FastMarginMaximizationviaDualAccelerationZiweiJi1NathanSrebro2MatusTelgarsky1AbstractMargin0.050.00Wepresentandanalyzeamomentum-basedgra-−0.05Alg1/eq(1.1)dientmethodfortraininglinearclassifierswit...
ConsistentStructuredPredictionwithMax-MinMarginMarkovNetworksAlexNowak-Vila1FrancisBach1AlessandroRudi1Abstractdictionmistakesarenotequallycostly.Insequencepre-diction,forinstance,thenumberofpossib...
OptimalMinimalMarginMaximizationwithBoostingAllanGrønlund1KasperGreenLarsen1AlexanderMathiasen1Abstractstructsalinearcombinationclassifiersign(Tαtht(x)),t=1Boostingalgorithmsiterativelyproducelin...
DoesDataAugmentationLeadtoPositiveMargin?ShashankRajput1ZhiliFeng1ZacharyCharles2Po-LingLoh3DimitrisPapailiopoulos2AbstractaugmentedImageNetusingtranslations,horizontalreflec-tions,andalteredintens...
Multi-ClassOptimalMarginDistributionMachineTengZhang1Zhi-HuaZhou1AbstractRecently,theMargintheoryforBoostinghasfinallybeendefended(Gao&Zhou,2013),andhasdisclosedthattheRecentstudiesdisclosethatmaxi...