The Disparate Benefits of Deep Ensembles

TL;DR

Deep Ensembles improve predictive performance but can unevenly benefit different social groups, affecting fairness metrics, and can be mitigated with post-processing methods.

Contribution

This work uncovers the disparate benefits effect of Deep Ensembles on fairness and proposes mitigation via Hardt post-processing.

Findings

Deep Ensembles unevenly favor different groups across fairness metrics.

Per-group diversity in ensemble predictions explains the disparate benefits.

Hardt post-processing effectively mitigates the fairness disparities.

Abstract

Ensembles of Deep Neural Networks, Deep Ensembles, are widely used as a simple way to boost predictive performance. However, their impact on algorithmic fairness is not well understood yet. Algorithmic fairness examines how a model's performance varies across socially relevant groups defined by protected attributes such as age, gender, or race. In this work, we explore the interplay between the performance gains from Deep Ensembles and fairness. Our analysis reveals that they unevenly favor different groups, a phenomenon that we term the disparate benefits effect. We empirically investigate this effect using popular facial analysis and medical imaging datasets with protected group attributes and find that it affects multiple established group fairness metrics, including statistical parity and equal opportunity. Furthermore, we identify that the per-group differences in predictive diversity of ensemble members can explain this effect. Finally, we demonstrate that the classical Hardt post-processing method is particularly effective at mitigating the disparate benefits effect of Deep Ensembles by leveraging their better-calibrated predictive distributions.