BapFL: You can Backdoor Personalized Federated Learning
ACM Transactions on Knowledge Discovery from Data · 31.08.2024 · 18 (7) · 1-17
DOI: 10.1145/3649316 ↗Kurz gefasst
Fachartikel in der Zeitschrift ACM Transactions on Knowledge Discovery from Data, online erschienen am 19.06.2024. Verfasst von Tiandi Ye und 4 weiteren Autor:innen. Die Arbeit wurde bisher 18-mal zitiert (Crossref) und nennt 37 Literaturangaben. Gefördert von: National Natural Science Foundation of China, Open Research Fund of KLATASDS-MOE, ECNU.
Abstract (Originalsprache)
In federated learning (FL), malicious clients could manipulate the predictions of the trained model through backdoor attacks, posing a significant threat to the security of FL systems. 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. [ 24 ] 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 parameter decoupling can significantly enhance robustness against backdoor attacks. However, in this article, 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
Förderung
- National Natural Science Foundation of China
- Open Research Fund of KLATASDS-MOE
- ECNU
- CCF-AFSG Research Fund
Zitieren
APA
Ye, T., Chen, C., Wang, Y., Li, X., & Gao, M. (2024). BapFL: You can Backdoor Personalized Federated Learning. ACM Transactions on Knowledge Discovery from Data, 18(7), 1-17. https://doi.org/10.1145/3649316
Vancouver
Ye T, Chen C, Wang Y, Li X, Gao M. BapFL: You can Backdoor Personalized Federated Learning. ACM Transactions on Knowledge Discovery from Data. 2024;18(7):1-17. doi:10.1145/3649316
BibTeX
@article{Ye2024,
author = {Ye, Tiandi and Chen, Cen and Wang, Yinggui and Li, Xiang and Gao, Ming},
title = {BapFL: You can Backdoor Personalized Federated Learning},
journal = {ACM Transactions on Knowledge Discovery from Data},
year = {2024},
volume = {18},
number = {7},
pages = {1--17},
doi = {10.1145/3649316}
}
EUVIMED