arXiv 2510.042172026-02-02cs.LGcs.AI

MLLMEraser: 通过激活引导在多模态大语言模型中实现测试时遗忘

MLLMEraser: Achieving Test-Time Unlearning in Multimodal Large Language Models through Activation Steering

University of Science and Technology of China（中国科学技术大学）；National University of Singapore（新加坡国立大学）；The Hong Kong Polytechnic University（香港理工大学）；Huawei（华为）

Chenlu Ding, Jiancan Wu, Leheng Sheng, Fan Zhang, Yancheng Yuan, Xiang Wang, Xiangnan He

阅读 PDF

arXiv 页面 ↗HTML 原文 ↗

AI总结：

MLLMEraser通过激活引导实现多模态大语言模型的测试时遗忘，无需训练且有效降低计算成本，同时保持保留知识的效用。

摘要：

多模态大语言模型（MLLMs）在视觉-语言任务中展现了显著的能力，但其大规模部署引发了对记忆中的私人数据、过时知识和有害内容的紧迫担忧。现有的MLLMs遗忘方法通常采用基于训练的策略，如梯度上升或偏好优化，但这些方法计算成本高、不可逆且常扭曲保留的知识。在本工作中，我们提出MLLMEraser，一种输入感知、无需训练的测试时遗忘框架。我们的方法利用激活引导，实现动态知识擦除而无需参数更新。具体而言，我们通过对比对抗扰动的知识回忆图像-文本对与知识擦除对应物，构建多模态擦除方向，捕捉文本和视觉差异。为防止不必要的干扰，我们进一步设计了输入感知的引导机制，适应性地确定何时以及如何应用擦除方向，在保留知识的同时强制遗忘指定内容。在LLaVA-1.5和Qwen-2.5-VL上的实验表明，MLLMEraser在性能上始终优于最先进的MLLMs遗忘基线，实现了更强的遗忘性能，计算成本更低且对保留知识的效用影响较小。

英文摘要：

Multimodal large language models (MLLMs) have demonstrated remarkable capabilities across vision-language tasks, yet their large-scale deployment raises pressing concerns about memorized private data, outdated knowledge, and harmful content. Existing unlearning approaches for MLLMs typically adapt training-based strategies such as gradient ascent or preference optimization, but these methods are computationally expensive, irreversible, and often distort retained knowledge. In this work, we propose MLLMEraser, an input-aware, training-free framework for test-time unlearning. Our approach leverages activation steering to enable dynamic knowledge erasure without parameter updates. Specifically, we construct a multimodal erasure direction by contrasting adversarially perturbed, knowledge-recall image-text pairs with knowledge-erasure counterparts, capturing both textual and visual discrepancies. To prevent unnecessary interference, we further design an input-aware steering mechanism that adaptively determines when and how the erasure direction should be applied, preserving utility on retained knowledge while enforcing forgetting on designated content. Experiments on LLaVA-1.5 and Qwen-2.5-VL demonstrate that MLLMEraser consistently outperforms state-of-the-art MLLM unlearning baselines, achieving stronger forgetting performance with lower computational cost and minimal utility degradation.

机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。