Double ramp loss based reject option classifier

Abstract

The performance of a reject option classifiers is quantified using 0 − d − 1 loss where d ∈ (0,. 5) is the loss for rejection. In this paper, we propose 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 difference of convex 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.