Fairness-aware Vision Transformer via Debiased Self-Attention

TL;DR

This paper introduces Debiased Self-Attention (DSA), a novel fairness-aware framework for Vision Transformers that mitigates bias by eliminating spurious features while preserving predictive accuracy.

Contribution

The paper proposes DSA, a new fairness method tailored for ViT that effectively reduces bias by masking spurious features using adversarial examples and attention regularization.

Findings

DSA improves fairness across multiple vision tasks.

DSA maintains high target prediction accuracy.

The framework outperforms existing fairness methods on ViT.

Abstract

Vision Transformer (ViT) has recently gained significant attention in solving computer vision (CV) problems due to its capability of extracting informative features and modeling long-range dependencies through the attention mechanism. Whereas recent works have explored the trustworthiness of ViT, including its robustness and explainability, the issue of fairness has not yet been adequately addressed. We establish that the existing fairness-aware algorithms designed for CNNs do not perform well on ViT, which highlights the need to develop our novel framework via Debiased Self-Attention (DSA). DSA is a fairness-through-blindness approach that enforces ViT to eliminate spurious features correlated with the sensitive label for bias mitigation and simultaneously retain real features for target prediction. Notably, DSA leverages adversarial examples to locate and mask the spurious features in the input image patches with an additional attention weights alignment regularizer in the training objective to encourage learning real features for target prediction. Importantly, our DSA framework leads to improved fairness guarantees over prior works on multiple prediction tasks without compromising target prediction performance. Code is available at \href{https://github.com/qiangyao1988/DSA}{https://github.com/qiangyao1988/DSA}.