FairSHAP: Preprocessing for Fairness Through Attribution-Based Data Augmentation

2025-05-16

AI Summary

Ensuring fairness in high-stakes machine learning applications demands interpretable and traceable preprocessing methods. Existing approaches lack transparent mechanisms to identify the root causes of unfairness—whether at the feature or sample level. This paper introduces FairSHAP, the first preprocessing framework that leverages Shapley value attribution to pinpoint fairness-critical samples and enable semantically aligned matching across sensitive groups. Its core innovation lies in quantifying sample-level discrimination risk via Shapley values, guiding low-perturbation, cross-group semantic matching. This simultaneously preserves data fidelity and model accuracy while jointly enhancing both individual and group fairness—specifically demographic parity and equal opportunity. Extensive experiments on diverse tabular datasets demonstrate that FairSHAP significantly improves fairness metrics, minimizes data perturbation, and—in some cases—even increases predictive accuracy.

Technology Category

Machine Learning: Ethics, Bias, and FairnessHumans and AI: Learning Human Values and PreferencesComputer Vision: Bias, Fairness & Privacy

Application Category

User Modeling, Personalization and Recommendation: Fairness-aware retrieval and rankingSearch and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for rankingSecurity and Privacy: Data transparency and provenance

Abstract

Ensuring fairness in machine learning models is critical, particularly in high-stakes domains where biased decisions can lead to serious societal consequences. Existing preprocessing approaches generally lack transparent mechanisms for identifying which features or instances are responsible for unfairness. This obscures the rationale behind data modifications. We introduce FairSHAP, a novel pre-processing framework that leverages Shapley value attribution to improve both individual and group fairness. FairSHAP identifies fairness-critical instances in the training data using an interpretable measure of feature importance, and systematically modifies them through instance-level matching across sensitive groups. This process reduces discriminative risk - an individual fairness metric - while preserving data integrity and model accuracy. We demonstrate that FairSHAP significantly improves demographic parity and equality of opportunity across diverse tabular datasets, achieving fairness gains with minimal data perturbation and, in some cases, improved predictive performance. As a model-agnostic and transparent method, FairSHAP integrates seamlessly into existing machine learning pipelines and provides actionable insights into the sources of bias.Our code is on https://github.com/youlei202/FairSHAP.

Problem

• Identifying features causing unfairness in ML models
• Improving fairness via interpretable data augmentation
• Reducing discriminative risk while preserving accuracy

Innovation

• Uses Shapley values for fairness attribution
• Modifies fairness-critical instances systematically
• Preserves data integrity and model accuracy