arXiv 2601.226012026-10-06cs.LG

\textsc{Lethe}: 用于联邦遗忘中持久知识擦除的原则性双流更新

Lethe: Principled Dual-Stream Update for Persistent Knowledge Erasure in Federated Unlearning

Department of Computer Science, College of Information Science and Technology, Jinan University（计算机科学系，信息科学与技术学院，暨南大学）；WMG, University of Warwick（沃森盖尔学院，沃里克大学）；School of Information and Communications Engineering, Xi’an Jiaotong University（信息与通信工程学院，西安交通大学）；Department of Computer Science, University of Warwick（计算机科学系，沃里克大学）；School of Data Science and Engineering, East China Normal University（数据科学与工程学院，华东师范大学）；School of Computer Science and Engineering, South China University of Technology（计算机科学与工程学院，华南理工大学）

Wentai Wu, Hanwei Tan, Yijun Quan, Haixia Peng, Ligang He, Bin Yang, C. L. Philip Chen

阅读 PDF

arXiv 页面 ↗HTML 原文 ↗

AI总结：

针对联邦遗忘后继续训练导致已遗忘知识重新浮现的问题，提出Lethe方法，通过遗忘流和保留流的反对齐更新实现持久知识擦除。

摘要：

联邦遗忘（FU）旨在从全局模型中擦除知识。现有研究通常假设遗忘后联邦协作终止，忽略了在删除请求完成后剩余客户端继续训练的实际部署场景。在这项工作中，我们识别出一个关键失败模式，称为知识重新浮现，揭示了仅对保留数据进行持续训练可以在几轮内重新激活已遗忘的知识。实验表明，许多最先进的FU方法容易发生知识重新浮现。我们随后提出Lethe，一种用于联邦设置中持久知识擦除的新型遗忘方法。在每次迭代中，Lethe操作来自遗忘客户端的遗忘流和来自保留客户端的保留流。它将遗忘更新重定向到两个流反对齐的区域，阻止保留数据训练移回遗忘知识。因此，Lethe在后续联邦训练期间确保更强的遗忘持久性。跨不同模型、数据集和遗忘级别的广泛实验验证了Lethe以统一方式支持CV和NLP任务中的所有遗忘级别，即使在极长后续训练时间后，大多数情况下也持续显示出低于1%的低重新浮现率。

英文摘要：

Federated unlearning (FU) aims to erase knowledge from a global model. Existing studies commonly assume that federated collaboration terminates after unlearning, overlooking a deployment-realistic scenario where training continues on the remaining clients after deletion requests are fulfilled. In this work, we identify a critical failure mode, termed knowledge resurfacing, revealing that continued training on retained data alone can reactivate unlearned knowledge in a few rounds. Empirically, we demonstrate that many state-of-the-art FU methods are prone to knowledge resurfacing. We then propose Lethe, a novel unlearning method for persistent knowledge erasure in federated settings. In each iteration, Lethe operates on a forget stream from the unlearning client and a retain stream from the retained clients. It redirects unlearning updates toward a region where the two streams are anti-aligned, discouraging retained-data training from moving back toward the forgotten knowledge. Consequently, Lethe ensures stronger unlearning persistence during subsequent federated training. Extensive experiments across diverse models, datasets, and unlearning levels validate that Lethe supports all levels of unlearning in a unified manner across both CV and NLP tasks, demonstrating consistently low RR, below 1% in most cases, even after an extremely long horizon of follow-up training.

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。