Privacy-Preserving Personalized Decentralized Learning With Fast Convergence
IEEE Transactions on Consumer Electronics · 2024-11 · 70 (4) · 6618-6629
DOI: 10.1109/tce.2024.3475370 ↗Kurz gefasst
Fachartikel in der Zeitschrift IEEE Transactions on Consumer Electronics, 2024-11. Verfasst von Jing Qiao und 6 weiteren Autor:innen. Die Arbeit wurde bisher 3-mal zitiert (Crossref) und nennt 37 Literaturangaben. Gefördert von: National Science Fund for Excellent Young Scholars of China, Major Basic Research Program of Shandong Provincial Natural Science Foundation, National Natural Science Foundation of China.
Abstract (Originalsprache)
Personalized decentralized learning aims to train individual personalized models for each client to adapt to Non-IID data distributions and heterogeneous environments. However, the distributed nature of decentralized learning is insufficient for protecting client training data from gradient leakage danger. In this paper, we investigate a p rivacy-preserving p ersonalized d ecentralized l earning optimization mechanism instead of traditional SGD. We design the P2DL mechanism to optimize our proposed objective function, whereby adjusting the regularization term parameter for a resilient local-global trade-off. Instead of exchanging gradients or models, auxiliary variables with knowledge can be transferred among clients to avoid model inversion and reconstruction attacks. We also provide theoretical convergence guarantees for both synchronous and asynchronous settings. Particularly, in case of synchronous communication, its convergence rate $\mathcal {O}\left ({{{}\frac {1}{k}}}\right)$ matches with the optimal result in decentralized learning, where k is the number of communication rounds. Extensive experiments are conducted to verify the effectiveness of newly proposed P2DL comparing with the state of the arts.
Förderung
- National Science Fund for Excellent Young Scholars of China
- Major Basic Research Program of Shandong Provincial Natural Science Foundation
- National Natural Science Foundation of China
- Shandong Provincial Natural Science Foundation of China
- Fundamental Research Funds for the Central Universities
Zitieren
APA
Qiao, J., Xie, Z., Zheng, Z., Zhang, X., Zhang, Z., Zhang, Q., & Yu, D. (2024). Privacy-Preserving Personalized Decentralized Learning With Fast Convergence. IEEE Transactions on Consumer Electronics, 70(4), 6618-6629. https://doi.org/10.1109/tce.2024.3475370
Vancouver
Qiao J, Xie Z, Zheng Z, Zhang X, Zhang Z, Zhang Q, et al. Privacy-Preserving Personalized Decentralized Learning With Fast Convergence. IEEE Transactions on Consumer Electronics. 2024;70(4):6618-6629. doi:10.1109/tce.2024.3475370
BibTeX
@article{Qiao2024,
author = {Qiao, Jing and Xie, Zhenzhen and Zheng, Zhigao and Zhang, Xiao and Zhang, Zhenyu and Zhang, Qun and Yu, Dongxiao},
title = {Privacy-Preserving Personalized Decentralized Learning With Fast Convergence},
journal = {IEEE Transactions on Consumer Electronics},
year = {2024},
volume = {70},
number = {4},
pages = {6618--6629},
doi = {10.1109/tce.2024.3475370}
}
EUVIMED