UnderstandingtheDynamicsofGradientFlowinOverparameterizedLinearModelsSalmaTarmoun12GuilhermeFranc¸a13BenjaminHaeffele14Rene´Vidal14AbstractathoroughanalysisoftheDynamicsofoptimizationmeth-ods,suc...
UnderstandingSelf-SupervisedLearningDynamicswithoutContrastivePairsYuandongTian1XinleiChen1SuryaGanguli12Abstractquiringexpensivetargetlabels(Devlinetal.,2018).Manystate-of-the-artSSLmethodsincompu...
Prediction-CentricLearningofIndependentCascadeDynamicsfromPartialObservationsMateuszWilinski1AndreyY.Lokhov1AbstractThismotivatesthedevelopmentofefficientalgorithmsthatcaninfertheeffectivespreading...
InferringLatentDynamicsUnderlyingNeuralPopulationActivityviaNeuralDifferentialEquationsTimothyDoyeonKim1ThomasZhihaoLuo1JonathanW.Pillow12CarlosD.Brody13Abstractaneuralpopulationcancoordinatemoveme...
GroundingLanguagetoEntitiesandDynamicsforGeneralizationinReinforcementLearningAustinW.Hanjie1VictorZhong2KarthikNarasimhan1AbstractunseenentitiesorDynamics(Narasimhanetal.,2018;Zhongetal.,2020).Whi...
GenerativeAdversarialNetworksforMarkovianTemporalDynamics:StochasticContinuousDataGenerationSungWooPark1DongWookShu1JunseokKwon1AbstractprobabilitydistributionXt∼pt,whichhasthefollowingformulation...
DatasetDynamicsviaGradientFlowsinProbabilitySpaceDavidAlvarez-Melis1Nicolo`Fusi1Abstract2019),datasetpooling,orinstancemixing(Zhangetal.,2018).Ontheotherhand—andourprincipalmotivationinVariousmach...
Data-drivenPredictionofGeneralHamiltonianDynamicsviaLearningExactly-SymplecticMapsRenyiChen1MoleiTao1Abstract1.IntroductionWeconsiderthelearningandpredictionofnon-Givenacollectionofsequences,eachbe...
ControllingGraphDynamicswithReinforcementLearningandGraphNeuralNetworksEliA.Meirom1HaggaiMaron1ShieMannor1GalChechik1AbstractFigure1.Aviralinfectionprocessonagraphandaninterventionaimedtostopitsspr...
BetterTrainingusingWeight-ConstrainedStochasticDynamicsBenedictLeimkuhler1TiffanyVlaar1Timothe´ePouchon1AmosStorkey2AbstractCurrentapproachestoenhancethegeneralizationperfor-manceofoverparameteriz...
Align,thenmemorise:theDynamicsoflearningwithfeedbackalignmentMariaRefinetti12Ste´phaned’Ascoli13RubenOhana14SebastianGoldt5AbstractIntroductionDirectFeedbackAlignment(DFA)isemergingTrainingadeepn...
AugmentedWorldModelsFacilitateZero-ShotDynamicsGeneralizationFromaSingleOfflineEnvironmentPhilipJ.Ball1CongLu1JackParker-Holder1StephenRoberts1Abstractproblems(Dulac-Arnoldetal.,2019),withthepotent...
UnderstandingandStabilizingGANs’TrainingDynamicsusingControlTheoryKunXu1ChongxuanLi1JunZhu1BoZhang1Abstractplesfromfakeones.Gaimstorecoverthedatadistributionbymaximizingthisratio.Thisframeworkisfo...
TowardsUnderstandingtheDynamicsoftheFirst-OrderAdversariesZhunDeng1HangfengHe2JiaoyangHuang3WeijieJ.Su2Abstractattackssuchasinputde-noising(Guoetal.,2017),random-izedsmoothing(Ilyasetal.,2019),grad...
ImplicitLearningDynamicsinStackelbergGames:EquilibriaCharacterization,ConvergenceAnalysis,andEmpiricalStudyTannerFiez1BenjaminChasnov1LillianRatliff1Abstractandhyperparameteroptimization(Maclaurine...
FractionalUnderdampedLangevinDynamics:RetargetingSGDwithMomentumunderHeavy-TailedGradientNoiseUmutS¸ims¸ekli12LingjiongZhu3YeeWhyeTeh2MertGu¨rbu¨zbalaban4Abstract1.IntroductionStochasticgradien...
EstimatingQ(s,s)withDeepDeterministicDynamicsGradientsAshleyD.Edwards1HimanshuSahni2RosanneLiu13JaneHung1AnkitJain1RuiWang1AdrienEcoffet1ThomasMiconi1CharlesIsbell2JasonYosinski13AbstractActionat+1...
DynamicsofDeepNeuralNetworksandNeuralTangentHierarchyJiaoyangHuang1Horng-TzerYau2Abstractproblems,thereisnoguaranteethatagradientbasedalgo-rithmwillbeabletofindtheoptimalparametersefficientlyTheevo...
Curvature-correctedlearningDynamicsindeepneuralnetworksDongsungHuh1AbstractWhilerecentadvancesinapproximateNGDmethods(e.g.K-FAC)havedramaticallyimprovedthecomputationalef-Deepneuralnetworksexhibitc...
Context-awareDynamicsModelforGeneralizationinModel-BasedReinforcementLearningKiminLee1YounggyoSeo2SeunghyunLee2HonglakLee34JinwooShin2Abstractetal.,2019;Hafneretal.,2019;2020;Kaiseretal.,2020).Mode...