Mitigating Unfairness via Evolutionary Multiobjective Ensemble Learning

Abstract

In the literature of mitigating unfairness in machine learning (ML), many fairness measures are designed to evaluate predictions of learning models and also utilized to guide the training of fair models. It has been theoretically and empirically shown that there exist conflicts and inconsistencies among accuracy and multiple fairness measures. Optimizing one or several fairness measures may sacrifice or deteriorate other measures. Two key questions should be considered: 1) how to simultaneously optimize accuracy and multiple fairness measures and 2) how to optimize all the considered fairness measures more effectively. In this article, we view the mitigating unfairness problem as a multiobjective learning problem, considering the conflicts among fairness measures. A multiobjective evolutionary learning framework is used to simultaneously optimize several metrics (including accuracy and multiple fairness measures) of ML models. Then, ensembles are constructed based on the learning models in order to automatically balance different metrics. Empirical results on eight well-known datasets demonstrate that compared with the state-of-the-art approaches for mitigating unfairness, our proposed algorithm can provide decision makers with better tradeoffs among accuracy and multiple fairness metrics. Furthermore, the high-quality models generated by the framework can be used to construct an ensemble to automatically achieve a better tradeoff among all the considered fairness metrics than other ensemble methods.