EUVIMEDEUVIMED entdecken
Crossref

A Server-Side Model Intellectual Property Protection Method for Federated Learning Against Model Theft

Wenxiong Chen, Xuantao Tang, Dan Wang, Ju Ren

IEEE Internet of Things Journal · 15.03.2026 · 13 (6) · 10727-10740

DOI: 10.1109/jiot.2025.3650344 ↗

Kurz gefasst

Fachartikel in der Zeitschrift IEEE Internet of Things Journal, erschienen am 15.03.2026. Verfasst von Wenxiong Chen und 3 weiteren Autor:innen. Die Arbeit wurde bisher 1-mal zitiert (Crossref) und nennt 31 Literaturangaben. Gefördert von: National Natural Science Foundation of China, China Postdoctoral Science Foundation, Natural Science Foundation of Hunan Province.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Federated learning (FL) has gained significant attention for enabling the collaborative model training while preserving data privacy. However, protecting the intellectual property (IP) of models in FL, particularly against model theft by malicious clients, remains a critical challenge. Existing works often employ watermarking techniques to embed watermarks into models for ownership verification, but most of them can only verify ownership after the model has been stolen and cannot proactively defend against malicious clients attempting to steal the model. To address this limitation, this article proposes FedLock, a novel server-side watermarking mechanism designed to resist model theft and safeguard the IP of global models. Specifically, FedLock leverages split FL (SFL) to partition the model, effectively preventing malicious clients from accessing the complete set of global model parameters. To enhance security, FedLock incorporates an autoencoder for label encoding, safeguarding the server-side model from reconstruction attacks. Furthermore, FedLock introduces an additional backdoor client to embed a black-box watermark into the global model, enabling remote verification of model ownership. Experimental results demonstrate that FedLock achieves robust watermarking with the minimal impact on model performance, effectively resisting various attacks, including model theft, pruning, and fine-tuning.

Förderung

  • National Natural Science Foundation of China
  • China Postdoctoral Science Foundation
  • Natural Science Foundation of Hunan Province
  • Hunan Province Education Science Planning Project
  • Hunan Province’s Major Science and Technology Research Project

Zitieren

APA

Chen, W., Tang, X., Wang, D., & Ren, J. (2026). A Server-Side Model Intellectual Property Protection Method for Federated Learning Against Model Theft. IEEE Internet of Things Journal, 13(6), 10727-10740. https://doi.org/10.1109/jiot.2025.3650344

Vancouver

Chen W, Tang X, Wang D, Ren J. A Server-Side Model Intellectual Property Protection Method for Federated Learning Against Model Theft. IEEE Internet of Things Journal. 2026;13(6):10727-10740. doi:10.1109/jiot.2025.3650344

BibTeX
@article{Chen2026,
  author = {Chen, Wenxiong and Tang, Xuantao and Wang, Dan and Ren, Ju},
  title = {A Server-Side Model Intellectual Property Protection Method for Federated Learning Against Model Theft},
  journal = {IEEE Internet of Things Journal},
  year = {2026},
  volume = {13},
  number = {6},
  pages = {10727--10740},
  doi = {10.1109/jiot.2025.3650344}
}
Kostenlos starten →