Marrying Fairness and Explainability in Supervised Learning. (arXiv:2204.02947v1 [cs.LG])

Machine learning algorithms that aid human decision-making may inadvertently
discriminate against certain protected groups. We formalize direct
discrimination as a direct causal effect of the protected attributes on the
decisions, while induced discrimination as a change in the causal influence of
non-protected features associated with the protected attributes. The
measurements of marginal direct effect (MDE) and SHapley Additive exPlanations
(SHAP) reveal that state-of-the-art fair learning methods can induce
discrimination via association or reverse discrimination in synthetic and
real-world datasets. To …

arxiv explainability fairness learning supervised learning