Instance-OptimalCompressedSensingviaPosteriorSamplingAjilJalal1SushrutKarmalkar2AlexandrosG.Dimakis1EricPrice2AbstractforameasurementmatrixA∈Rm×nandnoisevectorWecharacterizethemeasurementcomplexi...
CompressedMaximumLikelihoodYiHao1AlonOrlitsky1AbstractAnestimatorforfisafunctionfˆ:Z→QsuchthatwhenZ∼p,thelossd(fˆ(Z),f(p))issmall.TheMLplug-inMaximumlikelihood(ML)isoneofthemostfundamentalandge...
OnthePowerofCompressedSensingwithGenerativeModelsAkshayKamath1SushrutKarmalkar1EricPrice1AbstractoftheCompressedrepresentationofxthantoitsambientdimensionn.ThegoalofCompressedsensingistolearnastruc...
AccelerationforCompressedGradientDescentinDistributedandFederatedOptimizationZhizeLi1DmitryKovalev1XunQian1PeterRichta´rik1Abstract1.IntroductionDuetothehighcommunicationcostindistributedWiththepr...
LearningaCompressedSensingMeasurementMatrixviaGradientUnrollingShanshanWu1AlexandrosG.Dimakis1SujaySanghavi1FelixX.Yu2DanielHoltmann-Rice2DmitryStorcheus2AfshinRostamizadeh2SanjivKumar2AbstractIfd>...
FastDirectSearchinanOptimallyCompressedContinuousTargetSpaceforEfficientMulti-LabelActiveLearningWeishiShi1QiYu1Abstractretrievalandorganization.UsersfromQ&Awebsites,suchasstackoverflowandQuora,are...
DeepCompressedSensingYanWu1MihaelaRosca1TimothyLillicrap1Abstracttrainingatall.CShasbeensuccessfullyappliedinscenarioswheremeasurementsarenoisyandexpensivetotake,suchCompressedsensing(CS)providesan...
CompressedFactorization:FastandAccurateLow-RankFactorizationofCompressively-SensedDataVatsalSharan1KaiShengTai1PeterBailis1GregoryValiant1AbstractInthelatter,theuseofrandomprojectionsunderliesmanys...
SIGNSGD:CompressedOptimisationforNon-ConvexProblemsJeremyBernstein12Yu-XiangWang23KamyarAzizzadenesheli4AnimaAnandkumar12AbstractAlgorithm1SIGNSGDTraininglargeneuralnetworksrequiresdistribut-Input:...
ModelingSparseDeviationsforCompressedSensingusingGenerativeModelsManikDhar1AdityaGrover1StefanoErmon1Abstractcations,amongmanyothers,canbenefitsignificantlyfromCompressedsensingtechniquestoacquires...
ProtoNN:CompressedandAccuratekNNforResource-scarceDevicesChiragGupta1ArunSaiSuggala12AnkitGoyal13HarshaVardhanSimhadri1BhargaviParanjape1AshishKumar1SaurabhGoyal4RaghavendraUdupa1ManikVarma1Prateek...
FollowtheCompressedLeader:FasterOnlineLearningofEigenvectorsandFasterMMWUZeyuanAllen-Zhu1YuanzhiLi2Abstractsymmetricandsatisfies0AkI.1Boththeseassump-tionsareforthesakeofsimplicityandcanberelaxed.2...
CompressedSensingusingGenerativeModelsAshishBora1AjilJalal2EricPrice1AlexandrosG.Dimakis2Abstractoftheunknownvectorx⇤.Weneedtoassumethattheun-knownvectoris“natural,”or“simple,”insomeapplicatio...