InferenceandSamplingofK33-freeIsingModelsValeriiLikhosherstov1YuryMaximov12MichaelChertkov123Abstractoneofthemostsignificantcontributionsintheoreticalandmathematicalphysicsofthe20thcentury.Then,Kac...
HOList:AnEnvironmentforMachineLearningofHigher-OrderTheoremProvingKshitijBansal1SarahLoos1MarkusRabe1ChristianSzegedy1StewartWilcox1AbstractriSpeech(Panayotovetal.,2015)forspeechrecognition,theNet...
HierarchicalDecompositionalMixturesofVariationalAutoencodersPingLiangTan12RobertPeharz1AbstractSPNVariationalautoencoders(VAEs)havereceivedz1z1z2z2...zLzLconsiderableattention,sincetheyallowustolea...
Graphite:IterativeGenerativeModelingofGraphsAdityaGrover1AaronZweig1StefanoErmon1Abstractneuralnetworks(Gorietal.,2005;Scarsellietal.,2009;Brunaetal.,2013).Agraphneuralnetworklearnsnode-Graphsareaf...
GradientDescentFindsGlobalMinimaofDeepNeuralNetworksSimonS.Du1JasonD.Lee2HaochuanLi34LiweiWang54XiyuZhai6AbstractThesecondmysteriousphenomenonintrainingdeepneuralnetworksis“deepernetworksareharder...
GoodInitializationsofVariationalBayesforDeepModelsSimoneRossi1PietroMichiardi1Mauriziofilippone1AbstractAFTERPOORINITIALIZATIONAFTEROURINITIALIZATIONStochasticvariationalinferenceisanestablished22w...
GlobalConvergenceofBlockCoordinateDescentinDeepLearningJinshanZeng12TimTsz-KitLau3Shao-BoLin4YuanYao2AbstractinGogames(Silveretal.,2016).DeeplearninghasarousedextensiveattentiondueThepracticaloptim...
GenerativeModelingofInfiniteOccludedObjectsforCompositionalSceneRepresentationJinyangYuan1BinLi1XiangyangXue1Abstractcanautomaticallysegregatethevisualsceneintoconceptualentities(e.g.objects)throug...
First-orderAdversarialVulnerabilityofNeuralNetworksandInputDimensionCarl-JohannSimon-Gabriel12YannOllivier2BernhardSchölkopf1LéonBottou2DavidLopez-Paz2Abstractparticularifitcanspreadhomogeneously...
Fine-GrainedAnalysisofOptimizationandGeneralizationforOverparameterizedTwo-LayerNeuralNetworksSanjeevArora12SimonS.Du3WeiHu1ZhiyuanLi1RuosongWang3Abstractthatsufficientlypowerfulnets(withvastlymore...
Finite-TimeAnalysisofDistributedTD(0)withLinearFunctionApproximationforMulti-AgentReinforcementLearningThinhT.Doan12SivaThejaMaguluri1JustinRomberg2Abstractenvironment,oftenmodeledasaMarkovDecision...
FindingMixedNashEquilibriaofGenerativeAdversarialNetworksYa-PingHsieh1ChenLiu1VolkanCevher1Abstract2016).Indeed,thestochasticgradientdescent(SGD),alongwithitsvariants,remainthemethodofchoicefortrai...
FasterStochasticAlternatingDirectionMethodofMultipliersforNonconvexOptimizationFeihuHuang1SongcanChen23HengHuang14Abstract1.IntroductionInthispaper,weproposeafasterstochasticalter-Alternatingdirect...
FastandFlexibleInferenceofJointDistributionsfromtheirMarginalsCharlieFrogner1TomasoPoggio2Abstractjointobservationsandinsteadmustrelyonobservationsofthevariablestakenseparately.Politicalscientists,...
Fairwashing:theriskofrationalizationUlrichA¨ıvodji1HiromiArai23OlivierFortineau4Se´bastienGambs1SatoshiHara5AlainTapp67AbstractInparticular,toensuretransparencyandexplainabilityinalgorithmicdeci...
ExploringtheLandscapeofSpatialRobustnessLoganEngstrom1BrandonTran1DimitrisTsipras1LudwigSchmidt1AleksanderMa˛dry1AbstractThereisalonglineofworkonmethodsforconstructingadversarialperturbationsinvar...
ExploitingStructureofUncertaintyforEfficientMatroidSemi-BanditsPierrePerrault12VianneyPerchet23MichalValko14Abstracttimeandspacecomplexity)oftheircombinatorialgeneral-ization,inwhichtheagentselects...
EmpiricalAnalysisofBeamSearchPerformanceDegradationinNeuralSequenceModelsEldanCohen1J.ChristopherBeck1Abstractinferencealgorithmfordecodingneuralsequencemodelsisbeamsearch,asearchalgorithmthatgener...
EfficientOptimizationofLoopsandLimitswithRandomizedTelescopingSumsAlexBeatson1RyanP.Adams1Abstractovertime,eachlearningsteprequiresloopingovertimesteps.Morebroadly,inmanyscientificandengineeringWec...
EfficientTrainingofBERTbyProgressivelyStackingLinyuanGong1DiHe1ZhuohanLi1TaoQin2LiweiWang13Tie-YanLiu2Abstractespeciallyindomainsthatrequireparticularexpertise.Unsupervisedpre-trainingiscommonlyuse...