3.10966343e-01 3.77520949e-01 2.30299249e-01
-3.69963735e-01 -8.44756246e-01 -9.51111138e-01 4.04391944e-01
8.10851157e-01 -7.41555870e-01 7.73636997e-01 -4.89091486e-01
-8.54801893e-01 1.17097282e+00 -1.70416057e-01 -3.51591200e-01
2.41630316e-01 6.81745708e-01 2.60456037e-02 3.78778249e-01
-2.03830689e-01 1.73445001e-01 2.70794421e-01 -1.28641093e+00
-6.28116727e-01 -6.79057658e-01 6.95988059e-01 6.03565052e-02
-5.03405333e-01 -4.37690914e-02 -6.92380607e-01 -3.19065064e-01
2.61916101e-01 -1.28138053e+00 -8.30270350e-01 -2.30504930e-01
-4.07069355e-01 -3.40504684e-02 3.15410793e-01 -2.87962347e-01
1.87163770e+00 -1.71838892e+00 7.47441649e-01 8.78502131e-01
7.65095174e-01 1.69800110e-02 -4.20866430e-01 9.01688457e-01
-3.06010276e-01 3.02824587e-01 -1.99651539e-01 -2.63271434e-03
3.22511464e-01 3.07337433e-01 8.14478993e-02 5.70170879e-01
-4.94536161e-01 7.75474608e-01 -9.71919060e-01 -2.12552339e-01
-3.16906087e-02 -3.58127505e-02 -9.34188008e-01 -3.81683886e-01
1.92528009e-01 -4.19994444e-01 -2.87594020e-01 4.56926733e-01
1.57216299e+00 -2.23505348e-01 8.51406574e-01 9.05329511e-02
1.06050879e-01 -4.03215349e-01 -1.84155583e+00 1.45978308e+00
-5.34377873e-01 2.34688297e-01 9.04462099e-01 -1.43854952e+00
5.18760860e-01 -2.74596605e-02 9.55222011e-01 -5.14625549e-01
1.11849658e-01 2.22546145e-01 -2.68557996e-01 -2.37943694e-01
4.12409306e-01 -1.97417513e-01 -3.42611670e-01 8.34823012e-01
-3.29036623e-01 -6.18349575e-02 8.83076787e-01 9.77866888e-01
1.34096241e+00 -5.97087204e-01 9.85954851e-02 -6.44966960e-01
7.08937824e-01 -3.99204493e-02 4.59776640e-01 5.91497660e-01
9.55037996e-02 2.24700823e-01 1.12610948e+00 -4.63969886e-01
-1.01632440e+00 -1.00773478e+00 3.40365954e-02 1.15651572e+00
2.64955729e-01 -5.37548304e-01 -6.42270088e-01 -3.75006050e-01
2.76290774e-01 2.44349062e-01 -7.79806674e-01 -2.12026045e-01
-3.42628092e-01 -9.00026500e-01 1.35097399e-01 6.19969726e-01
-1.84631601e-01 -3.13462354e-02 6.09057955e-02 1.66998014e-01
-1.90099314e-01 -1.09011996e+00 -8.64582479e-01 2.62048662e-01
-7.42833018e-01 -1.24929643e+00 -6.80904686e-01 -1.01393592e+00
1.15459394e+00 6.35360599e-01 1.15593553e+00 2.98167020e-01
-3.17691714e-01 6.53825760e-01 -2.38811806e-01 1.61075756e-01
2.74389721e-02 1.35099426e-01 1.08048283e-02 -2.61654764e-01
1.08287521e-02 -5.86952150e-01 -5.81683457e-01 2.07057804e-01
-1.04805672e+00 -2.31359601e-01 1.83033228e-01 7.26040184e-01
7.08672404e-01 1.85668096e-01 1.56010896e-01 -1.28261626e+00
8.58714163e-01 -3.83668661e-01 -1.15234709e+00 4.34604138e-01
-8.69104683e-01 2.51248926e-01 5.99014461e-01 -6.47102892e-02
-6.07398450e-01 1.69281200e-01 -5.14504611e-02 -1.68099850e-01
7.26308167e-01 7.17249513e-01 -3.76030244e-02 -7.02135205e-01
2.81740576e-01 1.56934053e-01 -2.31566548e-01 -8.25257525e-02
7.07306147e-01 1.99992508e-01 3.76779974e-01 -7.75108695e-01
1.15913379e+00 6.94883168e-01 5.61220586e-01 -3.20404619e-01
-7.01363802e-01 -7.79626548e-01 -3.97625685e-01 -4.26919460e-02
3.42174411e-01 -4.59428847e-01 -1.28084457e+00 -3.01257446e-02
-9.22594547e-01 -1.08854927e-01 -2.16197059e-01 1.97103143e-01
-5.78985989e-01 8.36996794e-01 -6.13411844e-01 -1.03033054e+00
-2.80767858e-01 -8.92707765e-01 7.66293764e-01 -1.83346212e-01
3.84003878e-01 -1.13416684e+00 5.16167700e-01 4.67308521e-01
2.31808543e-01 3.59044403e-01 9.50205088e-01 -2.33766362e-01
-6.41981363e-01 -3.36409837e-01 -6.25958502e-01 -6.05111159e-02
-3.91718507e-01 6.18935861e-02 -4.64517884e-02 -6.55925870e-01
-5.78404307e-01 -7.27508813e-02 8.87305856e-01 3.61254781e-01
9.37781394e-01 -5.67619145e-01 -6.41609251e-01 7.13640094e-01
1.84336889e+00 -1.12395152e-01 6.64591014e-01 -2.73183640e-02
5.49258649e-01 2.56728828e-01 5.01320064e-01 8.52928340e-01
4.54780847e-01 4.65669185e-01 6.51310384e-01 -2.74691254e-01
4.37644958e-01 8.57604071e-02 -1.56450525e-01 6.99518740e-01
-2.42988870e-01 -5.16668320e-01 -8.17087054e-01 6.58890069e-01
-2.11106992e+00 -9.28782344e-01 -6.87560201e-01 2.51597166e+00
4.72948074e-01 -2.53983773e-02 5.63662708e-01 4.80437458e-01
8.25128019e-01 -3.05158161e-02 -2.24531218e-01 -1.08320355e+00
-9.93302558e-03 5.37777960e-01 1.10185313e+00 8.45961988e-01
-7.05304325e-01 6.99432433e-01 7.57661867e+00 9.08971906e-01
-3.44766617e-01 -2.95395032e-02 3.97210062e-01 -2.13349551e-01
-1.07661724e+00 2.64669329e-01 -4.07787442e-01 -5.31381629e-02
4.89941746e-01 -9.14208055e-01 8.68393779e-01 7.80185103e-01
1.23064160e-01 -1.31218106e-01 -8.22237611e-01 1.08430159e+00
-1.44069763e-02 -1.32006729e+00 -1.78483695e-01 4.76044685e-01
1.07589018e+00 -6.18580937e-01 1.18612675e-02 5.91153046e-03
3.91678661e-01 -5.22796810e-01 8.42233598e-02 -2.34652251e-01
8.14337254e-01 -1.31117678e+00 5.01639009e-01 3.19223269e-03
-1.76128030e+00 -2.61389196e-01 -3.88560146e-01 -1.14675589e-01
5.19662738e-01 8.99837136e-01 -1.97576955e-01 8.49304497e-01
3.82288426e-01 -2.00489350e-02 2.44048163e-01 1.16156340e+00
8.56495649e-02 1.91120982e-01 -7.61939228e-01 1.04418270e-01
3.31439078e-01 -7.85197020e-01 7.11328804e-01 1.10750735e+00
2.31134705e-02 4.27246153e-01 3.35085511e-01 6.16440654e-01
-3.57813597e-01 3.29137981e-01 -4.78323221e-01 -2.14186758e-01
3.57425600e-01 1.49688327e+00 -1.12313032e+00 -1.64311565e-02
-4.68333095e-01 8.88665259e-01 4.70219314e-01 3.26111823e-01
-1.23117805e+00 -6.72769666e-01 5.65929413e-01 1.40727058e-01
4.27201360e-01 -6.24109805e-01 -1.83741391e-01 -9.59292889e-01
1.15934305e-01 -6.41900361e-01 8.61620843e-01 -1.57873705e-01
-8.85613561e-01 3.34795803e-01 1.67959541e-01 -6.12931430e-01
1.19821325e-01 -6.15069985e-01 -4.95773524e-01 4.57797825e-01
-1.30706179e+00 -9.14361060e-01 -7.45010003e-02 6.48778558e-01
-2.02932432e-01 5.60648084e-01 4.74871010e-01 6.25991106e-01
-6.20222032e-01 9.10903811e-01 1.24486990e-01 -2.08929107e-01
8.74573439e-02 -1.40314531e+00 -2.55079530e-02 1.13880360e+00
-1.68833847e-03 2.75626928e-01 6.72089756e-01 -1.35827184e-01
-2.16299748e+00 -6.26921594e-01 7.02253282e-01 1.68217868e-01
9.55514729e-01 -1.65078789e-01 -3.93045813e-01 8.21127415e-01
2.05165654e-01 -4.82981391e-02 8.36402833e-01 4.91799176e-01
-3.79940242e-01 -4.05040801e-01 -1.30795670e+00 4.87268627e-01
1.63111222e+00 -1.55291170e-01 1.27951041e-01 7.12536514e-01
6.81237817e-01 -5.20832062e-01 -8.59132409e-01 2.17622608e-01
4.48516279e-01 -6.80037200e-01 1.09850740e+00 -6.69349194e-01
6.28212467e-02 -9.57712233e-02 -7.09948912e-02 -1.07867992e+00
-8.31156313e-01 -1.16115487e+00 -2.52199292e-01 9.12824452e-01
9.00148988e-01 -7.23281682e-01 9.80809212e-01 8.15766335e-01
1.53276667e-01 -1.05431843e+00 -8.49225938e-01 -9.47654486e-01
9.84344259e-02 -4.87446338e-01 4.56501305e-01 7.67540634e-01
6.54456496e-01 2.80467719e-01 -4.93637979e-01 2.94326216e-01
8.64371240e-01 5.14983714e-01 6.29335284e-01 -8.96817565e-01
-4.66111124e-01 -5.30812263e-01 -5.38752556e-01 -1.08282828e+00
2.52939075e-01 -1.28263593e+00 -5.94870508e-01 -1.90483952e+00
8.41088951e-01 -5.31612992e-01 -7.05680102e-02 3.90360355e-01
1.03063295e-02 4.77402776e-01 2.49357134e-01 -6.48101032e-01
-8.46260130e-01 1.66809365e-01 1.18482316e+00 -1.77645966e-01
-2.59453714e-01 2.55970746e-01 -9.77495670e-01 2.68465012e-01
3.97507071e-01 -3.63857180e-01 -4.77578729e-01 -7.62373924e-01
1.16136467e+00 4.82363075e-01 -1.38299644e-01 -4.43539262e-01
1.80062026e-01 -3.67865473e-01 -5.89470863e-01 -6.29803598e-01
-1.40500963e-01 -1.09277236e+00 2.07871169e-01 7.38013029e-01
-2.48459384e-01 2.56602496e-01 1.82473317e-01 6.89182162e-01
1.00938953e-01 -4.23837781e-01 7.25340247e-01 9.74595249e-02
-2.92511195e-01 8.69785786e-01 -2.76108712e-01 3.65945190e-01
1.43431914e+00 -4.75061893e-01 -2.53829241e-01 -6.63665354e-01
-9.98858273e-01 9.76544619e-01 1.97078958e-01 -2.03533769e-01
4.70066518e-01 -1.32163930e+00 -8.67280483e-01 -3.13883662e-01
-3.51116285e-02 -1.11604728e-01 5.65048039e-01 1.25375819e+00
-7.29747534e-01 3.89587492e-01 1.23123392e-01 -1.84309661e-01
-1.49616039e+00 1.19757152e+00 1.08684056e-01 -8.24071169e-01
-2.76751459e-01 1.02944732e+00 -2.79893223e-02 -2.62313187e-01
1.36316106e-01 -3.07952631e-02 5.83877861e-01 -3.34397443e-02
3.02108645e-01 9.25801516e-01 -1.11277886e-01 -2.37332925e-01
-6.26104951e-01 5.73474705e-01 9.86461248e-03 8.39048401e-02
1.28053641e+00 -4.18422490e-01 -5.48032999e-01 -3.52079749e-01
1.31812716e+00 4.74773914e-01 -3.80951822e-01 -1.16113558e-01
-2.96586186e-01 -6.53460443e-01 -8.64494406e-03 -1.87277138e-01
-1.56270838e+00 5.16264737e-01 -7.70728737e-02 8.06633830e-01
1.46185422e+00 1.44804806e-01 9.05601859e-01 4.54054117e-01
6.44445181e-01 -1.07965314e+00 -4.34019566e-01 3.51627082e-01
6.44457698e-01 -8.77969563e-01 2.77809501e-01 -1.16684353e+00
-2.25756019e-01 1.25199366e+00 3.42114210e-01 -2.15816543e-01
8.13473642e-01 7.14188755e-01 -8.22913349e-01 -4.87647690e-02
-6.81865215e-01 -5.66249549e-01 -4.14456055e-02 5.43332756e-01
1.68512076e-01 2.96871781e-01 -1.12766838e+00 3.53114843e-01
2.08842471e-01 -2.50746638e-01 6.49992168e-01 5.50931811e-01
-3.86462152e-01 -1.45415080e+00 -1.34932920e-01 3.31332088e-01
-2.52422780e-01 -6.52524307e-02 -5.42222619e-01 6.73421323e-01
-3.90172064e-01 9.34299290e-01 -3.10124904e-02 -3.16080183e-01
2.82652676e-01 -5.46032727e-01 9.30998921e-01 -3.53246927e-01
-2.19152465e-01 -7.35221729e-02 3.06815118e-01 -5.77203512e-01
-2.99501568e-01 -1.95179060e-01 -1.36462712e+00 -1.10254133e+00
-7.24682450e-01 4.85186726e-01 6.93696588e-02 5.83676100e-01
2.98291057e-01 2.63036966e-01 1.18414450e+00 -5.06337047e-01
-3.55094641e-01 -1.85854107e-01 -9.64677274e-01 1.93038270e-01
-2.40273714e-01 -3.00610423e-01 -5.26822746e-01 -7.10412979e-01] | [6.947055816650391, 5.126405715942383] |
69b7d7b2-26f2-4b88-9fed-33afc9f38409 | fine-tuning-distributional-semantic-models | null | null | https://aclanthology.org/2021.vardial-1.7 | https://aclanthology.org/2021.vardial-1.7.pdf | Fine-tuning Distributional Semantic Models for Closely-Related Languages | In this paper we compare the performance of three models: SGNS (skip-gram negative sampling) and augmented versions of SVD (singular value decomposition) and PPMI (Positive Pointwise Mutual Information) on a word similarity task. We particularly focus on the role of hyperparameter tuning for Hindi based on recommendations made in previous work (on English). Our results show that there are language specific preferences for these hyperparameters. We extend the best settings for Hindi to a set of related languages: Punjabi, Gujarati and Marathi with favourable results. We also find that a suitably tuned SVD model outperforms SGNS for most of our languages and is also more robust in a low-resource setting. | ['Ashwini Vaidya', 'Divyanshu Aggarwal', 'Kushagra Bhatia'] | null | null | null | null | eacl-vardial-2021-4 | ['word-similarity'] | ['natural-language-processing'] | [-1.92108542e-01 -2.60275543e-01 -3.42831731e-01 -3.71704370e-01
-1.03393745e+00 -8.72006476e-01 1.00984120e+00 2.81477302e-01
-9.38925326e-01 7.16158330e-01 9.43347156e-01 -6.76778078e-01
-4.34550792e-01 -5.11094928e-01 7.98674300e-02 -5.38804591e-01
-2.18849152e-01 6.61949456e-01 2.18892366e-01 -7.53665090e-01
4.25049454e-01 2.58633435e-01 -9.02040660e-01 2.29215756e-01
5.16222179e-01 -7.38526061e-02 4.42700721e-02 7.03650773e-01
3.06307614e-01 4.43574250e-01 -5.21853343e-02 -6.64271235e-01
4.26468223e-01 -1.27171189e-01 -1.12605238e+00 -3.98049086e-01
1.43667683e-01 -7.75466636e-02 -3.24950963e-01 6.93543136e-01
9.71847236e-01 4.55920279e-01 9.11889613e-01 -8.22518051e-01
-6.15829349e-01 9.62034464e-01 -5.59186935e-01 5.79117954e-01
4.95578289e-01 2.47101873e-01 1.50434017e+00 -1.00779402e+00
7.12624490e-01 1.56238699e+00 9.83017206e-01 1.08621055e-02
-1.05149066e+00 -3.75771195e-01 3.00441310e-02 7.90928900e-02
-1.38151073e+00 -5.20036757e-01 2.94122487e-01 -3.11926275e-01
1.60275578e+00 4.60120678e-01 1.85573012e-01 1.12971520e+00
1.27593234e-01 6.01647913e-01 1.12372375e+00 -6.49344087e-01
1.04506619e-01 2.31945202e-01 2.24503994e-01 1.54036537e-01
3.35328519e-01 -2.37576574e-01 -2.78747618e-01 -8.63910437e-01
5.49392819e-01 -3.00709516e-01 3.00025642e-02 -1.22874230e-01
-1.57247043e+00 1.29282093e+00 -1.02051787e-01 5.76236784e-01
-4.12737429e-01 -1.51535317e-01 6.18829668e-01 4.67513412e-01
5.33841074e-01 5.68031728e-01 -8.23833406e-01 -4.16576713e-01
-7.83928752e-01 3.10894698e-01 1.19755948e+00 6.81318104e-01
3.45428735e-01 -5.93736842e-02 -2.90062368e-01 1.40407836e+00
5.63386083e-01 4.38665420e-01 7.92104304e-01 -7.92256713e-01
6.77900255e-01 -5.69583289e-02 1.48031205e-01 -9.19894099e-01
-5.03114820e-01 -2.78897017e-01 -4.63374823e-01 -4.47154731e-01
3.07470381e-01 -2.01370448e-01 -7.83845663e-01 1.66194701e+00
2.59323776e-01 -2.87754595e-01 1.94115520e-01 9.16720569e-01
6.76436245e-01 5.25580287e-01 2.42933616e-01 -1.46137133e-01
1.43777943e+00 -8.80002916e-01 -5.25520980e-01 -5.64884007e-01
1.11085093e+00 -1.20599198e+00 1.12234819e+00 3.40032130e-01
-8.61216009e-01 -6.62311763e-02 -8.07662010e-01 1.89781606e-01
-1.90481082e-01 -2.49146372e-02 8.35018396e-01 9.02871847e-01
-1.20499790e+00 5.35983205e-01 -5.31996846e-01 -1.09209156e+00
-3.50652754e-01 3.46789956e-01 -3.88009906e-01 -9.43164304e-02
-1.78386950e+00 1.24389982e+00 5.48074186e-01 -3.05946916e-01
-2.08496124e-01 -3.52906913e-01 -8.32693577e-01 -4.85526830e-01
-1.44854844e-01 -4.63940859e-01 1.10934508e+00 -4.97787356e-01
-1.01150703e+00 9.03275788e-01 8.43282193e-02 -3.56306255e-01
4.20778364e-01 -6.11827783e-02 -5.02115786e-01 -2.25067869e-01
2.74913907e-01 2.87310719e-01 2.61807144e-01 -8.99872959e-01
-3.54792356e-01 -3.76625687e-01 2.25222316e-02 3.90697718e-01
-4.65380430e-01 6.83536470e-01 -5.51871479e-01 -9.71825182e-01
9.29614827e-02 -1.17110455e+00 -5.76779902e-01 -8.52954090e-01
-2.71368623e-02 -2.48869210e-01 1.58124551e-01 -1.02104628e+00
1.75552428e+00 -1.90311086e+00 -1.42895371e-01 4.89927322e-01
-3.40982735e-01 1.45754039e-01 -4.38546270e-01 1.01113057e+00
-1.67185768e-01 1.50614470e-01 -1.64841637e-01 1.10151105e-01
-4.57361192e-02 5.37333369e-01 1.29575759e-01 6.86911404e-01
-2.21284449e-01 6.19304121e-01 -9.77842569e-01 -4.71640855e-01
-7.23992884e-02 5.06298840e-01 -7.31170237e-01 -3.57576489e-01
3.83644730e-01 -1.74623802e-01 -7.54790604e-02 4.27399248e-01
4.17341590e-01 -2.16440354e-02 6.23269856e-01 -9.47440509e-03
-1.61044374e-01 5.50235093e-01 -1.32983899e+00 1.35568440e+00
-4.96040404e-01 2.33801126e-01 -5.20991609e-02 -5.99710047e-01
7.17864752e-01 4.18894231e-01 2.98775673e-01 -5.02635360e-01
1.89641580e-01 3.54579985e-01 3.83550555e-01 -4.07007515e-01
6.37122452e-01 -1.84587747e-01 -9.66875721e-03 8.16698551e-01
1.88236702e-02 -2.11450122e-02 5.42360902e-01 4.38199043e-01
8.05698633e-01 -2.11610392e-01 7.61729836e-01 -7.91648448e-01
4.58618999e-01 -3.15368414e-01 4.29650933e-01 9.07905281e-01
-3.09491992e-01 7.89303005e-01 1.36250451e-01 -2.41255779e-02
-1.25269604e+00 -7.14671671e-01 -2.69492835e-01 1.67442846e+00
-6.29895866e-01 -7.55053997e-01 -2.73383796e-01 -5.03645003e-01
-1.47420904e-02 1.07289267e+00 -5.33195794e-01 -8.27013925e-02
-5.34216881e-01 -1.36842847e+00 4.43325251e-01 4.40564811e-01
-3.82675952e-03 -8.25852573e-01 -7.65724629e-02 2.90757656e-01
-3.71927708e-01 -9.27724957e-01 -5.77977479e-01 4.14880335e-01
-8.49464595e-01 -4.38948214e-01 -9.26229715e-01 -8.17383170e-01
4.79326732e-02 2.62966961e-01 1.21249092e+00 -3.41480315e-01
1.14257261e-01 7.01013446e-01 -6.64366543e-01 -2.39211142e-01
-4.67166364e-01 2.85440952e-01 4.65002924e-01 -3.56775045e-01
6.38247252e-01 -4.39937115e-01 -5.75288236e-01 3.13071638e-01
-7.37369359e-01 -5.31830609e-01 5.34414291e-01 6.55195773e-01
1.50304232e-02 -2.54019499e-01 4.25034255e-01 -1.07492363e+00
9.95565355e-01 -8.49355340e-01 2.00227380e-01 1.97670326e-01
-8.13999653e-01 8.00128356e-02 -1.71387270e-02 -3.60889494e-01
-1.02952898e+00 -3.27227801e-01 -2.55220383e-01 3.09555262e-01
1.46917209e-01 7.63634384e-01 1.58421785e-01 -1.01520652e-02
9.33883250e-01 -5.39656132e-02 -1.29224718e-01 -6.90888405e-01
4.55630213e-01 1.08006620e+00 1.43441930e-01 -6.29986286e-01
4.64918226e-01 3.47449362e-01 -5.23383856e-01 -1.10252035e+00
-4.33014035e-01 -8.74048650e-01 -7.33142078e-01 1.91256404e-01
6.44793451e-01 -1.17766917e+00 -1.07611969e-01 -6.29328489e-02
-8.75037611e-01 -2.39367604e-01 1.92349419e-01 9.75294113e-01
-4.84514743e-01 8.24991286e-01 -1.00259316e+00 -9.24519777e-01
-5.13486564e-01 -8.55237246e-01 8.30747426e-01 -4.24250156e-01
-1.00297105e+00 -1.46610856e+00 6.44669712e-01 4.14641529e-01
3.69342834e-01 -2.94814944e-01 1.03627431e+00 -1.31614304e+00
5.21910489e-01 -1.51421487e-01 -9.42800194e-02 2.55254596e-01
1.27925321e-01 -6.18659630e-02 -5.97475588e-01 -6.31669879e-01
-1.21312745e-01 -2.68359393e-01 9.41984594e-01 2.01614723e-01
6.40766099e-02 -5.30223489e-01 8.58733896e-03 3.36434729e-02
1.41861796e+00 7.72029385e-02 4.12247449e-01 9.15549517e-01
5.47475100e-01 5.24602473e-01 6.20220780e-01 8.77319574e-01
5.38358808e-01 6.62634432e-01 -9.68556106e-02 4.33486253e-02
2.59362638e-01 -4.99861129e-02 4.03118491e-01 1.08883119e+00
-2.76820928e-01 -1.99880645e-01 -1.23722041e+00 9.97097671e-01
-1.98153389e+00 -9.82155144e-01 -4.07916307e-01 2.37720394e+00
9.76873755e-01 -7.43437484e-02 6.89201593e-01 1.43814892e-01
3.49676937e-01 2.47963995e-01 9.70166102e-02 -9.16449785e-01
-2.82464743e-01 3.97389643e-02 7.84608006e-01 7.73480654e-01
-1.13355505e+00 8.74852538e-01 7.38128567e+00 7.61607349e-01
-5.87452352e-01 1.94150165e-01 5.10141373e-01 -2.82420576e-01
-4.69825327e-01 1.31804630e-01 -6.50140762e-01 2.85961986e-01
1.28579986e+00 -1.20721579e-01 2.79645473e-01 6.03091121e-01
2.64741987e-01 2.68203896e-02 -8.14172924e-01 8.31126392e-01
5.87155968e-02 -6.43716991e-01 1.19925067e-02 2.74523124e-02
9.00964320e-01 6.82650506e-01 5.79482578e-02 2.85234898e-01
7.67390549e-01 -7.72325993e-01 4.31882828e-01 -2.17700712e-02
2.96606809e-01 -9.08369303e-01 8.67065668e-01 1.86702907e-01
-6.88705206e-01 -7.66385496e-02 -5.03692091e-01 -1.54642656e-01
1.01359412e-01 4.44168776e-01 -9.38910484e-01 3.15189302e-01
7.66233444e-01 5.27787268e-01 -6.98938370e-01 9.22987223e-01
8.03735554e-02 1.05851018e+00 -5.59704185e-01 -1.79350689e-01
8.04479420e-01 -2.93663383e-01 5.14922380e-01 1.86717081e+00
2.15208545e-01 7.10295327e-03 3.17704044e-02 -7.17490762e-02
6.01160467e-01 6.36406362e-01 -6.26817286e-01 2.79554985e-02
2.74660796e-01 1.09779084e+00 -6.36130571e-01 -1.68722644e-01
-7.96158731e-01 9.40750539e-01 1.67994559e-01 3.63859057e-01
-4.42821771e-01 -3.75097722e-01 7.15874016e-01 1.24524496e-01
4.84558731e-01 -2.68155605e-01 -2.17733860e-01 -1.04823995e+00
-4.03211445e-01 -1.20586753e+00 1.07575881e+00 -4.96655822e-01
-1.52873218e+00 4.04926419e-01 2.69354880e-01 -1.03354776e+00
-6.53001070e-01 -6.21936440e-01 -3.80124986e-01 1.00769830e+00
-9.49861944e-01 -9.28965509e-01 6.87891066e-01 4.45511818e-01
3.42215538e-01 -2.69640267e-01 9.34543371e-01 4.13101852e-01
-3.26070726e-01 7.14837253e-01 7.22657204e-01 3.29751112e-02
1.01043761e+00 -1.05471313e+00 8.30092847e-01 6.44608974e-01
2.74687171e-01 1.13025343e+00 1.11783099e+00 -6.84417188e-01
-1.19756699e+00 -6.92339540e-01 1.66306841e+00 -5.95308721e-01
1.13180876e+00 -1.44435540e-01 -7.19500303e-01 7.59479761e-01
4.73806888e-01 -6.31865382e-01 1.15482950e+00 5.69445729e-01
-4.65018898e-01 5.47616422e-01 -1.08693218e+00 7.64466405e-01
9.31703150e-01 -5.32728195e-01 -5.13610542e-01 7.80449629e-01
4.30892974e-01 2.88554519e-01 -1.08124757e+00 1.80945307e-01
6.99759007e-01 -8.11004758e-01 1.09479153e+00 -1.15417993e+00
7.52978325e-02 1.33158267e-01 -6.17356181e-01 -1.28177798e+00
-9.97809529e-01 -6.45684779e-01 3.66736740e-01 1.22789240e+00
7.30885625e-01 -6.64142549e-01 3.91552836e-01 7.43428409e-01
4.33973968e-01 -6.08605087e-01 -6.48502171e-01 -1.02565861e+00
3.36923420e-01 -5.67480385e-01 1.23775929e-01 1.33227837e+00
2.96985149e-01 6.80853128e-01 -4.87731308e-01 -3.22137386e-01
4.43226963e-01 -2.61901766e-01 3.51661533e-01 -1.02734852e+00
-6.12135053e-01 -2.07333654e-01 -3.64277422e-01 -5.61161637e-01
-5.16196387e-03 -1.09508812e+00 -4.10725206e-01 -1.56996703e+00
5.11814892e-01 -9.45370719e-02 -4.73596543e-01 4.31370199e-01
-3.30141813e-01 5.68149149e-01 2.18881711e-01 3.54490995e-01
-4.27930385e-01 1.20880857e-01 7.04350114e-01 1.44301459e-01
-1.27117068e-01 1.96753070e-02 -1.01376402e+00 6.67804062e-01
8.42134476e-01 -3.68738621e-01 -4.05317664e-01 -4.78266269e-01
4.98503327e-01 -1.52697578e-01 -2.28337452e-01 -3.51074070e-01
-1.64185703e-01 -3.15655500e-01 1.25976861e-01 -2.78471291e-01
5.15449122e-02 -5.10269582e-01 -9.17060673e-03 5.67719996e-01
-4.31677818e-01 4.59600359e-01 -3.33523098e-03 3.79878491e-01
1.53931111e-01 -4.57673103e-01 6.20375335e-01 -4.14238900e-01
-7.19617128e-01 2.35832073e-02 -7.18015909e-01 3.07769805e-01
4.29069787e-01 -2.68176466e-01 2.96112329e-01 -6.54199064e-01
-7.25943148e-01 1.45606980e-01 2.84047723e-01 6.95234120e-01
1.71965554e-01 -1.54582822e+00 -1.14518225e+00 7.85508603e-02
4.71751958e-01 -1.15964103e+00 -8.96957442e-02 9.23740447e-01
-3.18958819e-01 7.27629900e-01 1.17862515e-01 -4.78494130e-02
-1.39856231e+00 2.02694565e-01 -2.13373765e-01 -3.31603855e-01
-1.82136461e-01 7.44185328e-01 -2.05851749e-01 -9.58782434e-01
1.33001715e-01 2.15334788e-01 -4.16186512e-01 4.77556199e-01
4.03965622e-01 5.99355221e-01 1.87263772e-01 -1.15196776e+00
-6.40003622e-01 1.69992685e-01 -3.05600733e-01 -6.59681261e-01
1.38031483e+00 -2.81987786e-01 -2.78317630e-01 6.20371699e-01
1.52457964e+00 1.10017814e-01 -3.86813134e-01 -3.86849970e-01
1.69720814e-01 -3.15807134e-01 6.56314418e-02 -7.63021469e-01
-5.33058047e-01 3.68645191e-01 4.90947366e-01 -1.18750796e-01
5.25974870e-01 -1.43747017e-01 5.13730943e-01 6.65720999e-01
4.51138735e-01 -1.46599197e+00 -5.95107734e-01 8.79578173e-01
7.17556179e-01 -1.09076595e+00 2.49711648e-01 -3.98489311e-02
-1.12756205e+00 8.16353619e-01 9.69671085e-02 1.93604827e-01
8.90371859e-01 2.65010484e-02 2.51277298e-01 9.39352252e-03
-8.23756456e-01 -2.74941862e-01 2.70895243e-01 6.23770833e-01
1.21483600e+00 1.21909328e-01 -1.14552641e+00 5.18673301e-01
-3.84296060e-01 -4.93286341e-01 3.84925395e-01 8.43405008e-01
-4.10305858e-01 -1.10799336e+00 -4.66366380e-01 6.91098154e-01
-6.68956518e-01 -6.24167323e-01 -6.74601853e-01 8.94582570e-01
-5.46621680e-01 1.07064927e+00 -1.54157013e-01 -3.78072858e-01
2.15250671e-01 2.17169389e-01 3.86694402e-01 -5.61372817e-01
-9.70561922e-01 2.73273587e-01 6.35329545e-01 -2.71665245e-01
-2.31302649e-01 -1.07475197e+00 -8.17112148e-01 -3.32591593e-01
-3.45099032e-01 3.03293884e-01 4.20421481e-01 6.58508778e-01
7.67921135e-02 -3.60359699e-01 4.49222356e-01 -6.31079257e-01
-9.78360295e-01 -1.19584012e+00 -7.60329008e-01 5.51342905e-01
-2.16189936e-01 -2.75439888e-01 -4.76906508e-01 -3.04053903e-01] | [10.844442367553711, 9.97326374053955] |
6328020f-b6e3-4f29-b310-dafa96840e01 | hccl-at-semeval-2017-task-2-combining | null | null | https://aclanthology.org/S17-2033 | https://aclanthology.org/S17-2033.pdf | HCCL at SemEval-2017 Task 2: Combining Multilingual Word Embeddings and Transliteration Model for Semantic Similarity | In this paper, we introduce an approach to combining word embeddings and machine translation for multilingual semantic word similarity, the task2 of SemEval-2017. Thanks to the unsupervised transliteration model, our cross-lingual word embeddings encounter decreased sums of OOVs. Our results are produced using only monolingual Wikipedia corpora and a limited amount of sentence-aligned data. Although relatively little resources are utilized, our system ranked 3rd in the monolingual subtask and can be the 6th in the cross-lingual subtask. | ['Yonghong Yan', 'Xuemin Zhao', 'Junqing He', 'Long Wu'] | 2017-08-01 | null | null | null | semeval-2017-8 | ['multilingual-word-embeddings', 'stock-price-prediction'] | ['methodology', 'time-series'] | [-5.08241713e-01 5.85970245e-02 -5.25233686e-01 -1.00866437e-01
-1.13380456e+00 -7.82998085e-01 8.52485061e-01 3.65220577e-01
-1.10613930e+00 8.46663535e-01 5.55701613e-01 -8.06808174e-01
2.16426849e-01 -3.17008615e-01 -5.13515174e-01 3.14951539e-02
2.89333999e-01 8.16241443e-01 -1.11143075e-01 -8.21807265e-01
6.88891634e-02 -9.38421786e-02 -1.03044581e+00 -3.84855345e-02
1.46295345e+00 3.36545765e-01 3.13999325e-01 2.79481739e-01
-5.50106168e-01 1.09334894e-01 -1.37200370e-01 -8.44833553e-01
2.97008753e-01 -2.36126289e-01 -1.02484107e+00 -5.32794476e-01
9.23046291e-01 2.42616564e-01 -4.49648976e-01 1.17888737e+00
8.15199137e-01 8.09605718e-02 5.63936591e-01 -8.44727337e-01
-1.36373663e+00 8.50951850e-01 -1.03144579e-01 5.44832528e-01
5.98473132e-01 -7.74489418e-02 1.51010573e+00 -1.56640935e+00
1.08739698e+00 1.26170361e+00 6.09502316e-01 3.66063476e-01
-1.01601517e+00 -4.22860384e-01 -6.56452626e-02 6.46948457e-01
-1.28814280e+00 -5.32595873e-01 2.89581448e-01 -2.80705184e-01
1.56293333e+00 -9.06371698e-02 4.18191135e-01 1.24070215e+00
2.14072242e-01 5.85292459e-01 1.26121402e+00 -7.26235271e-01
-2.65744686e-01 3.32946092e-01 3.81873816e-01 4.81565028e-01
5.31051993e-01 -7.88738951e-02 -3.65245849e-01 -4.50939126e-02
8.81774127e-02 -3.56178433e-01 -3.12337697e-01 -1.13137521e-01
-1.44715476e+00 1.00327921e+00 1.81070358e-01 6.94267690e-01
-1.84007257e-01 -1.68024361e-01 6.93539023e-01 8.17935169e-01
8.59067738e-01 1.05292404e+00 -7.37118781e-01 -2.50162423e-01
-6.37224376e-01 1.26989082e-01 6.96170747e-01 9.64142144e-01
8.42339277e-01 -4.13404480e-02 8.92492607e-02 1.28789926e+00
-2.37779040e-02 7.81269908e-01 9.02233183e-01 -6.09243751e-01
6.51873231e-01 2.17615798e-01 -5.85114919e-02 -5.75951815e-01
-3.97122353e-02 -3.07057232e-01 5.74553534e-02 -4.62931037e-01
3.21965098e-01 -2.86072969e-01 -6.24493003e-01 1.57893121e+00
5.96663728e-02 -3.36915463e-01 3.14977825e-01 8.28495324e-01
8.13635111e-01 5.85832238e-01 1.11706421e-01 7.59232342e-02
1.59542429e+00 -1.39082301e+00 -8.87811124e-01 -5.06506860e-01
1.10585034e+00 -1.29906654e+00 1.36231685e+00 -4.34954949e-02
-9.24026191e-01 -5.98781407e-01 -1.06452608e+00 -4.60670054e-01
-8.18302155e-01 -1.33992240e-01 4.14886296e-01 2.22055405e-01
-1.25346482e+00 3.48247051e-01 -4.18371111e-01 -1.10370421e+00
-1.94935918e-01 -2.94256687e-01 -6.03340566e-01 -6.57356262e-01
-1.84997785e+00 1.85079837e+00 3.02877009e-01 -4.35597390e-01
-4.17436600e-01 -9.12237704e-01 -1.13919699e+00 -3.49120289e-01
-5.28565794e-02 -3.80423665e-01 9.10991549e-01 -4.90257442e-01
-9.75019991e-01 1.27557015e+00 -2.28830740e-01 -4.32357967e-01
3.25485438e-01 -4.70274895e-01 -8.15166891e-01 -2.39483431e-01
6.35426342e-01 6.48497343e-01 1.09227851e-01 -7.23136604e-01
-5.72060943e-01 -1.36096627e-01 -6.46111742e-02 5.07898390e-01
-7.06023276e-01 3.23484510e-01 -4.26500618e-01 -6.46857619e-01
-2.08407491e-01 -8.71596754e-01 -2.87449688e-01 -4.35212016e-01
6.19151480e-02 -7.00334311e-01 4.61911380e-01 -1.14155746e+00
1.07165980e+00 -1.80690873e+00 3.16356242e-01 -3.47463578e-01
-2.36290321e-01 1.75263390e-01 -6.24282002e-01 8.98063958e-01
-1.12366728e-01 3.52025479e-01 2.78527550e-02 -3.30331564e-01
2.08623320e-01 3.34247977e-01 -3.83154362e-01 3.77469659e-01
1.92044586e-01 1.07908690e+00 -1.36471891e+00 -5.34141958e-01
9.05002579e-02 2.03759044e-01 -3.73832285e-01 3.48415487e-02
2.56755114e-01 9.12581906e-02 -8.33813548e-02 6.12577140e-01
4.59977031e-01 4.12246913e-01 4.33329016e-01 -2.01054141e-01
-3.28185052e-01 1.04591405e+00 -3.26746076e-01 2.33065176e+00
-1.06216133e+00 8.60232949e-01 -3.80694449e-01 -7.82296598e-01
7.95565546e-01 3.42615366e-01 2.32409999e-01 -1.24293661e+00
-5.52150384e-02 7.42560625e-01 1.32249355e-01 -6.03296995e-01
9.04875755e-01 1.39294593e-02 -4.62918431e-01 6.45767689e-01
4.90276515e-01 -2.72597283e-01 5.45881152e-01 1.36672780e-01
9.04790401e-01 2.10814282e-01 4.88326639e-01 -9.21072125e-01
3.79470855e-01 4.02881622e-01 2.90809482e-01 2.56767571e-01
-2.71780849e-01 2.50870943e-01 -7.01424759e-03 -5.89948475e-01
-1.33900249e+00 -1.43552303e+00 -5.10909021e-01 1.27403605e+00
5.48535176e-02 -4.81233895e-01 -5.35060585e-01 -9.73823130e-01
1.39842853e-01 1.06441295e+00 -3.67674261e-01 7.72809982e-02
-7.59533286e-01 -5.19114614e-01 7.39492655e-01 1.16860032e-01
-2.12278605e-01 -8.24904680e-01 2.10443050e-01 2.93297648e-01
-4.81364161e-01 -1.54791069e+00 -9.96719897e-01 -1.54320486e-02
-5.78988433e-01 -9.73576367e-01 -7.30207801e-01 -1.30902851e+00
3.84445548e-01 1.63589686e-01 1.46819341e+00 -3.69302064e-01
-2.77134031e-01 2.62821671e-02 -5.22741675e-01 -2.00000241e-01
-3.01775247e-01 3.41781676e-01 6.49975181e-01 -6.45094812e-01
9.97336924e-01 -4.01688635e-01 -1.44508272e-01 -2.03001127e-01
-3.73474896e-01 -4.66310263e-01 2.31219068e-01 8.26902151e-01
4.45374995e-01 -8.88026178e-01 7.04068363e-01 -6.52111828e-01
1.03280246e+00 -6.49672508e-01 -2.70096421e-01 4.69255209e-01
-8.57699454e-01 2.29358017e-01 7.43025184e-01 -4.91164178e-01
-6.24584675e-01 -6.27989411e-01 -1.98929116e-01 -1.29332513e-01
3.11524451e-01 5.74011564e-01 8.70104730e-02 1.78688049e-01
8.09284925e-01 1.30930945e-01 -3.46832275e-01 -7.75901258e-01
9.39087331e-01 8.40333700e-01 3.79294366e-01 -8.75151575e-01
6.43682957e-01 -1.28843650e-01 -7.90065944e-01 -5.87632179e-01
-8.05606902e-01 -3.80773067e-01 -6.44321442e-01 6.82567954e-02
1.00487065e+00 -1.19228649e+00 1.07334815e-01 -1.05496079e-01
-1.50448775e+00 1.77457511e-01 -3.00215960e-01 9.04183149e-01
-2.30562408e-02 3.78507316e-01 -6.57129526e-01 -1.26989380e-01
-5.57930410e-01 -1.14771950e+00 7.15680838e-01 -3.93161327e-01
-7.40184724e-01 -1.37972879e+00 6.79862857e-01 5.41807532e-01
5.66310346e-01 -3.08796555e-01 1.16427255e+00 -1.06198275e+00
1.24070115e-01 -4.53006998e-02 -1.73749372e-01 5.39145112e-01
3.58653188e-01 -4.01156783e-01 -5.89089274e-01 -3.68959129e-01
-3.95087063e-01 -6.49240792e-01 6.62092030e-01 -1.32711008e-01
4.18681115e-01 -1.09224692e-01 -4.92557362e-02 3.51723194e-01
1.37877047e+00 -1.19589917e-01 3.63450199e-01 5.68653166e-01
7.64864862e-01 6.98051453e-01 5.17376661e-01 -2.59763211e-01
9.92179751e-01 7.73076832e-01 -1.75549790e-01 -4.94818427e-02
-5.46750724e-01 -4.58346635e-01 6.97443604e-01 2.18462133e+00
1.57703862e-01 1.47054018e-02 -1.22414649e+00 1.38657451e+00
-1.44183350e+00 -6.03933632e-01 -3.93284529e-01 2.01202369e+00
1.16097355e+00 -3.09743643e-01 -3.97993058e-01 -5.07480323e-01
6.01700246e-01 2.89311141e-01 -9.36151221e-02 -1.21640241e+00
-3.94581616e-01 7.80817211e-01 6.51738346e-01 9.37599182e-01
-7.64539659e-01 1.69974613e+00 7.10126734e+00 9.43546236e-01
-7.71947980e-01 7.82995582e-01 -1.84015349e-01 7.79744387e-02
-8.07944119e-01 1.93146557e-01 -7.49063432e-01 5.06962061e-01
9.71249521e-01 -5.41032314e-01 4.44716632e-01 5.99984646e-01
-2.18093872e-01 2.05032110e-01 -1.19027007e+00 7.47824490e-01
5.14206350e-01 -1.06158197e+00 2.33715609e-01 -1.96934342e-01
1.05304158e+00 7.34840512e-01 3.69429705e-03 6.86740816e-01
6.13602281e-01 -1.02462411e+00 4.00045097e-01 -7.47476667e-02
1.07492077e+00 -7.68942177e-01 8.79273593e-01 -1.50125906e-01
-1.06031191e+00 5.49222648e-01 -4.89883631e-01 -7.47952238e-02
3.22928458e-01 4.92116988e-01 -5.21425068e-01 8.23982954e-01
5.68828523e-01 9.59054887e-01 -4.20558989e-01 4.64851648e-01
-6.56040013e-01 4.04429346e-01 5.35120741e-02 -8.20900127e-02
4.59725320e-01 -3.06772709e-01 5.15894353e-01 1.47393787e+00
5.63681543e-01 -7.57985532e-01 2.43780971e-01 3.81995142e-01
-4.20691818e-01 7.35815525e-01 -8.45698297e-01 -2.89591402e-01
6.32124126e-01 1.03572392e+00 2.26027951e-01 -4.49634969e-01
-5.31984687e-01 1.33886576e+00 9.86387432e-01 3.53589416e-01
-5.85536182e-01 -7.27786958e-01 1.22518396e+00 -2.31483489e-01
5.04262140e-03 -4.30720598e-01 -2.03757241e-01 -1.47898388e+00
-3.12129315e-02 -9.38388705e-01 1.66770637e-01 -5.92961133e-01
-1.78248322e+00 9.27618086e-01 -2.56225765e-01 -1.01118994e+00
-3.16544682e-01 -7.71630466e-01 -4.17900056e-01 1.12333095e+00
-1.80889702e+00 -1.01210606e+00 4.71163362e-01 2.63926059e-01
6.73283815e-01 -4.51283574e-01 1.20006025e+00 6.64232969e-01
-2.23112658e-01 9.00758505e-01 3.97872299e-01 1.35311097e-01
1.29971337e+00 -1.05747318e+00 8.28339994e-01 7.47761130e-01
4.74841505e-01 7.31675446e-01 7.94355094e-01 -5.24536073e-01
-1.26577985e+00 -1.02055824e+00 2.26026940e+00 -7.35443890e-01
1.55793369e+00 -4.53736782e-01 -7.14605808e-01 7.39049315e-01
8.73705506e-01 -8.22250545e-02 7.59234369e-01 4.96027142e-01
-7.62545526e-01 1.15657046e-01 -8.43696773e-01 9.38867867e-01
1.20886302e+00 -1.10515344e+00 -1.41045272e+00 5.17950416e-01
1.08433247e+00 -1.31451681e-01 -1.21902490e+00 2.05888748e-01
4.01494741e-01 -4.57720421e-02 7.99209476e-01 -9.07004297e-01
5.99819541e-01 5.97038912e-03 -4.02581185e-01 -2.01601696e+00
-1.58827931e-01 -1.48473650e-01 2.76983559e-01 1.07122397e+00
8.76506209e-01 -8.03882003e-01 -1.12508997e-01 -2.65502393e-01
-4.37792212e-01 -5.70281744e-01 -1.19519436e+00 -1.26064217e+00
8.19431126e-01 -4.35622752e-01 3.13768089e-01 1.63371134e+00
3.97748590e-01 5.84739983e-01 -2.37711355e-01 -2.06677094e-01
5.29306531e-01 -1.13340341e-01 4.13359404e-01 -9.17090476e-01
1.55303562e-02 -3.02319080e-01 -2.77688444e-01 -7.77047038e-01
7.78463066e-01 -1.65054739e+00 -2.42293235e-02 -1.72637057e+00
1.50499552e-01 -1.60719752e-01 -5.48608363e-01 3.75645995e-01
-4.00161922e-01 2.99985230e-01 6.00167997e-02 -7.61647671e-02
-4.32075649e-01 6.25015676e-01 1.00185335e+00 -2.56664574e-01
2.69929200e-01 -1.19782543e+00 -5.80649972e-01 3.63460124e-01
7.36498475e-01 -6.85596049e-01 -1.06613353e-01 -1.27717352e+00
1.28511742e-01 -5.76140583e-01 -2.99286276e-01 -4.18950260e-01
-5.24807861e-03 6.71463646e-03 -2.35694915e-01 -1.62255943e-01
1.68127581e-01 -4.20857787e-01 -3.27694356e-01 4.72010374e-01
-2.98615724e-01 7.11823285e-01 2.50624061e-01 -5.01020029e-02
-4.66103226e-01 -1.61944181e-02 3.71217906e-01 -1.01707485e-02
-7.91063070e-01 1.90061167e-01 -6.07309103e-01 8.61424029e-01
7.22375214e-01 1.68525919e-01 -4.74379390e-01 8.04354325e-02
-4.10831362e-01 5.13734639e-01 6.28673136e-01 1.23894441e+00
2.80909270e-01 -1.82010996e+00 -1.12756562e+00 -3.37410793e-02
7.64012575e-01 -8.21798921e-01 -1.36878461e-01 7.60071337e-01
-3.59189928e-01 6.80985212e-01 -3.73368979e-01 -3.57800573e-02
-9.27019417e-01 2.96854198e-01 1.48477834e-02 -4.33803320e-01
-1.96652532e-01 6.95063531e-01 -3.47285390e-01 -1.25320244e+00
-2.38845110e-01 4.73122336e-02 -2.34657619e-02 3.10695708e-01
3.47149104e-01 2.94487983e-01 1.17587417e-01 -9.47014689e-01
-7.03102708e-01 6.77159488e-01 -2.61482954e-01 -3.14255953e-01
1.16740608e+00 -2.01930448e-01 -4.00357485e-01 6.34410501e-01
1.63985252e+00 2.73659408e-01 -1.20688908e-01 -5.10220170e-01
4.15374845e-01 -2.98242509e-01 -2.71205418e-02 -7.01181591e-01
-6.19975507e-01 1.02429056e+00 4.41986233e-01 -4.03318912e-01
3.03782314e-01 1.65890500e-01 1.27730680e+00 4.96253192e-01
5.77233613e-01 -1.55233359e+00 -2.90246457e-01 1.23407459e+00
7.64361143e-01 -1.36892724e+00 -1.87621519e-01 6.37123659e-02
-6.30726159e-01 8.76144469e-01 5.82148492e-01 -2.97609359e-01
3.92199785e-01 1.04803137e-01 5.10362089e-01 4.25440036e-02
-7.42298841e-01 -4.42300618e-01 5.13516188e-01 6.03335857e-01
8.66114974e-01 1.93660870e-01 -1.21381390e+00 3.75324726e-01
-5.13727725e-01 -4.71657068e-01 1.73205718e-01 4.10100698e-01
-2.94166923e-01 -1.51977813e+00 6.47160411e-02 -2.27197260e-02
-5.88617921e-01 -1.01109302e+00 -3.94821167e-01 9.22357202e-01
1.50725439e-01 8.51719797e-01 2.86126435e-01 -4.58184898e-01
3.49894434e-01 4.05714959e-01 5.11944413e-01 -6.50687635e-01
-4.77513283e-01 -5.73193729e-01 6.75989211e-01 -6.10709667e-01
6.77117780e-02 -5.19146144e-01 -9.80932534e-01 -3.93307298e-01
8.92184600e-02 4.10097629e-01 8.67351711e-01 1.03209615e+00
2.53878325e-01 2.63604254e-01 4.63038981e-01 -3.20778579e-01
-4.57237244e-01 -1.24232078e+00 -1.42174527e-01 4.70988810e-01
-1.53840572e-01 -2.85377085e-01 -4.00937945e-01 -3.23583245e-01] | [10.99714183807373, 9.870359420776367] |
8ce5b4e8-f257-4163-8d3a-37fbe12180f2 | intelligent-approaches-to-interact-with | 1303.02292 | null | http://arxiv.org/abs/1303.2292v1 | http://arxiv.org/pdf/1303.2292v1.pdf | Intelligent Approaches to interact with Machines using Hand Gesture Recognition in Natural way: A Survey | Hand gestures recognition (HGR) is one of the main areas of research for the
engineers, scientists and bioinformatics. HGR is the natural way of Human
Machine interaction and today many researchers in the academia and industry are
working on different application to make interactions more easy, natural and
convenient without wearing any extra device. HGR can be applied from games
control to vision enabled robot control, from virtual reality to smart home
systems. In this paper we are discussing work done in the area of hand gesture
recognition where focus is on the intelligent approaches including soft
computing based methods like artificial neural network, fuzzy logic, genetic
algorithms etc. The methods in the preprocessing of image for segmentation and
hand image construction also taken into study. Most researchers used fingertips
for hand detection in appearance based modeling. Finally the comparison of
results given by different researchers is also presented. | ['Ankit Chaudhary', 'Sonia Raheja', 'Karen Das', 'J. L. Raheja'] | 2013-03-10 | null | null | null | null | ['hand-detection'] | ['computer-vision'] | [ 1.88076288e-01 -5.11717677e-01 -5.50705306e-02 -3.22002053e-01
3.88571262e-01 -6.96428299e-01 3.87847573e-01 -3.97202849e-01
-5.03964067e-01 6.65931046e-01 -8.58890191e-02 -3.11423391e-01
-4.73623216e-01 -7.05786705e-01 1.31241366e-01 -7.23078012e-01
4.80438411e-01 7.26208866e-01 3.84748846e-01 -2.15811566e-01
9.23289061e-01 1.15913630e+00 -1.51039100e+00 -2.21189797e-01
3.20203632e-01 3.24423999e-01 2.94498622e-01 1.07641149e+00
-2.76270211e-01 5.17630696e-01 -2.54971325e-01 6.75835162e-02
5.21000385e-01 -3.14210832e-01 -6.92957997e-01 -2.56321803e-02
-2.82758772e-01 -2.82073408e-01 -8.27685446e-02 9.20697451e-01
9.37420487e-01 5.40323198e-01 9.38249648e-01 -1.10337257e+00
-4.21552628e-01 4.05861139e-01 -7.33758092e-01 -1.67124067e-02
5.03363013e-01 2.14666158e-01 1.61334202e-01 -5.80468059e-01
9.29883778e-01 1.41555107e+00 2.60015428e-01 7.94394255e-01
-5.61497509e-01 -5.69157064e-01 -1.67564467e-01 6.10649586e-01
-1.37688279e+00 1.21538900e-01 6.03799284e-01 -4.84071881e-01
1.14878535e+00 5.90414941e-01 2.23731145e-01 4.99845088e-01
-1.14469565e-01 6.14110768e-01 1.46052718e+00 -9.19771969e-01
1.70451313e-01 1.53590620e-01 8.78371477e-01 7.71543264e-01
1.68378115e-01 6.21907525e-02 1.96079209e-01 1.31071135e-01
1.15584445e+00 3.80506188e-01 -5.30037209e-02 3.85757089e-01
-9.48988259e-01 4.02592033e-01 1.14647999e-01 1.05147040e+00
-6.21750653e-01 -4.65112813e-02 1.51627705e-01 1.22400194e-01
-6.69876397e-01 5.52960001e-02 -4.58699673e-01 -2.79145300e-01
-6.17595971e-01 1.29394233e-01 8.04714084e-01 6.41506493e-01
6.74598143e-02 -8.10121223e-02 -1.29009083e-01 6.15398645e-01
6.42193317e-01 5.62562943e-01 5.44768453e-01 -7.22009838e-01
-3.87234129e-02 8.39459240e-01 1.37731940e-01 -7.93612123e-01
-3.54355216e-01 6.08640313e-01 -6.25082314e-01 1.01425445e+00
6.44743919e-01 -2.27832347e-01 -1.40457833e+00 6.43404305e-01
2.66431928e-01 -6.35438561e-02 -1.95135459e-01 9.68057454e-01
9.24631655e-01 4.98724818e-01 1.07006565e-01 8.86615440e-02
1.46794534e+00 -6.97649539e-01 -7.18785584e-01 6.13130629e-01
-3.05333078e-01 -1.22832870e+00 8.87529790e-01 1.00765491e+00
-5.69845676e-01 -4.01731789e-01 -6.60446525e-01 2.23990053e-01
-6.39986277e-01 2.85819858e-01 6.20122552e-01 1.09351301e+00
