UnderstandingInstance-LevelLabelNoise:DisparateImpactsandTreatmentsYangLiu1Abstractframework(Feldman,2020),wedemonstratethevaryingeffectsofmemorizingnoisyLabelsassociatedwithinstancesThispaperaimst...
SinkhornLabelAllocation:Semi-SupervisedClassificationviaAnnealedSelf-TrainingKaiShengTai1PeterBailis1GregoryValiant1Abstractsificationtasks.Self-training,alsoknownasself-Labeling,isanSSLmethodwhere...
Multi-DimensionalClassificationviaSparseLabelEncodingBin-BinJia12Min-LingZhang13Abstractexistinvariousfields,e.g.,textclassification(Shatkayetal.,2008),bioinformatics(Rodr´ıguezetal.,2012),resour...
LocallyAdaptiveLabelSmoothingforPredictiveChurnDaraBahri1HeinrichJiang1Abstractaswewillshowlaterintheexperiments.Thus,evenanoptimizedtrainingprocedurecanleadtohighpredictionTrainingmodernneuralnetw...
LeveragedWeightedLossforPartialLabelLearning1HongweiWen2JingyiCui1HanyuanHang2JiabinLiu3YisenWang1ZhouchenLin14Abstract2012),etc,andsubsequentlyattractsalotofattentiononmethodologystudies(Fengetal....
LAMDA:LabelMatchingDeepDomainAdaptationTrungLe1TuanNguyen1NhatHo2HungBui3DinhPhung13AbstractconsumingtocollectandLabelhigh-qualitydatasufficientlylargetotrainaccuratedeepmodels,suchasinthedomainDee...
LabelInferenceAttacksfromLog-lossScoresAbhinavAggarwal1ShivaPrasadKasiviswanathan1ZekunXu1OluwaseyiFeyisetan1NathanaelTeissier1AbstractthatPNi=1ui,k=1.Let2[K]NbeaLabeling.Then,theK-arylog-loss(or,c...
LabelDistributionLearningMachineJingWang12XinGeng12AbstractHAP:1.39descriptiondegree0.30SAD:2.030.25AlthoughLabelDistributionLearning(LDL)hasSUR:1.57witnessedextensiveclassificationapplications,itA...
ATheoryofLabelPropagationforSubpopulationShiftTianleCai12RuiqiGao12JasonD.Lee1QiLei1Abstracttees.Manyoftheworksarebasedonthecovariateshiftas-sumption(Heckman,1979;Shimodaira,2000),whichstatesOneoft...
VariationalLabelEnhancementNingXu1JunShu2Yun-PengLiu1XinGeng1AbstractrelevantLabelsinstancesirrelevantLabelssand(a)sunLabeldistributioncoversacertainnumberofla-(b)busbels,representingthedegreetowhi...
Self-supervisedLabelAugmentationviaInputTransformationsHankookLee1SungJuHwang234JinwooShin21Abstractsentationlearningforimages(Doerschetal.,2015;Noroozi&Favaro,2016;Larssonetal.,2017;Gidarisetal.,2...
LTF:ALabelTransformationFrameworkforCorrectingTargetShiftJiaxianGuo1MingmingGong2TongliangLiu1KunZhang3DachengTao1Abstract1.IntroductionDistributionshiftisamajorobstacletothede-Standardsupervisedle...
LearningwithBoundedInstance-andLabel-dependentLabelNoiseJiachengCheng12TongliangLiu1KotagiriRamamohanarao3DachengTao1AbstractLabelsmadebyhumanexpertsarelikelytobenoisyduetoconfusedpatternsandpercep...
DoesLabelSmoothingMitigateLabelNoise?MichalLukasik1SrinadhBhojanapalli1AdityaKrishnaMenon1SanjivKumar1AbstractInterestingly,therearetwocompetingintuitions.Ontheonehand,smoothingmightmitigatetheprob...
UnsupervisedLabelNoiseModelingandLossCorrectionEricArazo1DiegoOrtego1PaulAlbert1NoelE.O’Connor1KevinMcGuinness1Abstract3Clean4DespitebeingrobusttosmallamountsofLabelLoss2Noisy3noise,convolutionaln...
LearningContext-DependentLabelPermutationsforMulti-LabelClassificationJinseokNam1Young-BumKim1EneldoLozaMenc´ıa2SunghyunPark1RuhiSarikaya1JohannesFu¨rnkranz2AbstractofCCsisthattheyrequireafixeds...
HowdoesDisagreementHelpGeneralizationagainstLabelCorruption?XingruiYu1BoHan2JiangchaoYao3GangNiu2IvorW.Tsang1MasashiSugiyama24AbstractthesecomplexmodelscanfullymemorizenoisyLabels(Zhangetal.,2017;A...
ExploitingWorkerCorrelationforLabelAggregationinCrowdsourcingYuanLi1BenjaminI.P.Rubinstein1TrevorCohn1Abstractequalvotestowardsconsensus.Numerousprobabilisticmodelshaveemergedthatparameteriseworker...
CombatingLabelNoiseinDeepLearningUsingAbstentionSunilThulasidasan12TanmoyBhattacharya1JeffreyBilmes2GopinathChennupati1JamaludinMohd-Yusof1AbstractImageNetdatabase(Dengetal.,2009)usedfortrainingvis...