Fair Representations by Compression

TL;DR

This paper introduces a novel unsupervised compression method that creates fair data representations by filtering out sensitive attribute information, balancing accuracy and fairness effectively.

Contribution

The paper presents FBC, a new unsupervised compression approach that achieves state-of-the-art fairness-accuracy trade-offs and allows explicit control over representation entropy.

Findings

FBC outperforms existing methods in fairness-accuracy trade-off.

Explicit entropy control enables smooth adjustments along rate-distortion and rate-fairness curves.

The approach effectively filters sensitive attribute information in compressed representations.

Abstract

Organizations that collect and sell data face increasing scrutiny for the discriminatory use of data. We propose a novel unsupervised approach to transform data into a compressed binary representation independent of sensitive attributes. We show that in an information bottleneck framework, a parsimonious representation should filter out information related to sensitive attributes if they are provided directly to the decoder. Empirical results show that the proposed method, \textbf{FBC}, achieves state-of-the-art accuracy-fairness trade-off. Explicit control of the entropy of the representation bit stream allows the user to move smoothly and simultaneously along both rate-distortion and rate-fairness curves. \end{abstract}