-7.01173544e-01 6.43070340e-01 -1.03895879e+00 -1.06208444e+00
1.87051207e-01 9.12928581e-01 -2.36960605e-01 1.07007109e-01
-2.50228912e-01 1.07288241e+00 3.47296089e-01 6.36590794e-02
3.65107805e-02 1.84799716e-01 5.92960007e-02 -3.15445393e-01
1.68257281e-01 -1.28059909e-01 8.01411152e-01 -6.88245595e-01
-1.79881668e+00 9.84777629e-01 -1.63184464e-01 3.17317024e-02
6.32602751e-01 1.96414322e-01 -4.47365381e-02 -1.73750579e-01
-7.60643303e-01 4.34037387e-01 6.71243131e-01 -8.85436118e-01
-5.17595410e-01 -8.65072012e-01 -1.84453845e-01 1.59287199e-01
2.26729900e-01 5.79418063e-01 -1.87401116e-01 -5.69919109e-01
2.61382252e-01 -1.05644894e+00 -1.06605023e-01 -3.88749123e-01
-4.40936983e-01 -6.92437112e-01 1.18199563e+00 -9.24478054e-01
8.63627195e-01 -1.47994626e+00 2.42362678e-01 8.05005670e-01
-2.06529543e-01 6.31913781e-01 2.66477585e-01 5.09369433e-01
2.26916417e-01 5.57602420e-02 1.27826437e-01 5.33239126e-01
-1.00582419e-03 3.97771418e-01 2.16559112e-01 1.72994919e-02
-4.72726792e-01 5.13072491e-01 -4.28917080e-01 -6.46657407e-01
8.17712069e-01 6.94541514e-01 8.96799192e-02 1.85735404e-01
9.29869786e-02 7.58290291e-01 -7.62286901e-01 1.04454720e+00
5.26291549e-01 2.90958583e-01 1.43500149e-01 -7.88137764e-02
-3.66921276e-01 -4.60523635e-01 -1.69444001e+00 9.38662529e-01
-2.07584426e-01 6.58048451e-01 -7.49592856e-02 -8.52232814e-01
9.78918016e-01 6.24481142e-01 1.71750903e-01 -1.89854696e-01
8.37991595e-01 4.46753204e-01 2.61135191e-01 -1.07676387e+00
3.17708515e-02 1.17740877e-01 8.99678409e-01 7.91027546e-01
-3.10806453e-01 8.11729431e-02 3.88211496e-02 -6.32448554e-01
6.78854942e-01 4.84774679e-01 4.71851677e-01 8.68548304e-02
7.46830702e-01 1.86452419e-01 1.05357684e-01 5.22497118e-01
-3.36964458e-01 3.83866489e-01 -3.09915960e-01 -2.76399463e-01
-9.81842935e-01 -5.91422737e-01 5.17048966e-03 1.04612970e+00
-8.35958719e-02 5.76718330e-01 -8.57554376e-01 -1.77716821e-01
-3.77130732e-02 1.83896229e-01 -1.50344655e-01 7.34381437e-01
-1.00108361e+00 -5.67563295e-01 4.25907463e-01 5.82328975e-01
8.15095246e-01 -1.66042697e+00 -1.07559371e+00 1.96008950e-01
5.01770854e-01 -7.48700857e-01 6.08730242e-02 -1.33496717e-01
-1.06199706e+00 -9.67493057e-01 -1.13609421e+00 -1.01016665e+00
6.29615009e-01 6.17909022e-02 3.75626802e-01 5.60327172e-01
-1.06880581e+00 4.78732735e-01 -8.04488957e-01 -8.40539217e-01
-1.49146020e-01 5.17799817e-02 -1.06011920e-01 -4.12470907e-01
8.64866018e-01 -6.55018210e-01 -4.33621496e-01 2.97632992e-01
-5.02580941e-01 -3.11971009e-01 6.23103917e-01 5.69172382e-01
3.11777622e-01 9.39140990e-02 1.46826863e-01 -7.39075601e-01
7.01024652e-01 2.11140960e-01 -5.75401545e-01 6.61203861e-01
-5.48588574e-01 1.16854630e-01 -2.43791323e-02 -5.80963552e-01
-1.07068670e+00 4.68335003e-01 -1.49676844e-01 2.91490331e-02
-7.66116083e-01 -2.13635147e-01 -7.54131079e-02 -6.39854729e-01
5.29481947e-01 4.32051569e-01 4.44424385e-03 -7.16537237e-01
3.81783247e-02 1.38842142e+00 3.25756639e-01 -4.01220620e-01
4.99235630e-01 3.05430830e-01 5.15331365e-02 -1.25147319e+00
5.73733509e-01 -6.11365676e-01 -7.70231068e-01 -5.09893179e-01
1.21150494e+00 3.43566716e-01 -1.11630845e+00 9.60969627e-01
-1.17307997e+00 -1.77801132e-01 3.67017210e-01 7.56062567e-01
-2.59359758e-02 3.45567107e-01 -4.61667627e-01 -1.63112307e+00
-8.61401260e-01 -1.15340579e+00 6.40296400e-01 7.90150344e-01
-2.77198434e-01 -7.47379959e-01 -1.70880735e-01 2.46607497e-01
3.51960659e-01 2.51048982e-01 9.24741507e-01 -4.24668878e-01
-6.60562873e-01 -3.75766516e-01 -1.60778955e-01 9.42929462e-02
5.70715129e-01 5.31461895e-01 -7.30188668e-01 2.03132004e-01
-1.68631002e-01 1.62504032e-01 4.05196339e-01 5.69144189e-01
8.23859990e-01 -4.14786488e-02 -5.11362076e-01 -6.80566505e-02
1.79128385e+00 1.24378800e+00 1.02713871e+00 3.06353539e-01
6.70850754e-01 6.11777008e-01 5.97503185e-01 3.83041441e-01
-2.46576920e-01 5.97542346e-01 -8.27583894e-02 3.16692501e-01
-1.18786797e-01 4.40321535e-01 -1.25459097e-02 3.10379356e-01
-1.38392234e+00 -3.19787115e-01 -1.05474794e+00 7.80386105e-03
-1.71993089e+00 -1.20036268e+00 -4.81374770e-01 2.19490099e+00
3.91023755e-01 -4.31394517e-01 3.55610281e-01 8.35768938e-01
1.04549670e+00 -6.76171958e-01 -2.46050850e-01 -6.43268466e-01
2.46866062e-01 9.12845731e-01 6.21808827e-01 6.63586736e-01
-8.92074466e-01 1.02903998e+00 5.16263819e+00 4.01011795e-01
-1.53352261e+00 -1.05317337e-02 3.57741565e-02 2.66585678e-01
3.51461351e-01 -3.57177913e-01 -7.12160885e-01 4.41452533e-01
-1.86329409e-02 3.74364704e-01 1.08539164e+00 6.86402917e-01
4.05476928e-01 -4.59792048e-01 -6.20728195e-01 1.28625274e+00
-1.04586065e-01 -8.57438207e-01 -6.72369078e-02 6.09541349e-02
6.49479508e-01 -2.55409151e-01 -2.55057395e-01 -3.38258028e-01
6.89873323e-02 -1.37574899e+00 2.80649930e-01 7.57424951e-01
3.09009314e-01 -4.47348744e-01 7.71216691e-01 2.29130775e-01
-1.10515141e+00 -7.04585249e-03 -2.60553211e-01 -3.76836091e-01
1.42601013e-01 -2.23338544e-01 -1.12583995e+00 5.31414710e-02
6.38295889e-01 -1.44648626e-01 -2.71812022e-01 1.13952708e+00
-3.47220758e-03 2.70415097e-01 -5.75601935e-01 -8.80330145e-01
1.55856553e-02 -4.62974608e-01 5.34018040e-01 1.23316324e+00
1.28500298e-01 7.54117727e-01 -2.81019717e-01 6.15422308e-01
6.74592078e-01 4.22613353e-01 -5.02052426e-01 -2.95641631e-01
3.40472728e-01 1.12937176e+00 -1.49096262e+00 -3.13516527e-01
-5.24404179e-03 1.08816314e+00 -5.75519681e-01 1.35041669e-01
-2.34270185e-01 -9.84765530e-01 3.55562866e-01 1.22759081e-01
-1.16241733e-02 -4.99230146e-01 -6.17307186e-01 -7.05403209e-01
-1.86095774e-01 -7.42532551e-01 2.03535423e-01 -4.65487480e-01
-9.12803292e-01 4.17030782e-01 -1.30123198e-01 -6.51454329e-01
-1.49623409e-01 -1.39672649e+00 -9.14289415e-01 1.04519653e+00
-7.50664234e-01 -9.93059218e-01 -6.48043275e-01 6.68145299e-01
8.17409754e-01 -5.52037597e-01 7.94995666e-01 1.76006436e-01
-4.73915607e-01 3.71846944e-01 1.38538722e-02 2.87428916e-01
4.98634756e-01 -1.13475776e+00 -2.42075980e-01 6.97174549e-01
-2.18479857e-02 9.26660538e-01 5.23623466e-01 -7.81261206e-01
-1.58468950e+00 -1.74658746e-01 5.30526102e-01 -2.38031596e-01
-1.03998221e-01 3.96173626e-01 -5.00986099e-01 3.64407986e-01
3.07530135e-01 -4.07667637e-01 4.28650260e-01 -1.86809003e-01
1.91736132e-01 2.65998363e-01 -1.77992785e+00 4.72127855e-01
7.72435188e-01 -1.15968063e-01 -6.95947587e-01 2.65438020e-01
-3.91507059e-01 -3.16196769e-01 -5.54656386e-01 -4.92858887e-02
1.20138001e+00 -7.73717284e-01 8.79038036e-01 -7.61969447e-01
-2.74536729e-01 -3.95293206e-01 -1.16930142e-01 -4.39448327e-01
-3.96981463e-02 -4.52306271e-01 5.15298665e-01 1.25276017e+00
7.30705634e-02 -7.10918486e-01 9.73420084e-01 1.02040601e+00
4.25858349e-01 -3.49069655e-01 -4.61075366e-01 -5.89308083e-01
-3.17809016e-01 -2.04229057e-01 4.51774329e-01 7.78111339e-01
5.90190552e-02 1.15496807e-01 4.53176312e-02 -5.16011640e-02
6.68552697e-01 4.96936031e-02 4.77158308e-01 -1.57111669e+00
-3.52541268e-01 -6.06600285e-01 -7.41278887e-01 -2.03397527e-01
-3.76685560e-01 -5.36740005e-01 -8.78647491e-02 -1.87760258e+00
1.39538929e-01 -5.93712330e-01 1.43361231e-02 6.26672924e-01
2.28261098e-01 3.99391443e-01 3.74290079e-01 2.53815264e-01
1.64182827e-01 -6.50152504e-01 1.39029431e+00 -2.51594204e-02
-6.67787194e-01 1.89791918e-01 -4.55982164e-02 6.81785583e-01
1.00777411e+00 -1.31322280e-01 -2.74271935e-01 -1.48630366e-01
-1.17647082e-01 1.06923888e-02 1.87950388e-01 -7.88568854e-01
2.34846294e-01 -5.52729428e-01 3.96585882e-01 -3.82991374e-01
-4.57843095e-02 -1.17047119e+00 3.55071783e-01 8.70665669e-01
4.13930044e-02 -4.30381577e-03 -4.44667637e-01 1.97657030e-02
3.53771389e-01 -7.97619343e-01 7.81762004e-01 -5.22931457e-01
-9.47938919e-01 -1.59034982e-01 -6.31774068e-01 -7.45060623e-01
1.31794870e+00 -8.21905494e-01 9.49887633e-02 -1.77140862e-01
-1.01893449e+00 -1.43653437e-01 3.00761759e-01 1.67990610e-01
4.51036692e-01 -8.00486863e-01 -4.16772813e-01 -1.24129601e-01
-5.44664979e-01 -3.02341819e-01 5.78786619e-02 7.04452813e-01
-1.37714672e+00 6.72989964e-01 -9.73983228e-01 -2.03885019e-01
-2.22515345e+00 3.27455550e-01 1.52050778e-01 1.67272493e-01
-4.04019117e-01 7.25098610e-01 -8.08263063e-01 -2.98494488e-01
6.02082849e-01 -3.46702874e-01 -8.60430479e-01 -3.91084969e-01
6.56789541e-01 1.22198212e+00 -8.72670338e-02 -6.53577447e-01
-5.07379055e-01 1.25961661e+00 3.37561339e-01 -4.50879991e-01
1.23998046e+00 1.93490461e-01 -4.56093997e-01 1.33855417e-01
7.37033367e-01 -2.81083196e-01 -5.52815139e-01 3.86050582e-01
2.07375333e-01 -4.32184786e-01 -5.97579516e-02 -1.35398352e+00
-9.05123889e-01 1.14769220e+00 1.41306305e+00 -6.10506162e-02
1.04950655e+00 -4.22818631e-01 6.03883684e-01 9.23559904e-01
8.51782560e-01 -1.32563400e+00 -4.68063891e-01 4.29644853e-01
8.36260796e-01 -1.23004675e+00 -1.04797766e-01 -5.36282897e-01
-6.45563662e-01 1.60485351e+00 2.98167229e-01 -4.77579772e-01
8.74553084e-01 5.71219385e-01 2.04361439e-01 1.17178760e-01
2.15123832e-01 -4.69482660e-01 2.56924510e-01 1.04828370e+00
6.87965930e-01 2.58213937e-01 -1.18187296e+00 4.93853211e-01
8.29046369e-02 7.99040020e-01 2.79432178e-01 1.40093505e+00
-6.05848849e-01 -1.63312519e+00 -8.93483758e-01 6.17498815e-01
-5.84503889e-01 3.15905184e-01 -6.54757440e-01 6.07176721e-01
3.12567174e-01 9.52233374e-01 -2.71802515e-01 -6.98135376e-01
3.93584706e-02 3.33870739e-01 1.22374094e+00 -3.77089232e-01
-6.51333869e-01 -1.72454730e-01 -3.20388705e-01 -6.76829442e-02
-5.64012349e-01 -6.45759583e-01 -1.93843102e+00 -4.03108835e-01
-2.27361709e-01 -1.15975738e-01 1.18333709e+00 9.44849730e-01
-2.21636027e-01 1.40097961e-01 8.71875063e-02 -1.19464064e+00
-2.62225658e-01 -1.18078935e+00 -7.54796267e-01 -4.22012135e-02
-3.75998765e-01 -7.41054535e-01 2.63538986e-01 2.66526043e-01] | [6.4852986335754395, -0.29907140135765076] |
70e13169-efde-4b7d-951b-2dcc285fd737 | occlumix-towards-de-occlusion-virtual-try-on | 2301.00965 | null | https://arxiv.org/abs/2301.00965v1 | https://arxiv.org/pdf/2301.00965v1.pdf | OccluMix: Towards De-Occlusion Virtual Try-on by Semantically-Guided Mixup | Image Virtual try-on aims at replacing the cloth on a personal image with a garment image (in-shop clothes), which has attracted increasing attention from the multimedia and computer vision communities. Prior methods successfully preserve the character of clothing images, however, occlusion remains a pernicious effect for realistic virtual try-on. In this work, we first present a comprehensive analysis of the occlusions and categorize them into two aspects: i) Inherent-Occlusion: the ghost of the former cloth still exists in the try-on image; ii) Acquired-Occlusion: the target cloth warps to the unreasonable body part. Based on the in-depth analysis, we find that the occlusions can be simulated by a novel semantically-guided mixup module, which can generate semantic-specific occluded images that work together with the try-on images to facilitate training a de-occlusion try-on (DOC-VTON) framework. Specifically, DOC-VTON first conducts a sharpened semantic parsing on the try-on person. Aided by semantics guidance and pose prior, various complexities of texture are selectively blending with human parts in a copy-and-paste manner. Then, the Generative Module (GM) is utilized to take charge of synthesizing the final try-on image and learning to de-occlusion jointly. In comparison to the state-of-the-art methods, DOC-VTON achieves better perceptual quality by reducing occlusion effects. | ['Liang Lin', 'Tianshui Chen', 'Hao Li', 'Yukai Shi', 'Junyang Chen', 'Zhijing Yang'] | 2023-01-03 | null | null | null | null | ['virtual-try-on', 'semantic-parsing'] | ['computer-vision', 'natural-language-processing'] | [ 5.56030989e-01 3.11980426e-01 9.11262780e-02 -1.83812350e-01
-3.80560219e-01 -2.28169248e-01 2.17343673e-01 -4.90860969e-01
1.52171731e-01 2.91330397e-01 1.42468542e-01 4.24638279e-02
2.26696149e-01 -8.79421473e-01 -9.19781327e-01 -6.11866117e-01
5.87394655e-01 2.10818172e-01 3.67353499e-01 -3.32406253e-01
-2.06349492e-01 2.28508234e-01 -1.34319413e+00 4.77275610e-01
9.84193921e-01 9.86610472e-01 5.38873732e-01 3.29008818e-01
-1.13771379e-01 3.46393317e-01 -4.84314382e-01 -8.84479761e-01
3.93982053e-01 -5.44167340e-01 -3.08504909e-01 6.43630028e-01
6.30962610e-01 -4.53836292e-01 -4.60219145e-01 1.18814313e+00
2.85786837e-01 -6.95504248e-02 3.53800803e-01 -1.08415151e+00
-7.45320976e-01 3.73130143e-01 -8.04711461e-01 -5.86149335e-01
6.25846148e-01 2.13257700e-01 6.36482596e-01 -8.44706833e-01
9.25019860e-01 1.65541685e+00 7.21663952e-01 4.64787602e-01
-1.28829265e+00 -4.19419169e-01 4.63398844e-01 -3.71245705e-02
-1.22394001e+00 -1.41290173e-01 1.32457602e+00 -1.40840679e-01
2.58059353e-01 6.12998903e-01 1.07883275e+00 1.24810803e+00
4.42200840e-01 1.13320005e+00 1.09118247e+00 -4.85819191e-01
5.70154488e-02 6.46528378e-02 -2.56697118e-01 8.75093102e-01
-9.32475626e-02 1.34598851e-01 -2.60895461e-01 1.23544969e-01
9.93338823e-01 1.86235696e-01 -2.81972915e-01 -4.95818079e-01
-9.47671890e-01 3.49971980e-01 4.14299846e-01 -1.25820890e-01
-4.60607767e-01 1.43476307e-01 1.86272442e-01 4.74193972e-03
3.87760162e-01 2.72378605e-02 -2.66531944e-01 5.24489284e-01
-9.82193172e-01 3.78898919e-01 6.23925328e-01 1.21893799e+00
5.90120792e-01 9.34388787e-02 -2.82287657e-01 1.07674015e+00
4.33549941e-01 6.61138773e-01 -1.72870308e-01 -8.87815237e-01
3.90297741e-01 6.89947784e-01 -1.15140826e-02 -1.10593057e+00
-1.15224816e-01 -3.85334969e-01 -8.21467817e-01 1.46749780e-01
3.36548835e-01 3.45571898e-02 -1.35357106e+00 1.51247954e+00
5.77780545e-01 -3.18891667e-02 -2.50654727e-01 1.21025157e+00
1.07481551e+00 5.54746866e-01 3.26439530e-01 -1.69713393e-01
1.74402142e+00 -1.21234620e+00 -1.15443051e+00 -5.21478951e-01
-1.80151984e-01 -1.18608809e+00 1.23501265e+00 3.49729896e-01
-1.18896341e+00 -8.27147007e-01 -1.11766779e+00 -2.86692470e-01
-1.54152662e-01 2.21668929e-01 5.10737538e-01 7.05171227e-01
-6.22959435e-01 4.03078258e-01 -6.24297380e-01 -4.09387648e-01
3.85738313e-01 5.70158772e-02 -2.80788898e-01 -5.37298501e-01
-1.04699159e+00 6.95447326e-01 1.75580636e-01 3.52237374e-01
-8.39495957e-01 -6.36003494e-01 -1.08241308e+00 -2.45740309e-01
7.80607164e-01 -9.06453252e-01 6.78703606e-01 -1.18153298e+00
-1.62994015e+00 1.04414487e+00 -1.22377545e-01 2.04404950e-01
9.28702176e-01 -3.17970157e-01 -5.05917490e-01 1.17732733e-01
7.67937154e-02 5.65208375e-01 1.12305558e+00 -1.89162874e+00
-3.14475924e-01 -2.55352288e-01 1.58601001e-01 3.21719438e-01
9.52104628e-02 -1.15138546e-01 -1.29589391e+00 -1.34058797e+00
5.43928027e-01 -1.00118947e+00 -9.62876622e-03 3.21990550e-01
-8.80597532e-01 2.39255965e-01 1.03873289e+00 -1.27265966e+00
1.19329250e+00 -2.18049264e+00 4.54789132e-01 1.74748018e-01
1.13385879e-01 2.72569895e-01 -1.10397317e-01 3.16542298e-01
-1.46411583e-02 -1.91608950e-01 -1.88001260e-01 -8.92657459e-01
7.84619898e-02 4.79011089e-01 -1.88893393e-01 3.61800075e-01
-8.81436188e-03 1.02342522e+00 -7.29789913e-01 -6.93963826e-01
5.64790666e-01 4.99809742e-01 -4.76414323e-01 3.87996823e-01
-3.73244524e-01 4.57106411e-01 -3.01689804e-01 1.22766674e+00
1.07790351e+00 1.05627865e-01 4.05173808e-01 -8.38076770e-01
2.35443160e-01 -3.62589985e-01 -1.26106632e+00 1.88150299e+00
-2.22261220e-01 -1.22079095e-02 3.27568889e-01 -4.33014333e-01
8.60669911e-01 5.26272953e-02 3.70656371e-01 -8.03658426e-01
2.49663800e-01 6.93148561e-03 -4.46576566e-01 -6.61766827e-01
4.66919780e-01 -1.20929472e-01 -7.23809674e-02 -9.54778194e-02
-2.30667397e-01 -2.52428919e-01 6.31599175e-03 7.11876452e-02
4.43902224e-01 8.43481660e-01 -1.42193034e-01 -2.16621786e-01
4.24141973e-01 -1.76957007e-02 6.99485600e-01 4.97654259e-01
1.20501760e-02 1.05442464e+00 4.10514921e-01 -3.87811899e-01
-1.17810988e+00 -1.25153756e+00 -3.49530429e-02 8.63896430e-01
7.72514403e-01 -3.65638196e-01 -1.15778041e+00 -6.08577251e-01
2.82614841e-03 7.28439987e-01 -6.11364067e-01 -9.86824930e-02
-6.18951023e-01 -4.21461344e-01 5.96056804e-02 3.87243062e-01
8.43074679e-01 -1.01891100e+00 -2.71519929e-01 1.51275337e-01
-3.52776796e-01 -1.09880984e+00 -8.03421557e-01 -1.84425592e-01
-4.78334278e-01 -1.01913667e+00 -9.40699697e-01 -9.74495709e-01
1.01922727e+00 2.23532885e-01 9.87392426e-01 1.93714678e-01
-4.93355602e-01 3.25359106e-01 -2.94000179e-01 -2.61567354e-01
-4.45250750e-01 -3.45995754e-01 -2.85172731e-01 3.52941275e-01
-5.34342788e-02 -4.68824893e-01 -9.51056600e-01 6.62535012e-01
-9.72362995e-01 7.25680888e-01 5.53607106e-01 7.20118999e-01
8.31866562e-01 8.32089633e-02 -4.73524816e-02 -8.77986789e-01
3.05857509e-01 1.50215812e-02 -2.79841095e-01 4.92739081e-01
-8.84419978e-02 -4.21860754e-01 4.58488792e-01 -7.29325294e-01
-1.20254123e+00 2.80280620e-01 -3.01265419e-01 -8.49916041e-01
4.67358865e-02 -5.48296981e-02 -8.75886261e-01 1.79956071e-02
3.91620211e-02 2.43686944e-01 5.81254214e-02 -5.71265697e-01
4.75618780e-01 2.41671503e-01 7.25424826e-01 -6.04217410e-01
8.53593707e-01 4.88258392e-01 -1.82858750e-01 -6.37522042e-01
-7.93596566e-01 6.90839291e-02 -5.62969387e-01 -5.74901640e-01
1.07899010e+00 -8.26049924e-01 -5.60867846e-01 7.52436161e-01
-1.10074651e+00 -4.10956800e-01 -2.43932292e-01 4.87933755e-02
-3.85773718e-01 4.66615260e-01 -6.73269928e-01 -5.98740935e-01
-4.26252224e-02 -1.40137017e+00 1.34850848e+00 3.37204307e-01
-2.06640303e-01 -7.19813108e-01 -5.13465643e-01 7.57020712e-01
8.16042349e-02 4.16747332e-01 9.47724521e-01 2.36036256e-01
-6.59563839e-01 -1.70691311e-01 -2.82450140e-01 5.06425619e-01
1.45931080e-01 -7.01518282e-02 -8.72132480e-01 -1.39393017e-01
-1.48000062e-01 6.22686334e-02 6.48675501e-01 4.87231940e-01
1.07972848e+00 -2.30783969e-01 -3.38690042e-01 8.55447233e-01
1.37110996e+00 1.88660353e-01 1.11027956e+00 1.41167954e-01
1.06463420e+00 7.10303545e-01 7.77404785e-01 1.69510841e-01
4.32404906e-01 8.18671763e-01 5.06366670e-01 -6.88744426e-01
-7.23652720e-01 -7.77253926e-01 1.28182352e-01 5.77390432e-01
-1.99009448e-01 -3.35986495e-01 -2.76605844e-01 2.19682127e-01
-1.86812949e+00 -5.40803671e-01 -1.70090377e-01 2.02212715e+00
7.04725683e-01 6.42916337e-02 -2.06949618e-02 -6.18107840e-02
8.46653700e-01 3.27499539e-01 -4.84369040e-01 -1.32989138e-01
-2.79873490e-01 -9.76201594e-02 3.66351724e-01 5.32442868e-01
-1.10153091e+00 1.14479399e+00 5.61616325e+00 1.06698728e+00
-9.66638446e-01 1.37098759e-01 7.09284842e-01 1.42286584e-01
-4.95243877e-01 4.91984189e-02 -5.66844404e-01 7.28940547e-01
-3.16262305e-01 4.17807162e-01 4.58320081e-01 7.81903327e-01
1.67983025e-01 -1.20830983e-01 -8.19805026e-01 1.03043282e+00
2.63223112e-01 -8.61634314e-01 1.62125975e-01 3.26496661e-02
7.57550299e-01 -1.04005218e+00 1.12378590e-01 1.97796732e-01
-7.19739497e-02 -7.66238928e-01 1.28413165e+00 7.65597284e-01
1.01175535e+00 -5.68889081e-01 6.73180461e-01 2.23053377e-02
-1.22263443e+00 1.29184276e-01 -2.62152731e-01 3.29809368e-01
6.16543114e-01 6.65207684e-01 -7.25798011e-02 7.75190890e-01
5.74282229e-01 3.90600234e-01 -4.12042290e-01 6.28897548e-01
-5.52329719e-01 1.67940125e-01 -2.47461855e-01 4.31100667e-01
-2.22772971e-01 -4.55369055e-01 6.04676723e-01 1.07727969e+00
1.04092225e-01 3.20276678e-01 5.03018796e-01 1.09887803e+00
-4.12377194e-02 4.01686598e-03 -1.63670838e-01 2.68351227e-01
1.45390466e-01 1.25973284e+00 -8.16061974e-01 -4.65098619e-01
-2.77209669e-01 1.63204396e+00 -1.23793103e-01 5.80997884e-01
-1.21463406e+00 -4.36924249e-02 2.73063898e-01 5.27168214e-01
3.98009598e-01 2.71400306e-02 -4.12740022e-01 -1.11528969e+00
2.89920896e-01 -9.42182422e-01 4.44440404e-03 -1.07995582e+00
-1.25533044e+00 5.37398219e-01 1.93698499e-02 -1.13904607e+00
3.15054446e-01 -2.96687692e-01 -4.38197583e-01 8.29221189e-01
-1.06743956e+00 -1.84785533e+00 -7.05259860e-01 4.90428209e-01
7.99998283e-01 2.34363228e-01 4.55732495e-01 5.73990285e-01
-7.36104667e-01 6.22318149e-01 -3.92523497e-01 -1.07389145e-01
5.89410663e-01 -8.36439550e-01 2.23482132e-01 9.44768548e-01
-3.75162423e-01 4.81996268e-01 8.02596748e-01 -1.22523272e+00
-1.61513293e+00 -1.08399236e+00 5.77226520e-01 -1.30280614e-01
1.56888321e-01 -5.88879645e-01 -7.82032490e-01 5.48555315e-01
6.80473819e-02 5.71415387e-03 8.05002153e-02 -3.43327999e-01
-1.56876296e-01 -1.24857634e-01 -1.22295213e+00 8.92985106e-01
1.36970890e+00 -3.22945029e-01 -5.21174908e-01 2.66123414e-01
6.30585253e-01 -8.50841880e-01 -6.88852310e-01 5.83688259e-01
8.12250197e-01 -1.04323423e+00 1.31431115e+00 -1.15807764e-01
7.39522576e-01 -4.77039009e-01 -3.08068216e-01 -9.49805558e-01
-2.67217249e-01 -7.12492585e-01 -6.03775866e-03 1.42036414e+00
-1.88406736e-01 -2.69046992e-01 6.66705906e-01 8.67952406e-01
-2.16004416e-01 -8.38317215e-01 -5.90182126e-01 -5.02467752e-01
-5.37184358e-01 -3.22128534e-01 4.56415802e-01 8.08880866e-01
-5.67057431e-01 -7.28110000e-02 -9.01345074e-01 2.56786108e-01
9.18822825e-01 1.38845280e-01 9.29445028e-01 -6.25941396e-01
-3.76291513e-01 -6.74702153e-02 -6.14042059e-02 -1.37221169e+00
-2.39362210e-01 -4.72450525e-01 1.04872294e-01 -1.66879821e+00
1.65590435e-01 -4.55431402e-01 9.64229032e-02 5.19032598e-01
-4.25178945e-01 5.17557800e-01 4.77510393e-01 7.82699585e-02
-4.29104775e-01 6.71204567e-01 1.88767397e+00 -2.09935799e-01
-2.28127226e-01 -2.22870380e-01 -5.24511278e-01 9.60675895e-01
2.74547011e-01 -2.39009276e-01 -3.07505250e-01 -3.01238239e-01
-1.89299256e-01 1.80413365e-01 7.14285672e-01 -7.44096994e-01
-8.36535916e-02 -1.02543741e-01 6.60185099e-01 -6.60590529e-01
6.44706368e-01 -9.23291028e-01 6.52575433e-01 4.71590608e-01
-6.65365206e-03 -3.58462483e-01 1.44212395e-01 5.53137302e-01
4.52143475e-02 -7.35621750e-02 9.17790294e-01 -2.18497336e-01
-6.81256115e-01 1.60460353e-01 -1.87796783e-02 -3.94868165e-01
1.02492774e+00 -4.85043049e-01 -7.39577785e-02 -2.55095810e-01
-1.03330028e+00 7.11564198e-02 7.03484714e-01 5.67278445e-01
7.35163033e-01 -1.33603156e+00 -4.97184485e-01 5.33773541e-01
-2.11338550e-01 5.61093688e-02 8.52009416e-01 6.31148636e-01
-7.62893617e-01 -2.42157191e-01 -8.85791853e-02 -5.69856346e-01
-1.47248399e+00 7.46276319e-01 1.43206507e-01 -1.22526996e-01
-9.04394925e-01 6.75999105e-01 7.54226208e-01 -2.67589986e-01
3.12625319e-01 -1.71888635e-01 2.30713457e-01 -1.57040060e-01
2.11891592e-01 2.01150373e-01 -2.94928491e-01 -6.07898176e-01
-2.04868913e-01 8.72571111e-01 -6.80369213e-02 8.05883408e-02
1.03089976e+00 -4.29020673e-01 -6.81266636e-02 2.82233339e-02
8.72309506e-01 1.58414364e-01 -1.62965322e+00 -1.60342246e-01
-7.05145895e-01 -8.83742809e-01 -1.94061324e-01 -9.39828694e-01
-1.35877645e+00 5.42099059e-01 6.31180942e-01 -1.02070950e-01
1.29469192e+00 4.68236543e-02 1.11194897e+00 -4.38276201e-01
4.79397953e-01 -1.19948697e+00 2.49189690e-01 -9.71402079e-02
1.32383037e+00 -8.40587914e-01 2.95095474e-01 -1.28950095e+00
-8.68382394e-01 8.37797463e-01 6.27624631e-01 -2.33506888e-01
4.56882477e-01 1.93931431e-01 1.33774668e-01 -2.56538332e-01
1.04817282e-02 -7.76648968e-02 5.88281274e-01 6.57105863e-01
-1.35313347e-01 1.05658092e-01 -3.17546993e-01 7.58553863e-01
-1.27634063e-01 -1.25258729e-01 -3.08477301e-02 8.02234054e-01
3.06341890e-02 -1.14790809e+00 -6.75817668e-01 -1.87633540e-02
-1.62008986e-01 -3.79080363e-02 -1.77517593e-01 8.51316035e-01
7.10835338e-01 8.21200013e-01 -1.50095165e-01 -4.88873780e-01
7.30485499e-01 -2.13355690e-01 6.83763623e-01 -3.12602997e-01
-4.66785431e-01 7.24307895e-01 1.78452745e-01 -7.02925324e-01
-2.24249274e-01 -3.67293686e-01 -8.29570472e-01 -2.66464710e-01
-2.02608436e-01 -5.95020175e-01 6.72675967e-01 7.75447845e-01
1.88654989e-01 9.81123686e-01 3.01133484e-01 -1.12860417e+00
-5.44213392e-02 -5.77725470e-01 -7.39723206e-01 9.51958120e-01
-8.19445625e-02 -7.62172639e-01 -2.30973139e-01 4.00509447e-01] | [11.888586044311523, -0.8728317022323608] |
eba722f2-a54f-44cc-8d3e-5af99d7a9146 | yourtts-towards-zero-shot-multi-speaker-tts | 2112.02418 | null | https://arxiv.org/abs/2112.02418v4 | https://arxiv.org/pdf/2112.02418v4.pdf | YourTTS: Towards Zero-Shot Multi-Speaker TTS and Zero-Shot Voice Conversion for everyone | YourTTS brings the power of a multilingual approach to the task of zero-shot multi-speaker TTS. Our method builds upon the VITS model and adds several novel modifications for zero-shot multi-speaker and multilingual training. We achieved state-of-the-art (SOTA) results in zero-shot multi-speaker TTS and results comparable to SOTA in zero-shot voice conversion on the VCTK dataset. Additionally, our approach achieves promising results in a target language with a single-speaker dataset, opening possibilities for zero-shot multi-speaker TTS and zero-shot voice conversion systems in low-resource languages. Finally, it is possible to fine-tune the YourTTS model with less than 1 minute of speech and achieve state-of-the-art results in voice similarity and with reasonable quality. This is important to allow synthesis for speakers with a very different voice or recording characteristics from those seen during training. | ['Moacir Antonelli Ponti', 'Eren Gölge', 'Arnaldo Candido Junior', 'Christopher Shulby', 'Julian Weber', 'Edresson Casanova'] | 2021-12-04 | null | null | null | null | ['zero-shot-multi-speaker-tts'] | ['audio'] | [-1.51077479e-01 -5.20948768e-02 -8.36601928e-02 -4.04079407e-01
-1.72994256e+00 -5.43780982e-01 6.24128580e-01 -2.87844002e-01
-3.45403820e-01 3.15736264e-01 3.48150164e-01 -5.79828799e-01
2.32499033e-01 -1.36296570e-01 -3.47910404e-01 -3.82445961e-01
2.67579108e-01 8.10894787e-01 3.17616731e-01 -6.57681644e-01
-4.74350333e-01 1.93103909e-01 -1.71220517e+00 2.43509144e-01
6.37703538e-01 5.12600124e-01 5.40615737e-01 1.24134290e+00
-3.61305356e-01 3.76660079e-01 -6.36138141e-01 -3.74401242e-01
1.55476376e-01 -5.91428936e-01 -8.53969097e-01 -1.78936020e-01
8.37386370e-01 2.47463267e-02 -8.53379294e-02 6.49473548e-01
1.23414898e+00 4.56410974e-01 4.82915401e-01 -1.02442753e+00
-4.79686022e-01 6.94211423e-01 -1.63270663e-02 5.08461654e-01
5.79553068e-01 3.06620151e-01 1.12193811e+00 -9.75407660e-01
5.63674867e-01 1.57207465e+00 7.30407059e-01 7.81431377e-01
-1.20912623e+00 -6.11394882e-01 -2.70626962e-01 1.09483622e-01
-1.31390977e+00 -1.42291367e+00 1.58633932e-01 -2.86080241e-01
1.35096097e+00 5.00858963e-01 3.75356495e-01 1.15021920e+00
-1.65215552e-01 5.44554889e-01 8.99874151e-01 -8.10502768e-01
-9.43015143e-03 2.39076599e-01 -1.66895270e-01 4.21650201e-01
-7.62378693e-01 1.42206788e-01 -8.07327271e-01 6.65831715e-02
2.32989892e-01 -6.17654800e-01 -1.44972399e-01 1.60354510e-01
-1.45602858e+00 8.69837821e-01 -2.63067424e-01 7.32835650e-01
-2.86886673e-02 -9.14997086e-02 7.02778995e-01 8.29951525e-01
5.36224544e-01 2.91467428e-01 -6.16581142e-01 -7.63568580e-01
-1.47472405e+00 -1.94592699e-01 9.01030064e-01 1.28336954e+00
3.70848119e-01 7.83962250e-01 -2.61425346e-01 1.46678591e+00
1.21516049e-01 8.15118730e-01 8.52099478e-01 -8.81614804e-01
4.84493166e-01 -5.94487071e-01 -3.73516500e-01 2.49054655e-01
7.13979676e-02 -1.95764631e-01 -2.57197112e-01 1.02609813e-01
3.04042906e-01 -1.72323376e-01 -1.03878081e+00 1.66055095e+00
2.97854215e-01 2.14194700e-01 3.30159962e-01 5.65025747e-01
9.21149909e-01 1.04066992e+00 -2.11807758e-01 -5.32779276e-01
1.52152884e+00 -1.27629685e+00 -9.11935747e-01 -1.66479155e-01
4.58053678e-01 -1.38551664e+00 1.85472631e+00 1.46408513e-01
-1.12697446e+00 -6.99992359e-01 -7.82331884e-01 -1.37075633e-01
-2.81547338e-01 -3.22224885e-01 1.74803510e-01 1.12418437e+00
-1.19025075e+00 4.78210926e-01 -5.82526863e-01 -7.17613339e-01
-3.09312314e-01 1.88408643e-01 -1.49655223e-01 1.29216596e-01
-1.37966657e+00 1.06975782e+00 -3.04381549e-02 -6.30303323e-01
-1.02264094e+00 -1.04314303e+00 -9.32197630e-01 3.88118848e-02
2.59357303e-01 -3.15215588e-01 2.12814355e+00 -5.68976164e-01
-2.14284253e+00 7.09910691e-01 -5.18300474e-01 -2.85585761e-01
3.32848132e-01 1.07399477e-02 -1.06932449e+00 -1.01990998e-01
2.30221286e-01 5.75764477e-01 9.37125266e-01 -7.99608767e-01
-7.42930591e-01 3.68571281e-02 -5.64644992e-01 3.19643080e-01
-2.76579231e-01 6.22754335e-01 -3.81620675e-01 -4.90375429e-01
-4.16885138e-01 -9.32954371e-01 2.92849898e-01 -4.82262313e-01
-2.11827472e-01 -8.91873688e-02 9.33445990e-01 -8.40779603e-01
9.33294654e-01 -2.18351650e+00 6.66959584e-02 -3.93243879e-01
-4.88618672e-01 3.48929226e-01 -3.92473072e-01 4.70350832e-01
4.44282740e-02 4.03040349e-02 2.86885686e-02 -7.29028463e-01
1.05080403e-01 2.61513144e-01 -3.94597918e-01 1.47282541e-01
-2.91543871e-01 6.86791003e-01 -8.30756545e-01 -6.41965210e-01
4.53413308e-01 5.26327312e-01 -2.92121977e-01 3.86773914e-01
7.99882188e-02 5.22282660e-01 3.39744031e-01 5.70983410e-01
4.96504381e-02 6.19480729e-01 -1.16466902e-01 1.70134544e-01
-5.25868475e-01 5.86706877e-01 -1.05452275e+00 1.98209953e+00
-1.09050596e+00 9.09147620e-01 2.76805848e-01 -1.64054856e-01
6.57411575e-01 1.14213693e+00 2.60569453e-01 -6.16939306e-01
1.56533360e-01 5.04984856e-01 1.85344130e-01 -2.15496168e-01
6.29127681e-01 -7.66980290e-01 -6.38216659e-02 5.12238741e-01
8.10958147e-01 -8.82636309e-01 1.65571868e-01 8.63540452e-03
4.49425071e-01 -2.15787247e-01 1.55749887e-01 -4.16160464e-01
3.19595486e-01 -2.98198611e-01 2.67359793e-01 4.81842399e-01
-1.99755162e-01 7.55336642e-01 -1.75749674e-01 2.75155723e-01
-1.18404543e+00 -1.28015077e+00 -1.60212025e-01 1.77721381e+00
-5.54396868e-01 -3.81950587e-01 -8.13472331e-01 4.98487055e-03
-2.65751421e-01 1.27493584e+00 -9.94665474e-02 1.11659750e-01
-3.73766541e-01 -6.40398636e-02 1.17362761e+00 1.37286350e-01
-2.80714035e-01 -7.94621289e-01 -2.47678217e-02 4.25162852e-01
-5.56754470e-01 -1.34068382e+00 -1.09775043e+00 4.03343856e-01
-5.78083277e-01 -3.58779907e-01 -1.18709970e+00 -9.89466727e-01
-3.29765052e-01 2.02249393e-01 1.02914882e+00 -7.64935315e-01
-2.22516209e-01 4.09300864e-01 -2.42966056e-01 -5.91917157e-01
-1.13802838e+00 3.78034621e-01 7.50921488e-01 -2.23977387e-01
5.17418869e-02 -4.10741210e-01 1.05768353e-01 4.01359469e-01
-3.34318578e-01 -3.82970631e-01 3.89720760e-02 6.21343315e-01
3.78747493e-01 -3.52951556e-01 8.30096900e-01 -3.70387942e-01
7.97554195e-01 -1.93654850e-01 -3.34634125e-01 3.14847946e-01
-6.03700817e-01 4.05947194e-02 7.51960397e-01 -6.45902634e-01
-1.23963702e+00 -2.21582130e-01 -6.15650833e-01 -7.64150977e-01
-1.60090640e-01 -1.71803400e-01 -1.09902248e-01 4.96094227e-02
7.69434392e-01 2.64415413e-01 1.65778603e-02 -8.35368097e-01
9.81106460e-01 1.35535669e+00 6.92839086e-01 -4.63499099e-01
5.96392274e-01 -1.22330211e-01 -7.22032964e-01 -1.37831736e+00
-4.30753976e-01 -7.24684358e-01 -7.13710546e-01 -1.50080279e-01
7.33117044e-01 -1.13067913e+00 -4.70589221e-01 5.30119300e-01
-9.75118577e-01 -5.11543810e-01 -7.42630720e-01 6.74729228e-01
-8.26508641e-01 1.85702831e-01 -6.89695716e-01 -9.41834748e-01
-5.45216203e-01 -1.16058791e+00 1.14246392e+00 -7.86743686e-02
-4.48725551e-01 -1.02732861e+00 4.48629677e-01 3.95354569e-01
6.13478780e-01 -7.17991889e-01 7.25974977e-01 -7.30201602e-01
1.44738764e-01 1.84411615e-01 4.72548962e-01 4.41197366e-01
4.73684758e-01 1.94186851e-01 -1.46425033e+00 -4.89666522e-01
-3.19998525e-02 -3.95207793e-01 4.51782376e-01 3.96503389e-01
3.06428373e-01 -8.13817829e-02 1.60191834e-01 5.16872287e-01
8.03099155e-01 2.56128371e-01 2.14463174e-01 -9.83981565e-02
5.78720391e-01 4.83008385e-01 6.82006538e-01 2.31903747e-01
4.22862798e-01 9.96357739e-01 -1.84125870e-01 -1.88323021e-01
-7.46041775e-01 -1.22580878e-01 7.17699647e-01 1.77696192e+00
2.01836780e-01 -2.03358397e-01 -8.76369834e-01 9.03062999e-01
-1.19324398e+00 -1.05206943e+00 -1.55872598e-01 2.39602399e+00
1.11283779e+00 -9.82662514e-02 4.34432119e-01 1.60512701e-01
1.04711008e+00 4.62017357e-01 -1.85225189e-01 -1.13082623e+00
-1.49942309e-01 4.91154879e-01 2.72383869e-01 9.79370654e-01
-6.57495260e-01 1.50107110e+00 7.47895145e+00 1.11144304e+00
-1.43431485e+00 7.80525744e-01 1.35160968e-01 -6.44589186e-01
-2.85091072e-01 -1.15392946e-01 -1.03054595e+00 2.65461057e-01
1.90226674e+00 -5.53330123e-01 9.64314282e-01 6.17026985e-01
2.76098430e-01 3.46280545e-01 -1.04019976e+00 1.02173281e+00
2.48781130e-01 -9.43745911e-01 -1.75165311e-01 -7.40348455e-03
6.29850090e-01 6.22583270e-01 1.17509691e-02 7.16922164e-01
3.65488857e-01 -6.66324973e-01 1.07888412e+00 -1.34373128e-01
1.71480477e+00 -7.91359305e-01 1.45399660e-01 2.53332376e-01
-1.49401474e+00 1.59781545e-01 -2.84067482e-01 3.39743823e-01
5.29686511e-01 1.62470028e-01 -1.27758992e+00 5.21463335e-01
4.80293304e-01 1.51021451e-01 -2.04526439e-01 7.53309846e-01
8.74717832e-02 8.86128843e-01 -4.29102808e-01 1.07153244e-01
2.72855699e-01 2.15691939e-01 7.75744915e-01 1.54527676e+00
7.50534296e-01 -2.47371092e-01 3.65852402e-03 1.40437275e-01
-1.82617269e-02 5.87212324e-01 -5.96670508e-01 -1.14535600e-01
7.29719758e-01 1.11149657e+00 -2.11930692e-01 -5.80113351e-01
-4.04830456e-01 1.19971395e+00 2.39950884e-02 2.01001316e-01
-5.68012476e-01 -6.25377476e-01 9.25364435e-01 -1.54214855e-02
3.23916286e-01 -2.24731520e-01 6.40210062e-02 -1.14231229e+00
-4.04457420e-01 -1.08482194e+00 2.26457179e-01 -7.71553397e-01
-9.97185826e-01 1.01094520e+00 -2.61730731e-01 -1.28662419e+00
-9.35903072e-01 -2.67573088e-01 -8.38977993e-01 1.05654919e+00
-1.38535690e+00 -1.12810516e+00 2.63863564e-01 8.57994974e-01
1.23707664e+00 -6.44320130e-01 1.32046103e+00 5.21737337e-01
-1.88789159e-01 7.15130150e-01 4.14804846e-01 -3.49075437e-01
1.29963541e+00 -1.22724128e+00 1.07002974e+00 8.20236206e-01
5.42464912e-01 3.64659697e-01 8.62636924e-01 -3.29048306e-01
-1.36189687e+00 -9.46447611e-01 1.20788300e+00 -3.46510679e-01
8.35338175e-01 -4.11851853e-01 -7.84156382e-01 6.52724028e-01
6.07662797e-01 -1.48613974e-01 1.02054870e+00 6.26005113e-01
-4.28633869e-01 -1.73980206e-01 -1.04824519e+00 6.09994650e-01
8.57005715e-01 -1.13487232e+00 -9.64315236e-01 2.34119281e-01
1.24387085e+00 -2.71071404e-01 -9.45615649e-01 -2.02695757e-01
6.40254557e-01 -5.18957019e-01 7.53279924e-01 -4.79810119e-01
-2.70976216e-01 4.73756939e-02 -4.80926096e-01 -1.98285568e+00
-2.95430541e-01 -1.10701263e+00 2.21496522e-01 1.51994395e+00
6.74524367e-01 -4.80653644e-01 -2.15830281e-03 3.46787348e-02
-6.96454585e-01 5.68212755e-02 -1.48343599e+00 -1.41784561e+00
2.10867897e-01 -7.37595022e-01 6.72288954e-01 1.06441092e+00
1.15284473e-01 6.30321383e-01 -7.13092983e-01 -1.28474250e-01
5.50831258e-01 -2.96944231e-01 7.13862658e-01 -1.16663396e+00
-4.86405045e-01 -2.48826742e-01 1.04347356e-02 -6.11794412e-01
3.22687954e-01 -1.11272156e+00 3.93672377e-01 -1.32133949e+00
-2.86750913e-01 -5.95848188e-02 -2.71455526e-01 4.61760849e-01
7.36863092e-02 2.74418205e-01 6.11737788e-01 6.61169812e-02
-2.29959875e-01 6.93582296e-01 1.08271337e+00 -8.48632306e-02
-4.46752995e-01 -8.89378507e-03 -4.45014238e-01 3.71102780e-01
5.99807680e-01 -6.11037135e-01 -4.76714075e-01 -5.03401160e-01
-7.96642303e-01 3.63904595e-01 -3.04360241e-01 -1.01223695e+00
2.77754456e-01 -3.18395011e-02 -4.14317757e-01 -4.21219856e-01
9.00020003e-01 -3.38391513e-01 1.35456920e-01 2.66980231e-01
-1.25015408e-01 -1.19502239e-01 3.61418813e-01 1.16029054e-01
-2.87154853e-01 -1.69420019e-01 1.11765945e+00 -1.59203470e-01
-5.82183778e-01 7.06061274e-02 -8.69825721e-01 4.77085948e-01
5.53603113e-01 -1.77015767e-01 -2.48204004e-02 -4.79452938e-01
-8.06774437e-01 -1.12091592e-02 2.75956780e-01 9.77336764e-01
1.56822339e-01 -1.41007960e+00 -9.40622270e-01 3.03602010e-01
4.07344729e-01 -6.80392683e-01 1.75917163e-01 6.12486243e-01
-3.35832909e-02 6.87410951e-01 -1.04947269e-01 -6.96208715e-01
-1.73910141e+00 1.44232973e-01 4.07635957e-01 2.62631595e-01
-4.62362826e-01 9.41596270e-01 -2.31593028e-01 -9.59720850e-01
1.61921933e-01 -2.45129570e-01 2.80474901e-01 3.92899334e-01
5.54502547e-01 4.23655927e-01 3.91190231e-01 -1.02298677e+00
-4.99622375e-01 5.11964560e-01 -2.72375718e-03 -1.11444843e+00
9.97343242e-01 -3.95103097e-01 4.83961880e-01 1.41767240e+00
1.22434127e+00 4.31904733e-01 -9.02592719e-01 -3.84444207e-01
-2.79773951e-01 -4.13610756e-01 3.11868608e-01 -7.25201905e-01
-5.90141952e-01 1.18010783e+00 7.46704578e-01 -6.53327852e-02
5.38771093e-01 7.33230785e-02 1.22939157e+00 4.69133228e-01
3.14037144e-01 -1.40286899e+00 -2.10785106e-01 9.11995113e-01
6.35183632e-01 -1.17055738e+00 -5.74597538e-01 -8.78762305e-02
-1.05275559e+00 9.55170512e-01 4.09943499e-02 6.96444929e-01
4.11780447e-01 4.04628724e-01 7.07137644e-01 1.99222863e-01
-9.56728876e-01 -5.05727410e-01 3.53959352e-01 6.69683158e-01
7.88591385e-01 4.01990712e-01 2.48994619e-01 3.54717106e-01
-8.00184608e-01 -3.63073051e-01 3.24536532e-01 5.07755458e-01
-6.40171707e-01 -1.26821959e+00 -5.44654429e-01 5.77051751e-02
-5.73498547e-01 -6.63412631e-01 -2.92819291e-01 4.15807784e-01
-3.55255455e-01 1.38998461e+00 3.82173210e-02 -3.18873525e-01
5.45464218e-01 8.10568631e-01 3.91006082e-01 -1.04097652e+00
-8.91961098e-01 4.63395119e-01 3.88271660e-01 -3.25763613e-01
2.25440904e-01 -9.38053548e-01 -1.13393629e+00 -4.24245119e-01
-5.67520678e-01 3.63036633e-01 1.13574731e+00 9.54849243e-01
1.83581859e-01 7.37728715e-01 8.85428250e-01 -8.56837332e-01
-6.10190511e-01 -1.23611915e+00 -7.85783529e-01 -4.21586856e-02
4.64629471e-01 -2.45140567e-01 -3.88109088e-01 1.27535909e-01] | [14.778468132019043, 6.736785888671875] |
07e8e7a4-b8c7-49e4-92cc-f107f2a735b2 | datsing-data-augmented-time-series | null | null | https://dl.acm.org/doi/abs/10.1145/3340531.3412155 | https://dl.acm.org/doi/pdf/10.1145/3340531.3412155 | DATSING: Data Augmented Time Series Forecasting with Adversarial Domain Adaptation | Due to the high temporal uncertainty and low signal-to-noise ratio, transfer learning for univariate time series forecasting remains a challenging task. In addition, data scarcity, which is commonly encountered in business forecasting, further limits the application of conventional transfer learning protocols. In this work, we have developed, DATSING, a transfer learning-based framework that effectively leverages cross-domain time series latent representations to augment target domain forecasting. In particular, we aim to transfer domain-invariant feature representations from a pre-trained stacked deep residual network to the target domains, so as to assist the prediction of each target time series. To effectively avoid noisy feature representations, we propose a two-phased framework which first clusters similar mixed domains time series data and then performs a fine-tuning procedure with domain adversarial regularization to achieve better out-of-sample generalization. Extensive experiments with real-world datasets have demonstrated that our method significantly improves the forecasting performance of the pre-trained model. DATSING has the unique potential to empower forecasting practitioners to unleash the power of cross-domain time series data. | ['Chengcheng Bai', 'Mingjian Tang', 'Hailin Hu'] | 2020-10-19 | null | null | null | acm-international-conference-on-information-4 | ['univariate-time-series-forecasting'] | ['time-series'] | [ 2.33718663e-01 -3.21701467e-01 -1.48377508e-01 -5.16577065e-01
-9.20555949e-01 -7.80395031e-01 6.98697031e-01 -3.37486863e-01
1.99344143e-01 7.67079592e-01 1.62973076e-01 -6.75325334e-01
-2.25142568e-01 -6.91753566e-01 -9.32901502e-01 -7.99442410e-01
-3.18827182e-01 2.04282314e-01 -2.78916746e-01 -2.07992330e-01
-9.92969051e-03 4.96694058e-01 -1.19199729e+00 3.49236131e-01
1.19814706e+00 1.32769263e+00 -1.46165922e-01 1.04337901e-01
-3.80715691e-02 1.02302158e+00 -5.23679137e-01 -8.37954432e-02
3.94667268e-01 -2.45576993e-01 -3.20650250e-01 8.93059447e-02
-4.21182476e-02 -4.90370572e-01 -4.22722012e-01 6.65141761e-01
2.17105001e-01 3.25632900e-01 8.59058321e-01 -1.49881864e+00
