发表机构
The University of Osaka（大阪大学）
机构由 AI 辅助整理，请以论文原文为准。AI 中文总结
本研究提出VL-BICLE框架，系统评估六个大型视觉-语言模型在上下文学习中的性别偏见，发现性别化演示通过跨性别机制影响偏见，并用稳定扩散合成图像替换真实图像以有效去偏。
AI 中文摘要
上下文学习（ICL）使大型视觉-语言模型（LVLMs）能够通过遵循上下文示例中的模式来执行任务，但其放大社会偏见的潜力仍未得到充分探索。我们通过VL-BICLE系统性地研究了ICL如何影响LVLMs中的性别偏见，VL-BICLE是一个包含六种ICL设置、三个任务和四个数据集的评估框架。我们在六个LVLMs上的实验揭示，性别化的ICL演示起到了一种定向力的作用，通过一种跨性别机制将模型偏见转向所演示的性别，该机制不成比例地降低了相反性别的性能。这种效应出现在图像字幕生成和代词预测中，但未出现在视觉问答中，表明性别化的ICL仅在任务输出涉及性别化语言时才影响偏见。基于相似性的检索方法继承了训练池中的性别不平衡，并未提供去偏优势，而标准质量指标对这些偏见转变仍然视而不见。为缓解这种偏见，我们将真实的上下文图像替换为来自稳定扩散模型的合成图像，同时保持字幕不变。这种简单的干预减少了性别偏见，且未降低字幕质量。
英文摘要
In-context learning (ICL) enables large vision-language models (LVLMs) to perform tasks by following patterns from in-context examples, yet its potential to amplify societal biases remains underexplored. We systematically investigate how ICL influences gender bias in LVLMs through VL-BICLE, an evaluation framework comprising six ICL settings, three tasks, and four datasets. Our experiments on six LVLMs reveal that gendered ICL demonstrations act as a directional force, shifting model bias toward the demonstrated gender through a cross-gender mechanism that disproportionately degrades performance on the opposite gender. This effect appears in image captioning and pronoun prediction but not in visual question answering, indicating that gendered ICL influences bias only when the task output involves gendered language. Similarity-based retrieval methods inherit the training pool's gender imbalance and offer no debiasing advantage, while standard quality metrics remain blind to these bias shifts. To mitigate this bias, we replace real in-context images with synthetic ones from stable diffusion models while keeping captions unchanged. This simple intervention reduces gender bias without degrading caption quality.
CommentsECCV 2026
