DSO: Direct Steering Optimization for Bias Mitigation

TL;DR

This paper introduces DSO, a reinforcement learning-based method for inference-time bias mitigation in vision-language and language models, enabling controllable trade-offs between fairness and performance.

Contribution

DSO is the first to optimize steering transformations directly for bias mitigation, improving fairness-performance trade-offs in large models.

Findings

DSO outperforms existing methods in fairness-performance trade-offs.

DSO provides practitioners with inference-time control over bias and capabilities.

DSO achieves state-of-the-art results on both VLMs and LLMs.

Abstract

Generative models are often deployed to make decisions on behalf of users, such as vision-language models (VLMs) identifying which person in a room is a doctor to help visually impaired individuals. Yet, VLM decisions are influenced by the perceived demographic attributes of people in the input, which can lead to biased outcomes like failing to identify women as doctors. Moreover, when reducing bias leads to performance loss, users may have varying needs for balancing bias mitigation with overall model capabilities, highlighting the demand for methods that enable controllable bias reduction during inference. Activation steering is a popular approach for inference-time controllability that has shown potential in inducing safer behavior in large language models (LLMs). However, we observe that current steering methods struggle to correct biases, where equiprobable outcomes across demographic groups are required. To address this, we propose Direct Steering Optimization (DSO) which uses reinforcement learning to find linear transformations for steering activations, tailored to mitigate bias while maintaining control over model performance. We demonstrate that DSO achieves state-of-the-art trade-off between fairness and capabilities on both VLMs and LLMs, while offering practitioners inference-time control over the trade-off. Overall, our work highlights the benefit of designing steering strategies that are directly optimized to control model behavior, providing more effective bias intervention than methods that rely on pre-defined heuristics for controllability.