Fairness Through Computationally-Bounded Awareness

TL;DR

This paper introduces a new fairness concept called metric multifairness for classification, which ensures similar groups are treated similarly, even when the similarity metric is only partially known and queried a limited number of times.

Contribution

It proposes a novel fairness framework called metric multifairness that operates under limited access to the similarity metric and provides methods to achieve it.

Findings

Introduces metric multifairness as a new fairness criterion.

Develops algorithms to achieve metric multifairness with bounded metric queries.

Demonstrates that similar subpopulations are treated fairly under this framework.

Abstract

We study the problem of fair classification within the versatile framework of Dwork et al. [ITCS '12], which assumes the existence of a metric that measures similarity between pairs of individuals. Unlike earlier work, we do not assume that the entire metric is known to the learning algorithm; instead, the learner can query this arbitrary metric a bounded number of times. We propose a new notion of fairness called metric multifairness and show how to achieve this notion in our setting. Metric multifairness is parameterized by a similarity metric $d$ on pairs of individuals to classify and a rich collection ${\cal C}$ of (possibly overlapping) "comparison sets" over pairs of individuals. At a high level, metric multifairness guarantees that similar subpopulations are treated similarly, as long as these subpopulations are identified within the class ${\cal C}$.