AIM: Anchor Identity Features, Then Match for Multimodal Large Language Model Unlearning

2026-08-28

AI Summary

研究解决了在无法获取保留图像时，多模态大语言模型中的身份信息删除问题，提出AIM方法通过锚定身份遗忘目标并匹配视觉编码器来实现。

Abstract

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.

Problem

• Multimodal Large Language Models
• Identity Unlearning
• Privacy Risks

Innovation

• Identity Unlearning
• Multimodal Large Language Models
• Visual Perception Preservation
• Fisher-based Constraint