Bayes-Optimal Classifiers under Group Fairness

TL;DR

This paper develops a unified framework for deriving Bayes-optimal classifiers under group fairness constraints, introducing the FairBayes method to control disparity and optimize fairness-accuracy tradeoffs, supported by extensive experiments.

Contribution

It provides the first unified approach to characterize Bayes-optimal classifiers under various group fairness constraints using Neyman-Pearson theory.

Findings

FairBayes effectively controls disparity in classifiers.

The method achieves near-optimal fairness-accuracy tradeoffs.

Experimental results validate the theoretical advantages.

Abstract

Machine learning algorithms are becoming integrated into more and more high-stakes decision-making processes, such as in social welfare issues. Due to the need of mitigating the potentially disparate impacts from algorithmic predictions, many approaches have been proposed in the emerging area of fair machine learning. However, the fundamental problem of characterizing Bayes-optimal classifiers under various group fairness constraints has only been investigated in some special cases. Based on the classical Neyman-Pearson argument (Neyman and Pearson, 1933; Shao, 2003) for optimal hypothesis testing, this paper provides a unified framework for deriving Bayes-optimal classifiers under group fairness. This enables us to propose a group-based thresholding method we call FairBayes, that can directly control disparity, and achieve an essentially optimal fairness-accuracy tradeoff. These advantages are supported by thorough experiments.