持续机器遗忘中的可塑性崩溃

中文摘要 AI 辅助

机器遗忘使深度神经网络能够根据隐私和监管要求选择性去除特定数据的影响。现有研究大多关注单次遗忘，而现实系统需应对持续遗忘，即多个遗忘请求随时间序列发生。本研究识别出该场景的一个根本局限：可塑性崩溃，即模型有效遗忘能力的逐步退化。通过对持续遗忘动力学的理论分析，我们发现持续遗忘操作会在参数空间中累积几何约束，形成饱和子空间，限制后续更新。该结构效应引发两种不同失效模式：（1）正向失效——后续任务的遗忘质量下降；（2）反向失效——自发重新记忆已遗忘信息。在图像分类任务中，对多种架构、数据集和方法开展的大量实验证实，可塑性崩溃并非特定实现的人为产物，而是持续遗忘固有的普遍现象。我们的发现揭示了机器遗忘系统长期可靠性的关键障碍，并推动了保可塑性遗忘算法的发展。代码可在该httpsURL获取。

英文摘要

Machine unlearning enables deep neural networks to selectively remove the influence of specific data in response to privacy and regulatory requirements. While prior work largely studies single-shot unlearning, real-world systems must accommodate continual unlearning, where multiple unlearning requests occur sequentially over time. In this work, we identify a fundamental limitation of this setting: plasticity collapse, a progressive breakdown in a model's ability to effectively forget. Through theoretical analysis of continual unlearning dynamics, we show that continual unlearning operations accumulate geometric constraints in parameter space, leading to saturated subspaces that restrict future updates. This structural effect induces two distinct failure modes: (1) Forward failure -- diminishing forgetting quality for subsequent tasks, and (2) Backward failure -- spontaneous re-memorization of previously forgotten information. Extensive experiments across multiple architectures, datasets, and methods in image classification confirm that plasticity collapse is not an artifact of specific implementations, but a pervasive phenomenon inherent to continual unlearning. Our findings reveal a critical barrier to the long-term reliability of machine unlearning systems and motivate the development of plasticity-preserving unlearning algorithms. Our code is available at https://github.com/TIML-Group/Continual-Machine-Unlearning-Plasticity-Collapse

发表机构

• Illinois Institute of Technology（伊利诺伊理工学院）
• Amazon（亚马逊公司）
• Northwestern University（西北大学）
• University of Michigan（密歇根大学）

机构由 AI 辅助整理，请以论文原文为准。