Demographic Parity Tails for Regression

2026-04-02

AI Summary

This work addresses a key limitation in traditional regression fairness approaches, which often impose group fairness constraints over the entire outcome distribution at the expense of predictive accuracy, despite practical needs typically requiring fairness only in specific regions—particularly the tails. The paper proposes a novel group fairness framework for regression that uniquely focuses fairness constraints on the tail of the target distribution. Leveraging the geometric structure of optimal transport theory, the method constructs an interpretable and flexible constraint mechanism, efficiently solvable via modern optimization techniques. Theoretical analysis provides upper bounds on risk alongside formal fairness guarantees, while empirical results demonstrate that the approach significantly enhances group fairness in the tail regions without compromising overall prediction accuracy.

Abstract

Demographic parity (DP) is a widely studied fairness criterion in regression, enforcing independence between the predictions and sensitive attributes. However, constraining the entire distribution can degrade predictive accuracy and may be unnecessary for many applications, where fairness concerns are localized to specific regions of the distribution. To overcome this issue, we propose a new framework for regression under DP that focuses on the tails of target distribution across sensitive groups. Our methodology builds on optimal transport theory. By enforcing fairness constraints only over targeted regions of the distribution, our approach enables more nuanced and context-sensitive interventions. Leveraging recent advances, we develop an interpretable and flexible algorithm that leverages the geometric structure of optimal transport. We provide theoretical guarantees, including risk bounds and fairness properties, and validate the method through experiments in regression settings.

Problem

• Demographic Parity
• Regression
• Fairness
• Distribution Tails
• Sensitive Attributes

Innovation

• Demographic Parity
• Optimal Transport
• Fair Regression
• Distribution Tails
• Fairness-Aware Learning