A Unified Post-Processing Framework for Group Fairness in Classification

TL;DR

This paper introduces 'LinearPost', a versatile post-processing algorithm that ensures group fairness in classification tasks across multiple criteria and settings, with efficient computation and proven fairness guarantees.

Contribution

The paper proposes a unified post-processing framework for various group fairness criteria applicable to multiclass problems, extending existing methods with efficiency and theoretical guarantees.

Findings

Outperforms existing methods in high fairness regimes

Guarantees fairness when group membership predictor is multicalibrated

Efficiently computes parameters via linear programming

Abstract

We present a post-processing algorithm for fair classification that covers group fairness criteria including statistical parity, equal opportunity, and equalized odds under a single framework, and is applicable to multiclass problems in both attribute-aware and attribute-blind settings. Our algorithm, called "LinearPost", achieves fairness post-hoc by linearly transforming the predictions of the (unfair) base predictor with a "fairness risk" according to a weighted combination of the (predicted) group memberships. It yields the Bayes optimal fair classifier if the base predictors being post-processed are Bayes optimal, otherwise, the resulting classifier may not be optimal, but fairness is guaranteed as long as the group membership predictor is multicalibrated. The parameters of the post-processing can be efficiently computed and estimated from solving an empirical linear program. Empirical evaluations demonstrate the advantage of our algorithm in the high fairness regime compared to existing post-processing and in-processing fair classification algorithms.