GESD: Beyond Outcome-Oriented Fairness

TL;DR

The paper introduces GESD, a procedural fairness metric that assesses disparities in model explanation stability across groups, and integrates it into a multi-objective framework to improve fairness and utility.

Contribution

It proposes GESD, a novel explanation-based fairness metric, and a combined optimization framework FEU that enhances fairness and utility in machine learning models.

Findings

GESD effectively captures group-wise explanation disparities.

FEU improves both fairness and utility in benchmark datasets.

The approach bridges outcome-based and explanation-based fairness.

Abstract

Machine learning (ML) algorithms are increasingly deployed in high-stakes decision-making domains such as loan approvals, hiring, and recidivism predictions. While existing fairness metrics (e.g., statistical parity, equal opportunity) effectively quantify outcome-oriented disparities, they offer limited insight into the procedure or explanation behind biased decisions. To address this gap, we propose Group-level Explanation Stability Disparity (GESD), a \textit{procedural-oriented} fairness metric that measures disparities in the stability, robustness, and sensitivity of model explanations across different subgroups in a protected category. %GESD is explainer-agnostic, model-agnostic, and extends the scope of fairness analyses to the level of explainability. We further integrate GESD into a multi-objective optimization framework that jointly optimizes for utility, outcome-based fairness, and explanation-based fairness called FEU (Fairness--Explainability--Utility). Empirical results on multiple benchmark datasets show that GESD effectively captures group-wise discrepancies in explanation quality, and that FEU improves both utility and fairness over state-of-the-art methods. By bridging outcome-based and explanation-based fairness, GESD offers a comprehensive tool for diagnosing and mitigating bias in predictive modeling. Our code and datasets are available on GitHub {\hyperlink{https://github.com/horlahsunbo/GESD}{https://github.com/horlahsunbo/GESD}}