Blind Justice: Fairness with Encrypted Sensitive Attributes

TL;DR

This paper introduces secure multi-party computation techniques to train and evaluate fair machine learning models using encrypted sensitive attributes, ensuring privacy and fairness simultaneously.

Contribution

It presents novel methods to perform fairness-related tasks on encrypted data, preventing sensitive attribute disclosure during model training and assessment.

Findings

Encrypted methods enable fair model training without revealing sensitive attributes

Outcome-based fairness can be verified securely

Models can be audited for fairness without compromising privacy

Abstract

Recent work has explored how to train machine learning models which do not discriminate against any subgroup of the population as determined by sensitive attributes such as gender or race. To avoid disparate treatment, sensitive attributes should not be considered. On the other hand, in order to avoid disparate impact, sensitive attributes must be examined, e.g., in order to learn a fair model, or to check if a given model is fair. We introduce methods from secure multi-party computation which allow us to avoid both. By encrypting sensitive attributes, we show how an outcome-based fair model may be learned, checked, or have its outputs verified and held to account, without users revealing their sensitive attributes.