Double Ramp Loss Based Reject Option Classifier

TL;DR

This paper introduces a double ramp loss function for reject option classifiers, providing a continuous upper bound for the traditional loss and demonstrating improved performance over existing methods through experiments.

Contribution

The paper proposes a novel double ramp loss function and a DC programming-based approach for reject option classification, enhancing performance over prior methods.

Findings

Outperforms existing reject classifiers on benchmark datasets

Provides a continuous upper bound for the 0-d-1 loss

Effective on synthetic and real-world data

Abstract

We consider the problem of learning reject option classifiers. The goodness of a reject option classifier is quantified using $0-d-1$ loss function wherein a loss $d \in (0,.5)$ is assigned for rejection. In this paper, we propose {\em double ramp loss} function which gives a continuous upper bound for $(0-d-1)$ loss. Our approach is based on minimizing regularized risk under the double ramp loss using {\em difference of convex (DC) programming}. We show the effectiveness of our approach through experiments on synthetic and benchmark datasets. Our approach performs better than the state of the art reject option classification approaches.