Technical Challenges for Training Fair Neural Networks

TL;DR

This paper investigates the challenges of training fair neural networks, revealing overfitting to fairness objectives and unintended consequences in high-stakes applications like facial recognition and medical diagnosis.

Contribution

It highlights the limitations of current fairness methods in deep neural networks and provides empirical evidence of overfitting issues across multiple datasets.

Findings

Neural networks tend to overfit fairness constraints.

Unintended consequences arise from current fairness methods.

Overfitting impacts high-stakes applications like facial recognition and medical diagnosis.

Abstract

As machine learning algorithms have been widely deployed across applications, many concerns have been raised over the fairness of their predictions, especially in high stakes settings (such as facial recognition and medical imaging). To respond to these concerns, the community has proposed and formalized various notions of fairness as well as methods for rectifying unfair behavior. While fairness constraints have been studied extensively for classical models, the effectiveness of methods for imposing fairness on deep neural networks is unclear. In this paper, we observe that these large models overfit to fairness objectives, and produce a range of unintended and undesirable consequences. We conduct our experiments on both facial recognition and automated medical diagnosis datasets using state-of-the-art architectures.