ManifoldMixup:BetterRepresentationsbyInterpolatingHiddenStatesVikasVerma12AlexLamb2ChristopherBeckham2AmirNajafi3IoannisMitliagkas2DavidLopez-Paz4YoshuaBengio2AbstractThisisaworryingprospect,sinced...
LearningDistanceforSequencesbyLearningaGroundMetricBingSu1YingWu2AbstractAlthoughmetriclearninghasachievedaconsiderablematu-ritylevelbothinpracticeandintheory(Belletetal.,2013),Learningdistancestha...
ImprovingModelSelectionbyEmployingtheTestDataMaxWestphal1WernerBrannath1AbstractThisisinparticulartruewhenlabelleddataisexpensivetoacquireanddatasetsarethusonlyofmodestsizeortheem-Modelselectionand...
FlexiblyFairRepresentationLearningbyDisentanglementElliotCreager12DavidMadras12Jo¨rn-HenrikJacobsen2MarissaA.Weis23KevinSwersky4ToniannPitassi12RichardZemel12AbstractCurrentapproachestofairreprese...
EfficientTrainingofBERTbyProgressivelyStackingLinyuanGong1DiHe1ZhuohanLi1TaoQin2LiweiWang13Tie-YanLiu2Abstractespeciallyindomainsthatrequireparticularexpertise.Unsupervisedpre-trainingiscommonlyuse...
DiscoveringOptionsforExplorationbyMinimizingCoverTimeYuuJinnai1JeeWonPark1DavidAbel1GeorgeKonidaris1Abstractoptionsguaranteedtoreducetheexpectedcovertimeusingthetransitionfunctioneithergiventoorlea...
Curiosity-Bottleneck:ExplorationbyDistillingTask-SpecificNoveltyYoungjinKim12WontaeNam∗3HyunwooKim∗2Ji-HoonKim4GunheeKim2Abstractcontainsnovelbuttask-irrelevantinformation1.Forexam-ple,supposearo...
ConditioningbyadaptivesamplingforrobustdesignDavidH.Brookes1HahnbeomPark23JenniferListgarten4AbstractAdvancesinbiotechnology,chemistryandmachinelearningallowforthepossibilitytoimprovesuchdesigncycl...
BreakingInter-LayerCo-AdaptationbyClassifierAnonymizationIkuroSato1KohtaIshikawa1GuoqingLiu1MasayukiTanaka2Abstract0.431.18Thisstudyaddressesanissueofco-adaptation26betweenafeatureextractorandaclas...
AccelerationofSVRGandKatyushaXbyInexactPreconditioningYanliLiu1FeiFeng1WotaoYin1Abstractregularizerψ(x)isproper,closed,andconvex,butmaybenonsmooth.Anonzeroψ(x)isdesirableinmanyapplica-Empiricalri...
Non-LinearMotorControlbyLocalLearninginSpikingNeuralNetworksAdityaGilra12WulframGerstner1AbstractDadarlatetal.,2015).Forwardmodelsuseneuralmotorcommandstopredictbodymovement,whileinversemod-Learnin...
NetworkGlobalTestingbyCountingGraphletsJiashunJin1ZhengTracyKe2ShengmingLuo1AbstractForeachnode,weassignaProbabilityMassFunctionConsideralargesocialnetworkwithpossi-(PMF)πi=(πi(1),πi(2),···,...
Meta-LearningbyAdjustingPriorsBasedonExtendedPAC-BayesTheoryRonAmit1RonMeir1Abstractconstraints,e.g.,convolutionsandweightsharing(LeCunetal.,2015).However,oftentherelevantpriorinformationInmeta-lea...
LearningbyPlaying–SolvingSparseRewardTasksfromScratchMartinRiedmiller1RolandHafner1ThomasLampe1MichaelNeunert1JonasDegrave1TomVandeWiele1VolodymyrMnih1NicolasHeess1TobiasSpringenberg1Abstractsimul...
FastandScalableBayesianDeepLearningbyWeight-PerturbationinAdamMohammadEmtiyazKhan1DidrikNielsen1VootTangkaratt1WuLin2YarinGal3AkashSrivastava4AbstractusingBayes’rule.Unfortunately,thisisinfeasible...
DisentanglingbyFactorisingHyunjikKim12AndriyMnih1AbstractVAEWedefineandaddresstheproblemofunsuper-randomlypermute-Discriminatorvisedlearningofdisentangledrepresentationsoneachdimensiondatagenerated...
CurriculumLearningbyTransferLearning:TheoryandExperimentswithDeepNetworksDaphnaWeinshall1GadCohen1DanAmir1Abstractforcementlearning(e.g.Gravesetal.,2017).Althoughitremainedforthemostpartinthefringe...
ComputationalOptimalTransport:ComplexitybyAcceleratedGradientDescentIsBetterThanbySinkhorn’sAlgorithmPavelDvurechensky1AlexanderGasnikov234AlexeyKroshnin234Abstractclustering(Hoetal.,2017),textcla...
UnsupervisedLearningbyPredictingNoisePiotrBojanowski1ArmandJoulin1Abstractlearntheunderlyingdistributionofimages(Vincentetal.,2010;Goodfellowetal.,2014).Whilesomeoftheseap-Convolutionalneuralnetwor...
ScalingUpSparseSupportVectorMachinesbySimultaneousFeatureandSampleReductionWeizhongZhang12BinHong13WeiLiu2JiepingYe3DengCai1XiaofeiHe1JieWang3Abstracttionandvariableselectionby1-normpenalty.Thelast...