FairnessRiskMeasuresRobertC.Williamson1AdityaKrishnaMenon2Abstractabstractedearlierdefinitionsoffairnessbyframingtheprob-lemintermsofsubgrouplosses.Intuitively,theseworksEnsuringthatclassifiersaren...
TrainableCalibrationMeasuresForNeuralNetworksFromKernelMeanEmbeddingsAviralKumar1SunitaSarawagi1UjjwalJain1Abstractlearnedmodelsfeedintodecisionsystemsorarehumanin-terpreted.Alsocalibrationisuseful...
AUnifiedViewofMulti-LabelPerformanceMeasuresXi-ZhuWu1Zhi-HuaZhou1Abstracteachinstancecanbeassociatedwithmultiplelabelssimul-taneously.Forexample,itisdifficulttotellwhichmistakeMulti-labelclassifica...