A Reductions Approach to Fair Classification

TL;DR

This paper introduces a reduction-based method for fair binary classification that unifies multiple fairness definitions and optimizes for lowest error under fairness constraints by solving cost-sensitive classification problems.

Contribution

It proposes a systematic reduction approach that generalizes various fairness definitions and improves upon existing methods in terms of flexibility and performance.

Findings

Reductions effectively unify multiple fairness criteria.

The approach outperforms prior baselines on diverse datasets.

The method produces randomized classifiers with minimal empirical error.

Abstract

We present a systematic approach for achieving fairness in a binary classification setting. While we focus on two well-known quantitative definitions of fairness, our approach encompasses many other previously studied definitions as special cases. The key idea is to reduce fair classification to a sequence of cost-sensitive classification problems, whose solutions yield a randomized classifier with the lowest (empirical) error subject to the desired constraints. We introduce two reductions that work for any representation of the cost-sensitive classifier and compare favorably to prior baselines on a variety of data sets, while overcoming several of their disadvantages.