Fairness-aware Regression Robust to Adversarial Attacks

TL;DR

This paper introduces a novel fair regression model that is robust against adversarial attacks, optimizing performance and fairness even when datasets are intentionally corrupted.

Contribution

It formulates and solves a nonsmooth minimax problem to develop the first adversarially robust fair regression model.

Findings

Robust models outperform others on poisoned datasets.

Enhanced fairness and accuracy under adversarial conditions.

Effective in both synthetic and real-world data scenarios.

Abstract

In this paper, we take a first step towards answering the question of how to design fair machine learning algorithms that are robust to adversarial attacks. Using a minimax framework, we aim to design an adversarially robust fair regression model that achieves optimal performance in the presence of an attacker who is able to add a carefully designed adversarial data point to the dataset or perform a rank-one attack on the dataset. By solving the proposed nonsmooth nonconvex-nonconcave minimax problem, the optimal adversary as well as the robust fairness-aware regression model are obtained. For both synthetic data and real-world datasets, numerical results illustrate that the proposed adversarially robust fair models have better performance on poisoned datasets than other fair machine learning models in both prediction accuracy and group-based fairness measure.