A Novel Regularization Approach to Fair ML

2022/08/13 by Norman Matloff, Matloff, Norman, Wenxi Zhang +1 · 1 citation
Social Sciences · #Ethics and Social Impacts of AI #FOS: Computer and information sciences #Machine Learning (cs.LG)

paper · pdf · doi:10.48550/arxiv.2208.06557

openalex publication_date 2022/08/13 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28

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.

Cited by

• Bias Begins with Data: The FairGround Corpus for Robust and Reproducible Research on Algorithmic Fairness

Related

• FairPilot: An Explorative System for Hyperparameter Tuning through the Lens of Fairness
• Predicting Fairness of ML Software Configurations
• Fair Meta-Learning: Learning How to Learn Fairly
• Fairness Warnings and Fair-MAML: Learning Fairly with Minimal Data
• fairret: a Framework for Differentiable Fairness Regularization Terms