Individual Fairness In Strategic Classification

TL;DR

This paper explores individual fairness in strategic classification, demonstrating the limitations of deterministic thresholds and proposing a randomized classifier approach that improves fairness without sacrificing accuracy.

Contribution

It introduces conditions for achieving individual fairness with randomized classifiers and formulates an optimization method to find such classifiers, extending to group fairness.

Findings

Deterministic thresholds violate individual fairness in strategic settings.

A linear programming approach finds optimal, fair randomized classifiers.

Experimental results show improved fairness-accuracy trade-offs.

Abstract

Strategic classification, where individuals modify their features to influence machine learning (ML) decisions, presents critical fairness challenges. While group fairness in this setting has been widely studied, individual fairness remains underexplored. We analyze threshold-based classifiers and prove that deterministic thresholds violate individual fairness. Then, we investigate the possibility of using a randomized classifier to achieve individual fairness. We introduce conditions under which a randomized classifier ensures individual fairness and leverage these conditions to find an optimal and individually fair randomized classifier through a linear programming problem. Additionally, we demonstrate that our approach can be extended to group fairness notions. Experiments on real-world datasets confirm that our method effectively mitigates unfairness and improves the fairness-accuracy trade-off.