发表机构
Korea Advanced Institute of Science and Technology (KAIST)（韩国科学技术院（KAIST））
机构由 AI 辅助整理，请以论文原文为准。AI 中文总结
针对删除阶段无法获取保留图像的多模态大语言模型身份遗忘场景，提出两阶段方法AIM，通过通用视觉提示锚定遗忘目标并结合Fisher约束，实现身份遗忘的同时保留非删除身份、先验知识与视觉感知。
AI 中文摘要
多模态大语言模型（MLLMs）会在微调数据中记住与特定人物身份相关的事实，当该人物要求删除这些信息时会产生隐私风险。现有的MLLM遗忘方法通常假设在删除阶段可获取保留图像或真实答案，但这在许多实际场景中并不现实。我们研究了删除阶段无法获取保留图像时的身份遗忘问题。我们的分析表明，身份和视觉感知问题在微调后的隐藏状态中占据不同区域，且组织方式不同：身份问题按人物聚类，而感知问题按问题类型聚类。这表明可以在不抹去一般视觉感知的情况下抑制身份知识。基于这一观察，我们提出AIM，这是一种两阶段方法，它通过通用视觉提示锚定身份遗忘目标，然后在基于Fisher的约束下将视觉编码器与该目标匹配。大量实验表明，AIM在实现有竞争力的身份遗忘的同时，还能保留未删除的身份、先验知识以及相同图像上的视觉感知。
英文摘要
Multimodal large language models (MLLMs) can memorize identity-specific facts about people in their fine-tuning data, creating privacy risks when a person requests deletion. Existing MLLM unlearning methods often assume access to retain images or ground-truth answers during deletion, which is unrealistic in many practical scenarios. We study identity unlearning when retain images are unavailable at deletion time. Our analysis shows that identity and visual-perception questions occupy distinct regions in fine-tuned hidden states and are organized differently: identity questions cluster by person, whereas perception questions cluster by question type. This suggests that identity knowledge can be suppressed without erasing general visual perception. Building on this observation, we propose AIM, a two-stage method that anchors an identity-forgetting target with a universal visual prompt and then matches the vision encoder to that target under a Fisher-based constraint. Extensive experiments show that AIM achieves competitive identity forgetting while preserving non-deleted identities, prior knowledge, and visual perception on the same images.
CommentsAccepted to Findings of the Association for Computational Linguistics: EMNLP 2026
