Entropy Reweighted Conformal Classification

TL;DR

This paper introduces an entropy reweighting method for conformal classification that adaptively considers classifier uncertainty to improve the efficiency of prediction sets while maintaining coverage guarantees.

Contribution

It proposes an innovative entropy-based reweighting approach that enhances conformal classification efficiency by adaptively incorporating classifier uncertainty.

Findings

Significant improvement in prediction set efficiency

Maintains coverage guarantees with adaptive reweighting

Effective across multiple experimental datasets

Abstract

Conformal Prediction (CP) is a powerful framework for constructing prediction sets with guaranteed coverage. However, recent studies have shown that integrating confidence calibration with CP can lead to a degradation in efficiency. In this paper, We propose an adaptive approach that considers the classifier's uncertainty and employs entropy-based reweighting to enhance the efficiency of prediction sets for conformal classification. Our experimental results demonstrate that this method significantly improves efficiency.