REPAINT:KnowledgeTransferinDeepReinforcementLearningYunzheTao1SahikaGenc1JonathanChung1TaoSun1SunilMallya1Abstractimproveperformanceonothertasks.AcceleratinglearningprocessesforcomplextasksTransfer...
Zoo-Tuning:AdaptiveTransferfromaZooofModelsYangShu1ZhiKou1ZhangjieCao1JianminWang1MingshengLong1Abstractleveragemodelspretrainedonlarge-scaledatasets(Rus-sakovskyetal.,2015)andfine-tunethemodelonth...
Learn-to-Share:AHardware-friendlyTransferLearningFrameworkExploitingComputationandParameterSharingChengFu1HanxianHuang1XinyunChen2YuandongTian3JishenZhao1AbstractFigure1:RoadmapofGLUEscorev.s.total...
Group-SparseMatrixFactorizationforTransferLearningofWordEmbeddingsKanXu1XuanyiZhao1HamsaBastani1OsbertBastani1AbstractthisstrategyhasbeenusedinTransferringimageclassifi-cationmodels(Estevaetal.,201...
GlobalRhythmStyleTransferWithoutTextTranscriptionsKaizhiQian12YangZhang12ShiyuChang12JinjunXiong2ChuangGan12DavidCox12MarkHasegawa-Johnson3Abstractwhichthespeakerexpressesthecontent.Therearetwomajo...
FederatedContinualLearningwithWeightedInter-clientTransferJaehongYoon1WonyongJeong12GiwoongLee1EunhoYang12SungJuHwang12AbstractFigure1.Concept.Acontinuallearneratahospitalwhichlearnsonsequenceofdis...
CARTL:CooperativeAdversarially-RobustTransferLearningDianChen1HongxinHu2QianWang1YinliLi1CongWang3ChaoShen4QiLi5Abstracthumans.However,trainingsuchamodelrequiresawell-designednetworkarchitecture,ma...
ANewRepresentationofSuccessorFeaturesforTransferacrossDissimilarEnvironmentsMajidAbdolshah1HungLe1ThommenKarimpanalGeorge1SunilGupta1SantuRana1SvethaVenkatesh1Abstractintoindependentsub-domains.How...
UnsupervisedTransferLearningforSpatiotemporalPredictiveNetworksZhiyuYao1YunboWang1MingshengLong1JianminWang1Abstractdeeprepresentationsfromtheconsequencesofcomplexvideoevents.Thestudiesofthespatiot...
TransferLearningwithoutKnowing:ReprogrammingBlack-boxMachineLearningModelswithScarceDataandLimitedResourcesYun-YunTsai1Pin-YuChen2Tsung-YiHo1AbstractatargetdomainbasedontheknowledgeTransferredfroma...
SequentialTransferinReinforcementLearningwithaGenerativeModelAndreaTirinzoni1RiccardoPoiani1MarcelloRestelli1AbstractAkeyquestioniswhatandhowknowledgeshouldbetrans-ferred(Taylor&Stone,2009).Asforth...
RIFLE:BackpropagationinDepthforDeepTransferLearningthroughRe-InitializingtheFully-connectedLayErXingjianLi12HaoyiXiong1HaozheAn1ChengzhongXu23DejingDou1Abstract1.IntroductionFine-tuningthedeepconvo...
MutualTransferLearningforMassiveDataChing-WeiCheng1XingyeQiao2GuangCheng1Abstractdatadomainisthetargetandwhichdatawillbeusedasthesource.However,thisisnotnecessarilythecaseinIntheTransferlearningpro...
MetaVarianceTransfer:LearningtoAugmentfromTheOthersSeong-JinPark1SeungjuHan1JiwonBaek1InsooKim1JuhwanSong1HaeBeomLee2Jae-JoonHan1SungJuHwang2Abstractcepts.However,collectingacompletedatasetwithallp...
HierarchicallyDecoupledImitationforMorphologicalTransferDonaldJ.HejnaIII1PieterAbbeel1LerrelPinto12AbstractFigure1.Inthiswork,wefocusontheproblemofTransferringlong-horizonpoliciesacrossmorphologies...
Few-shotDomainAdaptationbyCausalMechanismTransferTakeshiTeshima12IsseiSato12MasashiSugiyama21AbstractFigure1:Nonparametricgen-erativemodelofnonlinearin-Westudyfew-shotsuperviseddomainadaptationdepe...
AQuantile-basedApproachforHyperparameterTransferLearningDavidSalinas1HuibinShen2ValerioPerrone2Abstractproblembyplacingaprobabilisticmodel,typicallyaGaus-sianprocess(GP),onthefunctiontominimize;the...
TransferofSamplesinPolicySearchviaMultipleImportanceSamplingAndreaTirinzoni1MattiaSalvini1MarcelloRestelli1Abstractagentissupposedtoreuseknowledgeacquiredfromasetofsourcetaskstoacceleratethelearnin...
TransferLearningforRelatedReinforcementLearningTasksviaImage-to-ImageTranslationShaniGamrian1YoavGoldberg12Abstractprocessofanewtaskhastobeperformedfromscratchevenforarelatedone.Recentworkshavetrie...
Parameter-EfficientTransferLearningforNLPNeilHoulsby1AndreiGiurgiu1StanisławJastrze¸bski2BrunaMorrone1QuentindeLaroussilhe1AndreaGesmundo1MonaAttariyan1SylvainGelly1AbstractAccuracydelta(%)50Fine...