SCOD: From Heuristics to Theory

arXiv:2403.16916v1 Announce Type: new
Abstract: This paper addresses the problem of designing reliable prediction models that abstain from predictions when faced with uncertain or out-of-distribution samples - a recently proposed problem known as Selective Classification in the presence of Out-of-Distribution data (SCOD). We make three key contributions to SCOD. Firstly, we demonstrate that the optimal SCOD strategy involves a Bayes classifier for in-distribution (ID) data and a selector represented as a stochastic linear classifier in a 2D space, using i) …

abstract arxiv classification cs.lg data designing distribution heuristics key paper prediction prediction models predictions samples stat.ml theory type uncertain