Understanding and Mitigating Accuracy Disparity in Regression

TL;DR

This paper investigates the causes of accuracy disparity in regression models across demographic groups, introduces an error decomposition theorem, and proposes algorithms to mitigate this disparity while preserving model accuracy.

Contribution

It presents a novel error decomposition framework for understanding accuracy disparity and introduces new algorithms for reducing disparity in regression tasks.

Findings

The error decomposition explains the sources of accuracy disparity.

Proposed algorithms effectively reduce disparity in experiments.

Algorithms maintain predictive power while mitigating disparity.

Abstract

With the widespread deployment of large-scale prediction systems in high-stakes domains, e.g., face recognition, criminal justice, etc., disparity in prediction accuracy between different demographic subgroups has called for fundamental understanding on the source of such disparity and algorithmic intervention to mitigate it. In this paper, we study the accuracy disparity problem in regression. To begin with, we first propose an error decomposition theorem, which decomposes the accuracy disparity into the distance between marginal label distributions and the distance between conditional representations, to help explain why such accuracy disparity appears in practice. Motivated by this error decomposition and the general idea of distribution alignment with statistical distances, we then propose an algorithm to reduce this disparity, and analyze its game-theoretic optima of the proposed objective functions. To corroborate our theoretical findings, we also conduct experiments on five benchmark datasets. The experimental results suggest that our proposed algorithms can effectively mitigate accuracy disparity while maintaining the predictive power of the regression models.