LearningContinuousHierarchiesintheLorentzModelofHyperbolicGeometryMaximilianNickel1DouweKiela1Abstractsystemsofconcepts.However,explicitinformationaboutsuchhierarchicalrelationshipsisunavailablefor...
invarianceofWeightDistributionsinRectifiedMLPsRussellTsuchida1FarbodRoosta-Khorasani23MarcusGallagher1Abstractplicationratherthananunderstandingofthecapabilitiesandtrainingofneuralnetworks.Recently...
inferenceSuboptimalityinVariationalAutoencodersChrisCremer1XuechenLi1DavidDuvenaud1AbstractFig.1forasimpleillustrationofthegaps.inFig.1,L[q]referstotheELBOevaluatedusinganamortizeddistribu-Amortize...
ImprovedRegretBoundsforThompsonSamplinginLinearQuadraticControlProblemsMarcAbeille1AlessandroLazaric2Abstracthasbeenmostlyaddressedfollowingtwomainapproaches:optimism-in-face-of-uncertainty(OFU)and...
ImportanceWeightedTransferofSamplesinReinforcementLearningAndreaTirinzoni1AndreaSessa1MatteoPirotta2MarcelloRestelli1Abstracttions,parameters,policies,etc.)andinthecriteriausedtoestablishwhethersuc...
ImplicitRegularizationinNonconvexStatisticalEstimation:GradientDescentConvergesLinearlyforPhaseRetrievalandMatrixCompletionCongMa1KaizhengWang1YuejieChi2YuxinChen3Abstract1.introductionRecentyearsh...
GEP-PG:DecouplingExplorationandExploitationinDeepReinforcementLearningAlgorithmsCe´dricColas1OlivierSigaud12Pierre-YvesOudeyer1AbstractDeepRLalgorithmsgenerallyconsistinapplyingStochas-ticGradient...
FastDecodinginSequenceModelsUsingDiscreteLatentVariablesŁukaszKaiser1AurkoRoy1AshishVaswani1NikiParmar1SamyBengio1JakobUszkoreit1NoamShazeer1Abstractlation(Sutskeveretal.,2014;Bahdanauetal.,2014;C...
FairnessWithoutDemographicsinRepeatedLossMinimizationTatsunoriB.Hashimoto12MeghaSrivastava1HongseokNamkoong3PercyLiang1AbstractJurgensetal.,2017),dependencyparsing(Blodgettetal.,2016),part-of-speec...
ExploringHiddenDimensionsinParallelizingConvolutionalNeuralNetworksZhihaoJia1SinaLin2CharlesR.Qi1AlexAiken1Abstracteachdevice.Anothercommonapproachismodelparal-lelism(Mirhoseinietal.,2017;Kimetal.,...
EssentiallyNoBarriersinNeuralNetworkEnergyLandscapeFelixDraxler12KambisVeschgini2ManfredSalmhofer2FredA.Hamprecht1Abstractformaconnectedmanifold.Moreprecisely,wearguethatthepartoftheparameterspacew...
EfficientBias-Span-ConstrainedExploration-ExploitationinReinforcementLearningRonanFruit1MatteoPirotta1AlessandroLazaric2RonaldOrtner3Abstractand,ateachstep,itexecutesthepolicywithhighestopti-mistic...
Discrete-ContinuousMixturesinProbabilisticProgramming:GeneralizedSemanticsandinferenceAlgorithmsYiWu1SiddharthSrivastava2NicholasHay3SimonS.Du4StuartRussell1Abstracttureofcontinuousanddiscreterando...
Detectingnon-causalartifactsinmultivariatelinearregressionmodelsDominikJanzing1BernhardScho¨lkopf2AbstractZZZWeconsiderlinearmodelswheredpotentialXYXYXYcausesX1,...,Xdarecorrelatedwithonetargetqua...
DeepReinforcementLearninginContinuousActionSpaces:aCaseStudyintheGameofSimulatedCurlingKyowoonLee1Sol-AKim1JaesikChoi1Seong-WhanLee2Abstract1992),andothello(Buro,1999).Recently,deepconvolu-tionalne...
DecompositionofUncertaintyinBayesianDeepLearningforEfficientandRisk-sensitiveLearningStefanDepeweg12Jose´MiguelHerna´ndez-Lobato3FinaleDoshi-Velez4SteffenUdluft1Abstractinthisworkweshowhowtoperfo...
CoordinatedExplorationinConcurrentReinforcementLearningMariaDimakopoulou1BenjaminVanRoy1Abstractandrefinesestimatesasdataisgathered.Atthestartofeachepisode,theagentsamplesanMDPfromitscurrentposte-W...
Closed-formMarginalLikelihoodinGamma-PoissonMatrixFactorizationLouisFilstroff1AlbertoLumbreras1Ce´dricFe´votte1AbstractwhereweusetheshapeandrateparametrizationoftheGammadistribution,i.e.,Gamma(x...
CharacterizingImplicitBiasinTermsofOptimizationGeometrySuriyaGunasekar1JasonLee2DanielSoudry3NathanSrebro1Abstractassociatedhyperparametercanchangetheimplicitbias.Forexample,Wilsonetal.(2017)showed...
BeyondtheOne-StepGreedyApproachinReinforcementLearningYonathanEfroni1GalDalal1BrunoScherrer2ShieMannor1Abstractsuggestedthatgreedyapproachesw.r.t.multiplestepsper-formbetterthanw.r.t.1-step.Notable...