A Novel Regularization Approach to Fair ML

Abstract

A number of methods have been introduced for the fair ML issue, most of them complex and many of them very specific to the underlying ML moethodology. Here we introduce a new approach that is simple, easily explained, and potentially applicable to a number of standard ML algorithms. Explicitly Deweighted Features (EDF) reduces the impact of each feature among the proxies of sensitive variables, allowing a different amount of deweighting applied to each such feature. The user specifies the deweighting hyperparameters, to achieve a given point in the Utility/Fairness tradeoff spectrum. We also introduce a new, simple criterion for evaluating the degree of protection afforded by any fair ML method. * NM is professor of computer science, UCD, and has served as an expert witness in various litigation cases involving discrimination.† WZ performed this work as an undergraduate at UCD.

Using Demographic Data as Predictor Variables: a Questionable Choice

“…To those with concerns about predicting using demographic variables, proxy variables raise similar concerns (e.g. Matloff & Zhang, 2022), and add the possible issue of being hard to identify a priori (Johnson, 2021).…”