arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2307.15971cs.CRcs.AI
收藏

你可以对个性化联邦学习植入后门

You Can Backdoor Personalized Federated Learning

East China Normal University(华东师范大学);Ant Group(蚂蚁集团)

Tiandi Ye, Cen Chen, Yinggui Wang, Xiang Li, Ming Gao

AI总结:

本文针对参数解耦个性化联邦学习,提出仅毒化特征编码器并噪声多样化分类器的 BapFL 后门攻击,证明其可突破现有防御。

摘要:

现有研究主要集中在通用联邦学习场景中的后门攻击与防御,在该场景中所有客户端协作训练单个全局模型。Qin 等人(2023)最近开展的一项研究标志着对个性化联邦学习(pFL)场景中后门攻击的初步探索;在该场景中,每个客户端基于其本地数据构建个性化模型。值得注意的是,该研究表明,采用参数解耦的 pFL 方法能够显著增强对后门攻击的鲁棒性。然而,本文揭示,采用参数解耦的 pFL 方法仍然容易受到后门攻击。参数解耦 pFL 方法的抵抗力归因于恶意客户端与良性客户端之间异构的分类器。我们分析了导致异构分类器的两个直接原因:(1)客户端之间固有地存在数据异构性;(2)恶意客户端的投毒进一步加剧了数据异构性。为解决这些问题,我们提出了一种双管齐下的攻击方法 BapFL,它包含两种简单而有效的策略:(1)仅对特征编码器投毒,同时保持分类器固定;(2)通过引入噪声使分类器多样化,以模拟良性客户端的分类器。在三个基准数据集上、不同条件下进行的大量实验证明了我们所提攻击的有效性。此外,我们评估了六种广泛使用的防御方法的有效性,发现即使存在最佳防御 Multi-Krum,BapFL 仍构成重大威胁。我们希望激发针对 pFL 场景中攻击与防御策略的进一步研究。代码可在 https://github.com/BapFL/code 获取。

英文摘要:

Existing research primarily focuses on backdoor attacks and defenses within the generic federated learning scenario, where all clients collaborate to train a single global model. A recent study conducted by Qin et al. (2023) marks the initial exploration of backdoor attacks within the personalized federated learning (pFL) scenario, where each client constructs a personalized model based on its local data. Notably, the study demonstrates that pFL methods with \textit{parameter decoupling} can significantly enhance robustness against backdoor attacks. However, in this paper, we whistleblow that pFL methods with parameter decoupling are still vulnerable to backdoor attacks. The resistance of pFL methods with parameter decoupling is attributed to the heterogeneous classifiers between malicious clients and benign counterparts. We analyze two direct causes of the heterogeneous classifiers: (1) data heterogeneity inherently exists among clients and (2) poisoning by malicious clients further exacerbates the data heterogeneity. To address these issues, we propose a two-pronged attack method, BapFL, which comprises two simple yet effective strategies: (1) poisoning only the feature encoder while keeping the classifier fixed and (2) diversifying the classifier through noise introduction to simulate that of the benign clients. Extensive experiments on three benchmark datasets under varying conditions demonstrate the effectiveness of our proposed attack. Additionally, we evaluate the effectiveness of six widely used defense methods and find that BapFL still poses a significant threat even in the presence of the best defense, Multi-Krum. We hope to inspire further research on attack and defense strategies in pFL scenarios. The code is available at: https://github.com/BapFL/code.

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。

↑