-7.83830822e-01 3.16285849e-01 -4.04154390e-01 2.92065024e-01
-4.48602848e-02 1.30621612e-01 6.21914804e-01 -6.11106753e-01
2.35772148e-01 1.04086328e+00 8.39711547e-01 4.47307020e-01
-1.37219238e+00 -1.28790927e+00 3.19829702e-01 -1.26261190e-01
-9.84375298e-01 -1.14081368e-01 1.33142257e+00 -7.13376284e-01
6.00996733e-01 -3.66915278e-02 1.75928727e-01 1.49393809e+00
3.55763167e-01 7.19173908e-01 1.31130159e+00 -1.17724642e-01
4.08050030e-01 -1.66574102e-02 -2.44412884e-01 -1.36499941e-01
-3.77888918e-01 2.98107296e-01 -2.78752357e-01 -2.59104788e-01
6.68261290e-01 5.81910372e-01 -6.61085248e-02 -3.91292930e-01
-1.19807780e+00 8.09537590e-01 5.82915962e-01 3.45476598e-01
-7.29709029e-01 -1.23797871e-01 6.13875329e-01 6.03475213e-01
1.25807881e+00 1.46857172e-01 -7.97100663e-01 5.02820089e-02
-8.78027201e-01 1.91020548e-01 7.15345025e-01 6.67600393e-01
4.50596720e-01 4.50298458e-01 1.09561868e-01 5.85976958e-01
-2.96758823e-02 6.68083608e-01 6.86342120e-01 -6.97648048e-01
7.13158906e-01 3.89746159e-01 2.01405555e-01 -9.81073022e-01
-1.27211273e-01 -5.59613883e-01 -1.02030826e+00 -4.21134457e-02
3.55312347e-01 -3.87946218e-01 -8.82772565e-01 1.74187505e+00
1.13753356e-01 7.47904003e-01 3.46157223e-01 7.74417400e-01
3.32270920e-01 1.05802608e+00 3.93252283e-01 -1.90701231e-01
7.76304364e-01 -6.36836648e-01 -5.73552847e-01 -1.82679921e-01
5.38978577e-01 -5.32647371e-01 9.12599444e-01 1.02373511e-01
-5.57089031e-01 -5.33174157e-01 -9.41660523e-01 3.80869508e-01
-5.77534318e-01 -8.44925418e-02 4.81046885e-01 2.93948770e-01
-4.10706669e-01 7.73541927e-01 -9.31211710e-01 -2.47030482e-02
6.30084038e-01 2.84554750e-01 -2.91221648e-01 -2.43461683e-01
-1.50560999e+00 7.22337246e-01 4.86475915e-01 -8.53068158e-02
-7.16886044e-01 -1.24792027e+00 -6.25222504e-01 2.19510794e-01
1.30667970e-01 -4.11954850e-01 1.21530640e+00 -1.10990620e+00
-1.47397327e+00 3.51021022e-01 6.43798113e-02 -6.92644596e-01
4.58041638e-01 -1.68331325e-01 -1.02145910e+00 -1.52922750e-01
1.34192601e-01 2.64132440e-01 1.16716290e+00 -1.04902112e+00
-4.93689567e-01 -3.77327174e-01 -5.02386808e-01 -1.31553039e-01
-5.88370442e-01 -1.81225702e-01 3.44458729e-01 -1.10070455e+00
4.66611758e-02 -8.44131112e-01 -2.03241587e-01 -3.85400146e-01
1.06860489e-01 -2.33212188e-01 1.13639581e+00 -9.42123234e-01
9.31946516e-01 -2.44149494e+00 -1.04590818e-01 3.40343446e-01
-6.12255465e-03 1.44635007e-01 -1.66958347e-01 6.52507424e-01
-2.65603036e-01 -2.17738718e-01 -2.28653193e-01 -1.61346883e-01
-3.39786895e-02 3.95181209e-01 -1.27841926e+00 4.03755188e-01
5.65112472e-01 8.14759254e-01 -8.85562003e-01 1.01235300e-01
2.27256104e-01 5.40835083e-01 -2.75815636e-01 3.69564325e-01
-4.19991672e-01 1.01292288e+00 -5.66965580e-01 5.39897859e-01
8.32729161e-01 -4.26934063e-01 9.75699350e-02 1.28722638e-01
1.04346827e-01 3.02501589e-01 -6.37980938e-01 1.26668918e+00
-6.13604605e-01 3.96782964e-01 -2.86514223e-01 -1.38973498e+00
1.39199543e+00 3.79335880e-01 8.53726149e-01 -9.11326230e-01
7.38237565e-03 4.64639038e-01 -2.05756202e-01 -1.96399450e-01
1.62633196e-01 -4.20706004e-01 -2.24097654e-01 2.17558742e-01
-9.13925543e-02 3.71677391e-02 -4.38736469e-01 -3.12348545e-01
7.38930762e-01 3.40707570e-01 2.14009546e-02 -9.30234194e-02
3.60628575e-01 2.18632519e-02 6.43525362e-01 2.55430609e-01
-6.35050163e-02 3.60159904e-01 3.94281983e-01 -5.80657363e-01
-1.06186831e+00 -1.11854708e+00 -1.25729173e-01 1.30642605e+00
-3.13440144e-01 2.77168304e-01 -3.08732331e-01 -7.48609900e-01
4.66591448e-01 9.50380385e-01 -6.97138965e-01 -3.78894329e-01
-5.26611269e-01 -4.34396982e-01 3.89660865e-01 8.27583492e-01
2.02593461e-01 -9.96519446e-01 -1.25836417e-01 5.68218291e-01
-9.05279890e-02 -1.02439106e+00 -3.93994242e-01 2.82523602e-01
-1.14523768e+00 -6.44903719e-01 -1.07182288e+00 -7.28580952e-01
4.66657549e-01 1.78460941e-01 1.00285423e+00 -7.36616910e-01
3.68096858e-01 1.96455643e-01 -3.22884172e-01 -6.57098234e-01
-4.46249187e-01 1.84704110e-01 4.96848859e-02 2.96226114e-01
7.51953602e-01 -9.35911655e-01 -4.61713195e-01 3.53325248e-01
-9.88141418e-01 -3.77736121e-01 3.81314158e-01 1.14788830e+00
3.37460607e-01 8.40310007e-02 1.37732637e+00 -8.20884347e-01
6.75396681e-01 -1.02267420e+00 -7.95286715e-01 2.04006150e-01
-6.43301785e-01 -2.09234342e-01 1.06546950e+00 -9.41185296e-01
-1.02388704e+00 -1.25252828e-01 4.08719003e-01 -9.56870317e-01
1.54319024e-02 8.15539837e-01 8.90468433e-02 2.34563172e-01
6.57060266e-01 4.77786601e-01 3.18664402e-01 -5.12192249e-01
3.12062562e-01 7.54649282e-01 6.13869429e-01 -5.63586056e-01
9.91699159e-01 3.60869735e-01 -1.84435397e-01 -3.81353229e-01
-8.01785052e-01 -3.56576532e-01 -5.14445782e-01 -5.79338800e-03
5.37509859e-01 -1.36839855e+00 -4.27829891e-01 4.32111025e-01
-8.00037324e-01 -4.41036850e-01 -2.45852247e-01 5.53314745e-01
-6.05413675e-01 -8.61453339e-02 -3.30120564e-01 -8.60203385e-01
-2.77203113e-01 -6.99555755e-01 8.88259351e-01 -3.31049226e-02
-2.45982949e-02 -1.44073546e+00 -1.86513993e-03 1.96458519e-01
7.12965965e-01 5.46423137e-01 8.36264193e-01 -1.12034547e+00
-4.03722495e-01 -4.55780029e-01 -6.39216155e-02 6.69383645e-01
2.78721899e-01 -4.50727969e-01 -9.81480360e-01 -3.95898044e-01
1.05955645e-01 -2.19543323e-01 6.82739139e-01 3.23066562e-01
1.36821425e+00 -2.99611211e-01 -2.63310999e-01 6.21793032e-01
9.67190146e-01 5.62911272e-01 4.70565796e-01 5.00252724e-01
4.68535870e-01 5.99048615e-01 6.95227027e-01 6.95949435e-01
3.06641012e-01 1.83430582e-01 1.74186215e-01 -1.01982966e-01
4.50740457e-01 -5.78273892e-01 3.58046830e-01 8.98532927e-01
2.27821589e-01 -1.14661410e-01 -1.09153557e+00 6.95080996e-01
-1.78722978e+00 -1.04135633e+00 4.61979002e-01 1.96916842e+00
6.35306299e-01 5.32168560e-02 1.68800548e-01 9.21316147e-02
6.02087677e-01 2.50273943e-01 -9.14759576e-01 -3.89508605e-02
3.66755500e-02 1.17028534e-01 5.33013821e-01 -2.22534791e-01
-1.35290408e+00 7.14714468e-01 5.72722054e+00 6.83180928e-01
-1.61017966e+00 -2.60086693e-02 7.92451084e-01 2.06408501e-01
-3.38550776e-01 -2.43872970e-01 -3.25321555e-01 8.15045714e-01
1.32926786e+00 -6.19335234e-01 5.76119244e-01 1.14626431e+00
1.66574344e-01 7.01224267e-01 -1.21157575e+00 6.35967076e-01
-2.16813445e-01 -1.22623289e+00 -1.95225507e-01 1.28695041e-01
1.01156235e+00 1.23644814e-01 5.83557963e-01 8.98972452e-01
2.85814345e-01 -9.97496009e-01 3.12294841e-01 5.03364623e-01
8.45067084e-01 -9.02297139e-01 6.70503378e-01 6.11967862e-01
-9.46427345e-01 -3.24284434e-01 -2.36566573e-01 -1.41367361e-01
-1.98192485e-02 5.63133597e-01 -9.61361289e-01 6.77832305e-01
6.74006939e-01 1.17170548e+00 5.07006831e-02 5.54129779e-01
3.67709339e-01 9.66749251e-01 -2.32624546e-01 3.18620920e-01
2.13847280e-01 -2.97796428e-01 2.79797465e-01 7.59195089e-01
6.21667504e-01 4.34460416e-02 4.89514619e-01 7.51813829e-01
-2.01818332e-01 -8.07925165e-02 -8.70801866e-01 -4.29028571e-01
5.98640561e-01 6.67024553e-01 -9.73105505e-02 -2.78772354e-01
-5.23003757e-01 7.20052660e-01 6.71810657e-02 6.82462633e-01
-8.37684333e-01 -2.55320072e-01 6.21184051e-01 9.71144438e-02
6.37388885e-01 -1.71553418e-01 -3.67360055e-01 -1.35179162e+00
1.45447522e-01 -1.04072738e+00 3.68370414e-01 -5.21000266e-01
-1.96029830e+00 5.74999213e-01 -1.64024010e-01 -1.95747149e+00
-7.39560425e-01 -4.01400268e-01 -6.75832570e-01 1.28102767e+00
-1.77818525e+00 -1.46821415e+00 -2.38027405e-02 6.92615092e-01
3.67014408e-01 -6.54129088e-01 7.53176212e-01 3.00904572e-01
-2.95289487e-01 5.60747385e-01 9.43388343e-01 2.52738327e-01
7.67337203e-01 -9.24418211e-01 5.05914211e-01 6.35845959e-01
-4.07291472e-01 5.51872313e-01 4.30336893e-01 -5.89483500e-01
-1.21877480e+00 -1.64146543e+00 6.12592816e-01 -2.67302066e-01
1.03969800e+00 -2.56455511e-01 -1.46317005e+00 1.02483428e+00
-1.96894675e-01 2.47400999e-01 8.20830524e-01 1.83325544e-01
-7.13490903e-01 -5.84077239e-01 -1.16217136e+00 3.01865518e-01
4.12644684e-01 -9.26826835e-01 -8.30993652e-01 3.34535033e-01
8.12683463e-01 -3.20151359e-01 -1.22559512e+00 5.56097925e-01
4.47401226e-01 -4.03429419e-01 1.06850517e+00 -9.43068624e-01
6.46776319e-01 5.58637530e-02 -3.75000894e-01 -1.65208805e+00
-3.13492864e-01 -7.08265126e-01 -3.36277217e-01 1.47130549e+00
2.04642281e-01 -1.20432687e+00 5.02036691e-01 6.56967282e-01
-1.80477612e-02 -4.29407716e-01 -9.64908421e-01 -1.16976476e+00
7.46681154e-01 -3.01521063e-01 9.72656786e-01 1.34728694e+00
-1.02625780e-01 -2.82388809e-03 -6.25131786e-01 2.78387159e-01
4.79253083e-01 4.65631306e-01 8.34243774e-01 -1.42344975e+00
-1.81201342e-02 -1.08675003e-01 -1.80056974e-01 -9.79176283e-01
5.71864545e-01 -7.87745059e-01 -1.66229352e-01 -9.43701565e-01
-3.89652252e-01 -6.12856805e-01 -7.81319797e-01 3.45340729e-01
-2.11714227e-02 -8.66866335e-02 1.63126245e-01 4.90947485e-01
-1.97570190e-01 8.30213785e-01 1.27612782e+00 -1.19531080e-01
-3.24036270e-01 3.67640823e-01 -5.55945873e-01 2.86538929e-01
9.86962736e-01 -4.84757990e-01 -5.51732421e-01 -1.57922387e-01
-4.02968526e-01 4.98099416e-01 4.67745960e-01 -8.47329140e-01
7.17747782e-04 -3.90439808e-01 5.75110137e-01 -6.94639146e-01
2.84136027e-01 -1.20403779e+00 1.13055229e-01 2.65613288e-01
-4.90626246e-01 -7.54532367e-02 4.24608469e-01 8.19774628e-01
-5.91622174e-01 4.31448638e-01 3.87315243e-01 2.25066513e-01
-6.85088992e-01 3.26081216e-01 -9.77595299e-02 2.83437520e-02
1.15022397e+00 6.01203926e-02 -2.06381559e-01 -5.06475925e-01
-5.86920261e-01 3.53707135e-01 3.42473611e-02 6.86014116e-01
2.31196180e-01 -1.62288380e+00 -7.44857490e-01 4.93278354e-01
2.10206494e-01 -9.94982123e-02 5.15174448e-01 7.12573111e-01
1.19345978e-01 4.46304053e-01 -3.07787329e-01 -6.78275526e-01
-5.24166822e-01 8.96000803e-01 1.66666225e-01 -2.53035277e-01
-6.79465711e-01 4.06025946e-01 3.30780864e-01 -7.73846865e-01
4.90581430e-02 -4.86079603e-01 8.15189350e-03 6.77532032e-02
4.25276577e-01 2.74184704e-01 8.87486562e-02 -4.28292632e-01
-3.01903278e-01 1.91427216e-01 -2.07598075e-01 1.51511833e-01
1.75797546e+00 -1.44203398e-02 2.65757799e-01 6.40045762e-01
1.48182988e+00 -3.01181823e-01 -1.79280770e+00 -6.13846779e-01
2.14555249e-01 -4.87067521e-01 3.06107868e-02 -8.62674117e-01
-8.81076872e-01 8.96537304e-01 4.75943863e-01 5.02878368e-01
1.40821695e+00 -3.62612128e-01 1.08037400e+00 1.48036823e-01
1.62342533e-01 -8.23400021e-01 5.84389828e-03 5.86284995e-01
9.51905727e-01 -1.51116335e+00 -4.41316903e-01 -3.70176733e-02
-8.62502992e-01 1.07973182e+00 3.15834016e-01 -4.85126495e-01
8.28545988e-01 1.37024283e-01 2.92093128e-01 3.56040955e-01
-8.90341759e-01 2.23119915e-01 6.13874674e-01 7.76749790e-01
3.04598421e-01 2.10279822e-01 3.47134620e-01 9.52699542e-01
8.14583339e-03 5.27042449e-01 -3.20550948e-02 7.25482762e-01
-1.96331162e-02 -9.76368129e-01 -5.11757672e-01 4.16643530e-01
-5.08560061e-01 1.62340492e-01 1.39112025e-01 8.50678027e-01
-2.05427811e-01 8.53750050e-01 2.35357851e-01 -4.32996809e-01
3.97837907e-01 4.46815103e-01 -2.40540907e-01 -7.86435157e-02
-5.31244636e-01 2.42358655e-01 -3.34031790e-01 -2.28413969e-01
-2.99056411e-01 -7.38989949e-01 -8.71550977e-01 -3.71364892e-01
1.52886167e-01 -9.43590254e-02 5.48464239e-01 1.01215827e+00
7.36166358e-01 6.99649274e-01 1.12023401e+00 -1.02114797e+00
-1.31757319e+00 -9.88461554e-01 -6.04938984e-01 7.41248965e-01
7.61308849e-01 -7.09856212e-01 -4.74332362e-01 1.83442935e-01] | [7.174508094787598, 2.994499444961548] |
78371278-4c9e-4012-b1d8-f7fa0b9911f4 | demystifying-the-base-and-novel-performances | 2206.10596 | null | https://arxiv.org/abs/2206.10596v1 | https://arxiv.org/pdf/2206.10596v1.pdf | Demystifying the Base and Novel Performances for Few-shot Class-incremental Learning | Few-shot class-incremental learning (FSCIL) has addressed challenging real-world scenarios where unseen novel classes continually arrive with few samples. In these scenarios, it is required to develop a model that recognizes the novel classes without forgetting prior knowledge. In other words, FSCIL aims to maintain the base performance and improve the novel performance simultaneously. However, there is little study to investigate the two performances separately. In this paper, we first decompose the entire model into four types of parameters and demonstrate that the tendency of the two performances varies greatly with the updated parameters when the novel classes appear. Based on the analysis, we propose a simple method for FSCIL, coined as NoNPC, which uses normalized prototype classifiers without further training for incremental novel classes. It is shown that our straightforward method has comparable performance with the sophisticated state-of-the-art algorithms. | ['Se-Young Yun', 'Jaehoon Oh'] | 2022-06-18 | null | null | null | null | ['few-shot-class-incremental-learning'] | ['methodology'] | [ 2.70520270e-01 -1.61616772e-01 -1.08837530e-01 -2.49717966e-01
-5.06706953e-01 -4.44698602e-01 7.40498960e-01 2.08988205e-01
-3.54889572e-01 7.71273971e-01 -3.68134558e-01 1.72627121e-02
-1.98315069e-01 -6.54477000e-01 -5.24117231e-01 -7.69614100e-01
1.05835177e-01 3.84129345e-01 8.43729258e-01 -1.36157885e-01
2.17124388e-01 4.41512972e-01 -2.02146554e+00 3.22713315e-01
8.83467197e-01 8.98053288e-01 5.77686019e-02 6.83028400e-01
-2.31313780e-01 5.77782631e-01 -6.86082661e-01 -2.89706260e-01
3.16867650e-01 -5.42132854e-01 -4.22366351e-01 1.09076455e-01
2.45194465e-01 -1.52445644e-01 -2.55929500e-01 9.89236593e-01
4.65223342e-01 4.78709310e-01 6.25415564e-01 -1.29796290e+00
-5.11458695e-01 5.72978437e-01 -3.71715426e-01 3.67189884e-01
1.89583972e-01 9.15987864e-02 5.85480511e-01 -1.21587896e+00
5.88260472e-01 1.08085072e+00 8.23893547e-01 7.54363835e-01
-1.03797758e+00 -6.51274979e-01 5.72367728e-01 8.02290082e-01
-1.41107857e+00 -5.09094000e-01 8.01385105e-01 -3.47517252e-01
5.48261821e-01 1.09328605e-01 5.71644664e-01 1.17208791e+00
1.63598973e-02 1.06441545e+00 1.02891707e+00 -7.07680285e-01
6.81946337e-01 5.12919009e-01 5.75695872e-01 4.04305130e-01
4.87710059e-01 1.39156595e-01 -3.50968927e-01 2.33588601e-03
-3.92960608e-02 3.54328781e-01 -1.71844229e-01 -5.13764977e-01
-7.45693266e-01 6.68053031e-01 3.10941674e-02 5.97144842e-01
-2.41381489e-02 -3.26148808e-01 3.21824580e-01 4.40544933e-01
4.63237882e-01 1.80744112e-01 -4.58329946e-01 -2.09228963e-01
-9.56306159e-01 1.22699693e-01 6.96768403e-01 1.03821611e+00
8.44406545e-01 8.63937959e-02 -1.85050383e-01 9.41693366e-01
-3.04778725e-01 2.56173640e-01 7.97093987e-01 -5.21627903e-01
-1.94370113e-02 5.34355819e-01 1.03436165e-01 -8.35300803e-01
-2.33831540e-01 -7.30191350e-01 -6.42359793e-01 1.60073694e-02
3.22915792e-01 -2.63826754e-02 -1.10870278e+00 1.64585817e+00
4.34400648e-01 6.35684550e-01 2.01676145e-01 2.85261065e-01
5.90134501e-01 7.54604995e-01 -6.65502101e-02 -6.64520323e-01
7.01831877e-01 -1.00341749e+00 -7.39312172e-01 -1.83818087e-01
1.73324987e-01 -4.39627230e-01 1.06319678e+00 4.09292042e-01
-6.70073926e-01 -8.87425244e-01 -1.27441895e+00 5.19759119e-01
-7.27372468e-01 -1.58662856e-01 4.31718022e-01 7.12939620e-01
-6.65868938e-01 7.25957274e-01 -8.01965475e-01 -6.89607441e-01
2.71415859e-01 -6.76474944e-02 2.13793576e-01 -2.37214923e-01
-1.16986418e+00 7.94331789e-01 6.94956064e-01 -1.10212125e-01
-9.53406394e-01 -7.55077958e-01 -4.96728569e-01 1.15725562e-01
6.22188449e-01 -3.62463117e-01 1.52980185e+00 -9.56488073e-01
-1.60127676e+00 4.35833454e-01 -3.60305518e-01 -4.53610301e-01
6.80624366e-01 -1.52248725e-01 -6.61162972e-01 -2.16434486e-02
-9.88924876e-02 2.93817312e-01 1.11905551e+00 -1.45567989e+00
-1.03023291e+00 -1.26413405e-01 6.95396587e-02 4.97366041e-02
-5.55945396e-01 -4.34458464e-01 -4.15627837e-01 -5.05942345e-01
4.37285155e-02 -7.29006886e-01 -6.98547214e-02 -7.65046328e-02
9.09865275e-03 -3.92034948e-01 1.13354099e+00 1.18177421e-01
1.43027127e+00 -2.25724220e+00 -1.79322436e-01 -5.29634170e-02
6.56247884e-02 6.26309335e-01 -1.49840504e-01 4.00567204e-01
6.19748458e-02 -1.78798527e-01 -2.60153413e-01 -6.92245141e-02
-6.79729953e-02 4.36166167e-01 -4.44550723e-01 9.33993682e-02
9.36663225e-02 7.88258553e-01 -1.36285424e+00 -3.57157052e-01
1.78651288e-01 1.34982079e-01 -3.29200327e-01 1.00009173e-01
-4.01473232e-02 9.40581411e-02 -1.92553386e-01 7.27482140e-01
7.71777809e-01 -2.77634084e-01 1.18225165e-01 -1.15491346e-01
-1.07979909e-01 -5.33211946e-01 -1.20376670e+00 1.32113135e+00
-7.26201087e-02 4.91859168e-01 -4.56517905e-01 -1.26197982e+00
8.00883651e-01 8.02999660e-02 2.34235883e-01 -5.33120811e-01
8.89637321e-02 2.24402875e-01 -3.96565311e-02 -5.05823672e-01
3.26749235e-01 -4.35166091e-01 1.32869557e-01 2.48315632e-01
2.73906171e-01 2.12082759e-01 4.67674166e-01 4.97843735e-02
1.05583096e+00 -1.58670023e-01 5.55407286e-01 4.99302596e-02
6.29245460e-01 -1.13749221e-01 7.85165191e-01 1.37059951e+00
-5.90668976e-01 4.20607507e-01 2.75702178e-02 -5.09884655e-01
-7.47138858e-01 -1.44404137e+00 -1.21025302e-01 1.13709497e+00
3.65327120e-01 -1.97062060e-01 -5.69816649e-01 -7.71237850e-01
-2.54515745e-02 1.07355189e+00 -6.24612212e-01 -6.28774643e-01
-4.41741526e-01 -9.04564559e-01 1.09364666e-01 3.32517773e-01
4.68107730e-01 -8.91471922e-01 -8.19826424e-01 3.96683544e-01
1.12981156e-01 -9.00016427e-01 -1.34458154e-01 2.02218771e-01
-9.34055865e-01 -1.23931694e+00 -8.23632419e-01 -8.21742117e-01
6.11111760e-01 5.12156904e-01 7.91466892e-01 -1.78207085e-01
-4.38264638e-01 5.87700248e-01 -7.60747910e-01 -6.11378431e-01
-3.15810323e-01 1.97024167e-01 2.86640823e-01 3.14107448e-01
4.92908627e-01 -6.81598186e-01 -3.55521053e-01 1.84102356e-01
-9.78774071e-01 -6.59940392e-02 5.92628837e-01 9.63455617e-01
6.31156862e-01 3.95488322e-01 8.88416648e-01 -1.08903599e+00
5.50303102e-01 -5.91931581e-01 -3.41349900e-01 7.26008475e-01
-9.11939919e-01 -9.63781253e-02 9.44669485e-01 -1.04450011e+00
-1.19908762e+00 -1.22858956e-01 1.60643652e-01 -5.35146594e-01
-1.13615677e-01 3.56884658e-01 5.43892086e-02 -3.61936465e-02
7.04777062e-01 6.21280730e-01 -3.52661043e-01 -5.02645254e-01
4.41443771e-01 7.16240287e-01 5.95329940e-01 -5.08786201e-01
1.04231453e+00 4.06369507e-01 -3.59611541e-01 -9.78225946e-01
-1.27664530e+00 -5.16305208e-01 -8.71373117e-01 -3.74394685e-01
2.24975139e-01 -5.68124115e-01 -2.05257460e-01 6.44680619e-01
-7.52647877e-01 -2.58156925e-01 -7.25805521e-01 3.27626228e-01
-4.77557003e-01 4.43990856e-01 -3.22094470e-01 -1.05308890e+00
-2.20659882e-01 -6.38613284e-01 4.36583668e-01 5.17477214e-01
1.47503316e-01 -8.00061166e-01 3.07672590e-01 -3.35027695e-01
5.56661010e-01 6.77984282e-02 8.84148300e-01 -9.26638484e-01
-3.79725456e-01 -4.63746727e-01 -2.13028677e-02 3.31617445e-01
3.94380867e-01 -5.61199188e-02 -1.11572683e+00 -5.79764068e-01
2.65375435e-01 -1.69616595e-01 9.04836118e-01 6.62468076e-02
1.18505335e+00 -1.80067345e-01 -3.26560408e-01 4.54203308e-01
1.40396929e+00 8.64789546e-01 4.67372954e-01 3.13046932e-01
2.07460716e-01 2.49285564e-01 6.74094975e-01 5.29295087e-01
1.54902972e-02 3.62970442e-01 -1.27627984e-01 3.19469333e-01
-1.29833609e-01 -2.76242584e-01 2.04146013e-01 1.10508108e+00
1.70059070e-01 -2.20191807e-01 -7.04238951e-01 6.16441190e-01
-2.02275038e+00 -1.22933888e+00 6.09340847e-01 2.30443859e+00
7.41042256e-01 4.38641340e-01 -2.17210338e-01 3.20042402e-01
9.02434647e-01 2.28117332e-02 -1.11089921e+00 -8.07004198e-02
-1.05296686e-01 2.76743323e-01 3.14755291e-02 3.06679368e-01
-1.06285703e+00 8.59791875e-01 7.17851257e+00 1.09870267e+00
-9.65382099e-01 2.06447467e-01 4.80495334e-01 -2.13034198e-01
2.97082942e-02 5.54826930e-02 -9.58604038e-01 4.75389183e-01
9.54821348e-01 -7.66322315e-01 3.32133651e-01 9.72824037e-01
-1.48655519e-01 -8.83953348e-02 -1.10716891e+00 1.14845729e+00
4.37687039e-01 -1.02227485e+00 2.89456367e-01 -6.90552235e-01
9.09030914e-01 -2.58534104e-01 -3.81953418e-02 9.62750494e-01
7.83365145e-02 -2.15999514e-01 6.06211424e-01 8.02622378e-01
6.47108257e-01 -6.98918700e-01 6.46388113e-01 5.62715232e-01
-1.23406541e+00 -4.60079551e-01 -4.77742165e-01 -1.87966395e-02
3.91708724e-02 5.40488064e-01 -6.03213727e-01 5.17011404e-01
5.83574533e-01 7.94293106e-01 -8.94061387e-01 1.41806412e+00
-1.73723903e-02 7.89096177e-01 -1.74579263e-01 -9.21442807e-02
8.71656165e-02 8.17390606e-02 6.02616787e-01 1.07894981e+00
4.33841676e-01 1.93012148e-01 3.45240325e-01 5.79195440e-01
9.49161053e-02 -4.06423956e-02 -4.35081095e-01 -2.57760473e-02
6.76050246e-01 8.72971117e-01 -8.33769202e-01 -7.81209350e-01
-5.00791907e-01 9.63905215e-01 2.94805616e-01 4.76755917e-01
-7.89637387e-01 -6.51886046e-01 3.62363368e-01 4.96048592e-02
4.81206596e-01 1.79092027e-02 4.76741269e-02 -1.44096673e+00
-3.54232267e-02 -6.04312062e-01 5.95010936e-01 -4.96524751e-01
-1.63609326e+00 6.11090302e-01 2.26228848e-01 -1.37115455e+00
-1.89505383e-01 -3.78890067e-01 -7.78733492e-01 1.01718232e-01
-1.55942297e+00 -8.01781356e-01 -4.76971120e-01 6.01360619e-01
9.69079196e-01 -3.18917066e-01 5.55265784e-01 3.02208245e-01
-7.54188299e-01 1.02235866e+00 5.11143506e-01 -2.72967070e-01
6.82868958e-01 -9.26592112e-01 1.50555238e-01 9.98137593e-01
7.54847052e-03 5.17597795e-01 8.22143376e-01 -6.36625528e-01
-1.04805994e+00 -1.26587462e+00 6.78454578e-01 -2.72546858e-01
6.34962976e-01 -3.09987992e-01 -1.13376009e+00 4.63809490e-01
-1.40902147e-01 5.17291166e-02 6.03570580e-01 -5.27656823e-02
-2.44515374e-01 -4.59354073e-01 -1.06680191e+00 5.71814954e-01
1.05639446e+00 -2.79338270e-01 -9.18492496e-01 2.05326393e-01
8.48832130e-01 -1.03731314e-02 -3.02592367e-01 5.69973826e-01
5.80460966e-01 -8.59886825e-01 8.93132806e-01 -6.65514529e-01
-1.87326744e-01 -3.57897878e-01 -2.70906329e-01 -1.19457805e+00
-5.74774265e-01 -3.20417792e-01 -6.33296669e-01 1.13661861e+00
3.77897471e-01 -8.88779819e-01 7.48725653e-01 3.06095868e-01
-4.56980802e-02 -6.91849887e-01 -9.47726548e-01 -1.37822819e+00
-9.36676860e-02 -3.43788147e-01 4.25031602e-01 8.76293480e-01
-1.26733348e-01 3.59741032e-01 -4.47292656e-01 -2.06623860e-02
6.75601900e-01 2.88487792e-01 6.53696239e-01 -1.43581831e+00
-1.84494063e-01 -2.36128524e-01 -4.48160708e-01 -9.11922693e-01
-1.38439521e-01 -6.78425610e-01 1.60623416e-01 -1.12423730e+00
5.28888106e-01 -3.61341327e-01 -9.04901147e-01 4.93694037e-01
-4.82421905e-01 -9.81729254e-02 3.51910800e-01 3.18087608e-01
-1.08437133e+00 7.92535007e-01 8.16334307e-01 -2.90800244e-01
-5.25428891e-01 1.83067530e-01 -6.90351605e-01 7.34636366e-01
8.20128143e-01 -5.11446714e-01 -7.40293682e-01 2.02617887e-02
-2.38119185e-01 -4.07164633e-01 -9.67326909e-02 -1.58288372e+00
5.29124498e-01 -3.20028633e-01 4.37134087e-01 -8.05531442e-01
1.97826073e-01 -7.71982551e-01 1.31796971e-01 6.60273373e-01
-2.65896052e-01 -2.39078984e-01 8.18929151e-02 9.90013480e-01
-1.25504076e-01 -5.07173657e-01 1.04095602e+00 -2.17399433e-01
-1.04735804e+00 4.01789993e-01 -4.82673943e-01 1.60271272e-01
1.44071746e+00 -4.06850040e-01 -2.29469851e-01 -2.22339347e-01
-9.30224359e-01 1.16022646e-01 3.17492664e-01 5.40031075e-01
7.11093187e-01 -1.31504083e+00 -4.83822435e-01 2.81009525e-01
4.32033122e-01 -3.24744403e-01 4.73392218e-01 5.10679185e-01
-1.19712710e-01 2.44249061e-01 5.24169765e-03 -5.68360567e-01
-1.07269359e+00 1.11612725e+00 1.92765564e-01 -1.69418633e-01
-6.33679509e-01 6.52568340e-01 1.07848652e-01 -2.99246371e-01
4.29128468e-01 -3.16616683e-03 -2.32517108e-01 3.82806540e-01
9.19827998e-01 5.07717013e-01 -7.27294432e-03 -1.74052238e-01
-1.78150788e-01 5.75727701e-01 -4.54520017e-01 2.96036243e-01
1.30341756e+00 -2.55589962e-01 3.73106003e-01 1.20171595e+00
9.65646267e-01 -3.81617963e-01 -1.32316518e+00 -5.54342866e-01
6.64943680e-02 -4.79438990e-01 -3.88950378e-01 -7.39402413e-01
-6.98336959e-01 6.97586656e-01 1.08187628e+00 3.96058522e-02
1.24856138e+00 -2.21092775e-01 7.33966410e-01 7.24300086e-01
7.08339691e-01 -1.28615940e+00 2.51177818e-01 6.97301030e-01
5.33324659e-01 -1.15277481e+00 -4.56559360e-02 -2.39426896e-01
-3.27258646e-01 1.16236150e+00 7.56075799e-01 4.61173132e-02
8.15548837e-01 -1.19661361e-01 -7.07856938e-02 1.71359941e-01
-1.01931596e+00 -3.96579802e-01 1.78095222e-01 5.56585073e-01
-2.78428733e-01 -4.70953360e-02 -3.24163020e-01 6.75464392e-01
2.88126860e-02 1.28217980e-01 4.91559446e-01 1.18529761e+00
-8.39590669e-01 -9.24756587e-01 -1.40797496e-01 4.94215459e-01
-6.69020191e-02 8.66173357e-02 -2.35835582e-01 7.79604256e-01
3.44270170e-01 9.18360353e-01 -5.11843944e-03 -6.14163935e-01
5.19993782e-01 4.38688725e-01 3.18777621e-01 -6.23120725e-01
-1.27115250e-02 -2.26267084e-01 -4.61396724e-01 -2.05940351e-01
-4.12818342e-01 -5.98364353e-01 -8.00795972e-01 -6.66987011e-03
-4.35040474e-01 2.02940255e-01 1.34186476e-01 9.58430350e-01
3.00314844e-01 4.31053460e-01 8.00080776e-01 -6.83282435e-01
-6.13812089e-01 -8.33066702e-01 -5.30787706e-01 3.59920055e-01
2.82211214e-01 -9.14743960e-01 -7.19261289e-01 -2.98194773e-03] | [9.92531967163086, 3.413419008255005] |
8acdc545-7174-491d-8b90-892313a1805d | towards-mapping-thesauri-onto-plwordnet | null | null | https://aclanthology.org/2018.gwc-1.6 | https://aclanthology.org/2018.gwc-1.6.pdf | Towards Mapping Thesauri onto plWordNet | plWordNet, the wordnet of Polish, has become a very comprehensive description of the Polish lexical system. This paper presents a plan of its semi-automated integration with thesauri, terminological databases and ontologies, as a further necessary step in its development. This will improve linking of plWordNet into Linked Open Data, and facilitate applications in, e.g., WSD, keyword extraction or automated metadata generation. We present an overview of resources relevant to Polish and a plan for their linking to plWordNet. | ['Maciej Piasecki', 'Marek Maziarz'] | null | null | null | null | gwc-2018-1 | ['keyword-extraction'] | ['natural-language-processing'] | [-3.85317445e-01 4.16161746e-01 -6.25294447e-01 1.61182284e-01
-5.60974240e-01 -7.14190900e-01 8.53326142e-01 7.58181512e-01
-8.90891790e-01 1.27111363e+00 7.74204135e-01 -1.19229637e-01
-6.92100883e-01 -1.15284145e+00 2.80816823e-01 -2.44326234e-01
4.52044040e-01 9.67365146e-01 8.01225305e-01 -7.62995124e-01
7.63281062e-02 3.85293961e-01 -1.83051550e+00 1.45953774e-01
6.28402412e-01 5.68997324e-01 6.30054116e-01 8.82709697e-02
-1.10192502e+00 3.06508213e-01 -6.14913464e-01 -6.89565897e-01
-3.47247094e-01 -3.47906873e-02 -1.17366016e+00 -2.63970912e-01
-1.55041572e-02 4.18958396e-01 -1.68708310e-01 1.43554509e+00
3.83176923e-01 -7.67859071e-02 2.08554924e-01 -8.76975596e-01
-4.29049522e-01 1.35740972e+00 2.85909116e-01 -2.22303599e-01
8.81537259e-01 -4.52401936e-01 1.19885623e+00 -6.45244062e-01
1.46548390e+00 9.86157596e-01 4.68343467e-01 1.97046459e-01
-6.94500983e-01 -5.11277735e-01 -4.33477014e-01 4.71513957e-01
-1.56705499e+00 -4.23107982e-01 4.26053494e-01 -1.75100192e-01
1.17634511e+00 5.02936602e-01 8.53405297e-01 8.05414200e-01
-5.45333661e-02 1.69444054e-01 9.01718140e-01 -1.10528564e+00
-4.22932487e-03 3.71279478e-01 4.33349818e-01 3.07761341e-01
1.30332482e+00 -4.48319256e-01 -6.83101535e-01 -5.30608058e-01
5.57205975e-01 -5.22110820e-01 -8.39105770e-02 -1.23458385e-01
-1.09838271e+00 3.71617258e-01 -5.34576476e-01 1.08216703e+00
-5.38690150e-01 -3.44017059e-01 5.61470330e-01 3.61012280e-01
5.31392813e-01 6.53066874e-01 -7.67373383e-01 -1.76255867e-01
-7.49563575e-01 6.06881857e-01 1.33218217e+00 1.29790699e+00
5.73235810e-01 -2.31961086e-01 4.41081405e-01 1.08971620e+00
5.74712515e-01 6.00272894e-01 8.45906317e-01 -7.78221667e-01
2.87554473e-01 9.24810529e-01 2.64430404e-01 -6.83557868e-01
-5.77699304e-01 2.88372010e-01 1.10982805e-01 -2.84805566e-01
1.02766760e-01 4.34929132e-02 -5.36246777e-01 1.08144557e+00
2.55761832e-01 -3.65873605e-01 5.62690079e-01 2.70095170e-01
1.64742863e+00 3.24001282e-01 4.33935314e-01 -5.63224792e-01
1.90067756e+00 -8.60712156e-02 -1.65926635e+00 3.58554780e-01
5.54499090e-01 -1.46204126e+00 5.34397066e-01 1.61556602e-01
-1.32948959e+00 2.97441836e-02 -1.06732130e+00 -2.72633154e-02
-1.42556965e+00 -3.05206865e-01 6.34570956e-01 9.53278422e-01
-7.48528957e-01 4.96762156e-01 -4.93561536e-01 -1.24521816e+00
-2.29334868e-02 1.49194365e-02 -6.45465195e-01 2.25300580e-01
-2.03029156e+00 1.32887518e+00 1.44408464e+00 -6.72004461e-01
4.45366621e-01 -6.65114224e-01 -1.09872198e+00 -1.86635211e-01
5.63046753e-01 -4.49220240e-01 9.21284854e-01 1.25899732e-01
-1.21405721e+00 1.47430933e+00 1.97788060e-01 -5.68553627e-01
-2.08991736e-01 2.68302202e-01 -1.66998887e+00 1.83873430e-01
3.83366466e-01 1.52894676e-01 -1.17766082e-01 -9.24434662e-01
-1.04365170e+00 -7.74365515e-02 -1.16926871e-01 1.03505664e-01
-3.96624595e-01 7.74600744e-01 -4.93203044e-01 -8.33857417e-01
-7.41642490e-02 -1.00281343e-01 -1.82503834e-02 -7.83998191e-01
5.47584370e-02 -6.88824832e-01 2.11103410e-01 -7.03673065e-01
1.91452467e+00 -1.40342808e+00 -3.65167618e-01 3.48797202e-01
2.96573043e-01 5.02742052e-01 1.00568421e-01 1.37607861e+00
-8.54051206e-03 3.73564214e-01 2.01086536e-01 5.91746986e-01
3.93334061e-01 7.31104553e-01 -1.98388010e-01 2.44901747e-01
-2.58514941e-01 8.36651862e-01 -1.09478140e+00 -8.56107235e-01
4.51569498e-01 1.47091508e-01 2.97645777e-01 -7.60396600e-01
-2.36872390e-01 -4.41869169e-01 -4.79426146e-01 6.16899371e-01
3.04114670e-01 4.21496063e-01 4.82569933e-01 -1.54473439e-01
-7.09243894e-01 7.52356470e-01 -1.43927896e+00 1.80448174e+00
-4.38679308e-01 3.51719022e-01 -1.34973437e-01 -4.70531523e-01
1.12902439e+00 1.14927268e+00 9.97860491e-01 -6.45684540e-01
2.06301123e-01 7.99526095e-01 -3.52300018e-01 -6.62090182e-01
1.06056654e+00 -3.54167894e-02 -3.40570807e-01 2.71902889e-01
7.99187899e-01 -1.18380405e-01 1.06249869e+00 2.12603249e-03
7.39490569e-01 2.76918769e-01 1.28823090e+00 -8.64793420e-01
9.84654844e-01 4.93796617e-01 4.42741156e-01 6.28263429e-02
2.65766025e-01 -1.75221302e-02 1.94399476e-01 -4.48436975e-01
-1.11058307e+00 -6.00938797e-01 -7.76229382e-01 5.35083711e-01
4.83382083e-02 -1.07976103e+00 -7.71440923e-01 -2.55324721e-01
-1.25765085e-01 8.99983168e-01 -3.99205238e-02 3.38629484e-01
-4.71896291e-01 -4.35460091e-01 7.13229299e-01 -4.18348104e-01
8.37860256e-02 -1.47691262e+00 -4.77749586e-01 5.56077898e-01
-2.93321162e-01 -1.35265505e+00 2.14374617e-01 2.41133105e-02
-3.47521514e-01 -1.27998221e+00 -3.37820739e-01 -1.00551474e+00
5.06631751e-03 -1.91910401e-01 1.29851317e+00 -8.39694664e-02
-1.47392541e-01 4.52111125e-01 -6.83747888e-01 -7.47853994e-01
-4.18609589e-01 5.29100537e-01 1.32089145e-02 -7.82005787e-01
1.15709317e+00 -4.35518116e-01 1.50207877e-01 2.22333550e-01
-1.30041742e+00 -5.68959892e-01 -1.17128260e-01 1.91550568e-01
4.75008398e-01 2.88558811e-01 6.89954102e-01 -1.17472076e+00
1.12963963e+00 -4.88153964e-01 -8.22982788e-01 4.76669699e-01
-8.18084300e-01 -1.30581418e-02 -4.35331166e-02 7.70662948e-02
-1.13984013e+00 -2.52028823e-01 -7.47573853e-01 5.24622560e-01
-3.67342442e-01 1.07915807e+00 -7.45242417e-01 -4.46245670e-02
5.66200554e-01 -1.80151269e-01 -2.40805656e-01 -1.01474619e+00
6.94326401e-01 7.52842486e-01 3.62720340e-01 -6.33275211e-01
4.85050619e-01 2.35742226e-01 -1.34565845e-01 -1.24017286e+00
-2.44482249e-01 -1.07715189e+00 -7.75568068e-01 -8.29049647e-02
7.03421354e-01 -8.13221872e-01 -4.37474757e-01 -1.71057716e-01
-1.18419576e+00 2.94914484e-01 -9.46567237e-01 4.12947625e-01
-4.40326959e-01 2.23901883e-01 -4.05626222e-02 -5.06680429e-01
-4.59243268e-01 -4.02717650e-01 4.34251994e-01 1.37768807e-02
-8.01205814e-01 -1.49028778e+00 5.54425299e-01 9.12904739e-03
-8.76609981e-02 1.15028612e-01 7.74116397e-01 -1.43564212e+00
5.14799535e-01 -7.20842242e-01 2.54788190e-01 -1.97193831e-01
3.10103238e-01 -1.80140398e-02 -4.05385166e-01 4.25239503e-01
-2.92339265e-01 2.07479537e-01 3.25502247e-01 -2.56272793e-01
2.64084697e-01 -3.54671508e-01 -4.84825134e-01 4.46381047e-02
1.88305044e+00 3.44067603e-01 7.91159868e-01 9.63312209e-01
4.37653586e-02 9.36708391e-01 7.82358825e-01 3.77016485e-01
4.44984227e-01 7.10401893e-01 -6.58941716e-02 4.91974235e-01
-6.95925236e-01 -1.84436977e-01 -9.13301110e-02 1.39145482e+00
-3.65090370e-01 -1.80036649e-01 -1.23213410e+00 7.73558319e-01
-1.84208810e+00 -9.83175814e-01 -4.61070627e-01 2.16136956e+00
1.01985490e+00 8.99431761e-03 4.54063639e-02 2.86562204e-01
7.59262621e-01 1.07917577e-01 5.17352998e-01 -3.24097276e-01
-7.06333578e-01 1.06264710e+00 9.66747999e-01 8.97579193e-01
-6.84094012e-01 1.46447420e+00 6.59557533e+00 1.13427246e+00
-3.68239611e-01 5.66623271e-01 -7.85928965e-01 3.42717201e-01
-5.88888586e-01 3.70871946e-02 -1.23557794e+00 4.58035290e-01
1.19624269e+00 -1.09733057e+00 1.41655102e-01 3.85329306e-01
1.87527686e-01 6.22598976e-02 -3.11606169e-01 5.92622399e-01
5.59292175e-02 -1.73814464e+00 2.64099628e-01 3.43363196e-01
5.72398424e-01 5.50200753e-02 -7.83177257e-01 3.19562443e-02
6.92871630e-01 -3.31185549e-01 6.16407275e-01 4.34210390e-01
9.98935461e-01 -8.11429679e-01 1.10843849e+00 -2.06390694e-01
-1.18724823e+00 6.09022260e-01 -7.10800231e-01 1.33864224e-01
5.38496792e-01 5.53490222e-01 -2.81999916e-01 1.21910405e+00
4.16303694e-01 4.51894909e-01 -9.89560932e-02 1.12652397e+00
-4.62926030e-01 1.64753512e-01 -2.68301606e-01 -2.98039973e-01
1.88286141e-01 -3.46503824e-01 7.99789846e-01 1.25018966e+00
3.79906118e-01 1.15003020e-01 3.79030891e-02 7.53903985e-02
-9.12436396e-02 9.59189713e-01 -5.57447255e-01 -2.39451125e-01
8.32347035e-01 1.17995250e+00 -7.28133917e-01 -4.92844850e-01
-6.75478876e-01 3.34715456e-01 -2.11208940e-01 3.22696753e-02
-1.19204648e-01 -1.15712023e+00 9.75661337e-01 4.30802375e-01
6.63521662e-02 -1.44895688e-01 -2.24556670e-01 -9.74645853e-01
-2.06719756e-01 -6.08794034e-01 8.45218718e-01 -5.07523656e-01
-9.47135448e-01 7.11851478e-01 5.12050033e-01 -1.18411195e+00
-4.59364206e-01 -7.25797653e-01 8.41297731e-02 1.13201809e+00
-1.23216081e+00 -1.24697602e+00 3.09852540e-01 4.42600816e-01
3.67299765e-01 -3.30062360e-01 1.51793218e+00 5.59890985e-01
6.35061786e-02 -1.04336224e-01 -3.15502076e-03 -5.13965897e-02
8.72939944e-01 -1.21991587e+00 3.15685630e-01 5.01683474e-01
3.34688395e-01 4.34784144e-01 8.81573498e-01 -1.24937892e+00
-7.34791279e-01 -6.12343490e-01 2.02799129e+00 -1.80850387e-01
1.72391212e+00 1.00379266e-01 -7.39200056e-01 4.14263397e-01
9.43872809e-01 -4.09625798e-01 1.06294584e+00 2.43605231e-03
3.77175980e-03 -1.29870335e-02 -1.20471334e+00 7.35129237e-01
1.14624190e+00 -2.25836620e-01 -1.31499124e+00 5.62567770e-01
8.49263251e-01 -1.02968112e-01 -1.77089322e+00 9.54148844e-02
6.77478850e-01 -9.87820998e-02 1.04034758e+00 -6.27883136e-01
-2.26405397e-01 -1.87670127e-01 -7.56345093e-02 -1.24450314e+00
-3.01014215e-01 -1.11560845e+00 1.59600347e-01 1.70359182e+00
5.85577548e-01 -9.97861743e-01 3.26564223e-01 1.54551595e-01
-2.93380588e-01 8.72332901e-02 -1.30120063e+00 -8.85883331e-01
3.00793201e-02 -7.31563687e-01 1.22441006e+00 1.15425968e+00
1.02990484e+00 1.35077357e-01 4.49200198e-02 -1.54605731e-01
4.82217193e-01 -6.26565695e-01 -3.54487002e-02 -1.80944216e+00
6.97828710e-01 -8.88102233e-01 -7.35342741e-01 4.35731336e-02
2.28910163e-01 -9.10229802e-01 -4.26603556e-01 -1.94241190e+00
-5.77666461e-01 -2.05938518e-01 -2.50668317e-01 2.81468838e-01
2.68031687e-01 3.27475488e-01 -1.07730038e-01 3.37930411e-01
-5.07850528e-01 -1.15444720e-01 9.02988851e-01 1.99148819e-01
-2.29144201e-01 -4.19986367e-01 -6.48666680e-01 9.46006656e-01
9.39915240e-01 -6.47716224e-01 -5.22714444e-02 -2.18702450e-01
9.94751811e-01 -3.03425968e-01 -4.93251175e-01 -8.38837743e-01
5.27569473e-01 -3.74585479e-01 -4.16962683e-01 -7.95101583e-01
1.98503155e-02 -1.24208915e+00 4.87039685e-01 4.32930052e-01
9.35884640e-02 1.39105901e-01 1.88108742e-01 -1.14046611e-01
-4.56337810e-01 -1.14835238e+00 3.10549706e-01 -5.34529090e-01
-1.03305054e+00 -7.53908008e-02 -7.62692630e-01 1.54568002e-01
9.26945388e-01 -3.34503770e-01 -3.13208789e-01 1.68698236e-01
-1.05765367e+00 1.35429874e-01 4.94077057e-01 4.76993769e-01
3.33323270e-01 -1.62951910e+00 -6.73749626e-01 -3.13435525e-01
7.00849771e-01 -8.08787942e-01 -1.69478059e-01 3.97691876e-01
-1.05840361e+00 1.18328762e+00 -4.81401294e-01 4.13861454e-01
-1.17368448e+00 5.02015412e-01 -2.09326848e-01 -4.93873388e-01
-7.73888350e-01 -9.95786041e-02 -8.53278399e-01 -4.12309676e-01
1.16692841e-01 1.49507439e-02 -1.30380952e+00 6.28942966e-01
7.44837105e-01 6.33037210e-01 5.06254435e-01 -9.72926676e-01
-4.22898054e-01 5.07163346e-01 4.28919643e-01 -5.26377618e-01
1.30076909e+00 -4.24010605e-01 -8.89746130e-01 8.45113993e-01
4.50854242e-01 5.02060473e-01 2.52934158e-01 -4.97715324e-01
8.83250654e-01 -5.59941679e-02 5.74377961e-02 -6.87329471e-01
-4.62912530e-01 -1.73178002e-01 2.83303231e-01 7.03987896e-01
7.78413296e-01 4.41141486e-01 7.15386927e-01 3.48410368e-01
6.56039894e-01 -1.44507694e+00 -1.18294358e+00 6.85631633e-01
8.33200693e-01 -4.97547179e-01 2.40674779e-01 -8.66940975e-01
-3.07877809e-01 1.45869362e+00 -3.07263523e-01 1.03209883e-01
1.15882409e+00 4.55012918e-01 2.69397587e-01 -5.61215341e-01
-5.72784364e-01 -1.12438500e+00 4.39729065e-01 9.26145256e-01
5.43875396e-01 2.11892296e-02 -1.74726045e+00 8.93880904e-01
-3.61539215e-01 1.21726528e-01 5.32331109e-01 9.89359915e-01
-6.48757041e-01 -1.94678450e+00 -5.21979392e-01 2.68454015e-01
-9.28168952e-01 -3.65301847e-01 -4.10172284e-01 1.24622524e+00
5.49493074e-01 7.98601925e-01 1.80330396e-01 -4.54689050e-03
6.81064248e-01 4.36118692e-01 4.77323800e-01 -9.57969248e-01
-7.43757665e-01 2.54781783e-01 9.82049882e-01 -3.04088652e-01
-9.34425890e-01 -8.48129809e-01 -1.02687752e+00 -4.54747319e-01
-3.00114959e-01 7.03000963e-01 9.49427009e-01 8.40517104e-01
-1.75838351e-01 3.73284310e-01 -2.59775013e-01 -1.31294653e-01
2.59157658e-01 -1.16851866e+00 -9.23687398e-01 1.89791117e-02
-6.07825577e-01 -5.92774928e-01 9.49554145e-02 2.04236642e-01] | [9.523280143737793, 8.734704971313477] |
5296521e-15db-449b-898e-6b0af899543a | deep-learning-from-parametrically-generated | 2302.05283 | null | https://arxiv.org/abs/2302.05283v1 | https://arxiv.org/pdf/2302.05283v1.pdf | Deep Learning from Parametrically Generated Virtual Buildings for Real-World Object Recognition | We study the use of parametric building information modeling (BIM) to automatically generate training data for artificial neural networks (ANNs) to recognize building objects in photos. Teaching artificial intelligence (AI) machines to detect building objects in images is the foundation toward AI-assisted semantic 3D reconstruction of existing buildings. However, there exists the challenge of acquiring training data which is typically human-annotated, that is, unless a computer machine can generate high-quality data to train itself for a certain task. In that vein, we trained ANNs solely on realistic computer-generated images of 3D BIM models which were parametrically and automatically generated using the BIMGenE program. The ANN training result demonstrated generalizability and good semantic segmentation on a test case as well as arbitrary photos of buildings that are outside the range of the training data, which is significant for the future of training AI with generated data for solving real-world architectural problems. | ['Wei Yan', 'Mohammad Alawadhi'] | 2023-01-03 | null | null | null | null | ['object-recognition'] | ['computer-vision'] | [ 6.33422136e-01 3.84468913e-01 6.07637584e-01 -5.96496820e-01
-8.22649479e-01 -5.04803121e-01 1.78928018e-01 -1.28132654e-02
-1.16531342e-01 3.50303739e-01 -1.73928648e-01 -5.02321422e-01
5.95590286e-02 -1.43574321e+00 -9.78007793e-01 -3.00287426e-01
1.12748612e-02 1.08335841e+00 2.44082332e-01 -3.35531294e-01
1.29636779e-01 8.48962009e-01 -1.92582238e+00 4.51195180e-01
9.47955251e-01 6.66192770e-01 5.87743461e-01 8.06330860e-01
