Fair Conformal Classification via Learning Representation-Based Groups

TL;DR

This paper proposes a fair conformal inference framework that ensures adaptive subgroup coverage in classification, addressing fairness issues in machine learning predictions.

Contribution

It introduces a novel method for constructing prediction sets with conditional coverage guarantees across subgroups, enhancing fairness and trustworthiness.

Findings

Effective in balancing compactness and informativeness of prediction sets.

Ensures adaptive equalized coverage across unfairly treated subgroups.

Demonstrated success on synthetic and real-world datasets.

Abstract

Conformal prediction methods provide statistically rigorous marginal coverage guarantees for machine learning models, but such guarantees fail to account for algorithmic biases, thereby undermining fairness and trust. This paper introduces a fair conformal inference framework for classification tasks. The proposed method constructs prediction sets that guarantee conditional coverage on adaptively identified subgroups, which can be implicitly defined through nonlinear feature combinations. By balancing effectiveness and efficiency in producing compact, informative prediction sets and ensuring adaptive equalized coverage across unfairly treated subgroups, our approach paves a practical pathway toward trustworthy machine learning. Extensive experiments on both synthetic and real-world datasets demonstrate the effectiveness of the framework.