EUVIMEDEUVIMED entdecken
Crossref

Towards Adaptive Privacy Protection for Interpretable Federated Learning

Zhe Li, Honglong Chen, Zhichen Ni, Yudong Gao, Wei Lou

IEEE Transactions on Mobile Computing · 2024-12 · 23 (12) · 14471-14483

DOI: 10.1109/tmc.2024.3443862 ↗

Kurz gefasst

Fachartikel in der Zeitschrift IEEE Transactions on Mobile Computing, 2024-12. Verfasst von Zhe Li und 4 weiteren Autor:innen. Die Arbeit wurde bisher 14-mal zitiert (Crossref) und nennt 43 Literaturangaben. Gefördert von: Shandong Provincial Taishan Scholar Program, China, Shandong Provincial Natural Science Foundation, China, National Natural Science Foundation of China.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Federated learning (FL) is an effective privacy-preserving mechanism that collaboratively trains the global model in a distributed manner by solely sharing model parameters rather than data from local clients, like mobile devices, to a central server. Nevertheless, recent studies have illustrated that FL still suffers from gradient leakage as adversaries try to recover training data by analyzing shared parameters from local clients. To address this issue, differential privacy (DP) is adopted to add noise to the parameters of local models before aggregation occurs on the server. It, however, results in the poor performance of gradient-based interpretability, since some important weights capturing the salient region in feature maps will be perturbed. To overcome this problem, we propose a simple yet effective adaptive gradient protection (AGP) mechanism that selectively adds noisy perturbations to certain channels of each client model that have a relatively small impact on interpretability. We also offer a theoretical analysis of the convergence of FL using our method. The evaluation results on both IID and Non-IID data demonstrate that the proposed AGP can achieve a good trade-off between privacy protection and interpretability in FL. Furthermore, we verify the robustness of the proposed method against two different gradient leakage attacks.

Förderung

  • Shandong Provincial Taishan Scholar Program, China
  • Shandong Provincial Natural Science Foundation, China
  • National Natural Science Foundation of China
  • Innovation Fund Project for Graduate Student of China University of Petroleum
  • Fundamental Research Funds for the Central Universities
  • Hong Kong Polytechnic University

Zitieren

APA

Li, Z., Chen, H., Ni, Z., Gao, Y., & Lou, W. (2024). Towards Adaptive Privacy Protection for Interpretable Federated Learning. IEEE Transactions on Mobile Computing, 23(12), 14471-14483. https://doi.org/10.1109/tmc.2024.3443862

Vancouver

Li Z, Chen H, Ni Z, Gao Y, Lou W. Towards Adaptive Privacy Protection for Interpretable Federated Learning. IEEE Transactions on Mobile Computing. 2024;23(12):14471-14483. doi:10.1109/tmc.2024.3443862

BibTeX
@article{Li2024,
  author = {Li, Zhe and Chen, Honglong and Ni, Zhichen and Gao, Yudong and Lou, Wei},
  title = {Towards Adaptive Privacy Protection for Interpretable Federated Learning},
  journal = {IEEE Transactions on Mobile Computing},
  year = {2024},
  volume = {23},
  number = {12},
  pages = {14471--14483},
  doi = {10.1109/tmc.2024.3443862}
}
Kostenlos starten →