-4.55276296e-02 5.44474185e-01 -5.86798012e-01 -1.01080798e-01
3.95991176e-01 -1.66562483e-01 -7.56358027e-01 6.08125865e-01
5.36032021e-01 -4.96056765e-01 -2.24660132e-02 5.86912096e-01
2.01022997e-01 1.13191903e-02 8.21297944e-01 -1.10126841e+00
-8.25197279e-01 3.51210833e-01 5.86700216e-02 -6.67917505e-02
1.25051096e-01 2.66317010e-01 7.88919091e-01 -7.18879342e-01
3.96840781e-01 9.64842677e-01 8.01726818e-01 3.67482603e-01
-1.20674205e+00 -1.34089664e-01 -1.92482367e-01 7.18098730e-02
-1.57012510e+00 -1.50575228e-02 1.07464397e+00 -8.08378041e-01
6.39526725e-01 4.55496699e-01 1.16262615e+00 2.89645582e-01
-4.11896169e-01 9.12007749e-01 9.33750033e-01 -8.24718058e-01
4.77875084e-01 1.73752084e-01 -2.34572142e-01 9.03105199e-01
-1.82141699e-02 2.67702997e-01 9.84692127e-02 3.23888510e-01
1.15493047e+00 -4.04376715e-01 5.10928594e-02 -2.87070155e-01
-1.05186725e+00 8.28028738e-01 8.35406840e-01 2.48840272e-01
-5.19333601e-01 1.18447550e-01 -2.57443786e-01 -1.32214710e-01
2.40232110e-01 7.79179335e-01 -2.92355508e-01 3.62434030e-01
-1.04668665e+00 3.12489569e-01 5.30526817e-01 8.37635279e-01
1.04681194e+00 6.31455302e-01 5.12844920e-01 8.48911166e-01
2.34899744e-01 5.15697837e-01 2.75237054e-01 -1.36386120e+00
8.63132328e-02 1.24394286e+00 -1.37525201e-01 -7.30162084e-01
-2.39445418e-01 -1.75110668e-01 -3.82113069e-01 4.24421906e-01
5.85315704e-01 -1.79071501e-02 -1.29458332e+00 1.20562172e+00
3.97963226e-01 1.27557036e-03 -3.85805629e-02 8.34848464e-01
7.82518148e-01 7.78218806e-01 9.07879323e-02 4.24649805e-01
1.11689425e+00 -6.14914298e-01 1.50552660e-01 -7.46487975e-01
6.41116679e-01 -4.70640689e-01 1.30556870e+00 8.87484998e-02
-1.02597010e+00 -7.72029579e-01 -1.24831128e+00 9.84838232e-02
-4.67402846e-01 2.05550492e-01 7.59574950e-01 7.90973544e-01
-1.04969323e+00 3.70815545e-01 -7.40710258e-01 -3.42882395e-01
5.53902566e-01 3.89182836e-01 -2.08541125e-01 -3.30802292e-01
-8.30898166e-01 9.07073140e-01 6.95636332e-01 1.81457371e-01
-1.21009684e+00 -6.66776597e-01 -9.32425320e-01 -4.40383032e-02
1.71496555e-01 -6.55571282e-01 1.30079341e+00 -1.44493210e+00
-1.00445032e+00 1.31657338e+00 4.70122665e-01 -3.41730952e-01
1.71616495e-01 1.68432117e-01 -1.74929500e-01 9.21537429e-02
1.96126550e-01 1.14451671e+00 4.92589414e-01 -1.72451961e+00
-4.37594205e-01 -5.42505980e-01 1.53882220e-01 2.14295581e-01
9.26305726e-02 -2.88934380e-01 -6.58643842e-02 -3.77957970e-01
4.39659476e-01 -9.78563726e-01 -5.24665117e-01 -5.07996534e-04
-1.97534427e-01 4.49419534e-03 6.62733436e-01 -1.01499557e+00
5.16117752e-01 -1.88105249e+00 -3.34417135e-01 5.59550285e-01
-3.72947246e-01 3.68055344e-01 -4.84569073e-02 2.37451255e-01
-1.15128137e-01 1.95107535e-01 -6.23113453e-01 3.37186754e-01
-1.59785673e-01 4.38547701e-01 8.57987814e-03 -2.48408645e-01
2.18865141e-01 8.56770098e-01 -6.33382380e-01 -5.44298053e-01
5.51767886e-01 3.70119750e-01 -5.76633453e-01 2.36728728e-01
-7.25179970e-01 4.28932607e-01 -5.04287302e-01 5.30065179e-01
1.29986390e-01 -1.15909889e-01 1.94542617e-01 -1.11403592e-01
4.33194041e-02 -6.19976372e-02 -1.20729899e+00 1.54167414e+00
-6.27371550e-01 7.18871057e-01 -3.53079557e-01 -1.12908876e+00
1.05482888e+00 2.46595964e-01 4.43911910e-01 -5.80308020e-01
1.46999760e-02 2.26021990e-01 -5.04648350e-02 -5.53859293e-01
3.70809615e-01 -2.92056918e-01 -5.72092161e-02 5.49364507e-01
-1.96233079e-01 -9.93268311e-01 2.19801828e-01 -2.42427230e-01
8.76841426e-01 2.92393506e-01 -2.25359648e-01 -1.16893232e-01
3.01687241e-01 8.70655537e-01 9.91949141e-02 4.26403612e-01
1.89823747e-01 8.97855699e-01 -2.84415036e-01 -8.65244091e-01
-1.59674037e+00 -1.11058927e+00 -1.20747732e-02 8.00294757e-01
-1.74024627e-01 6.72966167e-02 -1.23009539e+00 -3.70772153e-01
-2.46891826e-01 1.35663474e+00 -2.32128263e-01 -6.39573671e-03
-6.05585814e-01 -6.19105160e-01 2.70433545e-01 6.03016376e-01
5.92960060e-01 -1.34167683e+00 -1.12868524e+00 3.81687731e-01
-1.73900977e-01 -1.05551279e+00 2.19417319e-01 -3.68479155e-02
-8.84289801e-01 -1.14239955e+00 -1.81218371e-01 -1.16786444e+00
1.14195251e+00 6.58521727e-02 1.29070067e+00 4.96617258e-01
-6.78084135e-01 7.54114568e-01 -6.88395649e-02 -7.21454620e-01
-8.59860301e-01 -1.86882272e-01 -5.58772326e-01 -5.27008951e-01
4.11891222e-01 -7.46680439e-01 -3.78254652e-01 3.99307489e-01
-1.10074210e+00 6.92446351e-01 6.99317217e-01 3.72219771e-01
5.42141378e-01 6.81543291e-01 3.68358254e-01 -6.17389083e-01
2.54732370e-01 -3.34280804e-02 -6.84577465e-01 4.22037572e-01
-3.71900678e-01 1.44233974e-02 2.81020790e-01 -1.12697303e-01
-1.17890012e+00 6.62237108e-01 -2.01793551e-01 1.28617659e-01
-7.63974547e-01 6.09562695e-01 -2.55282015e-01 5.26599176e-02
1.03801465e+00 1.46090001e-01 -1.40941441e-01 -1.52363792e-01
3.81157547e-01 5.92437327e-01 6.54975712e-01 -8.30435574e-01
1.08342230e+00 2.49043837e-01 -7.68723562e-02 -1.13207698e+00
-8.84873271e-01 1.12709045e-01 -1.09679520e+00 -5.41830182e-01
1.05071414e+00 -9.23038602e-01 -1.71153530e-01 2.44885549e-01
-7.50856996e-01 -6.80138767e-01 -5.71750641e-01 3.23107570e-01
-8.09153795e-01 -4.06339288e-01 -1.98206335e-01 -8.18139672e-01
1.82516256e-03 -1.12319338e+00 8.92461836e-01 1.59090191e-01
-2.97747403e-01 -6.79003000e-01 -5.95846623e-02 1.11491287e+00
1.09761275e-01 4.54975039e-01 1.34448075e+00 -3.90575439e-01
-1.07702386e+00 -5.46111166e-01 7.29826614e-02 6.43713236e-01
3.38762738e-02 3.23589951e-01 -1.02464581e+00 3.86723995e-01
-2.30239853e-01 -4.33207572e-01 2.32713997e-01 4.27967221e-01
1.03458071e+00 -1.54343575e-01 -1.07692510e-01 -6.64658844e-02
1.54454935e+00 6.81428254e-01 8.20851088e-01 4.30904150e-01
5.92058539e-01 7.84092188e-01 3.00045580e-01 -9.97481793e-02
6.14482641e-01 4.38282907e-01 2.37867236e-01 -1.76980391e-01
-1.31466731e-01 -4.95371103e-01 -2.11540744e-01 4.79773283e-01
-2.69505590e-01 1.01506062e-01 -1.53904760e+00 9.10957277e-01
-1.41156733e+00 -8.63852203e-01 -2.67899215e-01 1.94140506e+00
6.34239793e-01 2.02685073e-01 1.81513522e-02 4.89234626e-01
6.46942377e-01 -3.70533556e-01 -3.66491824e-01 -6.81001917e-02
4.20947671e-02 3.45885992e-01 1.70493692e-01 3.22929919e-01
-1.06467354e+00 9.45760190e-01 6.35476208e+00 2.68292397e-01
-8.48079383e-01 -3.40641111e-01 8.84328485e-01 3.89733225e-01
-2.87849069e-01 9.43206176e-02 -2.68382311e-01 1.82941794e-01
1.04893029e+00 2.22718194e-02 5.38893521e-01 1.23120320e+00
4.05499756e-01 -2.90785581e-01 -9.79531348e-01 5.63827217e-01
2.91717015e-02 -1.53229403e+00 1.38707280e-01 -7.68984715e-03
9.52498674e-01 -4.10645366e-01 -3.26821446e-01 1.84807196e-01
8.52312922e-01 -1.15946162e+00 6.62898481e-01 2.94635385e-01
3.85893792e-01 -7.08287597e-01 4.88198042e-01 6.25414610e-01
-8.76278579e-01 -8.67473613e-03 -2.47813746e-01 -1.84967458e-01
1.07918911e-01 3.27436566e-01 -1.61098301e+00 2.84750462e-01
4.97856468e-01 1.10593066e-01 -9.87002134e-01 9.90484595e-01
-4.22046721e-01 6.98168933e-01 -6.17827713e-01 7.66217411e-02
2.39073589e-01 -3.01354021e-01 -1.46971466e-02 4.34368968e-01
4.78741705e-01 4.18339998e-01 2.64290214e-01 9.07012284e-01
1.72949865e-01 -1.07884765e-01 -5.68644106e-01 -1.91014886e-01
4.08194721e-01 9.98730659e-01 -1.01723051e+00 -8.63909274e-02
-1.03901036e-01 5.74207842e-01 7.14303330e-02 1.84340954e-01
-3.61177653e-01 2.48364419e-01 -1.57112002e-01 7.64309585e-01
3.14519942e-01 -3.75774682e-01 -6.47119045e-01 -4.39678639e-01
-1.52427927e-01 -6.75984740e-01 1.67697638e-01 -1.56526172e+00
-9.76001322e-01 3.46012741e-01 2.34430537e-01 -8.32454622e-01
-4.31115359e-01 -5.54750502e-01 -6.47276938e-01 4.05589908e-01
-8.24294984e-01 -1.81072986e+00 -6.95947587e-01 1.82947412e-01
5.62499702e-01 -6.78191334e-02 9.92217302e-01 1.55045182e-01
-1.99296713e-01 -2.12650344e-01 -2.55997956e-01 5.02270460e-01
-2.98526585e-01 -1.12460315e+00 3.34175318e-01 7.55546689e-01
7.18404412e-01 -8.95489659e-03 5.74200094e-01 -5.78722715e-01
-9.17098284e-01 -1.16009700e+00 4.53799039e-01 -5.95087409e-01
1.99354514e-01 -3.42653841e-02 -7.96420932e-01 1.03634715e+00
-2.08276197e-01 -4.99364406e-01 7.74210513e-01 -3.13590825e-01
1.63770586e-01 -6.60873801e-02 -1.17703021e+00 4.85773295e-01
9.05434906e-01 -4.04221982e-01 -8.21423233e-01 5.58607817e-01
2.56388158e-01 -2.41464838e-01 -9.25473750e-01 4.85183269e-01
3.56192827e-01 -9.00821447e-01 1.29825258e+00 -4.61383462e-01
5.46334982e-01 -5.96581638e-01 -4.73482609e-01 -1.29772878e+00
4.24228013e-02 3.02905172e-01 5.73640704e-01 1.03745902e+00
5.71435452e-01 1.80915549e-01 1.24347019e+00 1.23080301e+00
-4.87928003e-01 -3.74221355e-01 -6.39065027e-01 -3.76802236e-01
1.80013180e-02 -8.62269640e-01 7.76417732e-01 8.49392295e-01
-8.09732556e-01 2.70250201e-01 3.57818305e-01 3.90945524e-01
5.80450237e-01 1.85454905e-01 9.15203214e-01 -1.52004540e+00
3.42927948e-02 -7.29737952e-02 -8.17385018e-01 -6.15549445e-01
2.33843565e-01 -7.67778814e-01 2.50773072e-01 -2.02199244e+00
-1.33104518e-01 -8.99186254e-01 4.32684332e-01 5.70848048e-01
2.42400859e-02 3.01884145e-01 5.46153961e-03 1.02805801e-01
9.20001268e-02 3.50597650e-01 1.21007836e+00 -3.19374621e-01
-1.81618586e-01 2.38343894e-01 -5.26873589e-01 9.76565123e-01
7.29357839e-01 -3.79957616e-01 -7.26866186e-01 -6.31507933e-01
-3.82898701e-03 -9.61713865e-02 8.17030191e-01 -1.41736972e+00
1.53649412e-02 -1.65889248e-01 8.33940506e-01 -8.82677734e-01
4.86162633e-01 -1.39755857e+00 8.48632395e-01 3.57523620e-01
-1.91566959e-01 -1.34006828e-01 3.68326128e-01 1.36506528e-01
4.03011180e-02 -4.37569618e-01 6.70150399e-01 -8.21823537e-01
-1.20077193e+00 -5.64328171e-02 -4.70574170e-01 4.48562987e-02
1.30609322e+00 -7.14864314e-01 2.32520849e-02 -4.85177785e-01
-9.31256771e-01 -7.23583400e-02 7.00868368e-01 1.14644848e-01
9.09347594e-01 -1.33114195e+00 -6.70238018e-01 3.71974468e-01
-3.84912491e-02 7.67854273e-01 3.13137621e-01 -2.03284591e-01
-1.17419612e+00 5.62946200e-02 -3.89327675e-01 -8.36388648e-01
-1.00126767e+00 3.85883719e-01 5.04649341e-01 8.11314210e-02
-4.80872899e-01 6.41838193e-01 -2.20877156e-02 -8.09586644e-01
-2.52038509e-01 -1.42128110e-01 7.93806762e-02 -4.04074460e-01
1.14328064e-01 -2.24985424e-02 3.20663601e-02 -6.48007095e-01
2.43797973e-02 4.25385475e-01 5.69163084e-01 -2.00600252e-01
1.74720216e+00 3.01985592e-01 2.06940114e-01 4.22891438e-01
6.49346590e-01 -5.64939201e-01 -1.27273977e+00 -8.74406472e-02
2.83376604e-01 -4.48565602e-01 2.16009483e-01 -9.62174594e-01
-1.03135657e+00 8.40245903e-01 8.70720088e-01 1.87318712e-01
9.38367367e-01 8.94837454e-02 5.89124382e-01 5.17074764e-01
4.91043538e-01 -1.30002642e+00 3.46751451e-01 -1.10613979e-01
8.60043585e-01 -1.16045213e+00 2.32127979e-01 -4.01694626e-01
-5.87758899e-01 1.14645135e+00 8.26477885e-01 -7.83240348e-02
6.15671575e-01 1.78111315e-01 1.83518082e-01 -5.68583548e-01
-1.10371746e-01 -2.85651863e-01 4.00701195e-01 8.31520796e-01
1.53081924e-01 2.49326512e-01 5.67746699e-01 -5.68790026e-02
-6.08307898e-01 2.34153289e-02 4.87071842e-01 1.18827355e+00
-7.60494113e-01 -9.40534115e-01 -5.38750648e-01 4.42678303e-01
2.29404658e-01 -5.96412979e-02 -4.71641093e-01 1.11665571e+00
2.73825496e-01 8.06084692e-01 4.04722542e-01 -2.95540959e-01
4.24378186e-01 3.75545681e-01 6.25295520e-01 -8.20556402e-01
-3.38166028e-01 -4.19087440e-01 3.04827094e-01 1.66237786e-01
-7.20322907e-01 -5.99288523e-01 -1.27273762e+00 -1.82287559e-01
-1.60875887e-01 1.25819528e-02 1.00583529e+00 1.09965634e+00
3.49979177e-02 6.03549778e-01 3.87492836e-01 -9.46216464e-01
5.09175472e-02 -7.61668026e-01 -2.61267632e-01 6.12336516e-01
-3.76754671e-01 -3.31797630e-01 -1.36929778e-02 7.09697664e-01] | [9.65963077545166, 0.9172974824905396] |
0ab370d5-fc80-42cf-acfe-79ccef968bbe | automated-medical-coding-on-mimic-iii-and | 2304.10909 | null | https://arxiv.org/abs/2304.10909v1 | https://arxiv.org/pdf/2304.10909v1.pdf | Automated Medical Coding on MIMIC-III and MIMIC-IV: A Critical Review and Replicability Study | Medical coding is the task of assigning medical codes to clinical free-text documentation. Healthcare professionals manually assign such codes to track patient diagnoses and treatments. Automated medical coding can considerably alleviate this administrative burden. In this paper, we reproduce, compare, and analyze state-of-the-art automated medical coding machine learning models. We show that several models underperform due to weak configurations, poorly sampled train-test splits, and insufficient evaluation. In previous work, the macro F1 score has been calculated sub-optimally, and our correction doubles it. We contribute a revised model comparison using stratified sampling and identical experimental setups, including hyperparameters and decision boundary tuning. We analyze prediction errors to validate and falsify assumptions of previous works. The analysis confirms that all models struggle with rare codes, while long documents only have a negligible impact. Finally, we present the first comprehensive results on the newly released MIMIC-IV dataset using the reproduced models. We release our code, model parameters, and new MIMIC-III and MIMIC-IV training and evaluation pipelines to accommodate fair future comparisons. | ['Lars Maaløe', 'Tuukka Ruotsalo', 'Maria Maistro', 'Lasse Borgholt', 'Jakob D. Havtorn', 'Alexander Junge', 'Joakim Edin'] | 2023-04-21 | null | null | null | null | ['medical-code-prediction'] | ['medical'] | [ 4.50270802e-01 4.21184391e-01 -6.18085682e-01 -5.52067816e-01
-1.35052466e+00 -6.20715618e-01 3.38891745e-01 7.00879514e-01
-3.33275884e-01 7.48071492e-01 4.02861804e-01 -7.52165616e-01
-3.63738030e-01 -1.34183750e-01 -3.29978734e-01 -2.55712420e-01
-1.07067443e-01 9.54642177e-01 -1.08016059e-01 4.01495725e-01
1.62226826e-01 -6.10561892e-02 -1.18661308e+00 8.49477232e-01
1.06323028e+00 5.99380732e-01 -1.41804069e-01 8.68433952e-01
4.56452072e-02 1.37192333e+00 -5.26838481e-01 -8.03159952e-01
4.18805070e-02 -3.79306734e-01 -9.89782751e-01 -2.30539858e-01
3.55188549e-01 -4.21875060e-01 4.14401628e-02 7.87554026e-01
4.81380612e-01 -5.93780935e-01 1.04792154e+00 -1.09120393e+00
-6.79145873e-01 7.54894853e-01 -2.54426867e-01 -1.45416632e-01
4.87349778e-01 1.59313101e-02 9.47760642e-01 -3.90799522e-01
7.87843108e-01 7.23903716e-01 1.43847823e+00 7.77347803e-01
-1.25131798e+00 -5.58322191e-01 -2.84287989e-01 -4.89718094e-02
-1.49825573e+00 -5.24178505e-01 -4.11178246e-02 -1.08947015e+00
1.24407184e+00 3.92890662e-01 6.12481058e-01 1.04890680e+00
3.77209365e-01 6.04665935e-01 8.37285757e-01 -6.79632187e-01
3.08688551e-01 3.21798265e-01 2.90005565e-01 9.97669101e-01
4.76825476e-01 -2.20689438e-02 1.94314033e-01 -8.56133819e-01
2.72272825e-01 1.27100766e-01 -3.33670706e-01 -3.63727123e-01
-1.29725099e+00 8.99770319e-01 -4.11263518e-02 2.49828905e-01
-2.60946184e-01 -1.25996858e-01 5.89904904e-01 -5.01305610e-02
2.45201856e-01 6.42810762e-01 -8.12197924e-01 -5.39774358e-01
-1.35311234e+00 2.17477217e-01 9.45868790e-01 1.19025242e+00
2.49667913e-01 -5.68069696e-01 -6.19189739e-01 1.13318706e+00
4.51396629e-02 2.03353897e-01 7.81386912e-01 -1.12586761e+00
4.33127940e-01 5.99996984e-01 6.04135804e-02 -6.52910411e-01
-7.60968506e-01 -5.90299666e-01 -8.94611239e-01 -4.48569596e-01
1.68284461e-01 -3.01892728e-01 -1.10359097e+00 1.32812023e+00
-2.07147017e-01 1.49322050e-02 -1.48138320e-02 4.15849954e-01
8.73194993e-01 8.57146233e-02 3.35120201e-01 -2.60858208e-01
1.48898792e+00 -9.47363734e-01 -7.42108822e-01 2.56514680e-02
1.37573314e+00 -7.75072217e-01 7.82979429e-01 4.06307757e-01
-1.14094758e+00 -2.07639366e-01 -6.12735569e-01 -7.14768916e-02
-7.64751956e-02 6.58808708e-01 7.28597403e-01 9.01337504e-01
-1.03360379e+00 6.03538156e-01 -8.97457242e-01 -2.90993720e-01
6.97850585e-01 2.37627327e-01 -1.64206252e-01 -1.38648555e-01
-1.08381355e+00 8.51056099e-01 3.24584961e-01 -6.52117372e-01
-5.52263796e-01 -1.08637536e+00 -8.34736466e-01 1.03680052e-01
-1.12656161e-01 -1.00303662e+00 1.73382151e+00 -5.07634103e-01
-8.94671857e-01 1.41293252e+00 -1.01653002e-01 -6.86786056e-01
6.98885918e-01 1.12748981e-01 -4.46111411e-01 1.83719456e-01
3.05144787e-01 7.62744188e-01 5.69094270e-02 -9.25099492e-01
-6.12189054e-01 7.66038820e-02 -2.22592130e-01 -2.44354382e-01
-2.97117472e-01 4.93865460e-02 -5.32665372e-01 -7.76780725e-01
-3.40824723e-01 -1.02056468e+00 -3.58069539e-01 -2.54695892e-01
-4.48412836e-01 2.47176990e-01 -5.29255390e-01 -6.47944570e-01
2.00471759e+00 -2.14593124e+00 -3.44201952e-01 1.19689323e-01
3.58288765e-01 1.48397550e-01 1.45044461e-01 3.44077229e-01
-4.17639136e-01 4.70297903e-01 -6.33323014e-01 -5.88793218e-01
-1.64602846e-01 2.43952602e-01 -5.61555959e-02 4.14869219e-01
-6.08774181e-03 6.74408615e-01 -6.97827280e-01 -1.05736411e+00
-9.41811278e-02 2.31572360e-01 -1.16595805e+00 -3.99850756e-02
9.42328572e-02 -1.08198225e-02 -8.46625492e-02 8.26203287e-01
5.17335653e-01 -9.26347554e-01 2.51692504e-01 5.99175356e-02
6.93076849e-02 5.04539728e-01 -7.49109805e-01 1.50687921e+00
-3.27120274e-01 2.63318866e-01 -9.38890651e-02 -8.38861167e-01
3.45803648e-01 3.84014994e-01 8.02878857e-01 -2.67561644e-01
9.32738185e-02 2.72933632e-01 1.52061999e-01 -7.97800601e-01
1.70326546e-01 5.71785821e-03 4.85843010e-02 2.66868621e-01
5.23337582e-03 -7.54905194e-02 2.44002312e-01 2.91807920e-01
1.56480312e+00 -2.87299812e-01 9.03459728e-01 -2.35306486e-01
2.12715015e-01 3.98441434e-01 5.07691681e-01 1.02096224e+00
-2.29742378e-01 1.09466338e+00 6.84753835e-01 -3.15765649e-01
-1.01366603e+00 -6.84077263e-01 -8.37446451e-01 7.46144295e-01
-6.03007436e-01 -8.76189470e-01 -8.40905309e-01 -8.78275990e-01
3.12046707e-01 8.84460390e-01 -9.53773797e-01 -2.11804993e-02
-1.69101372e-01 -1.09304965e+00 9.99039352e-01 4.36110646e-01
-1.93794042e-01 -6.30750477e-01 -6.94458008e-01 2.28987917e-01
-3.10327947e-01 -8.21283996e-01 -3.39729279e-01 3.11858058e-01
-8.56108665e-01 -1.53734982e+00 -8.51803422e-01 -7.78617501e-01
6.67384982e-01 -2.27171585e-01 1.49020445e+00 4.30957675e-01
-5.95482290e-01 2.69889981e-01 -3.78175795e-01 -4.75711852e-01
-8.77729058e-01 4.84610945e-01 -3.13398600e-01 -6.14918411e-01
6.24901533e-01 6.63533285e-02 -5.69788396e-01 3.37023800e-03
-7.32381165e-01 2.79487044e-01 6.77349269e-01 1.01701117e+00
4.32833850e-01 -4.29680616e-01 2.08516285e-01 -1.54595554e+00
6.23176634e-01 -6.43632233e-01 -2.87847489e-01 4.63421315e-01
-1.01591063e+00 1.04766235e-01 4.23009455e-01 -3.49359848e-02
-6.19575262e-01 1.96292758e-01 -4.35366213e-01 -1.91620663e-01
-2.10289344e-01 7.02007592e-01 5.49542546e-01 3.72850090e-01
9.85809326e-01 -1.96414515e-01 4.38039117e-02 -4.85493630e-01
-1.07930869e-01 1.15725887e+00 1.20412618e-01 -2.42821753e-01
1.78441390e-01 3.97429377e-01 -3.61572474e-01 -1.57641590e-01
-9.72780704e-01 -6.65273547e-01 -4.45797086e-01 2.73182154e-01
7.80689061e-01 -1.12186301e+00 -6.36701405e-01 -6.17565624e-02
-1.12072957e+00 -4.39743102e-01 -2.05144718e-01 6.86537087e-01
-5.62309206e-01 2.91595489e-01 -8.20014715e-01 -5.28999209e-01
-2.84575969e-01 -1.27200079e+00 1.19279802e+00 -4.51407522e-01
-9.81898606e-01 -1.06139946e+00 4.51753050e-01 5.11277497e-01
2.41692528e-01 1.83194220e-01 1.11860633e+00 -9.25472796e-01
1.64353713e-01 -1.70051277e-01 -3.42738271e-01 7.70858973e-02
2.70534545e-01 2.51565903e-01 -9.72790420e-01 -2.34661102e-01
-3.74280363e-01 -2.22673431e-01 1.03369129e+00 7.36934900e-01
1.56393623e+00 -3.67330700e-01 -8.10628295e-01 7.93215871e-01
1.19861078e+00 2.43424371e-01 4.13203001e-01 4.30062383e-01
3.66914302e-01 4.68840867e-01 3.59204352e-01 6.56685770e-01
4.77525890e-01 4.66542572e-01 -5.21336542e-03 -2.16573551e-01
-1.48803610e-02 4.94864695e-02 -6.81220442e-02 8.06510866e-01
1.35512173e-01 7.22832084e-02 -1.56165671e+00 6.85172975e-01
-1.63956797e+00 -8.45387518e-01 -3.63656729e-01 2.06323862e+00
1.49054229e+00 1.59921378e-01 -6.83178008e-02 -3.96058671e-02
5.42785943e-01 -5.29206157e-01 -2.70044655e-01 -2.87671417e-01
1.25399902e-01 1.70155242e-01 7.29847312e-01 5.92335820e-01
-1.12470210e+00 3.69260877e-01 7.84367514e+00 6.47801399e-01
-7.72197962e-01 3.50805849e-01 8.40182126e-01 -3.88003141e-01
-3.07677656e-01 -2.16631606e-01 -6.46471262e-01 6.99371338e-01
1.31417179e+00 2.03459598e-02 1.40218452e-01 1.11341202e+00
-1.07480101e-01 1.34388939e-01 -1.32956469e+00 1.07248783e+00
1.19225584e-01 -1.56312323e+00 -1.03953861e-01 1.11149758e-01
8.60375345e-01 2.55329818e-01 -3.86726670e-02 5.35035670e-01
6.76387966e-01 -1.09440625e+00 6.46215200e-01 5.10499299e-01
1.41112530e+00 -3.21509480e-01 1.15931904e+00 1.73349082e-02
-6.46554351e-01 -4.30235833e-01 -1.46772072e-01 1.04328193e-01
-4.28535827e-02 8.49338472e-01 -1.20125234e+00 4.41631436e-01
5.75004935e-01 8.06350589e-01 -9.51878846e-01 1.34946072e+00
2.17844203e-01 9.26721036e-01 9.61195901e-02 2.38543183e-01
6.02652244e-02 3.74158680e-01 -1.76413313e-01 1.79393637e+00
3.92616689e-01 7.00403452e-02 4.79815826e-02 5.69223642e-01
-3.41504440e-02 2.05566093e-01 -2.81338722e-01 1.16189167e-01
4.53671098e-01 8.45914304e-01 -4.41150278e-01 -7.85293698e-01
-5.81083238e-01 5.38143039e-01 3.35466236e-01 7.59604499e-02
-1.03835022e+00 -3.13056797e-01 5.66797972e-01 2.92128175e-01
1.13111679e-02 4.52588975e-01 -6.24338627e-01 -1.24269450e+00
-2.68489778e-01 -1.00914061e+00 7.65082836e-01 -6.91265941e-01
-1.18442762e+00 5.01188517e-01 6.08399808e-02 -1.49622619e+00
-6.45512760e-01 -6.47900164e-01 2.93694019e-01 5.07841647e-01
-1.43517399e+00 -7.43563890e-01 -1.79602578e-01 2.34504670e-01
1.16569355e-01 -2.65588939e-01 1.24277604e+00 7.28118181e-01
-4.00555730e-01 9.51853216e-01 4.21902776e-01 4.14256871e-01
1.23101008e+00 -1.29428983e+00 2.08785936e-01 1.94362253e-01
-2.90970802e-01 9.71346855e-01 5.73352754e-01 -7.73762405e-01
-5.08506179e-01 -1.29591882e+00 1.27378094e+00 -8.47960591e-01
4.37330753e-01 -1.68172613e-01 -8.13087106e-01 8.02515924e-01
5.16703352e-02 -2.99790114e-01 1.40154552e+00 2.83149540e-01
-2.66832381e-01 2.02273846e-01 -1.28100491e+00 3.06241363e-01
9.37259972e-01 -2.58925110e-01 -7.67541468e-01 6.43541515e-01
5.87250888e-01 -4.99467552e-01 -1.04656601e+00 3.93717498e-01
7.28803575e-01 -9.81582403e-01 7.75428355e-01 -8.08894157e-01
8.63349140e-01 1.10726416e-01 1.12949610e-01 -1.13839507e+00
-6.72525227e-01 -2.74008870e-01 1.08727358e-01 8.97811592e-01
8.47741842e-01 -4.86536682e-01 6.15648448e-01 7.95802891e-01
-2.96073914e-01 -7.79052079e-01 -7.97216654e-01 -5.14822185e-01
1.38487801e-01 -3.54868263e-01 6.35767937e-01 1.54333007e+00
4.55529600e-01 -1.34933054e-01 -6.77788928e-02 -1.75257713e-01
3.63756806e-01 -8.40680525e-02 2.94685483e-01 -1.49413693e+00
-5.42199016e-01 -5.73192477e-01 -6.84404373e-02 -6.14861846e-01
-9.60155949e-02 -1.09941483e+00 -1.58077359e-01 -1.75390649e+00
8.02592695e-01 -6.64598942e-01 -2.50892997e-01 9.85191107e-01
-3.11082989e-01 1.35037839e-01 -8.57548043e-02 5.98965764e-01
-6.49997532e-01 -1.86639816e-01 6.62963092e-01 -2.13193282e-01
-3.85934785e-02 -6.74678199e-03 -8.98014247e-01 5.47045290e-01
7.90408373e-01 -8.91615272e-01 -2.89877504e-01 -4.78292733e-01
4.43828553e-01 1.31806210e-01 1.32127389e-01 -9.90162730e-01
2.29109392e-01 -2.84209494e-02 4.76379663e-01 -2.59506851e-01
-7.49148205e-02 -8.36006999e-01 2.36544892e-01 9.21196043e-01
-8.69533718e-01 2.40657344e-01 3.23072016e-01 2.81207025e-01
-1.15311481e-01 -5.53918481e-01 6.40089035e-01 -2.58253336e-01
1.63317934e-01 -6.99457526e-02 -6.52489901e-01 3.15639645e-01
8.76113892e-01 -1.41373679e-01 -2.99901813e-01 -1.21539906e-01
-7.61897504e-01 2.87659347e-01 5.90286970e-01 1.83137417e-01
1.89946935e-01 -1.07625961e+00 -9.73429620e-01 1.74468935e-01
6.54708505e-01 -3.66578728e-01 3.71465027e-01 9.74544168e-01
-9.56537068e-01 9.31772232e-01 9.84473675e-02 -6.17848158e-01
-1.48064852e+00 7.32404709e-01 3.43493193e-01 -7.05847383e-01
-3.04383993e-01 6.13887072e-01 -1.45210549e-01 -6.32274926e-01
3.50452095e-01 -8.93812418e-01 -2.41720513e-01 1.20582595e-01
6.30323172e-01 2.16250882e-01 4.84037369e-01 -2.62523234e-01
-5.72848797e-01 3.24131280e-01 -3.13443631e-01 1.11154236e-01
1.16749024e+00 1.30749598e-01 5.95087297e-02 1.34796113e-01
1.24352396e+00 1.79425813e-02 -5.19252062e-01 6.75283894e-02
-1.28371254e-01 -2.08300665e-01 -1.47072701e-02 -1.28559053e+00
-7.98458993e-01 6.83009148e-01 5.55006027e-01 2.16194943e-01
9.89071429e-01 -6.88618496e-02 2.98635572e-01 1.15615174e-01
1.10744126e-01 -9.97592449e-01 -5.90691566e-01 3.35852951e-01
4.71854389e-01 -1.23860407e+00 3.05130154e-01 -2.87635624e-01
-8.44046652e-01 8.37221205e-01 1.94006070e-01 4.93540496e-01
7.38300681e-01 5.34378171e-01 1.49596766e-01 -7.58484975e-02
-9.71072197e-01 1.63111448e-01 1.51235953e-01 4.87505794e-01
1.05291700e+00 2.70886183e-01 -3.37084085e-01 8.85543585e-01
-4.77023125e-01 5.17902911e-01 4.83518064e-01 8.91109228e-01
4.35310416e-02 -1.00445426e+00 -4.52733427e-01 1.01346219e+00
-9.23828661e-01 -5.94250560e-01 -1.90926880e-01 7.32951164e-01
2.50997573e-01 8.43613386e-01 1.07803479e-01 -2.39810079e-01
2.83789724e-01 2.60051906e-01 1.48718417e-01 -8.71129453e-01
-9.77411270e-01 -1.54279724e-01 1.83053106e-01 -4.14971352e-01
-3.45981270e-01 -8.74270260e-01 -1.29390359e+00 -2.12266177e-01
-1.26504153e-01 3.47504824e-01 3.57434481e-01 4.74366069e-01
9.02688622e-01 9.25171673e-01 -9.82449651e-02 -6.88334703e-02
-6.88578308e-01 -1.06215179e+00 -2.04447106e-01 4.37585860e-01
5.57380140e-01 -5.18078566e-01 -3.60055149e-01 4.99295026e-01] | [8.012947082519531, 6.806243419647217] |
2e317930-59e6-413f-9cba-1081cb808364 | a-generic-ensemble-based-deep-convolutional | 2004.07995 | null | https://arxiv.org/abs/2004.07995v1 | https://arxiv.org/pdf/2004.07995v1.pdf | A generic ensemble based deep convolutional neural network for semi-supervised medical image segmentation | Deep learning based image segmentation has achieved the state-of-the-art performance in many medical applications such as lesion quantification, organ detection, etc. However, most of the methods rely on supervised learning, which require a large set of high-quality labeled data. Data annotation is generally an extremely time-consuming process. To address this problem, we propose a generic semi-supervised learning framework for image segmentation based on a deep convolutional neural network (DCNN). An encoder-decoder based DCNN is initially trained using a few annotated training samples. This initially trained model is then copied into sub-models and improved iteratively using random subsets of unlabeled data with pseudo labels generated from models trained in the previous iteration. The number of sub-models is gradually decreased to one in the final iteration. We evaluate the proposed method on a public grand-challenge dataset for skin lesion segmentation. Our method is able to significantly improve beyond fully supervised model learning by incorporating unlabeled data. | ['Ruizhe Li', 'Dorothee Auer', 'Christian Wagner', 'Xin Chen'] | 2020-04-16 | null | null | null | null | ['semi-supervised-medical-image-segmentation', 'skin-lesion-segmentation', 'organ-detection'] | ['computer-vision', 'medical', 'medical'] | [ 6.65538371e-01 4.12306279e-01 -4.94482934e-01 -5.63530385e-01
-1.16719103e+00 -2.26557270e-01 2.88128257e-01 5.39757848e-01
-6.65175319e-01 6.37239635e-01 -2.44940650e-02 -1.24963202e-01
3.91563088e-01 -7.00358272e-01 -5.98622441e-01 -6.59480453e-01
2.45437846e-01 7.92304099e-01 5.24666488e-01 2.27740437e-01
-9.17133763e-02 2.08232641e-01 -1.12203944e+00 3.89144897e-01
1.10646272e+00 9.25889313e-01 2.87941277e-01 6.24924541e-01
-3.35435838e-01 9.12844956e-01 -3.00498009e-01 -2.15747282e-01
9.64077041e-02 -6.60322189e-01 -1.13368630e+00 5.33191860e-01
1.23047836e-01 -3.36671561e-01 1.58887893e-01 1.14283466e+00
4.28775430e-01 -2.62918383e-01 5.16387880e-01 -7.50665247e-01
-1.45183340e-01 6.46496773e-01 -5.13558090e-01 -2.32907087e-01
-3.79904985e-01 7.38851503e-02 7.50080168e-01 -5.14379442e-01
6.74864471e-01 7.52133250e-01 7.21434534e-01 8.57655108e-01
-1.13672280e+00 -3.28307956e-01 -1.77589767e-02 -3.54026482e-02
-1.17093492e+00 -9.03774500e-02 5.33583820e-01 -4.29240167e-01
3.65953296e-01 1.39927477e-01 6.97051346e-01 5.81693113e-01
-2.76382387e-01 1.05484271e+00 1.04130805e+00 -4.39196795e-01
3.88788909e-01 1.76264375e-01 2.06589893e-01 8.85715067e-01
1.57835305e-01 -3.24876010e-01 1.47484243e-02 6.16821796e-02
7.29123116e-01 -1.01615973e-01 -5.58966920e-02 -3.43728125e-01
-9.19654608e-01 8.48804355e-01 6.26372874e-01 4.08793449e-01
-5.02734840e-01 9.31186303e-02 6.69106603e-01 -2.22516075e-01
6.90715730e-01 2.81109720e-01 -6.15471303e-01 1.52180776e-01
-1.37135363e+00 -1.98118702e-01 6.06530666e-01 4.49387372e-01
8.41832995e-01 -3.18386793e-01 -3.50415260e-01 1.08225703e+00
2.92125791e-01 -3.34859565e-02 6.27557158e-01 -5.90668142e-01
1.66014224e-01 9.54925537e-01 -2.17608269e-02 -2.94599980e-01
-6.19105756e-01 -5.92722178e-01 -9.16714847e-01 1.18530974e-01
6.50752187e-01 -3.40801150e-01 -1.46571267e+00 1.44323730e+00
4.91714150e-01 1.50512397e-01 -1.65134408e-02 9.50570405e-01
9.43067789e-01 4.49396342e-01 4.07017648e-01 -1.88176841e-01
1.13898253e+00 -1.39348447e+00 -5.96201539e-01 -1.96495593e-01
1.01643264e+00 -5.72552025e-01 7.48427451e-01 4.08100635e-01
-9.55711305e-01 -4.57159460e-01 -8.16005766e-01 -1.07954852e-01
-9.00609642e-02 5.92330933e-01 4.34873521e-01 8.14716280e-01
-9.65646148e-01 3.66806924e-01 -1.16983116e+00 -4.29475814e-01
1.00655150e+00 6.29377246e-01 -3.99373919e-01 -2.38623708e-01
-8.02545547e-01 6.28918350e-01 5.08726597e-01 2.49537691e-01
-1.08574939e+00 -5.19379377e-01 -8.86840343e-01 -2.72110403e-01
4.42156136e-01 -4.42750126e-01 1.41682732e+00 -1.35728264e+00
-1.51362276e+00 1.19408345e+00 -1.26793772e-01 -6.67078555e-01
5.71790397e-01 -1.02373131e-01 -7.79926255e-02 2.62453020e-01
3.06877196e-02 8.88177633e-01 6.98432982e-01 -1.30003119e+00
-5.30281484e-01 -3.23269784e-01 -1.66486368e-01 -2.85883415e-02
-3.60459477e-01 2.50278070e-04 -8.21450174e-01 -3.90637875e-01
6.93302006e-02 -1.01547050e+00 -7.76146233e-01 1.48365453e-01
-8.67224455e-01 -5.37898466e-02 6.56061888e-01 -7.15962350e-01
1.12872469e+00 -1.85722876e+00 1.25786096e-01 2.48225987e-01
3.50683391e-01 7.90115058e-01 -1.61522537e-01 3.54839414e-02
1.80215854e-02 1.36001110e-01 -6.59696996e-01 -5.70396602e-01
-4.02208924e-01 1.46069378e-01 3.91573012e-01 3.89642447e-01
2.76944160e-01 8.52677464e-01 -9.98349190e-01 -7.98023999e-01
3.36926043e-01 3.57075870e-01 -4.99548912e-01 2.62474209e-01
-5.57066560e-01 8.21451068e-01 -3.46274555e-01 5.63049912e-01
6.11413777e-01 -6.32787108e-01 2.60102093e-01 -2.15261206e-01
3.56090106e-02 3.48220319e-02 -9.45803225e-01 1.88499558e+00
-4.24092084e-01 4.66883034e-01 8.90076458e-02 -1.30415428e+00
7.21207321e-01 3.16046149e-01 8.17015648e-01 -2.86671698e-01
2.77302235e-01 3.33461612e-01 1.42832562e-01 -7.80406535e-01
2.57218815e-02 -1.28791317e-01 1.98147207e-01 3.99728984e-01
1.78546786e-01 -1.22381924e-02 4.41071898e-01 3.36530544e-02
1.15547681e+00 1.05796814e-01 2.45532051e-01 6.72661513e-02
8.94708514e-01 3.56973469e-01 5.85578442e-01 4.23282087e-01
-2.69959301e-01 8.01991045e-01 5.39106131e-01 -4.64114249e-01
-1.02150798e+00 -5.10325432e-01 -3.23618859e-01 8.27552259e-01
-6.44726008e-02 -3.88917357e-01 -1.36745715e+00 -1.25767994e+00
-3.49945456e-01 3.80626082e-01 -8.31102669e-01 1.31565019e-01
-3.99427563e-01 -1.12303746e+00 5.35442412e-01 4.73726928e-01
6.41491771e-01 -1.15866768e+00 -5.82940638e-01 3.63341391e-01
-9.76696387e-02 -1.07308578e+00 -2.11768627e-01 2.17980683e-01
-8.24497640e-01 -1.21511781e+00 -9.64576066e-01 -9.61115122e-01
1.32725120e+00 -2.59745091e-01 8.44878852e-01 2.89040804e-01
-5.90349734e-01 -6.22161590e-02 -4.23509181e-01 -3.40445280e-01
-7.47055352e-01 3.81078780e-01 -5.46766341e-01 3.41622293e-01
2.92466640e-01 6.12682551e-02 -7.03339875e-01 1.92409679e-01
-1.14076400e+00 3.25457573e-01 8.02295864e-01 1.00972295e+00
1.00710046e+00 6.17666543e-02 6.22690797e-01 -1.48060691e+00
2.03198761e-01 -3.43207061e-01 -5.43175280e-01 1.55817389e-01
-2.37613261e-01 8.93810317e-02 6.86770260e-01 -2.56839484e-01
-1.12175560e+00 6.90472782e-01 -5.38272500e-01 2.79886578e-03
-3.21888298e-01 6.90602779e-01 -2.63131578e-02 -9.09800380e-02
6.04039907e-01 -8.65388289e-02 8.28760117e-02 -5.10972977e-01
2.91298509e-01 7.96577215e-01 3.34002912e-01 -3.47190201e-02
5.15392125e-01 4.16036010e-01 -9.44099054e-02 -4.81572270e-01
-1.13423610e+00 -5.42114317e-01 -9.27040339e-01 -2.45771036e-01
1.13841879e+00 -5.59061170e-01 -1.27277821e-01 6.07970774e-01
-1.00309277e+00 -8.64384890e-01 -2.19805181e-01 4.12157983e-01
-2.59683728e-01 3.56768668e-01 -6.86470330e-01 -5.66714883e-01
-4.12384748e-01 -1.39725220e+00 1.13776088e+00 4.23493624e-01
-5.97651973e-02 -1.10423517e+00 2.44350821e-01 5.79668581e-01
2.12481901e-01 2.99506336e-01 8.38436186e-01 -9.65884387e-01
-3.66598576e-01 -6.15380704e-01 -2.46505737e-01 5.67385674e-01
3.56834799e-01 -9.94834080e-02 -7.72109509e-01 -1.56573981e-01
-4.07797158e-01 -7.01071739e-01 1.09171391e+00 5.53864777e-01
1.54632807e+00 7.93416612e-03 -6.01522505e-01 4.78300750e-01
1.46994889e+00 -1.15262806e-01 4.81766820e-01 3.98084447e-02
7.85365880e-01 6.17672026e-01 6.05163336e-01 1.22476220e-01
3.32056582e-01 3.66229266e-01 5.42123318e-01 -6.74422741e-01
-1.04888327e-01 2.11031064e-02 -1.86353073e-01 5.31650424e-01
-9.38644353e-03 -1.45949706e-01 -1.10270250e+00 7.81643271e-01
-1.76347983e+00 -4.78175998e-01 -3.13292384e-01 2.23435354e+00
1.18376172e+00 2.30700135e-01 2.45692849e-01 2.63296574e-01
7.93521285e-01 -1.05314411e-01 -6.05601609e-01 -1.12586878e-01
3.16968739e-01 4.79508549e-01 5.05433559e-01 2.48418555e-01
-1.40328383e+00 1.06383586e+00 5.73634720e+00 7.96469748e-01
-1.28699172e+00 2.60762155e-01 1.00800729e+00 2.32855216e-01
7.63130561e-02 -1.82139724e-01 -6.79183483e-01 4.32706773e-01
8.59358847e-01 4.49243367e-01 -3.49426940e-02 7.51549304e-01
2.38282621e-01 -4.25836116e-01 -9.37875986e-01 6.12286389e-01
3.02209221e-02 -1.31521773e+00 -1.74673259e-01 -1.02806874e-02
1.07182908e+00 1.06255464e-01 -9.42588672e-02 9.05111879e-02
2.42580250e-01 -1.05245364e+00 3.24493684e-02 1.90331101e-01
8.90023172e-01 -6.05191469e-01 9.91190434e-01 5.87972045e-01
-8.02601933e-01 1.22993723e-01 -2.39240468e-01 5.49698055e-01
2.25517005e-01 9.39836562e-01 -1.00085270e+00 3.68471950e-01
2.89954931e-01 5.57329476e-01 -5.63861966e-01 1.38771403e+00
-2.08643377e-01 1.02977526e+00 -2.49823555e-01 7.61118606e-02
3.93359780e-01 -5.96369505e-02 9.01293010e-02 1.16855240e+00
-1.24747366e-01 -9.49186608e-02 3.47316623e-01 6.38870418e-01
-2.41394967e-01 4.08870608e-01 -4.32435796e-02 -9.09733027e-02
-4.14826684e-02 1.65632391e+00 -1.18548572e+00 -5.39622843e-01
-4.30838346e-01 9.92121100e-01 3.20146829e-01 4.86116065e-03
-7.32406318e-01 -2.17597842e-01 -1.78391263e-01 1.83972657e-01
1.72981828e-01 1.04543850e-01 -1.93710953e-01 -8.24646294e-01
-2.40677431e-01 -6.22149527e-01 3.57602000e-01 -3.34996462e-01
-1.01058447e+00 7.45642126e-01 -2.90061235e-01 -1.22147596e+00
-2.19899833e-01 -5.26091039e-01 -5.12379467e-01 6.80999994e-01
-1.61889458e+00 -1.21984768e+00 -5.25851130e-01 2.75641888e-01
6.34329855e-01 4.41097319e-02 9.97508407e-01 3.61411214e-01
-8.81748140e-01 4.65085924e-01 -6.67552650e-03 5.48594534e-01
6.39808714e-01 -1.51827574e+00 1.35214299e-01 7.17024207e-01
3.32944654e-02 1.58620819e-01 2.59094954e-01 -6.95600271e-01
-8.23871255e-01 -1.51785314e+00 6.51741982e-01 -5.93727045e-02
2.53567725e-01 -1.50935888e-01 -8.72697651e-01 5.47087133e-01
8.05627406e-02 4.92946386e-01 1.03831422e+00 -1.96611777e-01
1.07081302e-01 -4.34499718e-02 -1.37608540e+00 3.46678793e-01
4.57384437e-01 -1.94249302e-01 -6.16662912e-02 7.01094508e-01
3.75091791e-01 -5.86250603e-01 -8.48833323e-01 5.26042044e-01
2.46850401e-01 -7.15002537e-01 5.71364760e-01 -5.38267970e-01
5.62071204e-01 -2.73002714e-01 4.30598199e-01 -1.18366754e+00
8.21429491e-02 -5.05870283e-01 1.46645322e-01 1.01488781e+00
5.36283195e-01 -1.97341740e-01 1.27149165e+00 5.22619963e-01
-1.25842750e-01 -1.05928886e+00 -6.61398053e-01 -2.80500978e-01
4.83456347e-03 -3.31841946e-01 1.38645545e-01 7.25588143e-01
-6.17142208e-02 2.42520750e-01 -2.07292020e-01 1.29994901e-03
7.24555314e-01 -1.34174585e-01 5.78181684e-01 -1.08508682e+00
-1.02691785e-01 -1.57618299e-01 -2.78657496e-01 -9.48002636e-01
-4.76989597e-02 -9.59123671e-01 3.05466145e-01 -1.97185433e+00
3.69728297e-01 -5.78457952e-01 -2.48258516e-01 7.06698537e-01
-4.51691300e-01 7.60129094e-01 -1.53694421e-01 1.12732485e-01
-8.06528628e-01 4.53152470e-02 1.39074492e+00 -1.25124171e-01
-3.71900886e-01 3.46224219e-01 -4.23461527e-01 8.79300475e-01
9.42028224e-01 -4.92409468e-01 -2.84105569e-01 -2.89831698e-01
-2.21741229e-01 3.63375177e-03 1.67225122e-01 -9.69896615e-01
2.73117036e-01 3.44059840e-02 4.41518188e-01 -5.39305747e-01
4.11519520e-02 -8.15817773e-01 -1.32216722e-01 7.44548380e-01
-6.83684051e-01 -7.48668730e-01 1.23342603e-01 4.51241523e-01
-2.54172802e-01 -5.68560004e-01 1.13539898e+00 -3.28981668e-01
-4.52553928e-01 4.66023803e-01 -4.64930356e-01 -7.54183754e-02
1.07180440e+00 -6.03992231e-02 1.45413235e-01 -6.17820397e-02
-1.10055017e+00 2.73037702e-01 2.46140316e-01 -1.36600034e-02
2.41452292e-01 -1.05270696e+00 -7.74849951e-01 5.69131151e-02
9.91183817e-02 5.19577682e-01 2.29871422e-01 1.03772688e+00
-6.86232865e-01 3.25390458e-01 -3.47479731e-02 -7.83782959e-01
-1.23126352e+00 3.69455308e-01 5.33341885e-01 -7.59987831e-01
-5.04695594e-01 9.03386950e-01 -1.91489439e-02 -4.94309962e-01
2.33851716e-01 -4.41998363e-01 -3.99703771e-01 -1.70288563e-01
4.32278663e-01 1.71876922e-01 1.53018638e-01 -6.40841126e-01
-1.92607030e-01 4.68473762e-01 -3.98656666e-01 -2.23171730e-02
1.24355543e+00 2.21544102e-01 -2.15363339e-01 3.28154154e-02
1.26761758e+00 -2.37001836e-01 -1.26146936e+00 -2.60347933e-01
1.36167794e-01 -3.86285447e-02 2.00251967e-01 -9.95998681e-01
-1.36247492e+00 8.53874385e-01 6.47714376e-01 1.45435091e-02
1.34737313e+00 2.10252292e-02 9.62655306e-01 1.97202235e-01
1.79421097e-01 -1.15761840e+00 4.13238145e-02 6.49751574e-02
3.40578854e-01 -1.50303078e+00 -1.04695499e-01 -6.42403662e-01
-6.95792735e-01 1.05439746e+00 6.21255636e-01 -8.79188552e-02
5.69294810e-01 2.76032031e-01 2.53903389e-01 -8.29411373e-02
-3.61876726e-01 -5.58124602e-01 3.53404611e-01 4.54682976e-01
5.75385451e-01 -5.21012321e-02 -4.75034058e-01 5.99584341e-01
3.86344522e-01 4.04476166e-01 2.63031483e-01 8.25767577e-01
-4.21869099e-01 -1.61104560e+00 -1.40179068e-01 7.41413772e-01
-7.98026621e-01 -2.63126604e-02 -3.91988426e-01 6.18076622e-01
3.85265797e-01 7.77539372e-01 -8.39838609e-02 -1.89739652e-02
2.78235860e-02 -5.93620315e-02 4.33062911e-01 -1.04554880e+00
-7.38997459e-01 4.01947290e-01 4.55699712e-02 -2.71312803e-01
-7.24912167e-01 -6.19813740e-01 -1.54908323e+00 5.19590557e-01
-4.64642316e-01 6.89621866e-02 7.70813763e-01 1.06914365e+00
1.78279560e-02 5.89881718e-01 5.64156890e-01 -6.28695369e-01
-2.11400002e-01 -1.03730762e+00 -3.14146638e-01 4.08022940e-01
2.42980599e-01 -2.73753881e-01 3.89451869e-02 4.47523534e-01] | [14.738351821899414, -2.3668978214263916] |
5bbf144d-0c1a-4c40-b195-1ec6b9a6ed49 | near-optimal-representation-learning-for | 1810.01257 | null | http://arxiv.org/abs/1810.01257v2 | http://arxiv.org/pdf/1810.01257v2.pdf | Near-Optimal Representation Learning for Hierarchical Reinforcement Learning | We study the problem of representation learning in goal-conditioned
hierarchical reinforcement learning. In such hierarchical structures, a
higher-level controller solves tasks by iteratively communicating goals which a
lower-level policy is trained to reach. Accordingly, the choice of
representation -- the mapping of observation space to goal space -- is crucial.
