DifferentiableDynamicNormalizationforLearningDeepRepresentationPingLuo12ZhanglinPeng3WenqiShao23RuimaoZhang23JiaminRen3LingyunWu3AbstractForinstance,BNisusedaftereveryconvolutionallayerofaResNet101...
DiagnosingBottlenecksinDeepQ-learningAlgorithmsJustinFu1AviralKumar1MatthewSoh1SergeyLevine1AbstractwhichpotentialissueswithQ-learningmanifestinpractice.WeempiricallyanalyzeaspectsoftheQ-learningme...
DeepResidualOutputLayersforNeuralLanguageGenerationNikolaosPappas1JamesHenderson1Abstractbeddingstocapturethesimilaritystructureoftheoutputlabelspace,sothatdataforsimilarlabelscanhelpclassi-Manytas...
DeepFactorsforForecastingYuyangWang1AlexSmola1DanielleC.Maddix1JanGasthaus1DeanFoster1TimJanuschowski1Abstractmetricshavebeendevelopedinthecontextofforecastingindividualorsmallgroupsoftimeseries.Th...
DeepGenerativeLearningviaVariationalGradientFlowYuanGao1YulingJiao2YangWang3YaoWang4CanYang3ShunkangZhang3Abstract1.IntroductionWeproposeaframeworktolearnDeepgenerativeLearningthegenerativemodel,i....
DeepGaussianProcesseswithImportance-WeightedVariationalInferenceHughSalimbeni12VincentDutordoir2JamesHensman2MarcPeterDeisenroth12AbstractFigure1:Posteriordensityfromatwo-layermodel,illus-tratingno...
DeepCounterfactualRegretMinimizationNoamBrown12AdamLerer1SamGross1TuomasSandholm23Abstractintwo-playerzero-sumgames.FormsoftabularCFRhavebeenusedinallrecentmilestonesinthebenchmarkdomainCounterfact...
DeepCompressedSensingYanWu1MihaelaRosca1TimothyLillicrap1Abstracttrainingatall.CShasbeensuccessfullyappliedinscenarioswheremeasurementsarenoisyandexpensivetotake,suchCompressedsensing(CS)providesan...
ComplexityofLinearRegionsinDeepNetworksBorisHanin1DavidRolnick2AbstractFigure1.Howmanylinearregions?Thisfigureshowsatwo-dimensionalslicethroughthe784-dimensionalinputspaceofItiswell-knownthattheexp...
CombatingLabelNoiseinDeepLearningUsingAbstentionSunilThulasidasan12TanmoyBhattacharya1JeffreyBilmes2GopinathChennupati1JamaludinMohd-Yusof1AbstractImageNetdatabase(Dengetal.,2009)usedfortrainingvis...
CollaborativeChannelPruningforDeepNetworksHanyuPeng1JiaxiangWu2ShifengChen1JunzhouHuang3AbstractTherearealargecollectionofresearchpapersfocusingonacceleratingandcompressingDeepnetworkstoimprovetheD...
CharacterizingWell-Behavedvs.PathologicalDeepNeuralNetworksAntoineLabatie1Abstractinput-outputmappingsatinitializationextensivelyfoundtobepredictiveoftrainabilityandpost-trainingperformanceWeintrod...
CalibratedModel-BasedDeepReinforcementLearningAliMalik1VolodymyrKuleshov12JiamingSong1DannyNemer2HarlanSeymour2StefanoErmon1AbstractFigure1.Modernmodel-basedplanningalgorithmswithproba-bilisticmode...
BiasAlsoMatters:BiasAttributionforDeepNeuralNetworkExplanationShengjieWang1TianyiZhou1JefferyA.Bilmes2Abstractfunctions,utilizingmulti-layerarchitecturesandarichsetofpossiblehiddennon-linearities,m...
BayesianGenerativeActiveDeepLearningToanTran1Thanh-ToanDo2IanReid1GustavoCarneiro1Abstractbeperformedwithimmenseandcarefullylabeleddatasets,wherethelargerthedataset,themoreeffectivethetrainingDeepl...
BayesianActionDecoderforDeepMulti-AgentReinforcementLearningJakobN.Foerster12H.FrancisSong3EdwardHughes3NeilBurch3IainDunning3ShimonWhiteson1MatthewM.Botvinick3MichaelBowling3Abstract1.Introduction...
AKernelPerspectiveforRegularizingDeepNeuralNetworksAlbertoBietti1GrégoireMialon12DexiongChen1JulienMairal1Abstractsuchkernels,theRKHScontainsindeedDeepconvolutionalnetworkssimilartogenericones—up...
ADeepReinforcementLearningPerspectiveonInternetCongestionControlNathanJay1NogaH.Rotman2P.BrightenGodfrey1MichaelSchapira2AvivTamar3AbstractFigure1:MultipletrafficflowssharingalinkWepresentandinvest...
AConvergenceTheoryforDeepLearningviaOver-ParameterizationZeyuanAllen-Zhu1YuanzhiLi23ZhaoSong456Abstract1IntroductionDeepneuralnetworks(DNNs)havedemon-Neuralnetworkshavedemonstratedagreatsuccessinst...
Zero-ShotKnowledgeDistillationinDeepNetworksGauravKumarNayak1KondaReddyMopuri2VaisakhShaj3R.VenkateshBabu1AnirbanChakraborty1Abstractlentperformance,buttheycanbehugeandcomputationallyexpensive.Henc...