Loss-Aversively Fair Classification

TL;DR

This paper introduces loss-averse updates for fair classification, ensuring that algorithmic decisions improve outcomes relative to the current system while considering behavioral insights from prospect theory.

Contribution

It proposes a novel notion of fair updates based on loss aversion, with tractable proxy measures for training fair classifiers that account for behavioral perceptions.

Findings

Proxy measures effectively incorporate loss-averse fairness.

The approach improves fairness in both synthetic and real-world datasets.

It can be combined with existing fairness measures.

Abstract

The use of algorithmic (learning-based) decision making in scenarios that affect human lives has motivated a number of recent studies to investigate such decision making systems for potential unfairness, such as discrimination against subjects based on their sensitive features like gender or race. However, when judging the fairness of a newly designed decision making system, these studies have overlooked an important influence on people's perceptions of fairness, which is how the new algorithm changes the status quo, i.e., decisions of the existing decision making system. Motivated by extensive literature in behavioral economics and behavioral psychology (prospect theory), we propose a notion of fair updates that we refer to as loss-averse updates. Loss-averse updates constrain the updates to yield improved (more beneficial) outcomes to subjects compared to the status quo. We propose tractable proxy measures that would allow this notion to be incorporated in the training of a variety of linear and non-linear classifiers. We show how our proxy measures can be combined with existing measures for training nondiscriminatory classifiers. Our evaluation using synthetic and real-world datasets demonstrates that the proposed proxy measures are effective for their desired tasks.