To study this problem, we develop a notion of sub-optimality of a
representation, defined in terms of expected reward of the optimal hierarchical
policy using this representation. We derive expressions which bound the
sub-optimality and show how these expressions can be translated to
representation learning objectives which may be optimized in practice. Results
on a number of difficult continuous-control tasks show that our approach to
representation learning yields qualitatively better representations as well as
quantitatively better hierarchical policies, compared to existing methods (see
videos at https://sites.google.com/view/representation-hrl). | ['Sergey Levine', 'Shixiang Gu', 'Ofir Nachum', 'Honglak Lee'] | 2018-10-02 | near-optimal-representation-learning-for-1 | https://openreview.net/forum?id=H1emus0qF7 | https://openreview.net/pdf?id=H1emus0qF7 | iclr-2019-5 | ['2d-human-pose-estimation'] | ['computer-vision'] | [ 3.21413010e-01 5.21829545e-01 -5.67172825e-01 -3.59056257e-02
-9.43260550e-01 -5.86968839e-01 6.55892968e-01 1.80413350e-01
-1.99914977e-01 1.01902187e+00 4.28579926e-01 -2.28654951e-01
-3.27752173e-01 -6.67192638e-01 -7.46628225e-01 -8.31453979e-01
-4.09365416e-01 4.07677531e-01 -1.60436392e-01 -4.03257400e-01
3.85710984e-01 2.73201466e-01 -1.52758336e+00 1.23418905e-01
6.74356461e-01 8.83187175e-01 4.94618982e-01 7.61426449e-01
5.56060791e-01 9.81667399e-01 -4.34150696e-01 4.98677433e-01
3.42001498e-01 -7.89481938e-01 -9.57725167e-01 4.35750484e-01
3.67236733e-02 -2.33660460e-01 -2.19485983e-01 1.20943701e+00
1.49018437e-01 5.84983885e-01 8.31789911e-01 -1.08955967e+00
-5.40976703e-01 5.47604620e-01 -2.08432212e-01 1.38458058e-01
3.73660415e-01 2.42153972e-01 1.38736033e+00 -1.72782853e-01
5.53093493e-01 1.48556387e+00 1.04477562e-01 7.97435522e-01
-1.71210277e+00 -4.15771872e-01 4.15359169e-01 -6.63395971e-02
-1.04054976e+00 -3.76685590e-01 4.40026850e-01 -7.06534803e-01
7.94405639e-01 -6.21199869e-02 7.20625222e-01 1.01601899e+00
5.44920921e-01 6.89968348e-01 1.28940976e+00 -4.69126850e-01
5.33410788e-01 -1.78790599e-01 1.91421919e-02 8.51250648e-01
2.10021257e-01 8.47722471e-01 -3.15780520e-01 -2.92089675e-02
1.26706696e+00 -2.08964914e-01 -4.00561452e-01 -7.92305946e-01
-9.11756217e-01 1.21553588e+00 5.40560842e-01 2.11209148e-01
-6.49324358e-01 8.20736349e-01 2.26161748e-01 6.45937622e-01
-2.89204102e-02 9.00156379e-01 -2.99710542e-01 1.08027034e-01
-2.47275755e-01 5.37578762e-01 4.51942176e-01 8.85906577e-01
7.60108352e-01 5.06765604e-01 -3.15642238e-01 6.06304586e-01
2.82279789e-01 1.57473341e-01 3.97425681e-01 -1.60943723e+00
4.13837016e-01 2.12865219e-01 4.31471854e-01 -4.55083460e-01
-4.13189352e-01 -5.32862067e-01 -5.42447984e-01 6.36980414e-01
4.33799475e-01 -3.59147221e-01 -7.21769214e-01 2.20566678e+00
6.99789077e-02 -1.48260802e-01 1.98122889e-01 8.75428855e-01
-2.57193983e-01 7.65745461e-01 8.01854059e-02 -5.39339900e-01
1.03945160e+00 -6.71235383e-01 -7.30066955e-01 -1.80841893e-01
7.27482438e-01 -7.46026710e-02 1.04830337e+00 3.21848720e-01
-1.26002264e+00 -4.70766902e-01 -9.43533778e-01 2.92158276e-01
-5.55153638e-02 -6.15509925e-03 5.38512588e-01 1.37709066e-01
-1.16835666e+00 1.00129426e+00 -8.71334016e-01 -3.93969089e-01
1.70573637e-01 5.07776916e-01 -4.26626690e-02 3.76704097e-01
-1.14944077e+00 1.06383908e+00 7.56823123e-01 -2.18024105e-01
-1.81104755e+00 -2.74658978e-01 -8.34479749e-01 1.91856042e-01
5.48158288e-01 -8.34495962e-01 1.98555148e+00 -8.17139328e-01
-1.70387316e+00 4.94129658e-01 9.98241305e-02 -6.82069421e-01
5.34548350e-02 -2.44637087e-01 1.90300301e-01 1.79776609e-01
1.18704118e-01 6.48804247e-01 1.13911188e+00 -1.45896268e+00
-9.01185453e-01 -2.68950045e-01 6.53347015e-01 5.10381401e-01
-9.90441963e-02 -2.95920402e-01 1.93581223e-01 -5.80550075e-01
-1.69069678e-01 -9.90804911e-01 -7.71942914e-01 -2.20608860e-01
-2.19853088e-01 -3.44309568e-01 4.14315641e-01 -3.34453106e-01
1.10809970e+00 -1.78645396e+00 9.33254004e-01 1.20818257e-01
2.58404225e-01 -3.22969347e-01 -1.45830974e-01 5.59243202e-01
-1.25139058e-01 8.20395723e-02 -1.82560518e-01 -1.18469320e-01
2.74869084e-01 4.65005189e-01 -6.56416178e-01 5.34014761e-01
2.42825877e-02 6.84850872e-01 -9.74391997e-01 -1.76650137e-01
2.67861843e-01 5.19929267e-02 -8.26741517e-01 4.95789409e-01
-6.02838099e-01 7.70251274e-01 -9.74728405e-01 1.46971136e-01
-3.05794209e-01 -2.57554114e-01 5.30687571e-01 3.50831211e-01
-1.36581913e-01 5.01454771e-01 -8.61282945e-01 1.46784663e+00
-6.26551509e-01 3.23282361e-01 3.88696492e-01 -1.37381148e+00
6.69311643e-01 4.43418980e-01 4.97772127e-01 -5.60081542e-01
4.28734630e-01 -1.60654232e-01 1.08408332e-01 -1.72463953e-01
4.89272654e-01 -3.85093093e-01 -4.68454868e-01 3.05930287e-01
1.69233829e-01 -4.25644904e-01 2.22539440e-01 -1.53247709e-03
9.70055819e-01 3.25379342e-01 7.51953781e-01 -5.34024775e-01
3.44812512e-01 1.06102586e-01 4.36288744e-01 9.83567476e-01
-1.70841724e-01 9.63030942e-03 8.38492095e-01 7.08671138e-02
-9.92411554e-01 -1.00981295e+00 1.15386590e-01 1.39952040e+00
-9.86373127e-02 -4.93167788e-01 -7.24223673e-01 -3.48755896e-01
4.09411564e-02 8.48921478e-01 -8.98402095e-01 -3.34103227e-01
-6.21146560e-01 -9.79926214e-02 -5.10888882e-02 5.81293046e-01
3.10150087e-02 -1.30649281e+00 -1.02693021e+00 2.08020478e-01
2.99028367e-01 -6.75885081e-01 -4.09684509e-01 6.60535336e-01
-1.19467366e+00 -9.79303300e-01 -5.44856489e-01 -6.59851432e-01
6.73951626e-01 -1.33099714e-02 9.56063151e-01 -9.86405462e-02
4.73926850e-02 7.85907149e-01 -1.67791024e-01 -1.82195157e-01
-5.51380932e-01 5.36419339e-02 1.65186226e-01 -3.61348987e-01
-4.41131175e-01 -6.72545731e-01 -3.71449023e-01 -6.34898692e-02
-6.89282417e-01 1.34801775e-01 3.64474416e-01 9.05811369e-01
5.87530494e-01 4.91721667e-02 5.53530753e-01 -5.51683784e-01
1.05965149e+00 -3.64140362e-01 -1.27086318e+00 6.25631167e-03
-5.45207083e-01 5.49585104e-01 8.56488526e-01 -2.91820854e-01
-7.58830249e-01 1.94023743e-01 4.13614303e-01 -5.29890597e-01
-1.13730364e-01 4.55151021e-01 7.60414898e-02 2.12912694e-01
6.23785496e-01 2.02159792e-01 6.16012476e-02 -2.82475054e-01
6.85366213e-01 2.22522348e-01 3.64938766e-01 -1.22660589e+00
5.66429794e-01 -1.66370794e-02 1.25823840e-01 -6.38010085e-01
-9.85848844e-01 -1.38397351e-01 -4.39981371e-01 -2.26599962e-01
9.77980494e-01 -7.72742689e-01 -9.84917343e-01 -2.26672575e-01
-9.27946985e-01 -8.71204138e-01 -7.21185446e-01 4.04304415e-01
-1.59148800e+00 -2.01152951e-01 -6.22517467e-01 -1.11803925e+00
1.98639482e-01 -1.29731381e+00 9.19858456e-01 1.97688520e-01
-1.75960362e-01 -1.00621831e+00 2.70412982e-01 -1.38713494e-01
2.00539276e-01 2.27606788e-01 1.15176702e+00 -1.51326925e-01
-5.75988114e-01 3.66099715e-01 1.80721745e-01 1.59022897e-01
9.57915336e-02 -4.49066490e-01 -5.90766251e-01 -6.56985700e-01
-7.95585215e-02 -7.52076864e-01 7.89032161e-01 7.16552675e-01
1.31869996e+00 -5.49475968e-01 -1.89037964e-01 2.91924357e-01
1.55504012e+00 5.28524876e-01 2.84891009e-01 3.26626420e-01
3.96970719e-01 6.49253488e-01 7.25835145e-01 6.86330676e-01
1.30201802e-01 7.29606092e-01 7.34107971e-01 4.56170917e-01
2.40455791e-01 -5.00468671e-01 6.29282057e-01 5.38185060e-01
-2.25963891e-01 -1.81051474e-02 -5.56003034e-01 4.68675643e-01
-2.29953170e+00 -9.46500123e-01 4.99124318e-01 2.33655357e+00
7.54015923e-01 7.26376846e-02 4.10520256e-01 -1.72899127e-01
6.63790882e-01 1.98468775e-01 -8.29170465e-01 -6.15110934e-01
4.91456956e-01 1.81347355e-01 4.51406270e-01 8.16184163e-01
-1.02039647e+00 1.02245212e+00 7.33144855e+00 5.90292037e-01
-8.34734499e-01 1.77149847e-02 3.20018739e-01 4.78396798e-03
-8.97577256e-02 4.60724831e-02 -7.77688742e-01 8.18864405e-02
1.16900313e+00 -6.42449677e-01 8.83577406e-01 9.31354105e-01
5.41542888e-01 1.36257410e-01 -1.36435807e+00 5.63402891e-01
-6.30296707e-01 -1.28147137e+00 1.30777314e-01 3.77998084e-01
8.05887341e-01 -2.87487000e-01 1.59807220e-01 7.50696838e-01
1.11083949e+00 -1.08858597e+00 7.97532499e-01 3.08925152e-01
5.99868715e-01 -7.95578659e-01 -3.33686895e-03 4.97118890e-01
-1.29342055e+00 -6.38491035e-01 -3.80850136e-01 -2.98425853e-01
-8.93043634e-03 -2.79937476e-01 -4.05591667e-01 3.84155363e-01
2.80152440e-01 8.56316268e-01 -1.54059362e-02 7.40834892e-01
-6.03853524e-01 4.13525909e-01 6.30886406e-02 -8.64142030e-02
4.91235822e-01 -3.88726890e-01 4.18451488e-01 8.32247138e-01
2.79094934e-01 2.73861110e-01 7.33638823e-01 8.84328008e-01
-3.18811908e-02 -2.19052613e-01 -9.94757950e-01 -3.29138428e-01
2.17240170e-01 8.62038851e-01 -2.87729353e-01 -3.70035201e-01
1.01755500e-01 6.03068292e-01 9.06225324e-01 5.32715023e-01
-8.64102483e-01 2.68337457e-03 9.89924192e-01 -1.04573458e-01
3.82478178e-01 -5.01598418e-01 9.84562114e-02 -8.93118143e-01
-4.93749142e-01 -1.06097543e+00 4.52392310e-01 -6.59547269e-01
-8.81351888e-01 3.02213877e-01 3.46709073e-01 -1.27912688e+00
-7.73261189e-01 -5.83203793e-01 -4.14016247e-01 4.58739728e-01
-1.12070990e+00 -5.75115800e-01 2.85741180e-01 6.05721056e-01
5.96448839e-01 -1.11437626e-01 9.57638383e-01 -3.14210355e-01
-6.08723044e-01 1.64965943e-01 2.01169327e-01 -3.12045544e-01
-4.45017591e-02 -1.53761363e+00 -2.51272559e-01 4.91971463e-01
-1.05492420e-01 5.03421426e-01 9.46279287e-01 -2.97313064e-01
-1.34069788e+00 -1.07688129e+00 2.01899543e-01 -1.28031090e-01
1.08846092e+00 -7.95293450e-02 -6.27418101e-01 1.10527515e+00
3.65309566e-01 -3.29612404e-01 2.88576722e-01 8.04539546e-02
-1.21799171e-01 2.15177432e-01 -8.13193440e-01 8.29084933e-01
1.09626317e+00 -5.67436874e-01 -7.67388761e-01 4.12342161e-01
8.83722425e-01 -3.21588635e-01 -9.68971431e-01 1.45113304e-01
3.38071018e-01 -6.53838217e-01 8.10587764e-01 -1.16073298e+00
5.96056342e-01 -2.24103242e-01 -4.24177587e-01 -1.79289794e+00
-6.88544691e-01 -9.73285675e-01 -3.85211140e-01 4.36772406e-01
1.86076313e-01 -6.71871543e-01 6.12357855e-01 1.54570341e-01
-3.47611368e-01 -8.56783152e-01 -8.08934450e-01 -1.00192273e+00
4.99382406e-01 -5.93938306e-02 1.17512926e-01 6.46262109e-01
3.87880802e-01 5.64040720e-01 -3.87273163e-01 2.55338490e-01
7.32480347e-01 4.23508883e-01 4.33190823e-01 -9.09554183e-01
-5.28578937e-01 -7.34940350e-01 -1.36143183e-02 -1.35572433e+00
5.22958040e-01 -9.30594325e-01 3.09726685e-01 -1.53267217e+00
-4.43901271e-02 -2.19411552e-01 -5.88895917e-01 6.42865002e-01
1.33921638e-01 -5.64613879e-01 5.94375551e-01 2.10740507e-01
-6.99629128e-01 9.68028784e-01 1.65999472e+00 4.08316031e-02
-3.54685605e-01 -9.92613062e-02 -9.01137769e-01 6.43476129e-01
1.34711897e+00 -5.54979324e-01 -7.24689722e-01 -7.77377337e-02
-2.18643352e-01 7.08193064e-01 2.92223662e-01 -9.54979539e-01
-7.78518468e-02 -6.67259395e-01 6.85871243e-02 -1.43090487e-01
5.15869975e-01 -6.11074269e-01 -2.51885742e-01 9.13636088e-01
-1.18018460e+00 2.38894358e-01 6.09391220e-02 7.71318376e-01
2.55787708e-02 -3.98053288e-01 8.85357678e-01 -3.05403829e-01
-4.84736860e-01 1.05745971e-01 -9.45811689e-01 1.63595930e-01
1.08057570e+00 1.86642006e-01 -2.24955544e-01 -6.98926210e-01
-1.06225657e+00 6.00134015e-01 2.35820308e-01 2.60509819e-01
4.86124545e-01 -1.29914021e+00 -4.17933911e-01 -1.80782586e-01
-4.82490286e-02 -2.33960241e-01 -2.54186243e-01 5.27182519e-01
1.29890040e-01 6.80775046e-01 -5.35208642e-01 -3.43202919e-01
-8.44145954e-01 5.85412443e-01 5.78811049e-01 -6.82548940e-01
-5.80622315e-01 3.25364530e-01 5.83206713e-01 -2.79397100e-01
3.29162925e-01 -6.49078548e-01 -3.55803341e-01 -1.71091467e-01
3.20036709e-01 1.47976711e-01 -6.46832883e-01 -3.90699357e-01
1.46540150e-01 4.66432184e-01 -6.22878559e-02 -5.08209884e-01
1.04475939e+00 -6.62626699e-02 1.45136669e-01 4.97655988e-01
9.14905965e-01 -6.57708168e-01 -1.64584112e+00 -1.70792453e-02
1.08518220e-01 -2.38866746e-01 7.19998032e-02 -2.90159613e-01
-8.79126847e-01 6.88361585e-01 4.47364628e-01 5.28111696e-01
8.63211632e-01 -4.94506881e-02 6.77915961e-02 6.12733662e-01
7.85911739e-01 -1.06256318e+00 4.58257407e-01 7.77641833e-01
1.36105931e+00 -7.13431895e-01 -1.28592297e-01 2.34731138e-02
-6.05057299e-01 9.08312380e-01 6.87052965e-01 -7.26700425e-01
4.64266628e-01 2.71781068e-02 -4.82022613e-01 -2.00395882e-01
-1.32161975e+00 -6.46022618e-01 -3.32605354e-02 8.23533297e-01
2.50000924e-01 2.72832602e-01 -1.05923183e-01 1.51429787e-01
-3.59497666e-01 -8.90868083e-02 5.76522410e-01 1.11521864e+00
-9.37454641e-01 -1.21400034e+00 -3.53869528e-01 4.04821813e-01
-1.75039530e-01 3.33320588e-01 -6.73979968e-02 7.54723132e-01
-3.63987386e-01 9.27452266e-01 -1.75527316e-02 -2.42320925e-01
3.03359061e-01 -8.99199471e-02 8.99633050e-01 -9.98837471e-01
-4.76613581e-01 1.00175150e-01 1.82358682e-01 -8.94334912e-01
-3.33706588e-01 -5.79080820e-01 -1.50148487e+00 -3.01468670e-02
5.47919609e-02 3.61372381e-01 2.26540506e-01 7.33333826e-01
1.36339977e-01 9.51861382e-01 7.00596273e-01 -1.19525313e+00
-1.21692967e+00 -6.85440242e-01 -7.24896431e-01 1.61301702e-01
6.73995793e-01 -9.56762195e-01 -3.35110128e-01 -1.67912487e-02] | [4.167900085449219, 1.696533441543579] |
77cde550-a652-4b2d-a2cc-fba71f9858bf | a-study-on-passage-re-ranking-in-embedding | 1804.08057 | null | http://arxiv.org/abs/1804.08057v4 | http://arxiv.org/pdf/1804.08057v4.pdf | A Study on Passage Re-ranking in Embedding based Unsupervised Semantic Search | State of the art approaches for (embedding based) unsupervised semantic
search exploits either compositional similarity (of a query and a passage) or
pair-wise word (or term) similarity (from the query and the passage). By
design, word based approaches do not incorporate similarity in the larger
context (query/passage), while compositional similarity based approaches are
usually unable to take advantage of the most important cues in the context. In
this paper we propose a new compositional similarity based approach, called
variable centroid vector (VCVB), that tries to address both of these
limitations. We also presents results using a different type of compositional
similarity based approach by exploiting universal sentence embedding. We
provide empirical evaluation on two different benchmarks. | ['Md. Faisal Mahbub Chowdhury', 'Alfio M. Gliozzo', 'Vijil Chenthamarakshan', 'Rishav Chakravarti'] | 2018-04-22 | null | null | null | null | ['passage-re-ranking'] | ['natural-language-processing'] | [ 1.07807077e-01 -2.84208596e-01 -5.26644826e-01 -1.47457972e-01
-8.87823284e-01 -5.60304761e-01 1.09442282e+00 9.72525358e-01
-8.76891553e-01 4.70199943e-01 7.72676706e-01 -1.09710380e-01
-6.29962444e-01 -8.30173254e-01 -4.81563471e-02 -4.78106827e-01
8.00390169e-02 5.34666479e-01 6.57108963e-01 -5.33522904e-01
7.90054083e-01 2.39380494e-01 -1.53350139e+00 1.22436695e-01
6.17822766e-01 7.07092345e-01 3.72246474e-01 5.63382268e-01
-6.97605848e-01 3.98851335e-01 -4.65449452e-01 -5.48728853e-02
-6.48227427e-03 -7.58398354e-01 -1.16462052e+00 -4.70440030e-01
3.88106555e-01 2.34345376e-01 -1.86028644e-01 9.89351273e-01
5.66095054e-01 6.86618268e-01 6.39609575e-01 -7.74454117e-01
-7.56901979e-01 5.47882140e-01 -7.10341036e-02 5.67112207e-01
7.31105745e-01 -4.00195539e-01 1.72884965e+00 -9.97607946e-01
8.88975978e-01 1.14119339e+00 5.88299215e-01 3.35409641e-01
-1.24270761e+00 7.34135434e-02 4.58448678e-02 6.36625409e-01
-1.27818143e+00 -2.81929046e-01 1.05905330e+00 -2.48794794e-01
1.29758513e+00 4.64080662e-01 3.30934882e-01 1.10393107e+00
-2.97574103e-01 6.37291253e-01 1.08922255e+00 -8.30851436e-01
3.04716468e-01 2.89353490e-01 5.19773602e-01 4.52350616e-01
-4.93377075e-02 -2.67597958e-02 -4.43021595e-01 -5.95235348e-01
1.17003471e-01 -6.94084018e-02 -2.88519889e-01 -4.53847408e-01
-1.18330479e+00 1.26782787e+00 3.69349092e-01 1.02705181e+00
-1.32460192e-01 1.24717444e-01 6.59594178e-01 3.64856273e-01
3.85424405e-01 8.09292138e-01 -3.81394535e-01 -3.44610244e-01
-1.14277029e+00 2.31035471e-01 8.96278799e-01 7.05188334e-01
8.37078631e-01 -3.76625359e-01 -5.22128224e-01 9.10485029e-01
4.58219260e-01 7.33963251e-02 1.05211699e+00 -7.13111103e-01
1.66875526e-01 4.02596325e-01 6.13070838e-03 -1.24510276e+00
-2.43328989e-01 -3.15216511e-01 -7.50522614e-02 -3.21589708e-01
2.34925270e-01 4.78748739e-01 -6.69434130e-01 1.57733893e+00
3.34626079e-01 9.39218104e-02 1.94077313e-01 8.66403222e-01
1.06480360e+00 3.54031950e-01 -3.25551890e-02 -5.78496121e-02
1.55910420e+00 -1.23581588e+00 -7.54817128e-01 -1.85385123e-01
8.86596918e-01 -9.78440404e-01 1.27373827e+00 -3.13305110e-01
-8.97845566e-01 -4.10908371e-01 -1.06108916e+00 -2.92972893e-01
-8.85601938e-01 -4.75743443e-01 2.11806260e-02 4.39516366e-01
-1.11349916e+00 7.28007853e-01 -3.41658115e-01 -8.64066422e-01
-2.82981545e-01 7.33810477e-03 -2.89507538e-01 -8.16938803e-02
-1.61001384e+00 1.04066706e+00 4.29977417e-01 -6.78925991e-01
-2.09857836e-01 -4.54340100e-01 -1.12158728e+00 2.17083886e-01
3.68217587e-01 -7.00198054e-01 8.62022400e-01 -8.01042318e-01
-1.11334479e+00 8.43332052e-01 -5.73906898e-01 -5.19527555e-01
4.93720211e-02 -9.79206041e-02 -2.20986724e-01 5.17187297e-01
1.74803883e-01 4.51325417e-01 6.94292307e-01 -1.15395403e+00
-1.65785372e-01 -3.50190252e-01 3.14640254e-01 1.51306227e-01
-6.30994380e-01 4.78466868e-01 -2.72357464e-01 -9.93298292e-01
6.86748475e-02 -6.79835200e-01 -4.99161333e-02 -1.12308554e-01
-1.31796092e-01 -7.07728088e-01 8.74337494e-01 -4.29310679e-01
1.52734184e+00 -2.18437529e+00 2.77362347e-01 2.39912197e-02
-6.38785511e-02 3.05717051e-01 -2.85443306e-01 1.23635793e+00
9.11634266e-02 3.51082444e-01 -4.50110435e-01 -6.06839836e-01
2.49168471e-01 4.73108619e-01 -1.82581812e-01 1.61681488e-01
-1.34794125e-02 1.01193571e+00 -1.41919279e+00 -7.68723786e-01
-2.41023470e-02 3.91125560e-01 -5.34821928e-01 -1.99813932e-01
-1.35457456e-01 -2.93830540e-02 -5.45813560e-01 3.04806560e-01
1.77606114e-03 -1.27045736e-01 1.39119089e-01 -1.04851305e-01
2.63738334e-02 6.43814445e-01 -8.98287117e-01 1.91223192e+00
-6.07965887e-01 8.04981470e-01 -3.12056363e-01 -1.10844254e+00
8.34271669e-01 4.39170569e-01 4.40776229e-01 -7.38820732e-01
-8.06325004e-02 3.01242203e-01 -2.15959236e-01 -7.06095934e-01
9.20552135e-01 -3.13063711e-01 -7.56804794e-02 6.40886366e-01
1.40335321e-01 3.52817439e-02 2.56614000e-01 2.89741635e-01
1.29768121e+00 -5.25008480e-04 5.82409561e-01 -5.39718926e-01
7.53153324e-01 -1.69254765e-02 2.13395089e-01 7.99399078e-01
-3.95074099e-01 7.22811222e-01 2.71725029e-01 -7.10934773e-02
-1.00518894e+00 -1.05237579e+00 -1.64960667e-01 1.13341260e+00
3.73037547e-01 -1.04407215e+00 -4.09001231e-01 -8.78842592e-01
1.08205736e-01 7.67312050e-01 -8.81191492e-01 -2.80775398e-01
-5.26477933e-01 -1.84072435e-01 5.38139641e-01 5.65406382e-01
3.25596146e-02 -9.44608390e-01 -5.31832755e-01 2.35126764e-01
-3.71914446e-01 -1.01057816e+00 -6.79016173e-01 1.07743675e-02
-9.57168341e-01 -8.23320866e-01 -6.51991904e-01 -9.20492828e-01
2.81903625e-01 3.22083682e-01 1.20075846e+00 1.38912126e-01
-2.26727217e-01 9.08404231e-01 -9.20008957e-01 1.72701284e-01
-9.53916982e-02 8.35025981e-02 -4.80176434e-02 -1.35424018e-01
6.98243141e-01 -4.51808095e-01 -4.94293541e-01 2.26411238e-01
-1.17753530e+00 -7.05626249e-01 1.22251339e-01 1.23911858e+00
5.35677493e-01 -2.29112893e-01 4.57964361e-01 -5.59819460e-01
8.96160781e-01 -7.37669587e-01 -4.00976799e-02 3.42379123e-01
-6.84288561e-01 5.13987541e-01 5.08788526e-01 -4.70293403e-01
-5.54838181e-01 -6.05368853e-01 -1.62940636e-01 -3.92855406e-01
-1.29332826e-01 8.97280276e-01 3.22356999e-01 8.58133733e-02
5.57871282e-01 4.72221047e-01 -2.13358074e-01 -8.52259755e-01
4.68239456e-01 6.13078952e-01 1.35706350e-01 -5.23340225e-01
5.98586738e-01 5.23400307e-01 -8.96901265e-02 -9.71971631e-01
-5.77238202e-01 -1.04135621e+00 -6.68895185e-01 2.64237612e-01
9.94298935e-01 -2.97852725e-01 -3.84427309e-02 -3.87281328e-01
-1.12893212e+00 1.87328994e-01 -7.07654238e-01 6.37448549e-01
-4.34665054e-01 9.03218269e-01 -3.88190627e-01 -6.42225087e-01
-1.36250615e-01 -8.75598788e-01 1.10150611e+00 -1.88662820e-02
-6.07623994e-01 -1.49855232e+00 4.79626298e-01 3.35442752e-01
7.25050092e-01 1.05231218e-01 1.08035028e+00 -1.21765685e+00
-1.50933146e-01 -3.85317862e-01 -1.40896559e-01 2.03028992e-01
3.57736796e-01 -2.15057045e-01 -7.03378797e-01 -1.66943237e-01
1.30776465e-01 -5.60973259e-03 1.10163641e+00 -8.82174447e-02
5.08141160e-01 -2.50223249e-01 -3.43832284e-01 9.71863568e-02
1.64746070e+00 -1.37020946e-01 4.10443842e-01 6.33867025e-01
4.03470993e-01 7.38370776e-01 6.73448741e-01 3.06487352e-01
3.60290885e-01 1.14880538e+00 2.16002032e-01 4.09151077e-01
-2.64869064e-01 -1.93698913e-01 3.05489868e-01 9.99473572e-01
2.17667311e-01 -1.98200047e-01 -7.98495054e-01 1.04166853e+00
-1.76136196e+00 -1.11078751e+00 -1.34048536e-01 2.22613239e+00
8.80906641e-01 -7.04728365e-02 1.02354288e-02 1.72053754e-01
5.06694496e-01 4.80892181e-01 -2.52740085e-02 -8.06839943e-01
-1.47803441e-01 3.54766309e-01 1.41784772e-01 7.90925741e-01
-1.00652015e+00 8.71258616e-01 6.33205938e+00 1.10957718e+00
-6.12995863e-01 3.15020859e-01 -3.03707004e-01 1.22882076e-01
-7.18800485e-01 2.78902531e-01 -5.53496361e-01 4.88986224e-01
9.41554308e-01 -1.57384098e-01 1.10108435e-01 5.53507388e-01
-8.56555402e-02 -1.37936428e-01 -1.13499999e+00 7.44763792e-01
5.30679345e-01 -1.25682878e+00 8.20826292e-02 -2.16034219e-01
4.83088821e-01 5.13178706e-02 -5.31474166e-02 1.27965078e-01
-1.20680027e-01 -7.93149114e-01 4.76764470e-01 5.34313917e-01
1.58086017e-01 -3.72423559e-01 9.04723704e-01 8.93609673e-02
-1.36688471e+00 1.08620964e-01 -7.27256536e-02 3.13189737e-02
2.22732008e-01 3.03695291e-01 -3.98430675e-01 9.17953074e-01
5.74305952e-01 8.22821259e-01 -6.68155253e-01 9.24160302e-01
-2.22209692e-01 5.14185369e-01 -2.05724791e-01 -3.20867419e-01
8.21062028e-01 -1.19867936e-01 1.11658370e+00 1.54182768e+00
2.89377067e-02 -2.92464942e-01 1.16978675e-01 7.19325066e-01
2.22766742e-01 5.26496351e-01 -6.55314267e-01 -2.40295336e-01
5.47367811e-01 8.98622751e-01 -7.76057243e-01 -4.41683054e-01
-5.91447234e-01 1.20646536e+00 1.72760487e-01 2.45636016e-01
-4.19716418e-01 -7.14131594e-01 9.08123970e-01 -2.08925039e-01
6.39903605e-01 -4.72641319e-01 -2.67406972e-03 -1.06982231e+00
2.62839258e-01 -4.15619314e-01 6.42426491e-01 -5.57674170e-01
-1.42099392e+00 5.54801702e-01 1.02962583e-01 -1.02458239e+00
-3.39034140e-01 -3.07455748e-01 -5.01590312e-01 8.49166691e-01
-1.77030551e+00 -8.26421201e-01 8.32955539e-02 3.84939760e-01
7.29940712e-01 6.48269206e-02 1.14629054e+00 2.75723338e-01
1.96641646e-02 4.90805835e-01 4.24592167e-01 8.21377710e-02
8.23564768e-01 -1.52344716e+00 2.03248635e-01 5.16796589e-01
6.31300926e-01 8.54695976e-01 8.76997709e-01 -3.16001266e-01
-1.12596202e+00 -5.97458303e-01 1.79165566e+00 -7.38828003e-01
9.12885129e-01 -1.15648456e-01 -1.01499724e+00 2.21056134e-01
4.58088517e-01 2.16700640e-02 1.06876719e+00 3.40053916e-01
-6.59255266e-01 3.05150658e-01 -9.65116501e-01 6.77654326e-01
9.32661474e-01 -1.09038329e+00 -1.33110678e+00 2.28026465e-01
1.29703307e+00 3.70913297e-01 -9.38616872e-01 3.12873095e-01
4.75410342e-01 -8.83798778e-01 1.20671129e+00 -7.65877187e-01
4.06139016e-01 -2.18950436e-01 -6.39172614e-01 -1.21843612e+00
-1.50677413e-01 -2.51615167e-01 -2.73032278e-01 1.25858438e+00
2.88453877e-01 -7.87561297e-01 3.72141838e-01 2.59807557e-02
-1.36745483e-01 -1.05046868e+00 -1.01374304e+00 -1.25299311e+00
2.43754566e-01 -2.35679090e-01 4.84924853e-01 1.21466196e+00
3.17192078e-01 3.38969171e-01 1.15584597e-01 -2.30664805e-01
2.19948113e-01 3.84062290e-01 1.16439834e-01 -1.03066623e+00
-4.04984593e-01 -8.21936309e-01 -9.03822243e-01 -1.03517282e+00
2.27761924e-01 -1.04174566e+00 -8.88107941e-02 -1.75400865e+00
8.27675760e-02 -3.40495110e-01 -6.99899256e-01 8.37994516e-02
-4.14777875e-01 1.93307027e-01 1.83058321e-01 1.66643485e-01
-6.55390203e-01 8.48423719e-01 6.77764475e-01 -2.73523480e-01
2.05947272e-02 -6.34750903e-01 -4.67581898e-01 4.92884755e-01
6.30386651e-01 -6.53109670e-01 -3.88975829e-01 -5.85344970e-01
1.41768053e-01 -3.30107629e-01 3.86434972e-01 -7.47144282e-01
3.53893638e-01 -6.99245604e-03 -3.43044817e-01 -3.55627537e-01
5.40003002e-01 -7.09135592e-01 -3.93292069e-01 3.90991032e-01
-6.39501691e-01 2.93139935e-01 -9.05931816e-02 8.33252132e-01
-6.20149493e-01 -8.25699568e-01 3.70153368e-01 -7.61995837e-02
-7.72188663e-01 -2.19039679e-01 -2.78405607e-01 3.66137236e-01
6.68197334e-01 -3.51877093e-01 -1.18172608e-01 -2.28029415e-01
-6.36782527e-01 7.85563961e-02 6.64916158e-01 7.13083506e-01
6.84264064e-01 -1.46776712e+00 -6.04069471e-01 -1.53011784e-01
5.06669164e-01 -7.63963997e-01 -1.47582024e-01 1.04950845e+00
-2.03625783e-01 7.39584148e-01 4.26677644e-01 -3.47781658e-01
-1.38004744e+00 9.12782013e-01 -1.01186207e-03 -3.78328651e-01
-5.53595424e-01 8.36542785e-01 -1.36893675e-01 -5.42253077e-01
5.38928919e-02 -8.96578804e-02 -5.28139234e-01 5.91586411e-01
4.44286078e-01 3.82021844e-01 -1.68766081e-02 -1.04037535e+00
-4.70850945e-01 8.08076560e-01 1.02173276e-01 -5.16531408e-01
9.50151682e-01 -4.10148919e-01 -2.05278903e-01 8.37776899e-01
1.74183500e+00 1.97555378e-01 -3.00221324e-01 -8.73092413e-01
7.17296481e-01 -6.88580751e-01 1.69318944e-01 -4.48077619e-01
-5.70475936e-01 7.84772754e-01 3.32725525e-01 5.43028295e-01
9.01299894e-01 3.15147221e-01 1.07239318e+00 4.47217137e-01
1.76819295e-01 -1.20676100e+00 2.00436667e-01 6.07140422e-01
6.46051943e-01 -1.02349079e+00 -7.42931888e-02 -1.09664001e-01
-3.78554702e-01 1.13192761e+00 -5.82161471e-02 -1.45493850e-01
6.84715867e-01 -3.20553422e-01 3.80550176e-02 -4.66400146e-01
-5.99553108e-01 -7.30052531e-01 5.57047546e-01 3.19939464e-01
4.72281784e-01 -3.28354985e-01 -9.56287920e-01 2.50170350e-01
5.74024394e-02 -6.50476813e-01 8.99729598e-03 1.42653537e+00
-4.95393127e-01 -1.48645401e+00 -1.76885188e-01 3.09707850e-01
-4.07367408e-01 -5.47108889e-01 -7.29148328e-01 5.80837309e-01
-6.82713166e-02 1.08732915e+00 4.36565131e-02 -2.49670208e-01
5.63659929e-02 4.71198410e-01 3.46367329e-01 -7.88748860e-01
-6.39613152e-01 -2.02101231e-01 1.30532920e-01 -6.12922251e-01
-7.62508333e-01 -6.75060272e-01 -1.10197031e+00 1.69542909e-01
-5.07955253e-01 5.71433306e-01 7.88496017e-01 1.10952735e+00
3.56748462e-01 2.74360716e-01 6.04294896e-01 -5.50151944e-01
-4.50333476e-01 -9.67253983e-01 -5.55109203e-01 6.64868057e-01
2.97394544e-01 -7.66805410e-01 -5.76252282e-01 -2.61955053e-01] | [10.661491394042969, 8.77890396118164] |
460344a1-2014-4b17-8f4f-f48ab9c4930a | sparsity-and-robustness-in-face-recognition | 1111.01014 | null | http://arxiv.org/abs/1111.1014v1 | http://arxiv.org/pdf/1111.1014v1.pdf | Sparsity and Robustness in Face Recognition | This report concerns the use of techniques for sparse signal representation
and sparse error correction for automatic face recognition. Much of the recent
interest in these techniques comes from the paper "Robust Face Recognition via
Sparse Representation" by Wright et al. (2009), which showed how, under certain
technical conditions, one could cast the face recognition problem as one of
seeking a sparse representation of a given input face image in terms of a
"dictionary" of training images and images of individual pixels. In this
report, we have attempted to clarify some frequently encountered questions
about this work and particularly, on the validity of using sparse
representation techniques for face recognition. | ['John Wright', 'Arvind Ganesh', 'Zihan Zhou', 'Yi Ma', 'Allen Yang'] | 2011-11-03 | null | null | null | null | ['robust-face-recognition'] | ['computer-vision'] | [ 7.02101827e-01 -8.38494487e-03 1.98328774e-02 -3.85164350e-01
-4.93390739e-01 -7.07086846e-02 4.99604613e-01 -6.30859315e-01
1.56013653e-01 7.08295226e-01 2.81408966e-01 1.15211261e-02
-1.90915555e-01 -4.13727045e-01 -5.14587224e-01 -8.06531668e-01
-1.13055490e-01 -1.05842851e-01 -6.85223937e-01 -1.56118229e-01
4.47475404e-01 9.33255076e-01 -2.01749396e+00 3.01202118e-01
2.83492357e-01 9.99839127e-01 9.00704879e-03 5.15826583e-01
-1.33216381e-01 8.75339150e-01 -5.46715558e-01 -2.67508388e-01
3.26417953e-01 -6.95041597e-01 -6.68556094e-01 6.41242921e-01
8.53952587e-01 1.94867030e-02 -6.68705761e-01 1.44699788e+00
3.71060997e-01 1.62586644e-01 5.63610911e-01 -1.02258432e+00
-9.02463138e-01 1.22203052e-01 -7.23665595e-01 4.52881902e-01
7.26272523e-01 -3.16567600e-01 3.19440335e-01 -1.35793328e+00
4.92210597e-01 1.29654706e+00 8.08752418e-01 5.35220146e-01
-1.41022158e+00 -6.21265233e-01 -8.57287943e-02 2.38983512e-01
-1.85688293e+00 -1.26382351e+00 9.90067005e-01 -4.46195334e-01
8.46856356e-01 5.50868928e-01 3.08541268e-01 1.05784571e+00
4.88065109e-02 4.16291386e-01 1.20273089e+00 -8.76714408e-01
1.95886627e-01 1.23618849e-01 2.45852042e-02 9.28516209e-01
5.06666183e-01 2.24658236e-01 -7.91106880e-01 -4.79011834e-01
9.91476715e-01 -1.01378389e-01 -3.36324543e-01 -2.79042810e-01
-7.33631790e-01 1.02123284e+00 -1.07642792e-01 7.88244665e-01
-5.58588207e-01 -1.24266893e-01 5.50760962e-02 2.32543156e-01
5.38827598e-01 -1.11224681e-01 2.47887135e-01 2.09899321e-01
-1.26320112e+00 -1.81076854e-01 9.62938666e-01 8.38666558e-01
8.27903092e-01 8.29733014e-01 1.05768606e-01 9.54418421e-01
5.70354342e-01 8.68029594e-01 5.78294456e-01 -1.19536066e+00
8.58345032e-02 1.28966391e-01 -1.40400335e-01 -1.53203237e+00
6.69594854e-02 -2.27847770e-02 -1.14664268e+00 2.03283563e-01
1.49772717e-02 -1.07810341e-01 -7.69662142e-01 1.42373490e+00
1.53801322e-01 8.77604365e-01 1.77312285e-01 6.14430189e-01
8.51518035e-01 6.72312737e-01 -2.29819104e-01 -6.87324703e-01
1.11998975e+00 -2.28792801e-01 -9.56885636e-01 -1.55860350e-01
4.14035134e-02 -1.07605577e+00 1.80963382e-01 3.34189445e-01
-1.00115633e+00 -5.20536184e-01 -9.19346988e-01 3.03649276e-01
1.82164293e-02 3.70399117e-01 7.93097913e-01 9.26926494e-01
-1.27564752e+00 3.05587649e-01 -5.07506192e-01 -6.04460001e-01
4.28009450e-01 4.98008460e-01 -7.53414690e-01 -4.42432374e-01
-6.33506000e-01 8.26945841e-01 -3.45022351e-01 4.47185516e-01
-7.98765361e-01 -4.62598175e-01 -1.04675460e+00 -6.06644265e-02
4.49605696e-02 -4.26134199e-01 8.77464592e-01 -1.24654984e+00
-1.32567036e+00 1.15595877e+00 -9.35821772e-01 -2.93867975e-01
-3.15220386e-01 1.11206189e-01 -7.42960155e-01 3.78768086e-01
3.18914242e-02 4.91511859e-02 1.54177976e+00 -1.29573739e+00
-5.27580380e-02 -7.82873690e-01 -4.83060449e-01 -3.39738488e-01
6.57501966e-02 3.86110008e-01 -2.21900977e-02 -8.93942773e-01
4.08901215e-01 -9.14588749e-01 -1.62225693e-01 -1.18105039e-01
6.07071146e-02 8.17385986e-02 1.02955043e+00 -6.19241655e-01
1.00272036e+00 -2.54205894e+00 7.09984377e-02 6.55943394e-01
1.13992067e-02 2.70324260e-01 -3.74320567e-01 3.10205609e-01
-6.61356509e-01 -8.82172734e-02 -5.28193295e-01 -3.57172042e-01
-4.41067666e-01 3.35505545e-01 -5.10370672e-01 1.16389263e+00
3.38633567e-01 3.94083411e-01 -5.78350961e-01 -4.50990617e-01
1.85525343e-01 1.16565633e+00 -4.80155170e-01 2.99890507e-02
5.59076607e-01 3.49156320e-01 -3.03710282e-01 1.03543067e+00
8.39457691e-01 -2.69242451e-02 3.09040517e-01 -2.31054917e-01
-2.24898472e-01 -1.78972587e-01 -1.78218186e+00 1.52813888e+00
-2.27377549e-01 8.74149680e-01 6.45579517e-01 -1.33070600e+00
1.05436385e+00 6.45818770e-01 8.02749515e-01 -4.61893022e-01
2.50667304e-01 2.10835665e-01 -4.27309424e-01 -4.19623882e-01
2.73497224e-01 -3.79803419e-01 6.30815148e-01 4.25028950e-01
2.01560989e-01 -1.74226705e-02 -2.17816457e-02 1.23148970e-02
9.20037806e-01 -2.95825452e-01 4.22078580e-01 -4.27425444e-01
8.30363929e-01 -3.35581869e-01 2.98217207e-01 4.00668085e-01
-8.80134329e-02 7.89403558e-01 -2.57302262e-02 -3.27450424e-01
-6.56720757e-01 -6.39062822e-01 -3.99297893e-01 6.66492701e-01
-4.18041974e-01 -2.63033301e-01 -5.80933332e-01 -1.79802567e-01
7.45367482e-02 2.27464318e-01 -8.24582815e-01 5.49489744e-02
-5.84660947e-01 -7.33580291e-01 4.21349287e-01 2.79263347e-01
1.83556572e-01 -9.89617288e-01 -2.82543331e-01 -4.37769033e-02
-1.07253693e-01 -8.04606020e-01 -3.28327775e-01 1.81275681e-01
-1.06887436e+00 -1.12265158e+00 -8.05086017e-01 -1.03087974e+00
1.24629676e+00 8.02423656e-01 9.02075231e-01 3.04879606e-01
-5.84242404e-01 8.41133535e-01 -3.06580901e-01 -6.55517653e-02
-5.43809906e-02 -6.28657877e-01 3.72702599e-01 6.04281545e-01
5.61503768e-01 -6.54716909e-01 2.49314606e-02 1.30081892e-01
-8.96361232e-01 -8.19703102e-01 4.51788068e-01 9.60028231e-01
6.35251760e-01 -8.95998627e-02 3.84800255e-01 -9.28599894e-01
4.43782836e-01 -6.78710878e-01 -5.15573323e-01 1.23953961e-01
-4.80404019e-01 -1.55417889e-01 1.65001437e-01 -2.69286513e-01
-9.73356962e-01 4.28277135e-01 -3.03048074e-01 -6.39059246e-01
-1.56309661e-02 3.22164565e-01 -1.88403070e-01 -9.67290819e-01
8.02932382e-01 7.96805203e-01 4.97802496e-01 -5.07241607e-01
1.90993056e-01 7.00933278e-01 5.94117820e-01 -5.63647270e-01
8.44471097e-01 6.41007960e-01 1.66525751e-01 -1.56478751e+00
-5.12824833e-01 -6.74990475e-01 -5.11083961e-01 -2.34083068e-02
1.09923534e-01 -9.81671929e-01 -5.33188939e-01 1.04992948e-01
-1.10421741e+00 3.90213132e-01 -5.95159650e-01 4.35322553e-01
-6.85779691e-01 7.08097994e-01 -1.27807230e-01 -1.09944654e+00
-1.94402691e-02 -8.67975056e-01 1.03504908e+00 8.25664774e-02
-2.90737957e-01 -9.37748253e-01 8.61364082e-02 2.00546667e-01
5.88856280e-01 -5.32058440e-02 3.96930009e-01 -2.08484456e-01
-1.81456715e-01 -3.81921172e-01 -1.42722517e-01 5.32917798e-01
3.86517078e-01 -5.19573651e-02 -1.29363286e+00 -5.45731664e-01
7.88951457e-01 -1.26331076e-01 6.50780141e-01 4.94028330e-01
1.09830403e+00 -7.27568448e-01 -1.68199301e-01 7.35102832e-01
1.61636043e+00 2.50007838e-01 9.56127524e-01 -1.97643340e-01
3.76476437e-01 6.17580116e-01 -1.07599450e-02 5.57128906e-01
-3.39522183e-01 5.09386480e-01 9.69476998e-02 1.01260953e-01
-3.29077989e-01 -9.81533751e-02 3.52036297e-01 5.77251017e-01
-1.95363075e-01 2.65172631e-01 -4.35103476e-01 6.40142024e-01
-1.43670607e+00 -1.57079661e+00 5.58135659e-02 2.05457282e+00
3.44983339e-01 -9.19615746e-01 1.54779246e-02 4.89045799e-01
9.17036891e-01 2.98165858e-01 -2.58954197e-01 -2.87307560e-01
-4.35912371e-01 5.27037799e-01 2.67025292e-01 7.07262993e-01
-8.00745666e-01 4.94370162e-01 8.56490898e+00 5.73615313e-01
-1.18318784e+00 -8.75836611e-02 5.22098005e-01 2.13623792e-01
-3.32076430e-01 -5.37645668e-02 -7.87721515e-01 3.16022784e-01
9.83812153e-01 -2.19128743e-01 6.20279431e-01 7.50774324e-01
5.24821803e-02 -4.67718132e-02 -9.69035327e-01 1.49225402e+00
8.33811939e-01 -1.34479415e+00 3.25270265e-01 1.30043373e-01
9.29345071e-01 -4.14923549e-01 5.04200876e-01 -1.33671239e-01
-9.72169414e-02 -1.60924137e+00 1.64553761e-01 6.38342917e-01
6.98695958e-01 -4.54409540e-01 3.62222552e-01 -3.15044187e-02
-1.10298419e+00 -1.12492114e-01 -5.91266274e-01 -1.92040980e-01
-2.30134070e-01 8.25458825e-01 -5.38594186e-01 2.41542608e-01
3.81105751e-01 7.44634986e-01 -1.79915830e-01 9.56703246e-01
2.51763284e-01 7.36058593e-01 -4.25890535e-01 4.20646667e-01
7.93381943e-04 -3.11464638e-01 6.12883925e-01 1.22166300e+00
7.28156328e-01 5.00423968e-01 -9.82293412e-02 6.93555117e-01
6.49373904e-02 2.56585389e-01 -1.24375808e+00 -2.09723860e-01
4.08619016e-01 1.08710682e+00 -4.65748042e-01 -1.42612934e-01
-5.83267391e-01 7.72702873e-01 -1.06901653e-01 3.79516661e-01
-1.50652304e-01 -1.53500766e-01 7.78635979e-01 3.32077861e-01
5.49762487e-01 -2.35652089e-01 -4.51623380e-01 -1.20953035e+00
-7.90058076e-02 -1.12348449e+00 2.62171417e-01 -5.04428983e-01
-1.16161120e+00 5.91791689e-01 -2.07806453e-01 -1.02114606e+00
-2.45721593e-01 -4.59779114e-01 -4.79293883e-01 1.18634903e+00
-1.34638071e+00 -8.10871184e-01 1.91389527e-02 1.08799398e+00
5.50087750e-01 -7.09978223e-01 1.01595116e+00 2.93019503e-01
-4.88045186e-01 5.59103429e-01 3.25396925e-01 -5.34692258e-02
2.05254823e-01 -4.04364944e-01 -8.37157369e-02 9.90312934e-01
6.18337333e-01 9.62240517e-01 8.80326867e-01 -4.62139517e-01
-2.01083970e+00 -7.85995781e-01 1.36398292e+00 -2.97788858e-01
3.12764108e-01 1.29594445e-01 -9.62287366e-01 8.26145828e-01
1.50955588e-01 3.18476826e-01 1.03529775e+00 6.83314055e-02
-3.09633017e-01 -1.61462292e-01 -1.64979732e+00 8.62579886e-03
9.07371283e-01 -8.69714439e-01 -7.24769890e-01 3.03828388e-01
-3.18303436e-01 4.66198064e-02 -7.64421225e-01 1.42012298e-01
4.73584741e-01 -7.29582429e-01 1.19315386e+00 -4.41612959e-01
1.05734654e-01 -3.85645121e-01 -7.31627107e-01 -1.08834147e+00
-5.47429979e-01 -5.80012083e-01 -2.52398849e-01 1.04504836e+00
-1.93375260e-01 -4.21630323e-01 9.17343378e-01 2.53928244e-01
1.70214579e-01 -2.61101902e-01 -1.37260377e+00 -7.21993566e-01
-3.24655056e-01 -2.54518390e-01 3.72386128e-01 9.80550706e-01
6.24303939e-03 8.86148140e-02 -6.25635564e-01 2.93617487e-01
8.60209346e-01 1.30481735e-01 3.72126490e-01 -1.31133139e+00
1.10415824e-01 -1.36258096e-01 -8.12333941e-01 -7.53377795e-01
6.05946004e-01 -8.42653751e-01 7.19711930e-02 -1.00607514e+00
9.12153721e-02 -1.58420682e-01 -1.41799048e-01 4.11331296e-01
3.18915427e-01 6.87439084e-01 1.57265365e-01 4.12612975e-01
-1.23580448e-01 3.72688174e-01 7.94834673e-01 -3.41332346e-01
2.28667587e-01 -9.35524628e-02 -1.24886596e+00 5.78870296e-01
4.25313145e-01 -4.89617974e-01 -2.13148028e-01 -3.56112748e-01
-1.36366665e-01 2.16238916e-01 3.45962524e-01 -1.02023315e+00
3.48207742e-01 -2.02902973e-01 6.22754931e-01 -4.18755859e-02
6.23656869e-01 -1.14297473e+00 5.78017592e-01 4.17897075e-01
-4.15494069e-02 1.63449384e-02 2.29481876e-01 5.24093270e-01
-4.84426975e-01 -4.88220036e-01 9.62646723e-01 -2.57589072e-01
-7.41941929e-01 5.84051013e-02 -5.95252216e-01 -3.37507963e-01
9.59524214e-01 -5.29681206e-01 1.75651878e-01 -3.27990800e-01
-9.94612634e-01 -5.64372718e-01 2.37715706e-01 -6.28629625e-02
1.01699352e+00 -1.45682454e+00 -8.58464360e-01 9.39050376e-01
-1.56903163e-01 -6.15099072e-01 3.21171582e-01 7.46333897e-01
