VarianceReductionviaPrimal-DualAcceleratedDualAveragingforNonsmoothConvexFinite-SumsChaobingSong1StephenJ.Wright1JelenaDiakonikolas1Abstractvectorswithntypicallylarge;gi:R!R,i=1,2,...,n,bepossiblyn...
Zeroth-OrderNon-ConvexLearningviaHierarchicalDualAveragingAmélieHéliou1MatthieuMartin1PanayotisMertikopoulos21ThibaudRahier1Abstractalsorequiresthattheproblem’sobjectiveremainstationaryduringthe...
AcceleratingGossipSGDwithPeriodicGlobalAveragingYimingChen1KunYuan1YingyaZhang1PanPan1YinghuiXu1WotaoYin1AbstractMETHODEPOCHACC.%TIME(HRS.)CommunicationoverheadhindersthescalabilityPARALLELSGD12076...
SCAFFOLD:StochasticControlledAveragingforFederatedLearningSaiPraneethKarimireddy12SatyenKale3MehryarMohri34SashankJ.Reddi3SebastianU.Stich1AnandaTheerthaSuresh3Abstractclientdataoverthenetwork,ther...
OnlinemirrordescentanddualAveraging:keepingpaceinthedynamiccaseHuangFang1NicholasJ.A.Harvey1VictorS.Portella1MichaelP.Friedlander1Abstractthebenefitofhindsight.LettingTdenotethenumberofdecisions,th...
ASimplerApproachtoAcceleratedStochasticOptimization:IterativeAveragingMeetsOptimismPooriaJoulani1AnantRaj2Andra´sGyo¨rgy1CsabaSzepesva´ri13Abstractblynon-smooth)convexfunction.Whenφ=0,andgivena...
SWALP:StochasticWeightAveraginginLow-PrecisionTrainingGuandaoYang1TianyiZhang1PolinaKirichenko1JunwenBai1AndrewGordonWilson1ChristopherDeSa1Abstractandaccumulategradientinformationinhigherprecision...