Two Simple Ways to Learn Individual Fairness Metrics from Data

TL;DR

This paper introduces two straightforward methods to learn individual fairness metrics from data, improving fairness in machine learning tasks and providing theoretical guarantees for their effectiveness.

Contribution

It proposes novel, simple approaches to learn fair metrics from data, addressing the lack of widely accepted metrics for individual fairness in ML.

Findings

Fair training with learned metrics improves fairness on biased ML tasks

Methods are effective across various data types

Theoretical guarantees support the approaches' statistical performance

Abstract

Individual fairness is an intuitive definition of algorithmic fairness that addresses some of the drawbacks of group fairness. Despite its benefits, it depends on a task specific fair metric that encodes our intuition of what is fair and unfair for the ML task at hand, and the lack of a widely accepted fair metric for many ML tasks is the main barrier to broader adoption of individual fairness. In this paper, we present two simple ways to learn fair metrics from a variety of data types. We show empirically that fair training with the learned metrics leads to improved fairness on three machine learning tasks susceptible to gender and racial biases. We also provide theoretical guarantees on the statistical performance of both approaches.