-2.26381972e-01 6.40432239e-01 -4.56590801e-01 -3.04285258e-01
-1.49035418e+00 6.40827954e-01 1.80342585e-01 4.36658770e-01
-6.03927612e-01 8.51570427e-01 -2.08221182e-01 2.97645479e-01
2.19151005e-01 3.50717425e-01 -2.46251225e-01 7.45784566e-02
1.03384292e+00 5.01414776e-01 8.09978098e-02 -1.41563070e+00
-6.38727486e-01 8.43631268e-01 1.25730947e-01 7.87878856e-02
1.38389957e+00 -1.43683702e-01 -3.83656234e-01 2.44370565e-01
1.43445182e+00 2.65400767e-01 -7.99672902e-01 -2.65639454e-01
-2.29982018e-01 -8.88712823e-01 2.26105183e-01 -3.09341013e-01
-1.39093149e+00 5.90015352e-01 8.24379802e-01 9.28394943e-02
1.31184959e+00 -1.90393895e-01 2.48599887e-01 4.51982439e-01
4.56932575e-01 -6.88076615e-01 -2.38974705e-01 6.04142904e-01
1.22184396e+00 -1.23676109e+00 2.05747351e-01 -6.09460652e-01
-2.31284678e-01 1.11706138e+00 -7.22896680e-02 -4.98841047e-01
1.11867738e+00 3.47048312e-01 -2.19385237e-01 -3.69459987e-01
-3.86920780e-01 -1.57649383e-01 2.69204497e-01 1.09935522e+00
6.98357821e-01 -2.04159394e-01 -2.60663033e-01 1.39458522e-01
-5.77476621e-02 4.23212796e-01 4.44809109e-01 1.18536043e+00
-5.69099247e-01 -8.98493767e-01 -1.06073546e+00 4.78184313e-01
-5.18216848e-01 -5.21160252e-02 -4.36155438e-01 2.04653665e-01
5.86935170e-02 1.04405499e+00 -9.14133936e-02 -1.52682513e-01
2.21329451e-01 2.64018625e-01 7.78426766e-01 -8.25280368e-01
-2.87152499e-01 -2.40359992e-01 -2.09024578e-01 -5.39470613e-01
-7.82690465e-01 -9.91832912e-01 -7.14898288e-01 -6.06697083e-01
-2.19781455e-02 4.95730966e-01 8.41671944e-01 8.33417416e-01
1.81303412e-01 1.08335748e-01 8.58920813e-01 -8.74830961e-01
-5.50035000e-01 -7.58549154e-01 -1.20735991e+00 1.33670181e-01
4.55726266e-01 -6.17689013e-01 -5.93501151e-01 4.00308430e-01] | [12.543692588806152, 0.3831537067890167] |
4c55a379-421a-4cd1-8733-d2f669315b85 | audio-visual-grouping-network-for-sound | 2303.17056 | null | https://arxiv.org/abs/2303.17056v1 | https://arxiv.org/pdf/2303.17056v1.pdf | Audio-Visual Grouping Network for Sound Localization from Mixtures | Sound source localization is a typical and challenging task that predicts the location of sound sources in a video. Previous single-source methods mainly used the audio-visual association as clues to localize sounding objects in each image. Due to the mixed property of multiple sound sources in the original space, there exist rare multi-source approaches to localizing multiple sources simultaneously, except for one recent work using a contrastive random walk in the graph with images and separated sound as nodes. Despite their promising performance, they can only handle a fixed number of sources, and they cannot learn compact class-aware representations for individual sources. To alleviate this shortcoming, in this paper, we propose a novel audio-visual grouping network, namely AVGN, that can directly learn category-wise semantic features for each source from the input audio mixture and image to localize multiple sources simultaneously. Specifically, our AVGN leverages learnable audio-visual class tokens to aggregate class-aware source features. Then, the aggregated semantic features for each source can be used as guidance to localize the corresponding visual regions. Compared to existing multi-source methods, our new framework can localize a flexible number of sources and disentangle category-aware audio-visual representations for individual sound sources. We conduct extensive experiments on MUSIC, VGGSound-Instruments, and VGG-Sound Sources benchmarks. The results demonstrate that the proposed AVGN can achieve state-of-the-art sounding object localization performance on both single-source and multi-source scenarios. Code is available at \url{https://github.com/stoneMo/AVGN}. | ['Yapeng Tian', 'Shentong Mo'] | 2023-03-29 | null | http://openaccess.thecvf.com//content/CVPR2023/html/Mo_Audio-Visual_Grouping_Network_for_Sound_Localization_From_Mixtures_CVPR_2023_paper.html | http://openaccess.thecvf.com//content/CVPR2023/papers/Mo_Audio-Visual_Grouping_Network_for_Sound_Localization_From_Mixtures_CVPR_2023_paper.pdf | cvpr-2023-1 | ['object-localization'] | ['computer-vision'] | [-7.27756321e-03 -4.99116570e-01 -2.82083452e-01 -2.22525466e-02
-1.29403114e+00 -8.19343805e-01 3.32630038e-01 3.40752192e-02
2.65612453e-01 2.24967137e-01 4.99464959e-01 1.99250698e-01
-2.11596191e-01 -4.90553170e-01 -7.95862615e-01 -7.08377540e-01
-2.34010056e-01 -1.13840237e-01 5.55534720e-01 2.13856310e-01
2.09134355e-01 1.70754001e-01 -1.82191205e+00 4.88215685e-01
4.13137645e-01 1.21893477e+00 5.02754271e-01 8.33729625e-01
-2.29764327e-01 7.95917690e-01 -6.68240309e-01 3.17627221e-01
1.04066037e-01 -6.64108992e-01 -4.70420212e-01 -1.62589043e-01
7.97360599e-01 3.67613398e-02 -3.86627406e-01 1.06793833e+00
7.13037610e-01 2.01061711e-01 5.53576648e-01 -1.87425375e+00
-8.02881539e-01 9.14675295e-01 -8.09548497e-01 6.15672469e-01
5.44678748e-01 1.28852993e-01 1.48914456e+00 -1.08551431e+00
9.52428728e-02 1.21713209e+00 4.44317818e-01 2.34064698e-01
-8.84193897e-01 -1.14515901e+00 2.94852406e-01 4.24114436e-01
-1.57087862e+00 -6.14084244e-01 1.17070413e+00 -4.59984303e-01
5.12211502e-01 2.13649228e-01 5.87897360e-01 1.00045741e+00
-2.12751433e-01 7.52795875e-01 8.18928301e-01 -2.41184533e-01
2.23494038e-01 -1.49138823e-01 -1.58913329e-01 8.25221717e-01
3.69056575e-02 -2.72374928e-01 -1.28940022e+00 -2.63950139e-01
7.62595832e-01 2.51819659e-02 -4.67855752e-01 -2.45205328e-01
-1.34937918e+00 5.84038138e-01 8.08524668e-01 3.40358406e-01
-3.71250659e-02 6.67764008e-01 1.60200819e-01 -1.05058044e-01
5.90947986e-01 1.34418890e-01 -1.67832196e-01 -2.80891974e-02
-9.88991857e-01 6.78932108e-03 5.09529054e-01 8.10167134e-01
8.28621387e-01 2.16662318e-01 -1.58812970e-01 9.42327917e-01
5.02698958e-01 6.86356127e-01 2.30160445e-01 -8.92782331e-01
5.85254371e-01 4.25077379e-01 -1.59847692e-01 -1.38648295e+00
-3.07492435e-01 -6.92093909e-01 -5.72873294e-01 8.64388719e-02
2.93118626e-01 9.40858424e-02 -8.69004905e-01 1.86451280e+00
3.85332257e-01 1.05175126e+00 -4.65902686e-01 1.24159360e+00
1.27565575e+00 7.37352312e-01 4.92191613e-02 1.78856011e-02
1.10738659e+00 -1.19009769e+00 -4.65983808e-01 -2.28846401e-01
-2.03631297e-01 -8.18254769e-01 1.24303532e+00 1.74198478e-01
-8.85908663e-01 -7.58672297e-01 -8.34775209e-01 6.94215447e-02
-2.53689915e-01 1.94377631e-01 6.42915845e-01 3.84922862e-01
-1.04119670e+00 1.46508306e-01 -7.15025663e-01 -4.16577831e-02
7.26351798e-01 -1.12562418e-01 1.75280608e-02 4.93939668e-02
-9.42254424e-01 -8.87607411e-02 -1.54457122e-01 -2.97337007e-02
-1.64606106e+00 -1.08480096e+00 -7.94527113e-01 1.89660728e-01
5.28324842e-01 -5.66463351e-01 9.89983022e-01 -8.72052133e-01
-9.77997303e-01 3.91040415e-01 -3.57221425e-01 -4.52290438e-02
1.11161880e-02 -8.14011917e-02 -5.03167510e-01 4.16228175e-01
5.39152205e-01 6.85963154e-01 1.09892118e+00 -1.52346647e+00
-8.09747279e-01 -6.89484701e-02 1.35036916e-01 2.66426414e-01
-3.21099162e-01 1.69995382e-01 -6.15103245e-01 -8.80037904e-01
2.79755116e-01 -5.44073701e-01 1.50386512e-01 2.06155688e-01
-5.90574861e-01 -2.89894581e-01 5.66502750e-01 -2.97234625e-01
1.15075886e+00 -2.33286405e+00 1.75118059e-01 -1.64261237e-02
4.33899999e-01 -2.99710423e-01 -4.29675370e-01 1.88756883e-01
-2.11608466e-02 1.17117867e-01 1.21881224e-01 -4.80663151e-01
2.21159682e-02 -3.41353267e-01 -5.69499671e-01 4.57821548e-01
-1.41334400e-01 7.15860546e-01 -1.23162127e+00 -6.32783115e-01
1.06335759e-01 6.85240209e-01 -4.96867746e-01 1.59080461e-01
-1.21183820e-01 4.04033244e-01 -3.83019418e-01 1.14096546e+00
4.48138982e-01 -4.56864625e-01 -3.22927713e-01 -2.78636515e-01
2.01374665e-01 2.80788302e-01 -1.42028868e+00 2.10407233e+00
-5.43684483e-01 6.99841678e-01 3.05315018e-01 -6.71932697e-01
7.80287683e-01 3.48243326e-01 6.90708458e-01 -2.78626233e-01
-1.86730653e-01 3.57449979e-01 -2.54617989e-01 -3.59237403e-01
1.76936895e-01 1.03182383e-01 -1.71684921e-01 4.48198229e-01
5.23996592e-01 3.98997925e-02 -1.48600563e-01 4.02391344e-01
1.04895854e+00 -9.10077915e-02 4.68337983e-02 6.49536327e-02
2.51099050e-01 -3.34235579e-01 5.23551702e-01 8.08824539e-01
-1.79035336e-01 1.02500105e+00 2.74904996e-01 1.78823844e-02
-2.74139851e-01 -1.45232701e+00 1.50618598e-01 1.69297111e+00
4.82099533e-01 -7.11490571e-01 -4.86597478e-01 -6.91793621e-01
5.61288260e-02 1.00551970e-01 -3.52192640e-01 -3.61679457e-02
-2.95624375e-01 -4.22213435e-01 6.47383034e-01 5.61194181e-01
1.76338911e-01 -1.07593608e+00 -4.54372585e-01 -2.09290311e-02
-5.42151093e-01 -9.11253750e-01 -7.21445739e-01 5.46020381e-02
-2.73787409e-01 -1.12838376e+00 -7.67214715e-01 -8.23752046e-01
2.47354507e-01 7.61287332e-01 1.09705937e+00 -2.06978712e-02
-4.34091419e-01 7.93159604e-01 -3.26197296e-01 -5.45681894e-01
1.37871638e-01 -5.41542545e-02 7.82204866e-02 3.55615526e-01
4.23960947e-02 -9.37958062e-01 -9.51453626e-01 3.23617160e-01
-4.95934665e-01 -1.57255232e-01 2.02439845e-01 3.71166527e-01
8.12697649e-01 1.59341857e-01 9.75001633e-01 -2.12394640e-01
4.22894210e-01 -1.00367844e+00 -2.07545221e-01 8.68970975e-02
-5.81088737e-02 -6.05736315e-01 5.15985608e-01 -6.64592862e-01
-6.20794713e-01 6.60838187e-02 1.72190040e-01 -9.68481421e-01
-4.39194858e-01 2.27935925e-01 -3.48714709e-01 -1.36258140e-01
5.84533691e-01 1.84746727e-01 -5.12435853e-01 -7.44489729e-01
5.11190355e-01 5.72657287e-01 5.57014465e-01 -4.37900931e-01
7.05774605e-01 5.58498859e-01 8.12353380e-03 -6.46253943e-01
-1.13785577e+00 -8.21916401e-01 -2.66771168e-01 -4.92980689e-01
7.76018798e-01 -1.34402704e+00 -5.17644286e-01 3.71027827e-01
-9.73637879e-01 -2.89122194e-01 -2.00658858e-01 4.23477858e-01
-4.44984078e-01 6.43204600e-02 -2.42511436e-01 -8.15521538e-01
3.07003912e-02 -1.08476269e+00 1.40379202e+00 3.73495162e-01
-2.61273105e-02 -9.46095347e-01 7.01230019e-02 3.18434000e-01
2.34162822e-01 3.47402394e-02 5.79895139e-01 -4.42816973e-01
-8.43306720e-01 1.67125940e-01 -2.43832201e-01 9.36520472e-02
4.64846402e-01 -2.63392597e-01 -1.32937777e+00 -6.34520203e-02
-3.75714451e-01 -3.46544027e-01 1.27216506e+00 5.47456622e-01
1.38249075e+00 -2.69049525e-01 -4.43028778e-01 7.64725745e-01
1.26545799e+00 -4.21338864e-02 -6.13132864e-02 -7.99406320e-02
1.18308496e+00 2.38503084e-01 5.02941608e-01 4.90178108e-01
4.51956093e-01 7.21416652e-01 7.53890276e-01 -1.86040267e-01
-7.41816401e-01 -5.79670489e-01 4.13565993e-01 6.78954959e-01
6.53035790e-02 -3.86351019e-01 -7.10001349e-01 8.19231153e-01
-1.71963549e+00 -8.92861247e-01 -2.52454937e-03 1.86268139e+00
7.21704543e-01 -2.75114894e-01 4.61831927e-01 2.12920576e-01
8.36428404e-01 6.94470942e-01 -4.77524459e-01 4.36567634e-01
-1.28762990e-01 1.77198261e-01 8.55821520e-02 3.75133693e-01
-1.37111664e+00 7.58103549e-01 5.78727293e+00 1.23738766e+00
-1.11293030e+00 5.40912330e-01 3.57892185e-01 -6.94698095e-01
-3.32054675e-01 -1.29208952e-01 -7.10794270e-01 7.13757515e-01
5.58608770e-01 3.52699123e-02 5.80188274e-01 8.49860251e-01
6.71345517e-02 -3.34550254e-02 -8.45564842e-01 1.30753148e+00
4.10478592e-01 -1.42650449e+00 6.50213435e-02 -2.67396569e-01
4.94233310e-01 1.14272535e-01 2.26335198e-01 -4.85965498e-02
9.12361518e-02 -9.73970652e-01 1.16057813e+00 4.50899184e-01
8.06488812e-01 -7.12744176e-01 9.90506336e-02 6.99337572e-02
-1.89102924e+00 -1.82806075e-01 -3.26536447e-01 5.96365109e-02
2.43726999e-01 4.87129986e-01 -5.48339546e-01 4.57671642e-01
1.23835945e+00 8.85610580e-01 -6.99835598e-01 1.32435060e+00
-3.99900943e-01 1.06414759e+00 -3.76075834e-01 1.23168677e-01
-5.22912070e-02 2.82459646e-01 8.94304097e-01 1.19839060e+00
5.59176385e-01 -3.08597654e-01 5.81044972e-01 9.37057257e-01
-1.39540099e-02 1.10506892e-01 -4.93482739e-01 1.18388094e-01
7.79225111e-01 1.46539009e+00 -9.63141322e-01 1.16863558e-02
-4.08142745e-01 6.75483823e-01 1.65266916e-01 5.34070551e-01
-1.14858770e+00 -5.05040228e-01 8.85099709e-01 2.26755217e-01
3.89308363e-01 -8.43099505e-02 -1.89940199e-01 -9.83061373e-01
1.05728395e-01 -5.58941483e-01 5.73457122e-01 -1.10993576e+00
-1.30276906e+00 5.56888759e-01 -4.64030989e-02 -1.49421453e+00
1.62316367e-01 -3.06256652e-01 -8.14305305e-01 6.80396199e-01
-1.53320110e+00 -1.38838112e+00 -4.68443841e-01 9.53117251e-01
6.94122493e-01 -3.53651196e-01 5.34481585e-01 3.91853690e-01
-3.69864345e-01 6.92179263e-01 -3.46821457e-01 2.01464295e-01
8.37225676e-01 -1.20660770e+00 -1.18187033e-01 8.51376414e-01
1.09876776e+00 3.35910946e-01 3.90526295e-01 -5.33651590e-01
-1.26768863e+00 -1.33947730e+00 4.40465242e-01 -4.62419063e-01
8.88729930e-01 -6.29583538e-01 -8.05433095e-01 4.88810599e-01
1.26751915e-01 6.12335980e-01 7.55432606e-01 1.54030412e-01
-7.00942397e-01 -3.79444271e-01 -5.96272230e-01 2.53881156e-01
1.34276307e+00 -8.50129008e-01 -2.90823698e-01 4.10121262e-01
9.52468336e-01 -2.92429507e-01 -4.38058048e-01 2.73441114e-02
2.26038620e-01 -9.42420125e-01 1.43635213e+00 -1.61305010e-01
4.29813951e-01 -6.80426180e-01 -2.88435400e-01 -1.41760528e+00
-3.48038495e-01 -5.90991259e-01 -3.11187834e-01 1.68341959e+00
3.67943794e-01 -3.43772858e-01 3.20955008e-01 -3.06207836e-01
-3.13471198e-01 -5.17763078e-01 -1.12375665e+00 -7.59588957e-01
-4.08894330e-01 -8.65236044e-01 7.13265240e-01 7.71087766e-01
-9.21671987e-02 3.39731306e-01 -5.16430140e-01 5.51071405e-01
7.79182673e-01 5.75915694e-01 5.93971431e-01 -1.09008026e+00
-5.78553021e-01 -4.53603119e-01 -4.78320569e-01 -9.78155851e-01
3.76580715e-01 -1.23221517e+00 2.63777822e-01 -1.65970564e+00
2.47193828e-01 -6.62685215e-01 -8.52092862e-01 7.94979393e-01
-2.66578227e-01 6.97576463e-01 3.02737445e-01 3.11682642e-01
-1.03793657e+00 3.75786453e-01 1.07301927e+00 -2.61999875e-01
-2.44646430e-01 4.94446829e-02 -9.26630139e-01 8.69674087e-01
6.28337979e-01 -6.93508685e-01 -6.47131324e-01 -5.98928511e-01
2.76066035e-01 3.41065265e-02 1.06674170e+00 -1.25179851e+00
5.07524252e-01 -2.22549513e-01 3.77304167e-01 -5.05333483e-01
5.41839182e-01 -5.54121494e-01 -3.24906893e-02 -3.10747534e-01
-4.10678416e-01 -5.08829355e-01 1.24892280e-01 1.01982713e+00
-5.18372953e-01 1.48327976e-01 4.50025231e-01 -1.20409742e-01
-9.09313679e-01 4.94495273e-01 -5.33331186e-02 3.30046207e-01
7.60922015e-01 1.30035728e-01 -4.52059686e-01 -6.05349600e-01
-6.65170014e-01 3.17137428e-02 3.75346951e-02 7.74494767e-01
7.34260678e-01 -1.58609068e+00 -5.33552051e-01 6.90390021e-02
2.14042202e-01 8.34499311e-04 6.12655520e-01 7.72226572e-01
1.47429526e-01 -1.10069308e-02 1.73552576e-02 -9.87112641e-01
-1.32699752e+00 5.73852777e-01 3.00983191e-01 4.23089862e-01
-4.73854125e-01 1.38909578e+00 7.10571468e-01 -4.13306057e-03
5.10716558e-01 -2.46348113e-01 -2.53200322e-01 2.79957056e-01
6.26472235e-01 5.29321313e-01 -2.69976526e-01 -7.84770548e-01
-6.29547119e-01 9.17635858e-01 6.24155760e-01 -1.74895734e-01
1.22598827e+00 -3.20998311e-01 -4.96382527e-02 9.51750517e-01
1.24730003e+00 4.33770776e-01 -1.36834693e+00 -4.23062384e-01
-4.95594025e-01 -6.97981954e-01 1.70805454e-01 -7.93563247e-01
-1.56455219e+00 1.17750680e+00 5.55677891e-01 1.43450737e-01
1.44342923e+00 7.88721085e-01 5.00307143e-01 -1.24085441e-01
4.32381123e-01 -5.53669393e-01 5.90351641e-01 1.19586185e-01
1.01116753e+00 -1.10127413e+00 -2.80635774e-01 -6.07446492e-01
-5.89514196e-01 8.87305677e-01 6.21388018e-01 -1.35127291e-01
1.01035905e+00 4.36164975e-01 2.37051398e-01 -2.89387256e-01
-5.76993823e-01 -4.77641165e-01 7.29091823e-01 7.39577591e-01
2.63091892e-01 1.49819508e-01 6.91745877e-01 1.02441800e+00
-1.75463095e-01 -2.72831589e-01 1.78033561e-01 5.63496828e-01
-6.62348688e-01 -5.36521792e-01 -4.03521150e-01 1.66063920e-01
-4.74263519e-01 -2.06531256e-01 -4.04051721e-01 3.15831125e-01
4.25674289e-01 1.40420258e+00 1.82061434e-01 -5.92296720e-01
1.13363236e-01 -7.36618415e-02 2.51131833e-01 -8.02545011e-01
-4.18970793e-01 6.50955617e-01 -4.37860787e-01 -8.34571838e-01
-6.57250285e-01 -4.74482834e-01 -1.24383008e+00 3.40266049e-01
-9.56470370e-02 1.52418837e-01 3.05031270e-01 4.77006078e-01
5.26935816e-01 8.00408483e-01 7.36457467e-01 -1.11897492e+00
-8.82760212e-02 -7.58897007e-01 -8.62609744e-01 3.24199051e-01
6.69226289e-01 -9.93500590e-01 -7.70411491e-01 1.35022387e-01] | [14.799866676330566, 4.883406162261963] |
51f58ebd-0078-46fa-86da-0ad6897f78c2 | happy-people-image-synthesis-as-black-box | 2306.06684 | null | https://arxiv.org/abs/2306.06684v1 | https://arxiv.org/pdf/2306.06684v1.pdf | Happy People -- Image Synthesis as Black-Box Optimization Problem in the Discrete Latent Space of Deep Generative Models | In recent years, optimization in the learned latent space of deep generative models has been successfully applied to black-box optimization problems such as drug design, image generation or neural architecture search. Existing models thereby leverage the ability of neural models to learn the data distribution from a limited amount of samples such that new samples from the distribution can be drawn. In this work, we propose a novel image generative approach that optimizes the generated sample with respect to a continuously quantifiable property. While we anticipate absolutely no practically meaningful application for the proposed framework, it is theoretically principled and allows to quickly propose samples at the mere boundary of the training data distribution. Specifically, we propose to use tree-based ensemble models as mathematical programs over the discrete latent space of vector quantized VAEs, which can be globally solved. Subsequent weighted retraining on these queries allows to induce a distribution shift. In lack of a practically relevant problem, we consider a visually appealing application: the generation of happily smiling faces (where the training distribution only contains less happy people) - and show the principled behavior of our approach in terms of improved FID and higher smile degree over baseline approaches. | ['Margret Keuper', 'Claudia Schillings', 'Jan Christian Schwedhelm', 'Steffen Jung'] | 2023-06-11 | null | null | null | null | ['architecture-search'] | ['methodology'] | [ 3.75388205e-01 4.87976789e-01 -7.89462030e-02 -4.33943927e-01
-6.11226201e-01 -4.96423930e-01 8.59599352e-01 -1.74468264e-01
-1.34478718e-01 9.43415165e-01 7.04120100e-02 -7.51590803e-02
-1.71836659e-01 -8.50393116e-01 -8.51766527e-01 -8.98459077e-01
1.58040836e-01 7.91960895e-01 -4.39699501e-01 -1.85485989e-01
9.00501534e-02 5.56320190e-01 -1.66807377e+00 8.42972770e-02
1.04445004e+00 9.43058670e-01 6.35453835e-02 4.90396887e-01
4.21258919e-02 3.93685549e-01 -5.88039935e-01 -3.54917496e-01
2.41205364e-01 -7.26875424e-01 -5.31939089e-01 2.24622786e-01
5.25798678e-01 -1.27807245e-01 2.64197677e-01 1.06622708e+00
5.02213955e-01 1.96372047e-01 1.00899625e+00 -1.22029698e+00
-7.10083663e-01 3.01127195e-01 -2.87008077e-01 -3.32688361e-01
1.61148712e-01 2.56883502e-01 1.07780683e+00 -7.65482426e-01
9.33286488e-01 1.35359800e+00 2.61523008e-01 8.71340394e-01
-1.76011860e+00 -3.39191645e-01 -7.31429532e-02 -2.90102623e-02
-1.09681439e+00 -4.95252281e-01 1.03921843e+00 -4.73866165e-01
4.56410706e-01 4.64517862e-01 7.80556440e-01 1.30432224e+00
5.23835272e-02 9.44239318e-01 1.18889034e+00 -4.41054106e-01
6.48535907e-01 5.56418896e-01 -4.20378327e-01 5.14269888e-01
-4.63183224e-02 -9.34947189e-03 -4.89911377e-01 -1.92021906e-01
6.61958814e-01 -1.39808983e-01 -2.32453018e-01 -9.97444689e-01
-8.78950298e-01 1.16091752e+00 4.24734652e-01 1.85667709e-01
-4.81111646e-01 9.17779505e-02 4.61503305e-02 1.06646605e-01
6.91296041e-01 6.70281887e-01 -1.58040181e-01 1.07806765e-01
-1.41451919e+00 6.46378994e-01 7.10974157e-01 5.61246872e-01
8.07419181e-01 1.68194279e-01 -3.07639688e-01 7.14061201e-01
4.48354453e-01 3.57102364e-01 4.63260770e-01 -1.04074574e+00
-5.56329498e-03 4.49187040e-01 2.71903604e-01 -8.94742191e-01
2.79685874e-02 -6.31420791e-01 -7.36713946e-01 5.31623781e-01
4.93485510e-01 -1.49918988e-01 -7.69282758e-01 1.98812580e+00
5.28523982e-01 -8.91478807e-02 -1.10605666e-02 8.23250473e-01
3.36836517e-01 6.35445476e-01 -9.99379307e-02 -4.86992329e-01
9.31746304e-01 -5.52241087e-01 -5.20321250e-01 -9.38262790e-02
1.39603108e-01 -5.75045347e-01 1.28763020e+00 5.05070269e-01
-1.32232404e+00 -3.69752228e-01 -9.22329187e-01 7.08592832e-02
-9.56050009e-02 1.68274030e-01 5.36502063e-01 7.75658309e-01
-1.32968867e+00 8.17555547e-01 -5.28706074e-01 -3.66796106e-01
7.11964846e-01 2.87624300e-01 -1.50194556e-01 1.46343485e-01
-8.03499281e-01 6.70467138e-01 2.53088266e-01 1.11095168e-01
-1.01098335e+00 -5.63313603e-01 -7.60158896e-01 8.71788710e-02
2.50577569e-01 -1.11834121e+00 9.59843218e-01 -1.28303862e+00
-1.89824414e+00 9.34940875e-01 -1.81200489e-01 -5.50857008e-01
7.93202341e-01 -1.50479153e-01 9.19997245e-02 -1.12184279e-01
-1.34584844e-01 1.10246027e+00 1.39853358e+00 -1.29128766e+00
4.17460427e-02 -3.49338919e-01 -5.96351959e-02 1.65496960e-01
-2.86300212e-01 -3.68466258e-01 2.12690189e-01 -5.36381960e-01
-1.99424118e-01 -8.75831604e-01 -3.74344736e-01 1.03552885e-01
-5.26172221e-01 -1.89832568e-01 6.43151879e-01 -2.20909476e-01
9.39178705e-01 -1.85865796e+00 4.82186764e-01 2.63645411e-01
1.95588902e-01 1.26470160e-02 -1.17935032e-01 2.82433718e-01
-1.38979733e-01 7.28673637e-02 -2.36979231e-01 -5.05162954e-01
2.64928162e-01 1.01740360e-01 -7.03005970e-01 4.33973372e-01
3.31663430e-01 1.15885532e+00 -7.23613501e-01 -2.42897853e-01
8.68264362e-02 5.94065785e-01 -9.45385396e-01 2.18232512e-01
-7.25529075e-01 6.05780661e-01 -3.32979143e-01 2.46434227e-01
6.12289310e-01 -2.41194382e-01 2.03558952e-01 -5.18203117e-02
1.51269481e-01 -3.28442752e-02 -8.77746582e-01 1.63803101e+00
-4.61010516e-01 6.00439548e-01 -1.90601811e-01 -1.02377987e+00
1.23860812e+00 -9.77892876e-02 3.30478013e-01 -3.06724787e-01
1.09134980e-01 1.12772077e-01 -7.30634481e-02 -2.06800938e-01
4.60431278e-01 -6.40523851e-01 1.39131531e-01 3.08673888e-01
1.19828619e-01 -3.73597383e-01 -2.29059979e-02 -5.10028489e-02
4.94461656e-01 3.83844763e-01 3.73082101e-01 -2.73920566e-01
3.70372355e-01 -3.20560843e-01 3.54727395e-02 5.93731582e-01
1.00939043e-01 6.49970591e-01 7.09360301e-01 -3.40621829e-01
-1.31001389e+00 -1.21794534e+00 -1.41793162e-01 8.63411963e-01
-2.86078244e-01 -2.32840046e-01 -9.98669505e-01 -4.74004418e-01
-2.43486434e-01 9.57490861e-01 -7.77802348e-01 -4.09756079e-02
-3.36376160e-01 -7.16781139e-01 4.80927937e-02 9.92284790e-02
7.90659338e-02 -1.22425365e+00 -8.08813810e-01 6.04228452e-02
1.29492134e-01 -6.85314536e-01 -2.91154593e-01 -1.35161608e-01
-8.93563747e-01 -5.26411891e-01 -1.00979018e+00 -4.10248786e-01
7.13298082e-01 -4.44595873e-01 1.12756526e+00 -4.41132843e-01
-4.04290497e-01 4.46577460e-01 1.09994084e-01 -3.84753138e-01
-5.72790623e-01 -7.37043545e-02 -8.60447884e-02 4.63927329e-01
1.54840946e-01 -7.35300541e-01 -7.66313195e-01 7.51679540e-02
-8.57277155e-01 2.15191185e-01 3.63296956e-01 9.71765876e-01
6.17939770e-01 -3.33368689e-01 6.42390430e-01 -7.26411462e-01
8.83090079e-01 -3.70505095e-01 -7.12322891e-01 2.30836928e-01
-4.96241748e-01 4.63606328e-01 4.97516125e-01 -6.56283259e-01
-9.04648423e-01 2.30033025e-01 -7.02272952e-02 -5.42772412e-01
-2.64905959e-01 2.76890010e-01 -2.30023354e-01 2.37508669e-01
7.29126334e-01 3.45599264e-01 3.46142262e-01 -3.17370772e-01
7.55148351e-01 3.35855097e-01 3.86628419e-01 -7.60493994e-01
6.27268255e-01 5.49693286e-01 2.24350065e-01 -7.86745667e-01
-5.40678918e-01 5.20782098e-02 -3.70823860e-01 -2.84359515e-01
7.94284165e-01 -5.12764037e-01 -9.30599988e-01 4.86679561e-02
-1.14630389e+00 -4.30162042e-01 -8.22937548e-01 1.99578196e-01
-1.10702610e+00 1.82932168e-01 -9.68096405e-02 -1.17162848e+00
-3.85800719e-01 -1.18186295e+00 1.32732379e+00 1.88195199e-01
-4.93978322e-01 -1.01842439e+00 1.98328808e-01 3.72042991e-02
4.41673726e-01 5.43161690e-01 1.01420677e+00 -3.96203548e-01
-6.98349893e-01 -1.10034458e-01 1.74916297e-01 4.20957297e-01
7.13399202e-02 5.30350357e-02 -1.08758533e+00 -3.28444749e-01
1.29464030e-01 -4.25060153e-01 8.60132039e-01 7.22500682e-01
1.19440639e+00 -4.93151486e-01 -1.90413862e-01 6.16668284e-01
1.13344443e+00 -4.50857170e-02 8.16156387e-01 -1.53309777e-02
3.25697720e-01 8.84986639e-01 2.48665377e-01 6.07677221e-01
4.78037819e-02 9.48476613e-01 4.65891272e-01 1.07208177e-01
7.71768168e-02 -3.97842944e-01 4.55056608e-01 2.03266904e-01
-1.11962035e-01 -2.33930200e-01 -4.82288688e-01 4.36022520e-01
-1.74733651e+00 -1.17471552e+00 3.91366631e-01 2.41314530e+00
8.58766496e-01 1.65141039e-02 4.20908898e-01 -2.02361628e-01
4.97667313e-01 1.41582325e-01 -6.81898654e-01 -4.43533987e-01
-5.41162416e-02 4.83632028e-01 9.53129120e-03 5.26193678e-01
-8.02596867e-01 7.02144384e-01 6.57795334e+00 1.04033875e+00
-1.42380977e+00 -6.89907223e-02 1.07606769e+00 -4.00967658e-01
-8.98272038e-01 -1.33123267e-02 -6.88919544e-01 4.91240561e-01
7.45044291e-01 -2.31198817e-01 5.32897353e-01 8.79301131e-01
2.83366621e-01 2.59810220e-02 -1.28568375e+00 1.09802330e+00
4.32009399e-02 -1.47486591e+00 1.61787927e-01 4.05568033e-01
7.60388970e-01 -4.59999084e-01 7.98773706e-01 1.14404112e-01
8.94408207e-03 -1.42501080e+00 7.49249816e-01 8.11482668e-01
9.26484346e-01 -7.01492786e-01 1.06216528e-01 5.59253216e-01
-3.71779621e-01 9.19796079e-02 -4.01233763e-01 1.04701124e-01
1.11776277e-01 6.33815825e-01 -1.20717907e+00 1.69361621e-01
1.43244863e-01 4.35234755e-01 -3.25207472e-01 8.08801293e-01
-4.05457318e-02 4.75155085e-01 -3.68151546e-01 -2.76740819e-01
8.09594020e-02 -5.14588892e-01 7.14811683e-01 7.45755672e-01
7.14842916e-01 -4.51767385e-01 -1.64237931e-01 1.48602557e+00
-5.36899269e-02 3.37492973e-01 -7.46027470e-01 -7.55749419e-02
-5.51041253e-02 1.24216688e+00 -6.38473630e-01 -1.51756421e-01
1.66738972e-01 9.76392090e-01 3.88133466e-01 5.12680411e-01
-6.57492459e-01 1.06430024e-01 5.94448626e-01 3.09846789e-01
4.49224234e-01 -1.03632666e-01 -3.78333092e-01 -1.11381316e+00
-4.31525484e-02 -9.06969547e-01 -3.75389382e-02 -8.12508941e-01
-1.19943106e+00 6.21227860e-01 4.08918858e-02 -1.08528638e+00
-8.93790305e-01 -5.29236376e-01 -6.43061936e-01 1.00823283e+00
-1.09801388e+00 -1.07706118e+00 -3.31653878e-02 3.69378150e-01
4.73020315e-01 -4.43446368e-01 8.30338597e-01 -1.66472137e-01
-1.55463099e-01 6.03371441e-01 1.96151108e-01 -5.14536917e-01
4.92006332e-01 -1.33561969e+00 1.75211698e-01 5.90384126e-01
4.51306701e-01 7.54322231e-01 1.23689091e+00 -4.53451097e-01
-9.17014420e-01 -7.97566533e-01 5.76975703e-01 -4.72632408e-01
3.79957467e-01 -6.02283895e-01 -7.38377333e-01 3.11257571e-01
2.18074352e-01 -8.51232782e-02 5.41345179e-01 4.76801395e-02
-1.35685951e-01 -4.80539873e-02 -1.20604396e+00 7.45604157e-01
7.18330383e-01 -2.54950106e-01 -1.04479790e-01 3.23839933e-01
3.17608744e-01 -3.44365865e-01 -5.23445964e-01 2.13007405e-01
6.16005361e-01 -1.14756382e+00 9.24525499e-01 -7.69931257e-01
5.77668428e-01 -1.94177717e-01 -8.44549611e-02 -1.45700037e+00
8.76359418e-02 -9.30286169e-01 -2.14166537e-01 9.76978838e-01
2.37083077e-01 -5.05727410e-01 1.08621049e+00 4.34231579e-01
3.00846934e-01 -1.14715505e+00 -9.92396176e-01 -5.58993638e-01
2.49905944e-01 -3.65901738e-01 6.19362473e-01 5.86152613e-01
-1.77896559e-01 3.77266586e-01 -6.38091683e-01 -1.74468353e-01
6.69014156e-01 3.59210581e-01 1.07702482e+00 -1.03125060e+00
-8.25555623e-01 -6.34497583e-01 -3.66827756e-01 -1.36944199e+00
1.92108348e-01 -9.15848434e-01 1.46397129e-02 -1.25986350e+00
2.39637762e-01 -4.16514367e-01 2.53594946e-02 1.68544918e-01
-1.09820841e-02 1.36623800e-01 2.99694389e-01 1.27196148e-01
-3.48626375e-01 9.02035475e-01 1.43276250e+00 -1.24362387e-01
-2.64943063e-01 2.61890084e-01 -7.91768372e-01 4.85798270e-01
6.10101342e-01 -2.53027737e-01 -7.03830600e-01 8.93428400e-02
5.02995074e-01 1.11864991e-01 7.24938214e-01 -7.09543049e-01
-2.44508609e-01 -2.28251919e-01 3.96502733e-01 -3.03815514e-01
5.20422399e-01 -4.43847328e-01 2.76139230e-01 2.66902179e-01
-4.79523957e-01 -4.17126358e-01 -2.79514920e-02 4.79493707e-01
-4.81776930e-02 -3.43122929e-01 8.39457333e-01 4.57525402e-02
-1.89719260e-01 2.24463433e-01 -1.85126550e-02 -9.68351513e-02
1.04444909e+00 -2.51585513e-01 -1.48561492e-03 -8.13882589e-01
-8.76778185e-01 -2.14938700e-01 6.17584765e-01 2.22129360e-01
6.20498061e-01 -1.37314475e+00 -7.56081760e-01 2.56271809e-01
-6.87374324e-02 1.33360866e-02 3.02292436e-01 6.74889743e-01
-2.46286914e-01 1.11457020e-01 -3.51657569e-01 -8.04209352e-01
-1.04325628e+00 5.73101223e-01 4.67034608e-01 -2.43554085e-01
-2.68765748e-01 8.08996379e-01 6.18848205e-01 -2.80088454e-01
-1.17430203e-01 -2.49153867e-01 -7.41186440e-02 3.89257707e-02
2.75754631e-01 9.17111263e-02 -2.63806760e-01 -5.02101243e-01
-7.48383924e-02 3.50884467e-01 2.23138258e-02 -3.21828723e-01
1.44974160e+00 5.39309494e-02 1.57220289e-02 3.96668881e-01
1.15869188e+00 1.33275762e-01 -1.61033976e+00 1.67851180e-01
-2.97505677e-01 -4.08840507e-01 -2.70629019e-01 -7.27981746e-01
-7.39767551e-01 1.06799889e+00 7.31003046e-01 2.72160321e-01
1.15734148e+00 2.74337083e-01 4.60598528e-01 -3.65861319e-02
1.48589402e-01 -8.89501333e-01 4.14004505e-01 4.39393297e-02
1.04826283e+00 -1.13052595e+00 -6.79948926e-02 -3.65482606e-02
-5.98910868e-01 1.13226080e+00 2.17139348e-01 -2.12730780e-01
3.30899268e-01 5.06879836e-02 -1.81183919e-01 -9.68326256e-02
-7.63896942e-01 -5.87267540e-02 6.84453607e-01 7.45415151e-01
4.09682244e-01 1.55303687e-01 -2.40500942e-01 2.62416750e-01
-5.58854640e-01 -7.92541169e-03 3.87913376e-01 2.13451162e-01
-4.12204802e-01 -1.21773505e+00 -2.74605393e-01 3.65951896e-01
-2.39478171e-01 -1.21049032e-01 -1.58016041e-01 4.37274396e-01
1.30525962e-01 4.28365380e-01 8.43537375e-02 9.93519649e-02
-5.36358356e-02 2.22190499e-01 8.18605006e-01 -4.99017417e-01
3.87364253e-02 3.23334008e-01 -2.61630744e-01 -3.94728124e-01
-3.39475930e-01 -9.20187116e-01 -6.77054942e-01 2.36175880e-02
-1.00285105e-01 2.30112653e-02 7.22154260e-01 8.71902049e-01
3.21683228e-01 2.44341537e-01 5.31796277e-01 -9.69728708e-01
-7.33573675e-01 -7.65472770e-01 -4.87540454e-01 4.79164988e-01
3.34236264e-01 -5.42343736e-01 -1.41936138e-01 9.35397670e-02] | [11.644442558288574, -0.16647914052009583] |
cffa4099-4df7-4685-8a68-872a728f314f | one-agent-to-rule-them-all-towards-multi-1 | 2203.07665 | null | https://arxiv.org/abs/2203.07665v1 | https://arxiv.org/pdf/2203.07665v1.pdf | One Agent To Rule Them All: Towards Multi-agent Conversational AI | The increasing volume of commercially available conversational agents (CAs) on the market has resulted in users being burdened with learning and adopting multiple agents to accomplish their tasks. Though prior work has explored supporting a multitude of domains within the design of a single agent, the interaction experience suffers due to the large action space of desired capabilities. To address these problems, we introduce a new task BBAI: Black-Box Agent Integration, focusing on combining the capabilities of multiple black-box CAs at scale. We explore two techniques: question agent pairing and question response pairing aimed at resolving this task. Leveraging these techniques, we design One For All (OFA), a scalable system that provides a unified interface to interact with multiple CAs. Additionally, we introduce MARS: Multi-Agent Response Selection, a new encoder model for question response pairing that jointly encodes user question and agent response pairs. We demonstrate that OFA is able to automatically and accurately integrate an ensemble of commercially available CAs spanning disparate domains. Specifically, using the MARS encoder we achieve the highest accuracy on our BBAI task, outperforming strong baselines. | ['Jason Mars', 'Lingjia Tang', 'Yiping Kang', 'Walter Lasecki', 'Kevin Leach', 'Walter Talamonti', 'Karthik Krishnamurthy', 'Joseph Joshua Peper', 'Christopher Clarke'] | 2022-03-15 | null | https://aclanthology.org/2022.findings-acl.257 | https://aclanthology.org/2022.findings-acl.257.pdf | findings-acl-2022-5 | ['conversational-response-selection', 'multi-agent-integration'] | ['natural-language-processing', 'natural-language-processing'] | [-7.66981766e-02 1.95461512e-01 6.00335337e-02 -4.60931480e-01
-1.17781425e+00 -8.43246281e-01 1.02710676e+00 5.11151031e-02
-4.43838418e-01 6.30424142e-01 5.90348899e-01 -2.41488442e-01
5.92231564e-02 -4.30077046e-01 -3.88573498e-01 -3.44216377e-02
7.08926618e-02 9.94195342e-01 1.19081661e-01 -8.04032087e-01
-1.40409628e-02 -5.14009818e-02 -1.43195236e+00 5.18675148e-01
9.72983241e-01 7.56305993e-01 2.64153361e-01 1.04582298e+00
-1.16188668e-01 1.52640653e+00 -9.71031368e-01 -6.81232750e-01
1.83957532e-01 -4.40658987e-01 -1.37726438e+00 -2.01346725e-01
5.14322042e-01 -1.06859362e+00 -2.41056427e-01 2.86853433e-01
6.18564367e-01 4.13215697e-01 4.57386106e-01 -1.73589921e+00
-1.03401303e+00 7.46841073e-01 1.44211143e-01 -1.38943240e-01
1.12918067e+00 5.06981730e-01 1.44402754e+00 -6.60724282e-01
7.02207267e-01 1.12450957e+00 4.01126266e-01 8.57476294e-01
-1.06005764e+00 -2.28127688e-01 2.37585813e-01 2.65283406e-01
-7.13301778e-01 -7.18044519e-01 4.39198881e-01 -4.66269076e-01
1.50449431e+00 5.31790257e-01 3.41472328e-01 1.36759853e+00
-3.00988495e-01 1.00126994e+00 7.72674501e-01 -1.46375537e-01
1.72918856e-01 3.31265002e-01 4.00072187e-01 6.06063485e-01
-2.22229540e-01 -3.38317335e-01 -7.87422776e-01 -5.01374424e-01
3.59391898e-01 -3.33439618e-01 -1.16797216e-01 -2.94770032e-01
-1.32079172e+00 8.19563806e-01 3.40509504e-01 3.10361497e-02
-5.14917970e-01 1.53583568e-02 5.02587736e-01 6.56420529e-01
8.78848135e-02 1.02715814e+00 -5.45771897e-01 -5.36489248e-01
4.61438932e-02 9.71777141e-01 1.54597950e+00 1.05439389e+00
6.15157008e-01 -3.35689902e-01 -2.75779486e-01 1.06495965e+00
3.57951403e-01 3.51829350e-01 3.05691063e-01 -1.34903157e+00
5.64834297e-01 9.19203281e-01 6.33092225e-01 -5.15964031e-01
-6.78819239e-01 5.37558645e-02 -2.33178303e-01 3.35406959e-01
6.07798576e-01 -4.14205343e-01 -3.41462791e-01 1.89201570e+00
6.00178778e-01 -2.71109104e-01 5.41597843e-01 9.71746087e-01
1.31705058e+00 6.17173135e-01 3.35899830e-01 2.28525311e-01
1.73954201e+00 -1.56416035e+00 -5.96832156e-01 -5.92288852e-01
7.79318988e-01 -5.04078448e-01 1.38419116e+00 -3.40711698e-03
-1.31387854e+00 -1.54822484e-01 -1.09147584e+00 -4.62066293e-01
-2.56824404e-01 -2.99452335e-01 7.48130858e-01 2.13293955e-01
-1.21709788e+00 -7.30834156e-02 -3.73088866e-01 -4.24420595e-01
-6.03153035e-02 3.21180165e-01 -5.27815163e-01 1.67217493e-01
-1.24421787e+00 1.50159240e+00 -3.14831764e-01 -2.04806596e-01
-8.47859561e-01 -7.66798019e-01 -9.94254708e-01 1.46701545e-01
3.77355278e-01 -1.24860227e+00 2.19786620e+00 -7.69696474e-01
-2.05136895e+00 5.23876131e-01 5.73667400e-02 -3.35620821e-01
4.08406347e-01 -3.58345330e-01 -1.73633814e-01 2.33693365e-02
2.24229380e-01 6.73968613e-01 4.04813200e-01 -1.15280712e+00
-7.44092703e-01 -8.58759657e-02 9.42062378e-01 7.63391078e-01
-2.39660487e-01 3.84979844e-01 -4.64526080e-02 -7.92624131e-02
-6.65159285e-01 -9.37372446e-01 -6.30188435e-02 -3.57744306e-01
-5.88494120e-04 -8.11906159e-01 6.07658207e-01 -8.03209066e-01
8.14572573e-01 -1.81831288e+00 4.47581202e-01 -3.06240112e-01
4.66500819e-01 6.76911920e-02 -5.48240840e-01 9.81639445e-01
4.47497278e-01 -1.57834534e-02 -4.73462194e-02 -7.54168630e-01
5.10571539e-01 5.73984422e-02 -1.88329071e-01 4.03463747e-03
1.02597483e-01 9.59790945e-01 -8.94885361e-01 -2.96652168e-01
-4.40708399e-02 2.92479187e-01 -7.21445858e-01 8.96225572e-01
-8.02998543e-01 3.97073954e-01 -4.79984254e-01 5.26032865e-01
2.46635407e-01 -6.23934627e-01 1.69706747e-01 2.30986416e-01
1.56794459e-01 8.33172202e-01 -7.54467964e-01 2.12076092e+00
-7.77282000e-01 5.18848777e-01 6.00356936e-01 -1.32815167e-01
5.98648787e-01 6.54901981e-01 5.16950548e-01 -7.19108403e-01
-6.90776780e-02 2.86905199e-01 1.15292251e-01 -7.34349549e-01
7.76349604e-01 4.13658500e-01 -5.24327517e-01 1.08340573e+00
1.40945211e-01 -2.51114041e-01 1.88857377e-01 6.68248892e-01
1.47683895e+00 -2.31042076e-02 3.30117851e-01 3.26297343e-01
3.93272758e-01 2.07742259e-01 7.89310858e-02 9.18624699e-01
-5.64565957e-01 2.48001769e-01 3.48313093e-01 -4.28546488e-01
-9.79604006e-01 -8.48492205e-01 3.86432618e-01 1.84114993e+00
8.60676616e-02 -5.18953681e-01 -6.65212572e-01 -8.22423577e-01
1.99770615e-01 5.83787262e-01 -4.80325639e-01 1.75263077e-01
-7.39322305e-01 -2.95294851e-01 5.40749967e-01 3.78341824e-01
5.10838091e-01 -1.31247914e+00 -7.01443434e-01 2.33129010e-01
-5.44018269e-01 -1.26178050e+00 -5.93706191e-01 -3.40523809e-01
1.91595182e-01 -1.07592487e+00 -4.63042080e-01 -8.45201492e-01
-2.73891948e-02 4.55351591e-01 1.64489400e+00 2.21875951e-01
1.57169342e-01 8.35073948e-01 -5.94185889e-01 -1.59978464e-01
-8.50239933e-01 4.15264666e-01 -5.88536784e-02 -3.52498412e-01
3.01924288e-01 -5.28552830e-01 -7.97256470e-01 3.85929614e-01
-4.28070515e-01 3.00120890e-01 5.07145762e-01 5.58808684e-01
-6.78504050e-01 -1.00368750e+00 1.26027369e+00 -8.43492508e-01
1.37452507e+00 -8.62403691e-01 -1.80734426e-01 3.94147456e-01
-3.79771441e-01 -1.98079810e-01 4.81962860e-01 -2.71308273e-01
-1.23104310e+00 -3.38785678e-01 -3.38639826e-01 5.31220019e-01
-2.03045577e-01 4.51457113e-01 -4.56299223e-02 3.40085141e-02
9.90967989e-01 -3.70764248e-02 4.14760143e-01 -1.42700627e-01
7.91959584e-01 1.17998576e+00 5.91012836e-01 -7.11112559e-01
3.41283441e-01 -2.62681424e-04 -8.48815799e-01 -4.45334256e-01
-4.59330440e-01 -5.96155763e-01 5.61779067e-02 -4.69501734e-01
8.78223598e-01 -9.94788766e-01 -1.50601935e+00 4.15568769e-01
-1.46635675e+00 -7.43815064e-01 8.04791227e-02 1.26127481e-01
-8.40948701e-01 8.08755755e-02 -9.32848692e-01 -7.23364294e-01
-7.87407517e-01 -1.34865069e+00 9.29947197e-01 3.42522383e-01
-8.63641798e-01 -7.07615733e-01 4.89941716e-01 9.54099774e-01
