Repairing without Retraining: Avoiding Disparate Impact with Counterfactual Distributions

TL;DR

This paper presents a method to reduce fairness disparities in machine learning models by perturbing input distributions for disadvantaged groups, avoiding retraining and maintaining accuracy.

Contribution

It introduces a novel approach to mitigate disparate impact by learning and applying counterfactual input distributions without retraining the classifier.

Findings

Effective reduction of disparate impact in real datasets

Maintains model accuracy while improving fairness

Provides a practical algorithm for distribution perturbation

Abstract

When the performance of a machine learning model varies over groups defined by sensitive attributes (e.g., gender or ethnicity), the performance disparity can be expressed in terms of the probability distributions of the input and output variables over each group. In this paper, we exploit this fact to reduce the disparate impact of a fixed classification model over a population of interest. Given a black-box classifier, we aim to eliminate the performance gap by perturbing the distribution of input variables for the disadvantaged group. We refer to the perturbed distribution as a counterfactual distribution, and characterize its properties for common fairness criteria. We introduce a descent algorithm to learn a counterfactual distribution from data. We then discuss how the estimated distribution can be used to build a data preprocessor that can reduce disparate impact without training a new model. We validate our approach through experiments on real-world datasets, showing that it can repair different forms of disparity without a significant drop in accuracy.