Flexibly Fair Representation Learning by Disentanglement

TL;DR

This paper introduces a method for learning flexible, fair representations that can be adjusted at test time to satisfy subgroup fairness criteria across multiple sensitive attributes without needing sensitive attribute information during inference.

Contribution

The paper proposes a novel disentangled representation learning algorithm that produces compact, useful, and adaptable fair representations for multiple subgroup fairness scenarios.

Findings

Encoder enables adaptation to various fair classification tasks

Representation does not require sensitive attributes at inference

Method achieves flexible subgroup fairness across multiple attributes

Abstract

We consider the problem of learning representations that achieve group and subgroup fairness with respect to multiple sensitive attributes. Taking inspiration from the disentangled representation learning literature, we propose an algorithm for learning compact representations of datasets that are useful for reconstruction and prediction, but are also \emph{flexibly fair}, meaning they can be easily modified at test time to achieve subgroup demographic parity with respect to multiple sensitive attributes and their conjunctions. We show empirically that the resulting encoder---which does not require the sensitive attributes for inference---enables the adaptation of a single representation to a variety of fair classification tasks with new target labels and subgroup definitions.