7.89594650e-01 -2.13757977e-01 7.51334786e-01 -1.21569920e+00
-6.34070992e-01 -1.92071691e-01 -1.90208361e-01 -7.21351877e-02
9.30535719e-02 -4.13178861e-01 -8.77181470e-01 -2.62398303e-01
-1.32338956e-01 -9.46694016e-01 5.19145764e-02 -2.38070175e-01
2.33005702e-01 -5.68235636e-01 -3.33004780e-02 -5.34551665e-02
7.33322442e-01 2.63834029e-01 2.17406794e-01 6.57572627e-01
4.60218817e-01 6.83708608e-01 3.04290801e-01 4.65335399e-01
1.48910964e+00 9.17313814e-01 3.70822906e-01 1.94439515e-01
-6.21640161e-02 -1.63098983e-02 3.96297276e-01 6.56792283e-01
2.27385029e-01 -4.98692840e-01 -8.69711816e-01 5.33779681e-01
-2.25739312e+00 -8.93632770e-01 1.89061478e-01 1.68779635e+00
1.11343157e+00 -4.01147097e-01 6.02731109e-01 -6.00253284e-01
1.30011946e-01 2.63084441e-01 -7.46609867e-01 -5.23527145e-01
1.53405800e-01 -2.42726848e-01 -1.10121831e-01 7.27463603e-01
-8.80125701e-01 6.72085166e-01 6.80217886e+00 -5.37413247e-02
-5.22591591e-01 2.55238980e-01 1.13162704e-01 -2.33326331e-02
-4.13359016e-01 -1.15741611e-01 -4.65562105e-01 3.65556389e-01
1.04925656e+00 -2.33111411e-01 8.61104488e-01 8.84905577e-01
-1.42615795e-01 -4.28474359e-02 -1.48477781e+00 7.28091002e-01
2.89657414e-01 -1.36712861e+00 -3.83260213e-02 -2.75678039e-01
3.81630301e-01 1.56899780e-01 -8.76302794e-02 8.48187506e-01
1.10397315e+00 -9.30858016e-01 5.44561267e-01 1.82980120e-01
3.40216249e-01 -1.56264007e-01 5.23808241e-01 4.22354043e-01
-6.97937310e-01 -3.63249779e-01 3.53498310e-01 -4.23349679e-01
3.56403261e-01 -5.12240767e-01 -1.29209781e+00 3.00415933e-01
4.07232195e-01 2.32686356e-01 -2.65012830e-01 8.07707012e-01
3.47338617e-02 -1.93381924e-02 -2.05823272e-01 -4.11411107e-01
8.50862414e-02 -3.91176492e-02 5.88885486e-01 9.29775596e-01
-1.66031867e-01 1.16803899e-01 4.87841219e-01 5.06265521e-01
-4.34410691e-01 1.10162511e-01 -4.74705696e-01 6.44492134e-02
1.01817143e+00 1.35791934e+00 1.44264981e-01 -4.61803466e-01
-7.61720836e-01 1.05671012e+00 8.83801043e-01 3.93718868e-01
-6.96368992e-01 -1.16189547e-01 1.26644599e+00 -3.75347614e-01
-2.68854201e-01 -3.53512317e-01 -2.73605973e-01 -1.03673542e+00
1.48422658e-01 -1.64400804e+00 5.54518044e-01 -8.16176832e-01
-1.44475520e+00 8.58976841e-01 -2.41347387e-01 -7.61422098e-01
-9.15076911e-01 -3.13225269e-01 -6.54574990e-01 8.01974595e-01
-1.28999066e+00 -1.54171598e+00 -4.79106694e-01 4.95200574e-01
8.87591779e-01 -3.56196553e-01 1.15419984e+00 4.76789743e-01
-5.66193461e-01 6.24282122e-01 -3.58675927e-01 6.88798577e-02
1.11414599e+00 -1.46936750e+00 8.03433239e-01 3.22270364e-01
-1.67994902e-01 7.78527081e-01 7.72373378e-01 -4.59244430e-01
-1.67746389e+00 -4.27248299e-01 8.30432653e-01 -1.15512967e+00
7.58449614e-01 -5.87974489e-01 -8.65149856e-01 1.09384799e+00
9.39284563e-01 -5.96516788e-01 9.54859495e-01 4.67939854e-01
-6.21605277e-01 1.13451585e-01 -1.03829217e+00 8.61509800e-01
7.30867982e-01 -8.09590340e-01 -7.06427336e-01 4.79392916e-01
1.12468207e+00 -5.90882599e-01 -8.48325074e-01 8.89882594e-02
6.82893157e-01 -8.73509049e-01 1.04518902e+00 -8.37208450e-01
4.36184466e-01 -1.95009351e-01 6.58772066e-02 -1.57889760e+00
-1.60506830e-01 -9.38940763e-01 -3.16523015e-01 1.15646482e+00
6.11744106e-01 -8.25782418e-01 4.97254580e-01 1.51211715e+00
-3.96191299e-01 -4.66938406e-01 -7.08020210e-01 -1.59038097e-01
-4.76655625e-02 -1.02280192e-02 9.66268182e-01 8.45700383e-01
7.52479374e-01 9.56846535e-01 -5.00098169e-01 1.53549865e-01
2.24838182e-01 3.98221910e-02 1.36234117e+00 -9.61823761e-01
-6.46060884e-01 -7.12328911e-01 1.64421260e-01 -1.29797041e+00
1.86841980e-01 -7.01938450e-01 2.57673621e-01 -1.96931255e+00
8.88293013e-02 -6.20407224e-01 2.04241291e-01 4.14771527e-01
-3.62287998e-01 -1.78099826e-01 4.36182559e-01 3.42577279e-01
-9.52934206e-01 5.27070582e-01 1.05975711e+00 -5.06864823e-02
-3.39923561e-01 -1.67738006e-01 -9.86213148e-01 4.26579028e-01
6.89805865e-01 1.32768452e-01 -5.41919589e-01 -9.39731300e-01
5.29460192e-01 3.33203733e-01 2.54681259e-01 -7.65076280e-01
6.64204061e-01 -1.21178895e-01 -2.69528627e-01 -1.00852020e-01
7.44787037e-01 -5.55024922e-01 -1.54518202e-01 -1.83792382e-01
-8.76084685e-01 3.57301146e-01 -4.65154089e-02 3.45639348e-01
3.81549783e-02 -1.17159203e-01 2.11015582e-01 -1.13632374e-01
-6.71070695e-01 -5.00384681e-02 -5.65025687e-01 9.24205184e-02
1.02506018e+00 3.46681774e-01 -1.10782480e+00 -9.59838450e-01
-2.87137926e-01 8.49832118e-01 3.16161603e-01 7.05839276e-01
2.70554751e-01 -1.13375354e+00 -1.00880623e+00 -1.15519702e-01
3.65438789e-01 -3.61022279e-02 3.58924598e-01 4.03936863e-01
-4.02216554e-01 2.18096375e-01 -2.33352900e-01 -1.40899792e-01
-1.26105976e+00 2.69348621e-01 4.41790879e-01 -2.16314539e-01
-4.90896344e-01 9.18211818e-01 -3.46105173e-02 -9.87032235e-01
3.14822495e-01 1.11575030e-01 -4.72306401e-01 -1.66509058e-02
8.60824883e-01 3.21206391e-01 8.46503079e-02 -3.04340303e-01
-3.08845550e-01 -2.75957733e-01 -5.04612327e-01 -3.97669137e-01
1.21413612e+00 -2.86838740e-01 -2.56203026e-01 1.73961848e-01
9.13999856e-01 -1.82153713e-02 -1.14515305e+00 -3.56953293e-01
-5.65251969e-02 -3.14080715e-01 -5.12021661e-01 -1.35849297e+00
-5.94893992e-02 1.07803002e-01 -1.07904263e-02 6.95530772e-01
4.58543867e-01 2.13338777e-01 1.11834323e+00 7.93950617e-01
3.10297936e-01 -9.78785932e-01 3.72400850e-01 6.84281647e-01
1.28314590e+00 -1.51995003e+00 -3.56270105e-01 -1.75143063e-01
-1.07677090e+00 6.33751392e-01 1.12229204e+00 9.60100442e-02
-1.38995387e-02 2.52744317e-01 5.15309155e-01 -2.71709919e-01
-1.37007034e+00 -1.78128704e-01 -1.48734182e-01 6.73391819e-01
5.11304259e-01 6.37949109e-02 1.06361039e-01 5.82073927e-01
-3.82526159e-01 -2.16721833e-01 7.34273076e-01 1.09785914e+00
-2.41917908e-01 -1.23204768e+00 2.30259486e-02 3.35926235e-01
-5.28337546e-02 1.08124381e-02 -6.43504918e-01 5.41741967e-01
-7.82361507e-01 1.46169353e+00 -2.74617486e-02 -4.98628974e-01
6.38605475e-01 2.59580761e-01 1.61770627e-01 -5.71030200e-01
-1.19521976e+00 -7.15768635e-01 9.86617744e-01 -5.03440320e-01
-1.81082323e-01 -5.58196008e-01 -1.08590639e+00 -4.56711173e-01
3.36383581e-02 3.13917488e-01 6.87924087e-01 9.79302347e-01
7.75406241e-01 3.92224342e-01 5.20066082e-01 -6.66224480e-01
-9.47237313e-01 -1.20484638e+00 1.62008822e-01 5.75589299e-01
6.80205226e-01 -4.76233453e-01 -1.39990985e-01 -3.25875163e-01] | [12.544126510620117, 8.013285636901855] |
bd31d3c0-0ec4-4a23-8aec-6bbd0f8059ce | bayesian-optimisation-against-climate-change | 2306.04343 | null | https://arxiv.org/abs/2306.04343v1 | https://arxiv.org/pdf/2306.04343v1.pdf | Bayesian Optimisation Against Climate Change: Applications and Benchmarks | Bayesian optimisation is a powerful method for optimising black-box functions, popular in settings where the true function is expensive to evaluate and no gradient information is available. Bayesian optimisation can improve responses to many optimisation problems within climate change for which simulator models are unavailable or expensive to sample from. While there have been several feasibility demonstrations of Bayesian optimisation in climate-related applications, there has been no unifying review of applications and benchmarks. We provide such a review here, to encourage the use of Bayesian optimisation in important and well-suited application domains. We identify four main application domains: material discovery, wind farm layout, optimal renewable control and environmental monitoring. For each domain we identify a public benchmark or data set that is easy to use and evaluate systems against, while being representative of real-world problems. Due to the lack of a suitable benchmark for environmental monitoring, we propose LAQN-BO, based on air pollution data. Our contributions are: a) identifying a representative range of benchmarks, providing example code where necessary; b) introducing a new benchmark, LAQN-BO; and c) promoting a wider use of climate change applications among Bayesian optimisation practitioners. | ['Nigel H. Goddard', 'Christopher G. Lucas', 'Sigrid Passano Hellan'] | 2023-06-07 | null | null | null | null | ['bayesian-optimisation'] | ['methodology'] | [ 2.84740418e-01 -2.07614481e-01 7.72851706e-02 -1.83456719e-01
-6.21295869e-01 -7.24721253e-01 6.07807815e-01 8.74943137e-02
-1.61941037e-01 9.14170623e-01 -1.77796587e-01 -1.09515774e+00
-8.16423595e-01 -8.22216988e-01 -6.13763332e-01 -1.17564750e+00
-3.68086517e-01 4.42533195e-01 -3.49622332e-02 -2.35676169e-01
1.37865886e-01 5.48611045e-01 -1.64908516e+00 -3.75349283e-01
7.65123725e-01 6.35651171e-01 2.84451097e-01 1.13199532e+00
3.77361506e-01 -1.43870234e-01 -7.03076661e-01 1.74854577e-01
4.25705671e-01 -3.47686708e-01 -4.78715360e-01 -4.44816262e-01
-1.91734046e-01 4.94448766e-02 3.88871938e-01 7.32270598e-01
1.22815049e+00 4.45247293e-01 5.97703278e-01 -1.16646767e+00
2.21700757e-04 3.18006068e-01 -3.61738682e-01 3.35285336e-01
7.17185289e-02 7.66317427e-01 7.15944767e-01 -6.05050363e-02
-7.51951560e-02 1.29834270e+00 8.30152333e-01 3.17672998e-01
-1.58774912e+00 -7.43129134e-01 2.54157275e-01 -1.12268008e-01
-1.16405272e+00 -6.94824219e-01 1.78199887e-01 -3.98805946e-01
1.32871306e+00 9.08116162e-01 8.29371810e-01 7.99848497e-01
3.71192873e-01 3.09859902e-01 1.41008580e+00 -2.62762189e-01
7.75816023e-01 -1.59262702e-01 -3.08522224e-01 -2.18058690e-01
2.53330261e-01 1.06999362e+00 -2.22909451e-01 -5.56746006e-01
1.36511087e-01 -4.38421279e-01 -3.31653655e-01 -3.43475461e-01
-9.70095515e-01 8.19690824e-01 2.03022420e-01 -2.27753475e-01
-3.45124722e-01 5.20408034e-01 2.05406383e-01 2.91249454e-01
5.25575280e-01 6.64981782e-01 -1.01659298e+00 -3.32886338e-01
-8.75174880e-01 6.09205484e-01 1.15371561e+00 8.69756401e-01
3.43852192e-01 3.82066995e-01 -2.86220610e-02 8.58558595e-01
1.03128242e+00 1.40810061e+00 1.03983417e-01 -9.54728782e-01
1.42694847e-03 -3.15998465e-01 5.01963079e-01 -7.97712743e-01
-5.58321297e-01 -3.64562988e-01 -8.78863335e-01 5.28576672e-01
2.78882086e-01 -4.37623113e-01 -7.82438278e-01 1.45151949e+00
5.15437663e-01 5.76819554e-02 9.22596902e-02 5.93318105e-01
6.45509720e-01 1.00283182e+00 1.77829474e-01 -3.44714850e-01
1.29742932e+00 -3.07714283e-01 -5.40184557e-01 -4.17628139e-01
3.60116094e-01 -8.21653605e-01 6.02994442e-01 6.07766271e-01
-7.81964004e-01 -8.72985572e-02 -1.00478137e+00 7.95851231e-01
-7.47997165e-01 -5.14055192e-01 7.39858270e-01 1.49839842e+00
-9.41681504e-01 6.48210764e-01 -1.03396225e+00 -4.30883586e-01
1.52976573e-01 4.82887119e-01 3.09503853e-01 2.82119721e-01
-1.28726900e+00 1.30347753e+00 3.41761768e-01 5.77597618e-01
-1.18512523e+00 -1.01041448e+00 -9.11749363e-01 -5.14330305e-02
2.67655581e-01 -8.38927925e-01 1.40671146e+00 -3.36318493e-01
-2.02014399e+00 5.05918339e-02 7.71716014e-02 -3.08028817e-01
2.80931145e-01 -4.27767932e-02 -6.24359548e-01 -6.44804358e-01
-2.24828884e-01 4.68560457e-01 6.37921631e-01 -1.05937111e+00
-7.75112987e-01 -1.56824768e-01 -6.83464780e-02 2.54964083e-01
2.06731424e-01 2.18895122e-01 1.73867077e-01 -4.79202926e-01
-3.84670168e-01 -1.00129235e+00 -8.95046294e-01 -4.60862726e-01
-2.43872523e-01 -1.99358282e-03 6.17940485e-01 -6.02286041e-01
1.46428394e+00 -1.57846177e+00 -2.01002598e-01 6.14505708e-01
-4.98523414e-01 1.71816260e-01 -1.67775840e-01 4.61483359e-01
-3.19154859e-01 4.91599351e-01 -8.04171562e-01 2.83788353e-01
3.84475648e-01 4.93981719e-01 -3.81079048e-01 7.05964565e-01
2.95482934e-01 6.13630533e-01 -1.25495410e+00 1.31343037e-01
6.85148716e-01 2.66299665e-01 -2.83649921e-01 4.09560651e-02
-4.56323147e-01 5.55246234e-01 -2.99256772e-01 5.42029023e-01
7.51786828e-01 3.52545023e-01 -3.62667665e-02 3.22865367e-01
-4.35083777e-01 2.62823313e-01 -1.61823165e+00 1.11156797e+00
-8.06203067e-01 3.92253458e-01 2.40786687e-01 -9.29694176e-01
7.94378400e-01 3.65781605e-01 4.86449212e-01 -3.58950555e-01
-3.72970045e-01 1.31572783e-01 2.16212258e-01 -4.11164165e-01
1.64886281e-01 -1.94071054e-01 -3.84549499e-02 4.03341562e-01
-4.51937348e-01 -1.23747528e+00 3.62806380e-01 -5.46281755e-01
1.25286424e+00 4.09239411e-01 5.91548264e-01 -7.73617208e-01
2.88257867e-01 2.25181822e-02 5.82032859e-01 9.33250010e-01
-9.35870111e-02 1.74397260e-01 1.92991987e-01 -3.80796760e-01
-7.11085379e-01 -9.18312013e-01 -8.00236940e-01 1.21428037e+00
-2.63921976e-01 -3.48663360e-01 -2.09690779e-01 -3.59285809e-02
2.33233333e-01 1.13755572e+00 -2.49822542e-01 1.05209142e-01
-2.58742839e-01 -2.03943443e+00 3.95330161e-01 2.10250989e-01
2.22366631e-01 -8.37993026e-01 -9.95318413e-01 4.65445876e-01
2.18673289e-01 -2.92476356e-01 1.10382393e-01 8.11889350e-01
-1.14091718e+00 -8.84178817e-01 -5.22822738e-01 1.13305986e-01
1.11706316e-01 7.05192739e-04 1.62531257e+00 -3.33633095e-01
-3.63833308e-01 3.81420821e-01 -2.83213090e-02 -1.17287099e+00
-6.02788150e-01 -3.37829851e-02 -6.78669214e-02 -8.78911495e-01
2.01721713e-01 -4.53344882e-01 -6.36384606e-01 7.01661944e-01
-8.03169966e-01 -5.94849527e-01 4.16484386e-01 8.50417674e-01
3.66635948e-01 3.36131424e-01 6.75181091e-01 -5.68664372e-01
5.56384504e-01 -7.92093635e-01 -1.29558253e+00 1.04972512e-01
-1.03684449e+00 6.76181167e-02 7.83396885e-02 -2.55260337e-02
-9.45931196e-01 2.10780382e-01 -3.06632251e-01 3.81918430e-01
-4.72426385e-01 7.50318289e-01 -1.33330554e-01 4.12722304e-02
9.01250124e-01 -2.61327147e-01 -4.99668457e-02 -4.02027279e-01
4.13015574e-01 6.72332048e-01 1.82809755e-01 -8.09273481e-01
7.43018448e-01 2.32131749e-01 2.17359185e-01 -1.17968333e+00
-3.44857544e-01 -7.12376714e-01 -2.48699456e-01 -1.96683481e-01
5.03384531e-01 -8.39404106e-01 -7.17176914e-01 3.20175111e-01
-8.19534123e-01 -8.76784205e-01 -3.39224488e-01 5.10784924e-01
-6.46309495e-01 1.35287330e-01 2.36628830e-01 -1.33051324e+00
-3.12380135e-01 -1.34664166e+00 1.08183181e+00 1.07487537e-01
-3.30662191e-01 -1.24786317e+00 5.34252822e-01 -1.16557159e-01
8.25336814e-01 5.09860516e-01 6.07660592e-01 -1.99251220e-01
-2.73205936e-01 -1.74503364e-02 2.88253695e-01 2.27589175e-01
1.27425805e-01 6.39447868e-01 -1.28551781e+00 -4.68032777e-01
2.78469129e-03 1.63570270e-01 6.91026926e-01 8.53763223e-01
1.10797071e+00 -5.59582040e-02 -5.73189259e-01 5.49902618e-01
1.13137841e+00 3.01894546e-01 6.47750974e-01 5.51210642e-01
-4.63234552e-04 6.52166367e-01 5.62156498e-01 5.29453576e-01
1.40199289e-01 6.71952844e-01 7.33858824e-01 -2.51017034e-01
3.94540519e-01 6.74597859e-01 7.02500939e-01 5.85910201e-01
-9.77895558e-02 -3.14528704e-01 -1.16665602e+00 6.88028514e-01
-1.67538118e+00 -9.92352366e-01 -4.68228489e-01 2.63612342e+00
1.00806963e+00 -9.07159001e-02 5.83459400e-02 1.87946241e-02
6.31319761e-01 1.40912637e-01 -5.92367947e-01 -7.34619260e-01
1.17110603e-01 5.53278983e-01 1.05845582e+00 5.51653504e-01
-1.39517033e+00 2.58873731e-01 7.90747643e+00 8.47980559e-01
-7.06258893e-01 1.48254588e-01 4.96424019e-01 -1.43621027e-01
-2.26861447e-01 2.01220557e-01 -9.37896073e-01 4.72003639e-01
1.74606729e+00 -1.24421053e-01 5.98399878e-01 5.33846021e-01
8.41075659e-01 -6.12529457e-01 -1.02322412e+00 2.98586130e-01
-5.86257815e-01 -7.60743976e-01 -8.03248167e-01 1.37177974e-01
7.60207057e-01 4.43989605e-01 -1.67294726e-01 3.99989992e-01
1.13366485e+00 -1.33145559e+00 3.87022346e-01 5.23406506e-01
2.92292148e-01 -8.33504021e-01 8.62067342e-01 2.38646150e-01
-1.03245533e+00 -5.88600859e-02 -4.54667449e-01 -9.81098562e-02
1.43131763e-01 1.16756988e+00 -6.54254377e-01 8.25256586e-01
1.27033877e+00 5.29462516e-01 -3.45316321e-01 1.72058046e+00
-3.11264455e-01 1.14264703e+00 -1.06887591e+00 -2.21048757e-01
1.48944914e-01 -3.04744571e-01 8.48031044e-01 1.34864366e+00
5.05299747e-01 -1.60222411e-01 -1.17649563e-01 8.43937278e-01
8.21151435e-01 -1.75070569e-01 -7.43633747e-01 1.63933203e-01
5.01117468e-01 1.16224980e+00 -5.02764463e-01 2.01965436e-01
-1.99807838e-01 -2.32338738e-02 -8.18656921e-01 4.27871019e-01
-8.48851919e-01 -5.08422375e-01 1.01188159e+00 -4.63255793e-01
2.64897287e-01 -9.69470292e-02 -4.41705495e-01 -2.20821321e-01
-4.34748769e-01 -1.12900746e+00 4.99980360e-01 -8.31489682e-01
-1.16314793e+00 -1.98234096e-01 8.36894631e-01 -9.30550456e-01
-4.92412031e-01 -8.37329984e-01 -7.43434131e-01 1.44186556e+00
-1.70084965e+00 -4.59755957e-01 -1.86532035e-01 -2.97155291e-01
2.31740147e-01 1.90243684e-02 1.14707148e+00 5.83323464e-02
-6.78795576e-01 -1.69796571e-01 8.67930412e-01 -7.67176569e-01
6.97833180e-01 -1.60795534e+00 7.64622211e-01 5.76572061e-01
-4.12354499e-01 5.91820061e-01 1.17466187e+00 -5.34228623e-01
-1.35321796e+00 -1.16941822e+00 1.66963294e-01 -7.84872413e-01
7.79334068e-01 -1.77587271e-01 -5.07475615e-01 7.99489990e-02
2.28641331e-01 -4.07148123e-01 7.68726468e-01 2.99381107e-01
3.40902478e-01 -3.87833625e-01 -1.20027673e+00 4.79166806e-01
5.07434249e-01 -1.26177352e-02 -2.95795977e-01 6.77513182e-01
2.90931255e-01 -4.93108690e-01 -1.25777698e+00 7.36408770e-01
6.10003531e-01 -5.19721806e-01 1.17090726e+00 -4.01924580e-01
-7.10492805e-02 -8.31400335e-01 -1.34370610e-01 -1.84685206e+00
-1.86847717e-01 -1.08495033e+00 -1.61869079e-01 1.31610346e+00
5.69130659e-01 -1.04981863e+00 2.65027553e-01 5.83553016e-01
-1.54277831e-01 -5.76587915e-01 -1.10093129e+00 -9.35589314e-01
4.67470348e-01 -7.12667763e-01 1.02916670e+00 5.98953366e-01
-3.66660416e-01 -3.08051612e-02 -4.11049165e-02 5.81789732e-01
5.66971540e-01 -6.18630350e-02 9.04824734e-01 -1.26334214e+00
-3.61953348e-01 -6.64251387e-01 -1.11484736e-01 -7.05723345e-01
-2.65938133e-01 -5.06025255e-01 7.94267416e-01 -1.37273073e+00
-2.60580420e-01 -6.68864608e-01 -2.11930946e-01 4.62717772e-01
-2.16496632e-01 -1.08190946e-01 -4.40934479e-01 -2.66725391e-01
1.35045841e-01 5.80963552e-01 8.70621800e-01 -4.19726402e-01
-2.43795574e-01 5.63198149e-01 -6.82458997e-01 4.24793929e-01
9.43946362e-01 -8.39885592e-01 -3.99102181e-01 -1.68682069e-01
5.73415279e-01 -3.25174868e-01 4.74776089e-01 -9.28021491e-01
-8.93605649e-02 -7.01835752e-01 3.42450142e-01 -7.79103339e-01
-4.37527448e-02 -9.54576790e-01 7.61868715e-01 5.88871300e-01
7.93243945e-02 2.55468488e-01 8.49551320e-01 7.37863004e-01
3.94307882e-01 -7.06112742e-01 8.20438743e-01 -9.79768187e-02
-6.56536996e-01 -5.09164594e-02 -8.40587020e-01 1.37361828e-02
9.90537345e-01 -4.09355350e-02 -2.78947622e-01 -2.82335192e-01
-5.40235162e-01 8.13475490e-01 3.13117534e-01 2.56055921e-01
-5.37533239e-02 -9.58570123e-01 -9.25647378e-01 -1.82023924e-02
-1.47297550e-02 3.97640467e-01 -1.13931760e-01 8.15556228e-01
-5.56706786e-01 5.41974962e-01 1.55007347e-01 -8.94822061e-01
-9.47347045e-01 3.20226103e-01 7.30571628e-01 -1.24216631e-01
-8.42812061e-02 5.48713863e-01 -2.24065110e-02 -1.00366592e+00
1.72573384e-02 -6.53291106e-01 1.07481800e-01 -3.95481177e-02
2.93498755e-01 6.59679770e-01 4.92302388e-01 -1.40827119e-01
-3.65990341e-01 2.81547695e-01 8.78268421e-01 -2.02529579e-01
1.50372124e+00 -2.24306807e-01 -3.15776676e-01 5.75060844e-01
2.77833015e-01 -3.03825527e-01 -1.20686662e+00 3.80744964e-01
9.96059179e-02 -3.77349108e-01 4.84049916e-01 -9.78739679e-01
-7.46677756e-01 8.09227705e-01 1.04723144e+00 4.40359175e-01
1.11860406e+00 -7.13666856e-01 -8.21319502e-03 8.18801701e-01
1.19734883e-01 -1.11952198e+00 -6.72522664e-01 5.93627036e-01
1.16984963e+00 -1.01772130e+00 5.52935600e-01 3.94581929e-02
-4.86895740e-02 8.16859066e-01 1.48771107e-01 4.63221341e-01
1.08591771e+00 5.73302150e-01 1.04660839e-01 -1.29332766e-01
-8.00124466e-01 -3.07584524e-01 1.17294930e-01 1.17704022e+00
3.68981391e-01 1.87399551e-01 1.07539572e-01 1.63506925e-01
-3.07304353e-01 -2.34776035e-01 2.44839460e-01 1.00300479e+00
-3.83146107e-01 -1.16761220e+00 -1.08982658e+00 6.99771345e-01
-3.29590946e-01 -3.65637928e-01 1.32044675e-02 7.44611323e-01
-2.26046637e-01 1.50671172e+00 -2.40703180e-01 3.22985530e-01
3.83870691e-01 1.38182921e-04 1.13885008e-01 -6.47103608e-01
-7.91898668e-01 1.03307724e-01 5.13118625e-01 -4.21169847e-01
-5.58327258e-01 -9.71494913e-01 -3.88530672e-01 -4.46027786e-01
-8.35149407e-01 2.68675506e-01 1.10647666e+00 7.02214777e-01
1.35698363e-01 8.32814276e-01 5.50619781e-01 -1.31242895e+00
-8.60923290e-01 -9.47084725e-01 -6.58737183e-01 -4.25753623e-01
2.01137766e-01 -8.00384939e-01 -5.07972121e-01 -3.28165479e-02] | [6.116626262664795, 3.6142234802246094] |
6f94f00c-c371-4816-8794-c7a463a29024 | boosting-graph-structure-learning-with-dummy | 2206.08561 | null | https://arxiv.org/abs/2206.08561v1 | https://arxiv.org/pdf/2206.08561v1.pdf | Boosting Graph Structure Learning with Dummy Nodes | With the development of graph kernels and graph representation learning, many superior methods have been proposed to handle scalability and oversmoothing issues on graph structure learning. However, most of those strategies are designed based on practical experience rather than theoretical analysis. In this paper, we use a particular dummy node connecting to all existing vertices without affecting original vertex and edge properties. We further prove that such the dummy node can help build an efficient monomorphic edge-to-vertex transform and an epimorphic inverse to recover the original graph back. It also indicates that adding dummy nodes can preserve local and global structures for better graph representation learning. We extend graph kernels and graph neural networks with dummy nodes and conduct experiments on graph classification and subgraph isomorphism matching tasks. Empirical results demonstrate that taking graphs with dummy nodes as input significantly boosts graph structure learning, and using their edge-to-vertex graphs can also achieve similar results. We also discuss the gain of expressive power from the dummy in neural networks. | ['Xin Jiang', 'Yangqiu Song', 'Jiayang Cheng', 'Xin Liu'] | 2022-06-17 | null | null | null | null | ['graph-structure-learning'] | ['graphs'] | [-1.15840491e-02 5.64881861e-01 -5.42828381e-01 -2.73101658e-01
-1.82009399e-01 -6.34428859e-01 3.64588350e-01 3.77504379e-01
-1.33522958e-01 3.11695069e-01 1.78510740e-01 -4.66089249e-01
-3.31232920e-02 -1.15425611e+00 -8.47636521e-01 -6.06554806e-01
-7.86303222e-01 6.87988773e-02 1.32653564e-01 -1.13681190e-01
-1.19348384e-01 6.66125357e-01 -9.70656753e-01 5.97942658e-02
6.39866769e-01 3.84124994e-01 -7.42389485e-02 6.68729603e-01
-1.10315718e-01 8.56251657e-01 -1.27295047e-01 -5.32283664e-01
6.72756612e-01 -1.22558787e-01 -7.61282206e-01 1.31372465e-02
6.19612277e-01 -2.11178035e-01 -9.33246076e-01 1.22513115e+00
3.23024720e-01 1.58809453e-01 4.76824373e-01 -1.51129222e+00
-1.06695521e+00 1.02765632e+00 -7.97916770e-01 1.07743733e-01
2.16796532e-01 -5.44175170e-02 1.39057171e+00 -6.27116859e-01
5.62994421e-01 1.20269811e+00 9.47624028e-01 2.76043892e-01
-1.40030730e+00 -7.01878011e-01 3.03815633e-01 1.10163458e-01
-1.24149168e+00 -1.92304943e-02 1.10442710e+00 -2.52508104e-01
8.16050589e-01 2.68481791e-01 6.48452401e-01 6.36290789e-01
2.08489466e-02 6.21082604e-01 8.47941458e-01 -4.75263149e-01
-1.53717712e-01 6.54456541e-02 4.37366068e-01 1.48941135e+00
4.51947838e-01 1.64170817e-01 -3.53286712e-04 -3.12259048e-01
8.84843946e-01 1.46748543e-01 -4.47740972e-01 -8.76603484e-01
-7.51895368e-01 1.11333406e+00 1.02969420e+00 1.90150321e-01
-5.98037541e-02 4.49774176e-01 5.21461189e-01 7.31845140e-01
4.29608017e-01 4.62054253e-01 -3.14897656e-01 5.37935972e-01
-3.79565060e-01 -2.41610259e-01 9.98019755e-01 1.01767814e+00
1.31972575e+00 2.60420203e-01 1.61016330e-01 6.13023400e-01
-2.39199758e-01 1.82575896e-01 1.33422270e-01 -5.95266104e-01
4.45472300e-01 9.07248259e-01 -6.61342025e-01 -1.52467847e+00
-6.07181191e-01 -3.65984231e-01 -1.18286479e+00 -4.99609448e-02
3.18088740e-01 2.28795949e-02 -8.89614046e-01 1.97832179e+00
1.72312662e-01 4.99277174e-01 -8.35847035e-02 4.60544914e-01
8.37048888e-01 4.50189620e-01 -8.36324096e-02 9.11856890e-02
1.05287254e+00 -9.77834165e-01 -3.63346636e-01 -1.17269687e-01
1.27443779e+00 -1.72656178e-01 1.18118501e+00 -8.00940171e-02
-8.04790437e-01 -2.88110465e-01 -9.43514287e-01 -6.87358901e-02
-5.04041195e-01 -2.27456763e-02 1.32650876e+00 8.69334340e-01
-1.27958286e+00 8.04161012e-01 -7.92127252e-01 -3.63593966e-01
4.62953895e-01 6.01568818e-01 -8.99899960e-01 -1.90714434e-01
-1.08029807e+00 5.84675610e-01 4.93986398e-01 -2.30021417e-01
-5.66588938e-01 -9.50043023e-01 -1.33284676e+00 2.53510535e-01
5.42051613e-01 -5.99716425e-01 8.06808949e-01 -1.03224325e+00
-1.06298256e+00 8.15511465e-01 1.91583619e-01 -6.91274285e-01
3.06166075e-02 3.26992005e-01 -2.80661911e-01 3.49742085e-01
-7.34406412e-02 3.89321148e-01 7.54158437e-01 -9.01873648e-01
-3.59588981e-01 -2.89552569e-01 5.69692016e-01 -6.24165162e-02
-7.87918448e-01 -2.93171108e-01 -4.76598740e-01 -6.22935891e-01
8.48316997e-02 -1.08455527e+00 -4.26030278e-01 -1.42703513e-02
-3.63324612e-01 -1.93558559e-01 7.62177348e-01 -6.39418542e-01
1.22169793e+00 -1.94146109e+00 8.25621337e-02 5.30888438e-01
9.40865397e-01 2.31073067e-01 -4.82486486e-01 6.41082942e-01
-5.97235858e-01 2.69808024e-01 -8.74402523e-02 1.81152344e-01
-2.21468866e-01 1.25340104e-01 -2.52775729e-01 7.09567904e-01
9.13456529e-02 1.24273098e+00 -8.77117336e-01 -2.11331397e-01
1.50348410e-01 3.96782160e-01 -7.86412895e-01 2.43228320e-02
1.76761568e-01 -2.09241405e-01 -3.08668852e-01 4.50695634e-01
8.44444931e-01 -6.86587811e-01 5.16343236e-01 -4.15157080e-01
5.47897875e-01 -1.47736371e-02 -1.09124291e+00 1.27633083e+00
-5.04235387e-01 5.42042017e-01 1.48524314e-01 -1.45710278e+00
7.43636012e-01 5.85862575e-03 4.15048808e-01 -6.26972854e-01
-2.10687649e-02 -1.86169580e-01 1.65525019e-01 -2.27987364e-01
1.81375474e-01 -5.40495887e-02 5.13697788e-02 4.86717135e-01
5.00680842e-02 3.09217423e-01 1.99645206e-01 6.90004170e-01
1.45257163e+00 -2.84240246e-01 4.54445451e-01 -3.11472595e-01
5.03338695e-01 -4.30965990e-01 3.75147164e-01 6.02353394e-01
5.56457862e-02 2.12791950e-01 9.86164927e-01 -6.68621719e-01
-9.66279626e-01 -1.08274150e+00 4.55759257e-01 1.30756760e+00
2.07627527e-02 -7.98322797e-01 -6.85928047e-01 -9.40662920e-01
1.96624875e-01 2.47154415e-01 -8.37171376e-01 -7.44243979e-01
-6.81080878e-01 -6.25772238e-01 5.80259144e-01 7.57834911e-01
4.39745545e-01 -7.16437638e-01 1.39938086e-01 8.72093365e-02
1.73460037e-01 -1.06253660e+00 -1.00535142e+00 1.96874082e-01
-9.42114830e-01 -1.48162961e+00 -4.85459298e-01 -1.23295236e+00
1.17574453e+00 6.76431775e-01 1.02396500e+00 5.46447575e-01
-2.16497630e-01 5.29396534e-01 -1.89587981e-01 1.50019705e-01
-5.32314181e-01 3.38429332e-01 -7.20261857e-02 -6.82764500e-03
-4.08977419e-02 -9.92821455e-01 -1.36401117e-01 1.88218266e-01
-9.92103100e-01 1.41296327e-01 5.32474637e-01 8.83891046e-01
2.15604767e-01 1.90519154e-01 3.49336028e-01 -1.47592199e+00
7.15144515e-01 -3.86186242e-01 -7.87900984e-01 3.52299094e-01
-8.09174597e-01 5.68612874e-01 1.09375942e+00 -5.40281177e-01
-4.98260081e-01 1.51718065e-01 1.90624952e-01 -7.19485521e-01
2.73731232e-01 7.31129766e-01 -3.09825242e-01 -6.19418323e-01
6.80135489e-01 2.04261184e-01 1.92963421e-01 -3.86098176e-01
7.17763186e-01 -5.22558130e-02 3.69177759e-01 -6.09301388e-01
1.19249392e+00 4.54801828e-01 3.90920639e-01 -7.71830320e-01
-5.33601284e-01 -3.46846759e-01 -6.30161047e-01 2.28435457e-01
3.48679602e-01 -8.10397983e-01 -8.16561401e-01 2.25838527e-01
-8.22958469e-01 -3.28258306e-01 -1.02986470e-01 2.55272925e-01
-3.44618708e-01 8.94367337e-01 -9.56161797e-01 -2.67419815e-01
-3.09576124e-01 -7.98458278e-01 4.80765343e-01 -1.00975029e-01
2.92369366e-01 -1.43054986e+00 7.87224323e-02 -2.26851657e-01
1.82082310e-01 2.02310205e-01 1.41726434e+00 -7.16575980e-01
-6.07979715e-01 -4.63737190e-01 -5.34971237e-01 3.13343138e-01
3.38904291e-01 -3.41020763e-01 -5.61478496e-01 -7.24561334e-01
-4.03073072e-01 -1.35098755e-01 1.01886415e+00 2.19566345e-01
1.15439939e+00 -7.03358114e-01 -4.76108223e-01 1.11617506e+00
1.46469414e+00 -3.41093421e-01 4.94293869e-01 -6.18415289e-02
1.40445399e+00 5.92164874e-01 -1.85437933e-01 5.35876080e-02
3.84334505e-01 2.03908235e-01 2.87254810e-01 -3.77454817e-01
-2.45364040e-01 -5.99920273e-01 4.22605991e-01 9.98536170e-01
-1.00006208e-01 -1.45986184e-01 -7.53619611e-01 3.70670676e-01
-1.64315844e+00 -7.61533141e-01 -1.69402823e-01 2.21285772e+00
5.49102962e-01 -1.65655464e-02 2.61374801e-01 -1.18293636e-01
1.10826480e+00 3.36426079e-01 -3.50343704e-01 -2.71442294e-01
-7.05296174e-02 4.45772707e-01 9.57784116e-01 6.43953800e-01
-1.00457382e+00 1.17839992e+00 6.56393242e+00 7.46250451e-01
-9.17640209e-01 -2.72777285e-02 4.17216003e-01 5.40538967e-01
-5.58780611e-01 4.01116371e-01 -2.62388259e-01 3.00598331e-02
6.90583348e-01 -4.25098270e-01 7.45110989e-01 9.83566642e-01
-4.24815893e-01 5.58587551e-01 -1.21714389e+00 9.81116891e-01
-4.67206985e-02 -1.48971546e+00 3.23655695e-01 2.82757431e-01
6.28646016e-01 -1.95457384e-01 -1.60577402e-01 4.92652923e-01
6.86439872e-01 -1.12904549e+00 8.85440409e-02 4.40980159e-02
7.33348429e-01 -9.23219621e-01 4.72513348e-01 8.53614435e-02
-1.76158273e+00 4.71777245e-02 -6.07661784e-01 -1.56159312e-01
-4.04912412e-01 4.61982459e-01 -1.00682282e+00 8.44881952e-01
2.41022572e-01 8.30699325e-01 -7.81293213e-01 7.47560740e-01
-4.35479552e-01 6.56764925e-01 -2.23596409e-01 5.63833788e-02
2.91159064e-01 -5.20277977e-01 4.33182389e-01 1.15855849e+00
5.96859027e-03 -1.05522618e-01 4.50404912e-01 8.20149541e-01
-4.76037234e-01 3.29241723e-01 -1.15679061e+00 -2.95709878e-01
2.52506435e-01 1.49621749e+00 -8.81044984e-01 -3.25853795e-01
-7.81526208e-01 9.64972079e-01 9.91866708e-01 4.65726584e-01
-7.19007671e-01 -6.18214309e-01 5.87017477e-01 2.76099026e-01
4.02286112e-01 -2.72913426e-01 1.96936890e-01 -1.26817691e+00
6.10377043e-02 -8.29223454e-01 7.79149175e-01 -2.72825569e-01
-1.36489916e+00 3.60978812e-01 -8.16748198e-03 -8.50904644e-01
1.74594745e-01 -6.85969651e-01 -8.11435580e-01 5.40868223e-01
-1.36405003e+00 -1.31755352e+00 -2.06221417e-01 7.21692264e-01
-2.44538009e-01 -9.74931642e-02 7.91134596e-01 2.40107194e-01
-3.93341064e-01 9.97124076e-01 -1.57934785e-01 6.23559475e-01
4.04955506e-01 -1.45925426e+00 7.73517489e-01 8.86477649e-01
3.96756321e-01 8.00705492e-01 3.10760587e-01 -7.60362983e-01
-1.91359234e+00 -1.11825156e+00 4.03176278e-01 -1.91580474e-01
9.58187640e-01 -5.09341180e-01 -1.12724185e+00 1.16544628e+00
2.98671145e-02 4.45719689e-01 3.55732501e-01 4.29090768e-01
-9.84886229e-01 -1.48572460e-01 -9.61240351e-01 8.40117395e-01
1.31194186e+00 -8.17354918e-01 -2.89676815e-01 4.43288505e-01
9.53531802e-01 -2.09388539e-01 -8.43906462e-01 3.98793727e-01
3.01176310e-01 -6.66408181e-01 1.12664700e+00 -1.05155969e+00
7.82232806e-02 -1.43776044e-01 9.44044217e-02 -1.36693466e+00
-7.03854799e-01 -8.51125658e-01 -1.55071467e-01 1.01388979e+00
3.51555258e-01 -9.55476403e-01 1.12803209e+00 5.34967244e-01
9.16107371e-03 -5.86936653e-01 -5.18216729e-01 -9.96230364e-01
1.56945363e-01 -9.02536884e-02 5.42195797e-01 1.24242532e+00
5.24177104e-02 4.91533607e-01 -5.79320669e-01 3.65781963e-01
6.60083771e-01 2.06019297e-01 9.97188210e-01 -1.22522008e+00
-4.18403357e-01 -4.95173693e-01 -8.55958402e-01 -8.76691878e-01
5.43580830e-01 -1.74258065e+00 -5.32530010e-01 -1.42682135e+00
3.29735279e-01 -2.64362812e-01 -4.04222548e-01 7.95109689e-01
-1.96770743e-01 3.90687771e-02 1.04838610e-01 -1.49587646e-01
-3.24756712e-01 3.82668734e-01 1.22486269e+00 -3.15428108e-01
-2.08123416e-01 -8.94351900e-02 -9.67800021e-01 6.09778762e-01
7.69259930e-01 -4.49937373e-01 -6.25609934e-01 -9.44249704e-02
2.75545388e-01 -3.55632119e-02 4.14009511e-01 -6.91066444e-01
1.50016472e-01 6.05273694e-02 4.19075228e-02 2.20606755e-02
5.69446944e-03 -7.66342700e-01 1.10898152e-01 7.01359153e-01
-3.05960923e-01 2.55819976e-01 9.26536918e-02 8.48037899e-01
1.25245318e-01 -4.21731882e-02 8.18393707e-01 -2.02366322e-01
-6.81562722e-01 7.90996253e-01 5.78395352e-02 2.35929951e-01
9.21964049e-01 -3.12541813e-01 -2.68044263e-01 -6.03233457e-01
-7.04051137e-01 1.69029161e-01 6.01635635e-01 2.38823369e-01
5.20730257e-01 -1.55524921e+00 -5.54806769e-01 3.87994140e-01
1.89435676e-01 -4.27008927e-01 4.75451872e-02 7.54325271e-01
-5.83281517e-01 9.41402689e-02 -1.79899052e-01 -2.25491852e-01
-1.45235515e+00 1.01963472e+00 2.24491298e-01 -4.35639679e-01
-1.01216292e+00 7.37640023e-01 6.97504818e-01 -7.80676186e-01
3.56666148e-01 -4.79267091e-01 1.12557523e-01 -1.32159710e-01
2.98586965e-01 3.04735959e-01 -7.52786547e-02 -3.18243563e-01
-2.04402134e-01 5.47382534e-01 -4.10951763e-01 4.25589174e-01
1.39807439e+00 1.74836487e-01 -2.76712149e-01 9.20997642e-04
1.67931759e+00 2.19441682e-01 -9.98879731e-01 -4.50708151e-01
1.03069693e-01 -4.95272666e-01 -7.46592507e-02 -5.70067577e-02
-1.51574898e+00 7.68360734e-01 1.89168364e-01 4.32891279e-01
9.70025361e-01 1.88282102e-01 7.18396664e-01 7.94390678e-01
3.13825727e-01 -7.01750755e-01 1.94250971e-01 3.38845253e-01
5.92950761e-01 -1.05570662e+00 1.90624252e-01 -9.90201831e-01
-3.74136686e-01 1.22325790e+00 5.95877528e-01 -5.49600065e-01
8.38867009e-01 1.35174528e-01 -4.27252203e-01 -3.88283014e-01
-4.90815103e-01 -2.56811619e-01 4.19374973e-01 8.22288215e-01
2.08008707e-01 2.60167301e-01 7.17351660e-02 3.15082967e-01
-1.02861084e-01 -4.32497680e-01 6.39133751e-01 6.48046017e-01
-2.83807248e-01 -1.35156524e+00 8.00245106e-02 6.80843055e-01
-2.38369510e-01 -4.31177676e-01 -4.83002603e-01 1.12384880e+00
-4.62432683e-01 3.79004240e-01 -1.36504307e-01 -5.86627185e-01
5.21763824e-02 -2.16686696e-01 7.37035215e-01 -7.79125988e-01
-3.13481897e-01 -3.47552776e-01 5.78039251e-02 -5.94785094e-01
-1.62224367e-01 -1.43823892e-01 -1.35793996e+00 -7.01607645e-01
-4.82280046e-01 1.25368938e-01 1.18253641e-01 4.22722936e-01
4.91928875e-01 3.19847673e-01 6.92415178e-01 -5.14325321e-01
-5.26098788e-01 -5.90202153e-01 -9.09529626e-01 5.55381238e-01
3.70446026e-01 -4.11622554e-01 -6.48493171e-01 -8.29536244e-02] | [6.9976277351379395, 6.224408149719238] |
2731f493-5cec-4d2c-bc18-273b18f4bfee | scene-labeling-using-beam-search-under-mutex | null | null | http://openaccess.thecvf.com/content_cvpr_2014/html/Roy_Scene_Labeling_Using_2014_CVPR_paper.html | http://openaccess.thecvf.com/content_cvpr_2014/papers/Roy_Scene_Labeling_Using_2014_CVPR_paper.pdf | Scene Labeling Using Beam Search Under Mutex Constraints | This paper addresses the problem of assigning object class labels to image pixels. Following recent holistic formulations, we cast scene labeling as inference of a conditional random field (CRF) grounded onto superpixels. The CRF inference is specified as quadratic program (QP) with mutual exclusion (mutex) constraints on class label assignments. The QP is solved using a beam search (BS), which is well-suited for scene labeling, because it explicitly accounts for spatial extents of objects; conforms to inconsistency constraints from domain knowledge; and has low computational costs. BS gradually builds a search tree whose nodes correspond to candidate scene labelings. Successor nodes are repeatedly generated from a select set of their parent nodes until convergence. We prove that our BS efficiently maximizes the QP objective of CRF inference. Effectiveness of our BS for scene labeling is evaluated on the benchmark MSRC, Stanford Backgroud, PASCAL VOC 2009 and 2010 datasets. | ['Anirban Roy', 'Sinisa Todorovic'] | 2014-06-01 | null | null | null | cvpr-2014-6 | ['scene-labeling'] | ['computer-vision'] | [ 8.04953814e-01 2.97425926e-01 -4.73113686e-01 -7.84056783e-01
-7.53365517e-01 -5.00265241e-01 3.56297344e-01 4.81307432e-02
-3.59566450e-01 9.42017019e-01 -1.33525729e-01 -1.42561853e-01
-9.49721038e-02 -9.00708318e-01 -9.65219855e-01 -7.77112484e-01
2.69145608e-01 6.59018278e-01 3.46462697e-01 4.16673541e-01
3.13367337e-01 4.87037897e-01 -1.46145117e+00 4.07783836e-02
1.08592463e+00 9.79842782e-01 8.12173963e-01 4.70435232e-01
-2.89389584e-02 1.39480388e+00 -1.87931776e-01 -3.01722556e-01
9.74509567e-02 -1.55840352e-01 -1.38582122e+00 5.72147965e-01
8.36225092e-01 -5.41927852e-02 1.02391131e-01 1.42558467e+00
1.53119527e-02 3.05705130e-01 7.18127429e-01 -1.20802307e+00
-4.85459208e-01 3.29101443e-01 -7.59423912e-01 -5.07183187e-03
7.28047341e-02 -2.22087979e-01 1.13503933e+00 -7.80969799e-01
7.57544756e-01 1.13644016e+00 5.06759882e-01 3.27293813e-01
-1.53607476e+00 -3.34548324e-01 3.27608168e-01 1.25232160e-01
-1.54644120e+00 -1.21818297e-01 4.67697322e-01 -6.44188344e-01
6.54709995e-01 3.30997258e-01 4.26245928e-01 4.34761167e-01
-5.78742251e-02 8.55745733e-01 1.15459359e+00 -6.23589694e-01
5.26898265e-01 1.90520957e-01 3.39493573e-01 8.13490033e-01
-6.48038909e-02 -1.22705027e-01 -4.28857177e-01 -3.52698825e-02
7.83221424e-01 -2.96979666e-01 -1.14206880e-01 -3.84016305e-01
-1.07614279e+00 8.63559365e-01 6.06851339e-01 -1.58631980e-01
-3.55980784e-01 3.51643533e-01 -1.19869187e-01 -4.22957063e-01
4.84710604e-01 2.84482151e-01 -4.09568667e-01 4.99742776e-01
-8.91073287e-01 1.34269401e-01 5.59172392e-01 1.33342934e+00
1.25356376e+00 -3.72334123e-01 -3.32820356e-01 1.00986075e+00
5.25894105e-01 6.73570633e-01 -3.07018727e-01 -1.49207675e+00
2.80660182e-01 4.18699771e-01 1.42084002e-01 -8.74597311e-01
-2.11755127e-01 -4.39946026e-01 -7.37397254e-01 3.02226916e-02
1.86788172e-01 -9.93359759e-02 -1.16420090e+00 1.79944110e+00
7.33570755e-01 3.86214703e-01 -1.23625197e-01 8.50372732e-01
7.94935703e-01 7.62087584e-01 5.20440876e-01 -3.06292921e-01
1.24004495e+00 -1.00936973e+00 -6.26670361e-01 -3.84281129e-01
3.29168528e-01 -6.56182051e-01 6.87759459e-01 3.38523090e-01
-9.04704332e-01 -6.22538865e-01 -5.89383245e-01 -2.29699209e-01
-2.00172588e-01 3.04247171e-01 8.51080000e-01 5.94349205e-01
-1.12292933e+00 2.35453308e-01 -7.37641871e-01 -1.90326169e-01
6.05162084e-01 5.21733165e-01 -1.11079633e-01 -1.19544901e-01
-7.78867483e-01 8.02055657e-01 6.93498790e-01 2.82862484e-01
