TARA: Training and Representation Alteration for AI Fairness and Domain Generalization

TL;DR

TARA introduces a dual approach combining representation learning alteration and intelligent data augmentation to effectively mitigate bias in AI models, improving fairness and accuracy across datasets.

Contribution

The paper presents a novel dual strategy for AI fairness that integrates adversarial independence and generative augmentation, advancing bias mitigation techniques.

Findings

TARA significantly reduces bias and improves accuracy in image analytics.

It outperforms existing debiasing methods with higher accuracy and lower bias gaps.

Proposes new metrics for better assessment of debiasing effectiveness.

Abstract

We propose a novel method for enforcing AI fairness with respect to protected or sensitive factors. This method uses a dual strategy performing training and representation alteration (TARA) for the mitigation of prominent causes of AI bias by including: a) the use of representation learning alteration via adversarial independence to suppress the bias-inducing dependence of the data representation from protected factors; and b) training set alteration via intelligent augmentation to address bias-causing data imbalance, by using generative models that allow the fine control of sensitive factors related to underrepresented populations via domain adaptation and latent space manipulation. When testing our methods on image analytics, experiments demonstrate that TARA significantly or fully debiases baseline models while outperforming competing debiasing methods that have the same amount of information, e.g., with (% overall accuracy, % accuracy gap) = (78.8, 0.5) vs. the baseline method's score of (71.8, 10.5) for EyePACS, and (73.7, 11.8) vs. (69.1, 21.7) for CelebA. Furthermore, recognizing certain limitations in current metrics used for assessing debiasing performance, we propose novel conjunctive debiasing metrics. Our experiments also demonstrate the ability of these novel metrics in assessing the Pareto efficiency of the proposed methods.