MABR: Multilayer Adversarial Bias Removal Without Prior Bias Knowledge

TL;DR

This paper presents a novel adversarial training method that detects and mitigates social biases in models without needing prior bias or demographic information, improving fairness in classification tasks.

Contribution

Introduces a bias removal approach that operates without prior bias knowledge by using auxiliary models at multiple feature levels during training.

Findings

Effectively reduces racial and gender biases in sentiment and occupation classification.

Outperforms or matches existing bias mitigation methods that require demographic data.

Operates without demographic annotations, broadening applicability.

Abstract

Models trained on real-world data often mirror and exacerbate existing social biases. Traditional methods for mitigating these biases typically require prior knowledge of the specific biases to be addressed, such as gender or racial biases, and the social groups associated with each instance. In this paper, we introduce a novel adversarial training strategy that operates independently of prior bias-type knowledge and protected attribute labels. Our approach proactively identifies biases during model training by utilizing auxiliary models, which are trained concurrently by predicting the performance of the main model without relying on task labels. Additionally, we implement these auxiliary models at various levels of the feature maps of the main model, enabling the detection of a broader and more nuanced range of bias features. Through experiments on racial and gender biases in sentiment and occupation classification tasks, our method effectively reduces social biases without the need for demographic annotations. Moreover, our approach not only matches but often surpasses the efficacy of methods that require detailed demographic insights, marking a significant advancement in bias mitigation techniques.