Abstract

A prediction rule in binary classification that aims to achieve the lowest probability of misclassification involves minimizing over a nonconvex, 0-1 loss function, which is typically a computationally intractable optimization problem. To address the intractability, previous methods consider minimizing the cumulative loss – the sum of convex surrogates of the 0-1 loss of each sample. We revisit this paradigm and develop instead an axiomatic framework by proposing a set of salient properties on functions for bi nary classification and then propose the coherent loss approach, which is a tractable upper-bound of the empirical classification error over the entire sample set. We show that the proposed approach yields a strictly tighter approximation to the empirical classification error than any convex cumulative loss approach while preserving the convexity of the underlying optimization problem, and this approach for binary classification also has a robustness interpretation which builds a connection to robust SVMs.