AI 中文总结
针对现有MLLM遗忘方法无法应对持续场景的问题，本文提出MCU方法，通过动态合并遗忘适配器缓解跨任务依赖的干扰，在多个基准测试中实现了更优的遗忘效果与知识保留。
AI 中文摘要
多模态大语言模型（MLLM）遗忘方法旨在从预训练模型中移除私人、敏感或专有信息。然而，现有大多数MLLM遗忘方法针对单次请求设计，无法充分应对持续场景，因为反复应用单次操作会导致效用累积下降、遗忘反弹及保留知识漂移。本文提出持续遗忘合并方法（MCU），该方法在接收每一项新遗忘请求时，动态将多个单次遗忘适配器合并为统一适配器。通过留一法合并分析，本文发现这些遗忘适配器存在强跨任务依赖，这类依赖兼具双重影响：既可以促进跨任务遗忘迁移能力，也会引入严重干扰，降低遗忘有效性并损害保留知识。为应对这一挑战，MCU将适配器投影到共享表示空间，保留其主导方向，抑制过度集中的坐标，并重构跨任务依赖，以减轻干扰同时增强迁移能力。在ICU-Bench和MLLMU-Bench上的实验表明，MCU在实现卓越遗忘有效性的同时，能兼顾保留知识与通用多模态效用。
英文摘要
Multimodal large language model (MLLM) unlearning methods have been proposed to remove private, sensitive, or proprietary information from well-trained models. However, most existing MLLM unlearning methods are designed for one-shot requests and fail to adequately address continual scenarios, as repeatedly applying one-shot operations leads to cumulative utility degradation, unlearning rebound, and retention drift. We introduce Merging for Continual Unlearning (MCU), an approach that dynamically merges multiple one-shot unlearning adapters into a unified adapter upon receiving each new unlearning request.Through a leave-one-out merging analysis, we reveal that these unlearning adapters exhibit strong cross-task dependencies. Such dependencies have two contrasting effects: they can facilitate cross-task unlearning transferability, but they can also introduce severe interference that degrades unlearning effectiveness and compromises retained knowledge. To address this challenge, MCU projects the adapters into a shared representation space, preserves their dominant directions, suppresses over-concentrated coordinates, and reconfigures cross-task dependencies to mitigate interference while enhancing transferability. Experiments on ICU-Bench and MLLMU-Bench demonstrate that MCU achieves superior unlearning effectiveness while preserving both retained knowledge and general multimodal utility.
Comments17 pages, 5 figures
