RobustDebias: Debiasing Language Models using Distributionally Robust Optimization

TL;DR

RobustDebias introduces a novel fine-tuning method using Distributionally Robust Optimization to effectively reduce biases in pretrained language models like BERT without sacrificing performance.

Contribution

The paper proposes RobustDebias, a new fine-tuning approach that applies Distributionally Robust Optimization to mitigate biases across demographics during language model training.

Findings

Significant bias reduction across multiple demographics.

Minimal impact on language model performance.

Generalizes to various datasets and tasks.

Abstract

Pretrained language models have been shown to exhibit biases and social stereotypes. Prior work on debiasing these models has largely focused on modifying embedding spaces during pretraining, which is not scalable for large models. Fine-tuning pretrained models on task-specific datasets can both degrade model performance and amplify biases present in the fine-tuning data. We address bias amplification during fine-tuning rather than costly pretraining, focusing on BERT models due to their widespread use in language understanding tasks. While Empirical Risk Minimization effectively optimizes downstream performance, it often amplifies social biases during fine-tuning. To counter this, we propose \textit{RobustDebias}, a novel mechanism which adapts Distributionally Robust Optimization (DRO) to debias language models during fine-tuning. Our approach debiases models across multiple demographics during MLM fine-tuning and generalizes to any dataset or task. Extensive experiments on various language models show significant bias mitigation with minimal performance impact.