-1.11404645e+00 -5.89520156e-01 -9.97659743e-01 -2.07925007e-01
5.62030911e-01 -6.65467799e-01 1.15924501e+00 -9.94120181e-01
-1.34915125e+00 1.26932204e+00 -5.07538021e-01 -3.42363209e-01
2.14930147e-01 -1.25590935e-01 -5.11021167e-03 1.05819456e-01
5.76067567e-01 1.26175106e+00 6.40802503e-01 -1.61785662e+00
-1.04729366e+00 -2.06113935e-01 2.23239511e-01 3.56279761e-01
2.12656409e-01 1.98193714e-02 -7.16178715e-01 -2.37404525e-01
4.51384604e-01 -1.09425282e+00 -6.19370282e-01 -2.35424832e-01
-1.09482062e+00 -2.32692897e-01 5.23993015e-01 -3.81820858e-01
9.94394720e-01 -1.88139808e+00 1.35230437e-01 2.98836797e-01
-8.42093080e-02 -2.69664735e-01 6.81734011e-02 -1.25249684e-01
5.96040897e-02 -5.74013256e-02 -6.29698634e-01 -4.34179604e-01
-6.04347624e-02 4.57476467e-01 -3.57971162e-01 6.06977582e-01
1.03287451e-01 8.59527409e-01 -9.19726372e-01 -1.06926882e+00
5.03207207e-01 1.17064215e-01 -6.45964861e-01 1.86529040e-01
-7.36027539e-01 6.23504460e-01 -5.17446995e-01 7.25828767e-01
1.00947881e+00 -6.95790052e-01 2.28636250e-01 -4.49513733e-01
-2.26465002e-01 3.88124362e-02 -1.30119395e+00 1.57572746e+00
-2.48737127e-01 4.49413061e-01 -2.65635364e-02 -7.99948037e-01
8.07074964e-01 -1.60168752e-01 4.89153802e-01 -4.14066553e-01
-1.05056472e-01 -2.18835920e-01 -6.47174895e-01 -4.18354124e-01
6.07164085e-01 -9.08222497e-02 1.33777494e-02 -7.26830307e-03
7.46522769e-02 -5.06368220e-01 3.12040180e-01 3.19022328e-01
6.57214165e-01 2.89025396e-01 3.12866986e-01 -5.49189389e-01
5.11001468e-01 3.60697001e-01 7.69824862e-01 9.25067902e-01
-6.10840991e-02 4.60442334e-01 3.68733346e-01 -4.56368327e-02
-6.84154510e-01 -1.27391315e+00 -6.14156127e-01 1.13789511e+00
5.53164601e-01 -6.90517351e-02 -8.48409891e-01 -6.43193603e-01
-1.58224180e-01 8.62428725e-01 -5.50600827e-01 3.47357750e-01
-4.47012693e-01 -7.44372725e-01 8.72218907e-02 3.43078643e-01
5.83270133e-01 -1.23735845e+00 -6.83478713e-01 6.02754653e-02
-5.00308871e-01 -1.41215980e+00 -3.34855169e-01 4.27321315e-01
-7.12957740e-01 -1.02135706e+00 -4.63089854e-01 -1.19078267e+00
9.85156000e-01 3.44473347e-02 1.28916824e+00 -3.18273216e-01
-3.02497715e-01 3.09583277e-01 -9.40928459e-02 -1.09831482e-01
-7.17985556e-02 -7.02246502e-02 -4.11730707e-01 1.31916881e-01
1.07590936e-01 -1.00856826e-01 -4.37109292e-01 3.90940189e-01
-6.84171259e-01 5.06258905e-01 2.68361241e-01 6.72051013e-01
1.41481972e+00 2.86582440e-01 1.13609046e-01 -1.24446428e+00
-2.01638326e-01 -4.32613641e-01 -1.20276082e+00 5.49096286e-01
-2.59895444e-01 -1.39053151e-01 9.81775820e-02 4.15405855e-02
-1.61846709e+00 7.93721616e-01 2.49041706e-01 -2.85241336e-01
-5.05325198e-01 3.22347105e-01 -2.70892888e-01 -8.00164565e-02
3.57229352e-01 9.46734026e-02 -7.86013901e-01 -2.86176085e-01
6.76952958e-01 4.98262405e-01 9.24807191e-01 -8.62085462e-01
3.10459584e-01 7.23907113e-01 2.34191269e-01 -8.33418727e-01
-1.42493868e+00 -7.55146742e-01 -8.87993038e-01 -3.75728965e-01
1.35796106e+00 -9.25781846e-01 -7.34214902e-01 2.62231618e-01
-1.23308885e+00 -5.95800996e-01 -2.17852369e-01 2.90799081e-01
-7.36326814e-01 2.63701200e-01 -4.40874487e-01 -8.86242986e-01
9.68993604e-02 -1.05445385e+00 1.36122918e+00 4.57800120e-01
4.39802893e-02 -1.02208316e+00 8.89299139e-02 6.09138191e-01
-2.81358421e-01 5.06897390e-01 8.39841247e-01 1.69414744e-01
-1.02998877e+00 2.38527313e-01 -5.26244938e-01 2.88300186e-01
3.91694345e-02 6.54735491e-02 -1.16737509e+00 1.14948247e-02
-2.95832038e-01 -3.05620193e-01 9.12369430e-01 8.60795498e-01
1.48010051e+00 -2.65312731e-01 -6.21421874e-01 7.16025591e-01
1.92622948e+00 8.48665237e-02 5.73142409e-01 1.16751201e-01
8.69553685e-01 7.38567472e-01 9.51548219e-01 4.24447745e-01
4.42459434e-01 5.75643659e-01 7.16556847e-01 -2.10386351e-01
-6.01060875e-02 -2.61303902e-01 -9.09730047e-02 2.76810288e-01
1.74801588e-01 -4.39952880e-01 -9.71985459e-01 6.27503693e-01
-1.88539612e+00 -6.72442794e-01 -6.17566645e-01 2.01473498e+00
8.76743317e-01 -1.27795547e-01 -2.95965791e-01 -2.88705736e-01
1.06134880e+00 -5.21695018e-02 -5.66712081e-01 -1.89210789e-03
-2.28647098e-01 2.08612636e-01 8.36220503e-01 7.14116693e-01
-1.52409673e+00 1.32324779e+00 6.50273037e+00 8.54398668e-01
-6.62165403e-01 1.61926150e-01 9.58119154e-01 2.94143140e-01
-7.55752176e-02 3.31726432e-01 -1.05457866e+00 2.42148772e-01
3.58856767e-01 3.14900368e-01 3.01457018e-01 8.24007273e-01
1.02537267e-01 -7.25043714e-01 -1.08989143e+00 7.88452804e-01
-2.57534534e-01 -1.36229038e+00 -3.47788721e-01 -4.66657281e-02
1.38124442e+00 2.61034131e-01 4.35370207e-02 -1.35657400e-01
9.15484309e-01 -1.04093647e+00 9.22610879e-01 4.49445128e-01
8.87695312e-01 -5.25079787e-01 4.14057702e-01 3.80172491e-01
-1.19308388e+00 1.68879867e-01 -4.71981496e-01 2.87904859e-01
4.32478338e-01 8.98861051e-01 -8.65630448e-01 2.78438658e-01
7.22829282e-01 6.65641904e-01 -4.28587615e-01 9.54524517e-01
-4.70197856e-01 6.04015708e-01 -3.43162954e-01 1.42606989e-01
1.60490885e-01 -5.02080798e-01 1.79346114e-01 1.41440368e+00
-3.11215341e-01 2.65630245e-01 5.53714216e-01 1.20698726e+00
4.95342351e-02 -2.55046034e-04 -2.77343810e-01 4.08901662e-01
5.71089804e-01 1.33846772e+00 -1.13614964e+00 -4.29404348e-01
-6.36259317e-02 9.44455862e-01 5.85843444e-01 5.15253603e-01
-9.10284936e-01 7.28556365e-02 1.91728204e-01 -2.19703689e-01
4.15085047e-01 6.61629438e-02 -6.17451966e-01 -7.39213347e-01
-3.08365703e-01 -3.98141652e-01 4.56538022e-01 -1.03031158e+00
-1.24351478e+00 3.47310901e-01 1.84622899e-01 -7.97927797e-01
1.42652854e-01 -3.04062188e-01 -2.89712906e-01 7.64527917e-01
-1.56480610e+00 -1.18992221e+00 -3.96150917e-01 6.34561300e-01
4.32144493e-01 6.32672906e-01 6.43918276e-01 1.52506024e-01
-6.01904273e-01 1.15892433e-01 -3.42706330e-02 8.40438977e-02
1.89491704e-01 -1.58698535e+00 -1.40919805e-01 8.84952009e-01
1.04833253e-01 4.03503001e-01 6.15433574e-01 -8.76210988e-01
-8.93528283e-01 -1.48567426e+00 9.89648104e-01 -3.21019590e-01
3.05858582e-01 -2.07562983e-01 -5.86355507e-01 8.46827984e-01
6.80081099e-02 1.73674673e-01 3.97774786e-01 -1.53739322e-02
-7.17375800e-02 4.92152497e-02 -1.17259789e+00 2.81396329e-01
9.68557060e-01 -3.95480007e-01 -1.09767452e-01 9.42922890e-01
5.72721422e-01 -5.88215768e-01 -8.98045897e-01 5.58531225e-01
1.06548414e-01 -7.60870218e-01 1.05900431e+00 -3.09256345e-01
5.59217751e-01 -6.33591235e-01 -4.76772577e-01 -7.04558849e-01
-3.58358890e-01 -1.91430867e-01 2.97568858e-01 1.32776880e+00
4.34153885e-01 -2.40747765e-01 9.09089804e-01 7.16643870e-01
-8.30964968e-02 -5.67749918e-01 -8.46897244e-01 -5.56352198e-01
-2.67286777e-01 -4.15312976e-01 3.32765907e-01 9.92764473e-01
-3.91280293e-01 3.72384489e-01 -3.72526258e-01 5.59797823e-01
1.15419412e+00 4.48308825e-01 5.01608729e-01 -1.16067493e+00
-3.40381712e-01 -5.95694669e-02 -1.57576669e-02 -1.26657796e+00
5.29349446e-01 -9.40932751e-01 6.87295198e-01 -1.68327999e+00
6.06963754e-01 -9.27955985e-01 7.70546682e-03 6.91604018e-01
-1.26000151e-01 3.73135388e-01 1.60985887e-01 1.81855522e-02
-1.03257799e+00 3.82872581e-01 1.42555988e+00 -1.99052513e-01
-8.45752582e-02 1.45443529e-01 -3.17632943e-01 1.00104117e+00
6.01804435e-01 -6.52079701e-01 -4.22145545e-01 -3.33603472e-01
1.89886600e-01 3.06180805e-01 5.14920771e-01 -6.03952527e-01
4.42178845e-01 -6.30805135e-01 1.39560789e-01 -9.53916788e-01
3.63931686e-01 -6.47727370e-01 3.89705479e-01 1.65274218e-01
-5.83929121e-01 -6.06114924e-01 -1.27920331e-02 6.07280433e-01
-1.52844802e-01 -4.95827675e-01 1.12391210e+00 -2.12838113e-01
-1.06853116e+00 3.48038614e-01 -2.19698235e-01 7.74316564e-02
1.09894574e+00 -3.09534997e-01 1.60213057e-02 1.68233607e-02
-8.63691688e-01 4.88381058e-01 2.95519143e-01 -2.06548832e-02
3.60736936e-01 -1.08726454e+00 -5.25227010e-01 -1.83733806e-01
2.81371940e-02 7.21914649e-01 1.80406064e-01 5.52884340e-01
-5.33090889e-01 4.92312878e-01 2.01908007e-01 -1.04355943e+00
-1.26019061e+00 2.98566699e-01 3.59938741e-01 -1.66569784e-01
-3.13792706e-01 1.32682037e+00 6.66574538e-01 -5.68552673e-01
2.43086055e-01 -2.81988680e-01 -2.89948404e-01 -2.33695164e-01
-4.19449434e-03 5.15794992e-01 -2.30250448e-01 -8.72349083e-01
-4.90263611e-01 7.73847938e-01 1.65817887e-01 -1.27395153e-01
1.06999445e+00 -3.71759117e-01 -5.23510098e-01 2.20146090e-01
1.05413854e+00 -2.33035833e-01 -1.44486713e+00 -3.97120386e-01
9.75599885e-02 -5.14621258e-01 2.56939948e-01 -7.84824491e-01
-1.06177628e+00 4.59457397e-01 4.68586117e-01 -1.08090460e-01
1.15868020e+00 4.01102841e-01 2.49369398e-01 1.96422741e-01
4.97303784e-01 -1.14449430e+00 -1.49983317e-01 5.40213287e-01
4.64801282e-01 -1.27071500e+00 1.57564640e-01 -1.11183810e+00
-6.74065173e-01 7.56149650e-01 7.30635881e-01 -5.06911501e-02
6.71752810e-01 2.72757113e-01 -2.90553868e-01 -2.65719533e-01
-4.13977504e-01 -4.44403440e-01 3.27517211e-01 5.52018106e-01
1.21328384e-01 4.96765465e-01 2.51193065e-02 -8.41067452e-03
-1.05289318e-01 -9.86542739e-03 2.16782331e-01 6.78323030e-01
-6.79185331e-01 -7.98613369e-01 -4.15761262e-01 3.74527335e-01
-1.89457327e-01 -2.41308510e-01 -3.67464833e-02 4.16609615e-01
7.40617752e-01 1.15594196e+00 1.62445337e-01 9.37256068e-02
-1.41704500e-01 -4.51490253e-01 6.25408947e-01 -8.75502467e-01
-1.20507367e-01 1.30528912e-01 -7.49920234e-02 -4.45183784e-01
-8.11014891e-01 -8.51810992e-01 -1.48015189e+00 9.74894017e-02
-8.35588634e-01 -9.75066051e-03 6.77708745e-01 1.04407907e+00
-1.09343687e-02 3.62372786e-01 5.37624300e-01 -5.63105166e-01
5.69255045e-03 -7.01576531e-01 -7.89302945e-01 1.31694928e-01
1.15435235e-01 -5.91743708e-01 1.57571547e-02 6.45096540e-01] | [9.531890869140625, 0.482128769159317] |
584fd857-49ab-4378-9757-7baf289295e4 | primitive-generation-and-semantic-related-1 | 2306.11087 | null | https://arxiv.org/abs/2306.11087v1 | https://arxiv.org/pdf/2306.11087v1.pdf | Primitive Generation and Semantic-related Alignment for Universal Zero-Shot Segmentation | We study universal zero-shot segmentation in this work to achieve panoptic, instance, and semantic segmentation for novel categories without any training samples. Such zero-shot segmentation ability relies on inter-class relationships in semantic space to transfer the visual knowledge learned from seen categories to unseen ones. Thus, it is desired to well bridge semantic-visual spaces and apply the semantic relationships to visual feature learning. We introduce a generative model to synthesize features for unseen categories, which links semantic and visual spaces as well as addresses the issue of lack of unseen training data. Furthermore, to mitigate the domain gap between semantic and visual spaces, firstly, we enhance the vanilla generator with learned primitives, each of which contains fine-grained attributes related to categories, and synthesize unseen features by selectively assembling these primitives. Secondly, we propose to disentangle the visual feature into the semantic-related part and the semantic-unrelated part that contains useful visual classification clues but is less relevant to semantic representation. The inter-class relationships of semantic-related visual features are then required to be aligned with those in semantic space, thereby transferring semantic knowledge to visual feature learning. The proposed approach achieves impressively state-of-the-art performance on zero-shot panoptic segmentation, instance segmentation, and semantic segmentation. Code is available at https://henghuiding.github.io/PADing/. | ['Wei Jiang', 'Henghui Ding', 'Shuting He'] | 2023-06-19 | primitive-generation-and-semantic-related | http://openaccess.thecvf.com//content/CVPR2023/html/He_Primitive_Generation_and_Semantic-Related_Alignment_for_Universal_Zero-Shot_Segmentation_CVPR_2023_paper.html | http://openaccess.thecvf.com//content/CVPR2023/papers/He_Primitive_Generation_and_Semantic-Related_Alignment_for_Universal_Zero-Shot_Segmentation_CVPR_2023_paper.pdf | cvpr-2023-1 | ['panoptic-segmentation', 'zero-shot-segmentation'] | ['computer-vision', 'computer-vision'] | [ 4.28381979e-01 1.92150339e-01 -2.77853638e-01 -4.53794688e-01
-5.00185847e-01 -7.68233240e-01 5.82568228e-01 7.01945499e-02
-2.49203920e-01 5.17451644e-01 1.18808128e-01 1.65668763e-02
-9.56593305e-02 -1.15346301e+00 -7.18355894e-01 -7.62733340e-01
2.66061068e-01 3.30121547e-01 4.93133128e-01 -2.92376459e-01
1.00869916e-01 9.30822864e-02 -1.95773196e+00 2.76651114e-01
9.47455883e-01 1.00389099e+00 5.00731349e-01 4.85735655e-01
-5.19428730e-01 1.78222016e-01 -4.53493983e-01 -9.92983952e-02
2.41922483e-01 -6.61589265e-01 -8.45114350e-01 4.73307282e-01
3.36697578e-01 3.77139747e-02 -1.44366637e-01 1.22039950e+00
-2.60123443e-02 5.77242792e-01 9.22193468e-01 -1.46691489e+00
-1.08282959e+00 1.95977420e-01 -4.76315081e-01 2.15295285e-01
1.04119256e-01 2.69771874e-01 1.24346328e+00 -8.34239244e-01
7.70589828e-01 1.22877431e+00 4.71639521e-02 8.93911719e-01
-1.15017068e+00 -5.88676214e-01 2.39037499e-01 2.35077113e-01
-1.19376576e+00 -1.99835345e-01 8.44986022e-01 -6.87993765e-01
5.00997543e-01 2.85537153e-01 8.70188177e-01 1.16320443e+00
-2.55356669e-01 9.33264494e-01 9.16090786e-01 -3.20781887e-01
3.04882288e-01 2.71413207e-01 3.03414911e-01 5.13693392e-01
1.55323148e-01 1.46046340e-01 -3.20769608e-01 3.52727830e-01
8.16244066e-01 4.00863469e-01 -2.65625983e-01 -6.90846443e-01
-1.08626413e+00 1.04421520e+00 7.94834614e-01 5.14690280e-01
-1.45974904e-01 -1.40471477e-02 2.36024722e-01 6.48733005e-02
3.72648686e-01 5.25619149e-01 -4.55855638e-01 2.33811170e-01
-8.06291401e-01 -8.84522349e-02 4.14512813e-01 1.23465598e+00
1.30789220e+00 7.71391988e-02 -4.11974758e-01 1.00606775e+00
2.50258714e-01 3.93974006e-01 6.95088029e-01 -7.18296409e-01
3.03091437e-01 7.89173841e-01 -3.05767328e-01 -8.85050356e-01
-1.42593235e-01 -2.36128122e-01 -7.64856160e-01 -5.27165197e-02
2.73394823e-01 -3.27719115e-02 -1.72228730e+00 1.87859976e+00
4.67567086e-01 3.35276335e-01 2.63736933e-01 1.00695121e+00
1.31807053e+00 7.32756376e-01 2.30642274e-01 -5.91383036e-03
1.48589635e+00 -9.45957959e-01 -3.35135520e-01 -5.53775907e-01
3.23206753e-01 -5.09872496e-01 1.36955154e+00 -2.84985870e-01
-6.30300820e-01 -7.94583499e-01 -1.05199790e+00 -2.45897129e-01
-8.35137486e-01 -2.68062770e-01 6.76124036e-01 3.54513705e-01
-7.68196344e-01 3.84217024e-01 -5.27036011e-01 -6.67611301e-01
6.83966935e-01 8.14555287e-02 -2.71056622e-01 -1.62434101e-01
-1.33850336e+00 4.12323087e-01 9.15752649e-01 -4.24223989e-01
-1.01209736e+00 -6.23633742e-01 -1.28935134e+00 1.31659135e-01
6.96287811e-01 -6.92121148e-01 8.92158031e-01 -1.06573641e+00
-9.61060882e-01 1.06216276e+00 -5.79723641e-02 -6.97062761e-02
4.55058366e-02 2.43719429e-01 -3.95957142e-01 3.02675307e-01
5.87456703e-01 1.16009820e+00 1.09422302e+00 -1.49886906e+00
-8.95966709e-01 -4.21022236e-01 7.74201239e-03 3.65567625e-01
-4.17720616e-01 -6.07251227e-01 -5.86385548e-01 -7.25915432e-01
4.44826126e-01 -7.18983769e-01 -1.85184836e-01 -1.04702264e-01
-5.33341408e-01 -3.00951481e-01 9.14588034e-01 -3.26340467e-01
6.64932728e-01 -2.29947209e+00 1.95026889e-01 1.71630681e-01
2.96477497e-01 1.25247523e-01 -2.62773901e-01 6.61910251e-02
7.09236413e-02 1.08223230e-01 -4.78742957e-01 1.63972661e-01
-7.20670968e-02 5.36552012e-01 -3.81101042e-01 3.98515351e-02
2.67353684e-01 1.21643114e+00 -1.22094846e+00 -5.49943268e-01
5.42080402e-01 1.70660049e-01 -4.25795466e-01 2.86520392e-01
-3.19492459e-01 6.69975877e-01 -7.56506026e-01 8.35483074e-01
5.39422154e-01 -2.99057931e-01 -1.11383341e-01 -1.23699874e-01
1.95771679e-01 -1.63793653e-01 -9.44094181e-01 1.83090711e+00
-3.11457455e-01 3.89745772e-01 -3.75943363e-01 -1.25814164e+00
1.00957286e+00 1.05956700e-02 2.49792531e-01 -7.21325696e-01
3.33406478e-01 -2.40754951e-02 -1.89411581e-01 -5.01627147e-01
4.55833077e-01 -4.41325068e-01 -4.35326785e-01 2.38841966e-01
6.40634358e-01 -4.83984262e-01 3.38711888e-01 2.80261934e-01
4.19425219e-01 9.79073271e-02 4.62451607e-01 -1.65928677e-01
3.96616876e-01 1.68647349e-01 6.42613113e-01 6.12513423e-01
-2.23072901e-01 7.76688099e-01 4.19260651e-01 -1.64236769e-01
-9.26631868e-01 -1.44809639e+00 -1.31155223e-01 1.38457263e+00
7.65756905e-01 -1.78176299e-01 -6.62360489e-01 -7.95533776e-01
8.01977608e-03 9.05446470e-01 -9.85618532e-01 -4.12866056e-01
2.50874227e-03 -5.16618073e-01 1.08388044e-01 6.92122698e-01
4.07380313e-01 -1.33709145e+00 -6.24552250e-01 -1.91478975e-04
1.05850864e-02 -8.60858321e-01 -3.68319541e-01 3.08187276e-01
-6.10267818e-01 -1.25006449e+00 -8.54317725e-01 -1.03614640e+00
7.18664467e-01 6.16612971e-01 1.03119242e+00 4.10815813e-02
-5.33687472e-01 4.26777929e-01 -5.67257047e-01 -8.63516033e-02
-1.92063898e-01 -1.14948757e-01 -3.37823540e-01 1.64075851e-01
5.52610755e-01 -6.13267720e-01 -6.24645650e-01 2.71302968e-01
-1.02372444e+00 3.11531693e-01 4.29832220e-01 9.37643409e-01
8.54566097e-01 -2.57527024e-01 5.61437368e-01 -9.47924376e-01
2.85817474e-01 -7.25342155e-01 -2.47333944e-01 2.95574516e-01
-2.34889001e-01 9.15918052e-02 4.98319656e-01 -4.50622886e-01
-9.12298501e-01 -6.38277456e-02 1.41249791e-01 -7.00989068e-01
-4.49267894e-01 7.43208975e-02 -4.26733524e-01 3.19302142e-01
5.55519283e-01 4.73130941e-01 -3.43769073e-01 -3.10973525e-01
1.01208472e+00 5.10363638e-01 5.83016276e-01 -6.79724336e-01
7.99523771e-01 5.52379489e-01 -2.48390302e-01 -9.70856011e-01
-1.04551911e+00 -7.79628098e-01 -8.93088758e-01 2.49998737e-02
1.33226478e+00 -7.10931420e-01 4.03779894e-02 1.65544242e-01
-7.51747549e-01 -1.34585664e-01 -8.72638762e-01 1.90892443e-01
-6.95170820e-01 2.43336573e-01 -3.22048962e-01 -4.15813088e-01
-2.50245363e-01 -1.08807397e+00 1.14442205e+00 6.76505625e-01
-6.97105974e-02 -9.53299761e-01 -8.67746696e-02 2.25301057e-01
-1.35663807e-01 3.56725156e-01 1.08868945e+00 -8.38789582e-01
-4.77549344e-01 1.83469411e-02 -6.39348269e-01 3.73017550e-01
2.79720634e-01 -1.48198500e-01 -1.05434346e+00 -2.50249803e-01
-1.78716108e-01 -4.85726178e-01 1.17110872e+00 3.73317897e-01
1.19110262e+00 -9.83486623e-02 -5.03454924e-01 8.49049687e-01
1.46394038e+00 2.84784913e-01 4.85852599e-01 2.77030580e-02
1.07497954e+00 6.87273502e-01 7.10795939e-01 1.78645149e-01
3.21768343e-01 3.11481178e-01 2.37999693e-01 -1.29565969e-01
-2.97338903e-01 -5.40370584e-01 -1.39160931e-01 6.11233890e-01
1.71217740e-01 -1.14226922e-01 -7.17593193e-01 7.75538743e-01
-1.75334704e+00 -7.31354415e-01 9.43003446e-02 2.00791812e+00
6.15168869e-01 2.56966706e-02 1.93687603e-02 -1.47434309e-01
9.90381479e-01 2.95152545e-01 -7.68346727e-01 -2.40016177e-01
-2.27534212e-02 2.86703378e-01 2.18782306e-01 1.97003514e-01
-1.15943670e+00 1.48507941e+00 4.89805746e+00 1.10466337e+00
-9.16699648e-01 3.05249512e-01 5.88948667e-01 2.49832362e-01
-5.25001824e-01 1.41197145e-01 -6.24410212e-01 4.27361608e-01
2.69001961e-01 -4.25662905e-01 2.58988738e-01 9.95582998e-01
-3.99652660e-01 5.79085909e-02 -1.08583188e+00 8.53589833e-01
2.41137743e-01 -1.21658957e+00 4.39307004e-01 -4.32594717e-02
9.09953892e-01 -3.23112428e-01 7.99375996e-02 5.49404323e-01
4.12454545e-01 -9.76254106e-01 4.73761201e-01 3.72330368e-01
9.38068688e-01 -6.93634689e-01 3.72571766e-01 9.88138095e-02
-1.51185262e+00 -3.74308527e-02 -6.34398580e-01 2.51172155e-01
-3.84564325e-02 2.22104341e-01 -6.99693739e-01 4.53086346e-01
5.79626560e-01 1.05340707e+00 -6.51814461e-01 9.72032964e-01
-6.67222679e-01 2.64530599e-01 8.93615931e-03 1.12026908e-01
6.04527175e-01 -1.95299253e-01 4.81721163e-01 9.09170270e-01
3.72799456e-01 1.98366776e-01 4.79375571e-01 1.19177592e+00
-8.04734528e-02 1.34796381e-01 -7.24376678e-01 -8.76491442e-02
4.00000781e-01 1.38387549e+00 -1.14619040e+00 -7.57653594e-01
-4.94245023e-01 1.13499820e+00 2.42358699e-01 5.73434114e-01
-7.06348896e-01 -5.25519788e-01 8.12237978e-01 -1.13050431e-01
3.90006185e-01 1.83785632e-01 -2.94598430e-01 -1.24170506e+00
-3.05270374e-01 -3.76396894e-01 6.50734484e-01 -7.75648952e-01
-1.37936509e+00 4.45073783e-01 3.75698470e-02 -1.34935451e+00
-2.65191138e-01 -4.46757019e-01 -7.72029042e-01 7.80575216e-01
-1.18646598e+00 -1.41230404e+00 -4.71248269e-01 8.10155153e-01
9.58887815e-01 -8.04334283e-02 7.24571705e-01 -1.83291614e-01
-2.11262882e-01 4.36615318e-01 -1.30239725e-02 1.86249077e-01
4.17065859e-01 -1.31364977e+00 2.95426220e-01 8.22751462e-01
3.35477054e-01 4.10618305e-01 4.34784323e-01 -6.87069237e-01
-7.63769448e-01 -1.24544561e+00 4.25307781e-01 -3.66809189e-01
5.91709733e-01 -5.30987740e-01 -1.18713975e+00 5.38044095e-01
-9.43386778e-02 2.09791183e-01 6.95195138e-01 9.79560912e-02
-4.96648252e-01 1.62360162e-01 -1.10063946e+00 6.41967714e-01
1.21741402e+00 -5.76751709e-01 -1.12722099e+00 2.35445812e-01
1.13990176e+00 -5.26892953e-02 -5.12665033e-01 2.92841822e-01
2.95773596e-01 -6.68136716e-01 1.07388055e+00 -7.27738023e-01
4.76729751e-01 -3.07327062e-01 -3.68263900e-01 -1.50107503e+00
-3.52014095e-01 -7.77726248e-02 1.81162655e-01 1.22060692e+00
3.42357606e-01 -5.31576395e-01 7.14198530e-01 3.01101953e-01
-3.35562855e-01 -5.25040627e-01 -7.29275227e-01 -7.56526709e-01
1.80167198e-01 -2.86349028e-01 5.52330554e-01 1.17857385e+00
-1.22380883e-01 6.56600714e-01 -1.88117743e-01 -9.88331437e-02
6.15004241e-01 6.60592198e-01 6.19542181e-01 -1.34722531e+00
-2.32104763e-01 -3.81824374e-01 -6.18190944e-01 -8.06344271e-01
2.09856108e-01 -1.24642718e+00 1.85255647e-01 -1.73069322e+00
4.44846928e-01 -3.47240686e-01 -5.53981960e-01 6.81683242e-01
-4.06840324e-01 4.57175821e-01 4.06809121e-01 1.30765900e-01
-6.18289769e-01 7.38927782e-01 1.80690658e+00 -3.38402987e-01
-4.12169307e-01 -1.51405901e-01 -9.37158823e-01 6.37087345e-01
7.16651142e-01 -3.57805580e-01 -7.93232381e-01 -1.02754347e-01
-3.93567652e-01 -1.46463975e-01 5.81698000e-01 -9.58255053e-01
-6.17397297e-03 -4.90986615e-01 5.97103953e-01 -4.62507933e-01
3.94628882e-01 -5.42753577e-01 -3.23498428e-01 2.64300376e-01
-1.99662507e-01 -5.90796173e-01 1.94338784e-02 7.14743495e-01
-2.75024742e-01 -2.58915067e-01 9.44163740e-01 -4.95529562e-01
-1.51951241e+00 4.84770775e-01 -7.18754232e-02 5.57980955e-01
1.26992106e+00 -5.43271899e-01 -4.38654333e-01 -4.78704087e-02
-1.10404420e+00 3.63742530e-01 5.62283039e-01 8.73393714e-01
7.50727594e-01 -1.20159471e+00 -3.74975830e-01 5.17548680e-01
6.06453359e-01 1.47996038e-01 5.64613521e-01 3.28643590e-01
-1.44058034e-01 1.77797660e-01 -5.93538523e-01 -8.37721527e-01
-9.54123378e-01 9.69585538e-01 7.11020734e-03 2.56748289e-01
-7.84055471e-01 1.08751011e+00 1.06403601e+00 -4.19751525e-01
-1.09495483e-01 -2.84745276e-01 -3.98195505e-01 4.34810102e-01
3.87175322e-01 6.36373758e-02 -3.74828786e-01 -8.60485911e-01
-2.67592639e-01 8.11006904e-01 6.39175624e-02 7.18928799e-02
1.11820507e+00 -2.99132079e-01 1.04633406e-01 6.52070045e-01
1.38873017e+00 -5.11820376e-01 -1.34361267e+00 -2.87607014e-01
1.61964782e-02 -5.36200821e-01 -1.54710039e-01 -5.97056448e-01
-1.19287992e+00 1.21000886e+00 6.81578279e-01 1.75705865e-01
9.95540559e-01 5.76580286e-01 8.35270762e-01 -1.48704469e-01
2.58885801e-01 -1.10732770e+00 3.17741901e-01 4.91984487e-01
5.49555242e-01 -1.32770526e+00 -2.93086767e-01 -6.46130681e-01
-9.16539907e-01 8.12269509e-01 8.53543341e-01 -2.93826163e-01
4.87637818e-01 -4.08848614e-01 -2.26741806e-02 -3.80683810e-01
-3.41071695e-01 -9.68500495e-01 7.67572105e-01 9.49429989e-01
-6.14134856e-02 4.70949769e-01 -6.96012378e-02 8.07185471e-01
-1.96189001e-01 -3.69521886e-01 2.14550346e-01 7.36169875e-01
-8.08671355e-01 -8.57681453e-01 -7.43067637e-02 5.20245075e-01
1.05440922e-01 -2.49330491e-01 -3.01029772e-01 9.33445036e-01
5.17229557e-01 7.38034844e-01 5.02668440e-01 -3.78733844e-01
1.82061195e-01 1.39880717e-01 3.31825703e-01 -1.13434744e+00
-7.35246763e-02 1.70519114e-01 -2.70023197e-01 -4.05618340e-01
-1.22130170e-01 -2.82646924e-01 -1.42247677e+00 1.81312203e-01
-1.51072349e-02 2.18089163e-01 3.41746897e-01 9.97884095e-01
1.08485103e-01 7.11771309e-01 5.41113496e-01 -8.22988093e-01
3.14659737e-02 -7.46684134e-01 -7.53986001e-01 8.27805817e-01
2.72281736e-01 -1.02746809e+00 -4.13328856e-01 4.65943813e-02] | [9.834799766540527, 1.7229254245758057] |
4102fd9d-aa5f-410c-9ee3-53870f2cf37c | learning-parametric-distributions-for-image | null | null | http://openaccess.thecvf.com/content_iccv_2015/html/Li_Learning_Parametric_Distributions_ICCV_2015_paper.html | http://openaccess.thecvf.com/content_iccv_2015/papers/Li_Learning_Parametric_Distributions_ICCV_2015_paper.pdf | Learning Parametric Distributions for Image Super-Resolution: Where Patch Matching Meets Sparse Coding | Existing approaches toward Image super-resolution (SR) is often either data-driven (e.g., based on internet-scale matching and web image retrieval) or model-based (e.g., formulated as an Maximizing a Posterior estimation problem). The former is conceptually simple yet heuristic; while the latter is constrained by the fundamental limit of frequency aliasing. In this paper, we propose to develop a hybrid approach toward SR by combining those two lines of ideas. More specifically, the parameters underlying sparse distributions of desirable HR image patches are learned from a pair of LR image and retrieved HR images. Our hybrid approach can be interpreted as the first attempt of reconciling the difference between parametric and nonparametric models for low-level vision tasks. Experimental results show that the proposed hybrid SR method performs much better than existing state-of-the-art methods in terms of both subjective and objective image qualities. | ['Xuemei Xie', 'Yongbo Li', 'Guangming Shi', 'Weisheng Dong'] | 2015-12-01 | null | null | null | iccv-2015-12 | ['patch-matching'] | ['computer-vision'] | [ 5.20473123e-01 5.56046441e-02 -3.18049163e-01 -2.29950041e-01
-1.21076190e+00 -1.85915083e-01 5.55976212e-01 -2.56839991e-01
2.74537709e-02 6.90855742e-01 3.01809311e-01 1.60473436e-01
-3.81174982e-01 -7.01725960e-01 -4.52174455e-01 -9.11099315e-01
3.49137604e-01 -4.74433750e-02 2.77381510e-01 -2.35980406e-01
5.94264448e-01 4.94301558e-01 -1.86742795e+00 1.65307015e-01
1.00007665e+00 1.17006075e+00 4.53837901e-01 3.49353880e-01
-1.11972369e-01 6.89193785e-01 -1.98305145e-01 -5.57949916e-02
2.94095248e-01 -6.29567027e-01 -5.79847932e-01 3.86905700e-01
4.65905249e-01 -2.40390927e-01 -3.11466306e-01 1.29806149e+00
5.21342158e-01 8.38822126e-02 4.52014029e-01 -8.55105102e-01
-9.83426690e-01 1.46640703e-01 -8.95293832e-01 3.30588639e-01
5.09039700e-01 -2.64645189e-01 8.36249530e-01 -1.19071841e+00
4.57875848e-01 1.24190104e+00 3.58754963e-01 2.42074147e-01
-1.53816998e+00 -3.95535499e-01 -6.41980097e-02 3.00074667e-01
-1.65879130e+00 -7.56628215e-01 1.00659549e+00 -3.62631947e-01
2.68745601e-01 1.84546992e-01 2.79389650e-01 8.45526338e-01
-1.82279143e-02 5.73466003e-01 1.59275007e+00 -7.58752644e-01
2.85109133e-01 4.44206029e-01 4.67822291e-02 4.44576412e-01
2.74228841e-01 3.61913472e-01 -8.13177288e-01 -4.10879910e-01
1.24556231e+00 -2.82752395e-01 -5.13742745e-01 -6.70792520e-01
-8.81133258e-01 7.54539967e-01 1.42182916e-01 3.65007132e-01
-3.98788959e-01 -3.13307792e-01 -1.72174066e-01 2.01446339e-01
5.89791536e-01 1.17675781e-01 6.15911521e-02 3.41529250e-01
-1.14036965e+00 2.58409768e-01 4.17418897e-01 8.16013455e-01
6.95288599e-01 1.09449603e-01 -1.17507935e-01 1.23693132e+00
4.87348825e-01 3.46924335e-01 3.33992720e-01 -1.18928325e+00
-3.71961817e-02 -1.27594009e-01 5.27271569e-01 -1.04203951e+00
9.53586251e-02 -4.29635376e-01 -6.28713906e-01 2.27926716e-01
3.04390758e-01 3.98655176e-01 -6.19835198e-01 1.64799440e+00
3.44763815e-01 2.47891665e-01 2.98730358e-02 1.06694376e+00
6.68452740e-01 6.15254819e-01 -2.07652628e-01 -7.30460227e-01
1.20064473e+00 -8.30429018e-01 -9.16758657e-01 -3.65882441e-02
-4.60655093e-01 -8.85745347e-01 9.54532921e-01 6.11275136e-01
-1.41613138e+00 -9.06822503e-01 -9.23811615e-01 -2.96622552e-02
1.35467723e-01 1.31998941e-01 1.58159479e-01 6.90460801e-01
-1.37327456e+00 4.50365484e-01 -4.28397179e-01 -4.42046046e-01
7.40217790e-02 -1.06205255e-01 -6.30277768e-02 -1.90803379e-01
-8.33522677e-01 7.58967519e-01 2.25945562e-01 -1.88787170e-02
-4.92655605e-01 -4.78009492e-01 -7.43440390e-01 -5.93619514e-03
3.82276654e-01 -6.49819553e-01 1.02565372e+00 -9.37604666e-01
-1.64683425e+00 1.00496888e+00 -4.15521353e-01 -3.26573849e-01
2.74137944e-01 -2.46603236e-01 -5.36657631e-01 5.02169132e-01
4.94258804e-03 3.67406338e-01 1.39618003e+00 -1.96598637e+00
-7.35188484e-01 -4.15716916e-01 -1.93600595e-01 2.80836016e-01
-2.84605652e-01 1.13549016e-01 -3.59312624e-01 -6.04776204e-01
4.50978309e-01 -6.59718633e-01 -1.38031214e-01 2.42368609e-01
-1.11033686e-01 -1.14500441e-01 7.18318939e-01 -7.34613121e-01
1.12987876e+00 -2.21353483e+00 2.07898542e-01 7.66124576e-02
9.41386595e-02 1.00369908e-01 -6.97643906e-02 4.37419862e-01
-1.44680977e-01 -1.17553584e-01 -5.49255013e-02 -2.36027539e-01
-2.01920763e-01 -1.59502983e-01 -5.40549755e-01 6.51925981e-01
1.21112994e-03 4.28514838e-01 -7.38584936e-01 -7.04077244e-01
2.66816199e-01 6.92436039e-01 -1.87293753e-01 2.74364352e-01
1.28994614e-01 6.45149589e-01 -4.01565313e-01 6.37688339e-01
8.29922557e-01 -2.82972246e-01 3.23331058e-01 -4.95150805e-01
-3.67771417e-01 6.18028603e-02 -1.29781175e+00 1.58080184e+00
-3.94534498e-01 3.87964398e-01 7.50324503e-02 -9.46785331e-01
1.10094702e+00 2.18394607e-01 5.94606936e-01 -6.86833799e-01
-3.05065662e-01 3.02845508e-01 -4.35496181e-01 -3.93536270e-01
6.90717876e-01 -3.17780942e-01 3.38718861e-01 2.12110400e-01
-1.83876753e-01 -1.07217945e-01 -1.37512416e-01 -1.79726049e-01
5.72105229e-01 4.16665405e-01 7.36794412e-01 -3.40563625e-01
6.87660217e-01 -2.04443544e-01 4.71753448e-01 8.80142450e-01
-9.67872366e-02 8.89934361e-01 7.15630949e-02 -8.16116259e-02
-1.20766211e+00 -1.42671514e+00 -4.58003342e-01 6.70575202e-01
4.81768459e-01 -3.61422151e-02 -7.19885647e-01 -8.80758390e-02
-2.49941990e-01 5.27391791e-01 -2.27907136e-01 1.39924049e-01
-4.17899162e-01 -4.53467041e-01 -1.73204914e-02 3.00597340e-01
6.60684168e-01 -8.61605644e-01 -5.04315913e-01 1.78542823e-01
-4.29535776e-01 -1.32890856e+00 -2.21084431e-01 -4.28436041e-01
-9.75406945e-01 -7.35601544e-01 -9.19370711e-01 -6.44008934e-01
5.28520048e-01 8.58632922e-01 1.11609674e+00 -4.27303687e-02
-2.07608998e-01 5.85359573e-01 -4.64409322e-01 1.02589903e-02
-1.44943431e-01 -4.68260705e-01 7.80690014e-02 4.96046305e-01
8.70481730e-02 -7.30067074e-01 -7.75037944e-01 3.43909770e-01
-9.87458348e-01 -7.50129446e-02 7.77152538e-01 9.25615013e-01
1.08702266e+00 3.00316036e-01 5.78056157e-01 -7.04599261e-01
4.83732373e-01 -3.58101338e-01 -7.71018803e-01 2.93761134e-01
-7.77197361e-01 -1.32487163e-01 4.14698869e-01 -4.67948884e-01
-1.43506098e+00 1.47411346e-01 9.89198089e-02 -6.30850434e-01
-2.70468414e-01 3.20152462e-01 -7.80924410e-02 -4.02044982e-01
5.89132309e-01 7.28955686e-01 -2.48823054e-02 -6.76601231e-01
2.53740937e-01 7.77759314e-01 7.58143008e-01 -5.93916118e-01
1.04641092e+00 7.62517393e-01 -3.17121558e-02 -1.15357113e+00
-1.04701924e+00 -5.98404169e-01 -4.31525856e-01 -2.59484559e-01
7.78115869e-01 -1.03195775e+00 -3.49507034e-01 3.68132532e-01
-7.41392374e-01 3.61693488e-03 -3.34024280e-01 3.95094067e-01
-9.79269385e-01 6.87126040e-01 -6.12499297e-01 -1.12894797e+00
-3.19747552e-02 -9.47229922e-01 1.33833075e+00 4.89731491e-01
2.63475478e-01 -6.41827524e-01 3.22428830e-02 6.12371504e-01
6.38331234e-01 1.46528736e-01 7.01455116e-01 -5.01998626e-02
-7.58167207e-01 2.82786903e-03 -6.22578382e-01 2.77468294e-01
1.88314512e-01 -3.40623081e-01 -9.87477779e-01 -4.47902232e-01
5.41246533e-01 -3.19739789e-01 5.92760444e-01 6.74638093e-01
1.21203244e+00 -1.42142221e-01 -1.06952749e-01 4.67370868e-01
1.93028414e+00 1.58103220e-02 1.05890119e+00 4.39525455e-01
2.15799943e-01 8.20931017e-01 1.13583529e+00 5.19245207e-01
3.23220521e-01 1.11382067e+00 1.69179425e-01 -2.27993235e-01
-2.88363397e-01 -2.61133254e-01 1.03356533e-01 5.01322448e-01
-1.94971129e-01 5.33560514e-02 -4.26937342e-01 4.25765157e-01
-1.81116891e+00 -1.10342288e+00 1.36927441e-01 2.44333744e+00
6.81360483e-01 -2.15383604e-01 2.33464524e-01 -3.64644974e-02
8.29602182e-01 3.14694494e-01 -4.74821985e-01 1.03280589e-03
-2.43762657e-01 8.77285451e-02 3.00045729e-01 4.78851646e-01
-8.07384729e-01 6.23325229e-01 6.24516916e+00 1.15128410e+00
-8.71003628e-01 2.24565253e-01 5.86991429e-01 2.09945217e-01
-3.45014006e-01 1.77668463e-02 -7.42291152e-01 3.24976355e-01
6.66596353e-01 -2.98684239e-01 6.23687506e-01 6.31421745e-01
3.92161965e-01 -3.25788528e-01 -7.55444646e-01 1.25070429e+00
2.74610668e-01 -1.14985931e+00 -7.21877590e-02 3.49992692e-01
7.12418437e-01 -3.23933631e-01 3.09264749e-01 -1.95092395e-01
-2.40973253e-02 -9.83183742e-01 8.16619217e-01 7.69706547e-01
8.02864671e-01 -5.22822380e-01 4.93357003e-01 3.70967269e-01
-1.27569675e+00 -8.37474614e-02 -4.35977995e-01 4.03318822e-01
1.77858919e-01 5.94507158e-01 4.00526971e-02 8.61644149e-01
9.80204701e-01 4.54142332e-01 -3.09521317e-01 1.02926838e+00
8.31695460e-03 2.89756477e-01 -6.20444342e-02 5.68390667e-01
-3.06760490e-01 -5.71345150e-01 7.45299935e-01 8.15611899e-01
3.70570242e-01 3.10597539e-01 2.80700892e-01 1.09631634e+00
4.01584327e-01 6.98764548e-02 -6.97864354e-01 3.53801072e-01
5.47441661e-01 1.34339464e+00 -5.00809014e-01 -5.78925610e-02
-6.07520998e-01 8.83397162e-01 1.66972548e-01 5.17300785e-01
-6.05765343e-01 2.53656097e-02 3.32324266e-01 2.53358454e-01
5.59368789e-01 -3.03167850e-01 -2.15384290e-01 -1.25063729e+00
1.40266657e-01 -1.04492700e+00 2.88610488e-01 -1.10433149e+00
-1.34968007e+00 6.58288598e-01 1.74107686e-01 -1.58569837e+00
-2.68466353e-01 -2.82532215e-01 -2.24563137e-01 9.76063192e-01
-2.10864902e+00 -1.13561535e+00 -3.57220709e-01 7.83683479e-01
5.94982684e-01 1.03295803e-01 6.44723415e-01 1.12722032e-01
-1.16246372e-01 3.30426276e-01 2.14756370e-01 -4.53241348e-01
6.89137161e-01 -1.01903713e+00 -1.80531681e-01 1.01951528e+00
1.97007775e-01 4.85172004e-01 9.64778721e-01 -4.37052995e-01
-1.45910907e+00 -6.83508933e-01 6.34458721e-01 -1.09457798e-01
4.18448448e-01 2.39227097e-02 -9.86675739e-01 2.20404163e-01
5.60219027e-02 8.68391842e-02 5.43775439e-01 -1.49945468e-01
-3.78484249e-01 -1.42373264e-01 -1.42444587e+00 4.60781366e-01
8.92374396e-01 -5.91075778e-01 -4.97849375e-01 2.37502649e-01
4.03063834e-01 -2.27205023e-01 -1.15626156e+00 4.61960286e-01
5.07204711e-01 -1.38374925e+00 1.39055049e+00 -6.53781071e-02
3.65375221e-01 -5.42444527e-01 -5.76136589e-01 -9.62526977e-01
-4.69081014e-01 -3.96978021e-01 -8.88513848e-02 1.24004054e+00
8.11363664e-03 -6.12082064e-01 5.29686093e-01 2.15905532e-01
1.16024621e-01 -5.82580090e-01 -7.17729926e-01 -9.88627553e-01
-3.76043350e-01 2.76026540e-02 1.94411486e-01 8.28711689e-01
-2.63509452e-01 1.01003155e-01 -6.12873435e-01 5.47136247e-01
1.30707371e+00 5.54676771e-01 5.52007437e-01 -1.11574984e+00
-5.85311413e-01 -1.72539502e-01 -1.33126050e-01 -1.09975910e+00
-4.44234014e-02 -2.56530732e-01 2.14527667e-01 -1.31454432e+00
3.91364515e-01 -3.56586367e-01 -2.68315345e-01 -1.31391585e-01
-3.28324437e-02 3.76905859e-01 5.35525531e-02 5.98113418e-01
-3.81915271e-01 4.55368340e-01 1.04445291e+00 3.64394128e-01
-1.96586698e-01 -5.14372215e-02 -8.05573285e-01 6.20820642e-01
6.19266272e-01 -2.10623428e-01 -4.08903539e-01 -6.39956445e-02
-8.46984610e-02 6.19347811e-01 5.81352592e-01 -6.79110944e-01
2.81386614e-01 -3.59882921e-01 1.76525116e-01 -5.99311709e-01
6.65670335e-01 -7.08631337e-01 2.79346645e-01 1.00821532e-01
-3.19928080e-01 -3.89756680e-01 -2.79300541e-01 9.44408834e-01
-4.95730072e-01 -4.80893627e-02 1.27291667e+00 -2.26847127e-01
-6.18967533e-01 7.21379071e-02 -2.65016612e-02 -2.36068785e-01
6.34131432e-01 -4.68887419e-01 -3.80625755e-01 -6.12266898e-01
-5.82356691e-01 -2.82533973e-01 6.37112379e-01 2.90474087e-01
9.13982689e-01 -1.22654402e+00 -7.33573318e-01 1.17982186e-01
1.27351835e-01 -3.97335231e-01 3.98601711e-01 9.84572291e-01
-6.91349953e-02 3.11535567e-01 -1.37365222e-01 -6.44096434e-01
-1.21825480e+00 6.16593242e-01 2.35861152e-01 -2.80962378e-01
-7.20682085e-01 4.28965658e-01 3.68361622e-01 5.38842566e-02
6.53612837e-02 3.73071462e-01 -3.98604810e-01 -2.79131949e-01
7.41069019e-01 3.13173354e-01 -2.82798439e-01 -8.48964810e-01
-1.04132675e-01 9.59835768e-01 -1.10348277e-02 -2.66236395e-01
1.38687444e+00 -7.20134854e-01 -1.87806055e-01 4.05566931e-01
8.88961375e-01 1.79974377e-01 -1.15702808e+00 -6.79662824e-01
4.38079797e-02 -1.02072191e+00 3.54014665e-01 -5.28631747e-01
-8.85356545e-01 4.14011806e-01 9.21743989e-01 5.22401512e-01
1.60123599e+00 1.85291767e-01 5.91962457e-01 -4.35854495e-02
6.82000637e-01 -1.02887750e+00 1.32134616e-01 -2.65466481e-01
9.88656640e-01 -1.25949228e+00 1.80578589e-01 -8.72153699e-01
-4.19953525e-01 1.09760594e+00 3.76831144e-01 -3.24637324e-01
5.31620204e-01 2.89349575e-02 -1.22229226e-01 -1.24758266e-01
-5.83995640e-01 -3.72539014e-01 4.70995724e-01 6.93689287e-01
3.70212853e-01 -2.20086619e-01 -5.22259653e-01 2.73428082e-01
1.76938668e-01 1.86798811e-01 4.55837965e-01 8.03842247e-01
-6.81337118e-01 -1.02040863e+00 -6.49470031e-01 1.65772185e-01
-5.00171363e-01 -1.45652369e-01 2.13343967e-02 5.54537058e-01
-3.33545625e-01 1.14976239e+00 -2.48021543e-01 -3.36159393e-02
3.09322596e-01 -3.48252922e-01 6.87275708e-01 -4.24628347e-01
1.91569656e-01 5.52751243e-01 -9.14480686e-02 -7.21884787e-01
-9.12316680e-01 -8.26499403e-01 -5.91189802e-01 1.47912890e-01
-4.56924617e-01 -9.90297273e-02 6.57263100e-01 7.64320552e-01
2.53852934e-01 3.81613411e-02 9.71101046e-01 -1.00402784e+00
-7.46584177e-01 -6.34636700e-01 -9.11455333e-01 1.68553501e-01
3.49706501e-01 -6.85692430e-01 -5.36053181e-01 1.85234949e-01] | [11.069877624511719, -2.0517947673797607] |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.