EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

Personalized Federated Learning Based on Dynamic Parameter Fusion and Prototype Alignment

Ying Chen, Jing Wen, Shaoling Liang, Zhaofa Chen, Baohua Huang iD

Sensors · 15.08.2025 · 25 (16) · 5076

DOI: 10.3390/s25165076 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Sensors, online erschienen am 15.08.2025. Verfasst von Ying Chen und 4 weiteren Autor:innen. Beteiligte Einrichtung: School of Computer and Electronic Information, Guangxi University, Nanning 530004, China. Die Arbeit wurde bisher 4-mal zitiert (Crossref) und nennt 30 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: Guangxi Zhuang Autonomous Region Information Center, Open Project Program of Guangxi Key Laboratory of Digital Infrastructure.

AbstractSchlagwörterBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

To address the limitation of generalization of federated learning under non-independent and identically distributed (Non-IID) data, we propose FedDFPA, a personalized federated learning framework that integrates dynamic parameter fusion and prototype alignment. We design a class-wise dynamic parameter fusion mechanism that adaptively fuses global and local classifier parameters at the class level. It enables each client to preserve its reliable local knowledge while selectively incorporating beneficial global information for personalized classification. We introduce a prototype alignment mechanism based on both global and historical information. By aligning current local features with global prototypes and historical local prototypes, it improves cross-client semantic consistency and enhances the stability of local features. To evaluate the effectiveness of FedDFPA, we conduct extensive experiments on various Non-IID settings and client participation rates. Compared to the average performance of state-of-the-art algorithms, FedDFPA improves the average test accuracy by 3.59% and 4.71% under practical and pathological heterogeneous settings, respectively. These results confirm the effectiveness of our dual-mechanism design in achieving a better balance between personalization and collaboration in federated learning.

Schlagwörter

Stichwörter der Autor:innen

Non-IID datadynamic parameter fusionfederated learningprototype alignment

Autor:innen und Institutionen

  • School of Computer and Electronic Information, Guangxi University, Nanning 530004, China.
  • Guangxi Key Laboratory of Digital Infrastructure, Guangxi Zhuang Autonomous Region Information Center, Nanning 530000, China.

ORCID-Profile: Baohua Huang

Förderung

  • Guangxi Zhuang Autonomous Region Information Center
  • Open Project Program of Guangxi Key Laboratory of Digital Infrastructure

Zitieren

APA

Chen, Y., Wen, J., Liang, S., Chen, Z., & Huang, B. (2025). Personalized Federated Learning Based on Dynamic Parameter Fusion and Prototype Alignment. Sensors, 25(16), 5076. https://doi.org/10.3390/s25165076

Vancouver

Chen Y, Wen J, Liang S, Chen Z, Huang B. Personalized Federated Learning Based on Dynamic Parameter Fusion and Prototype Alignment. Sensors. 2025;25(16):5076. doi:10.3390/s25165076

BibTeX
@article{Chen2025,
  author = {Chen, Ying and Wen, Jing and Liang, Shaoling and Chen, Zhaofa and Huang, Baohua},
  title = {Personalized Federated Learning Based on Dynamic Parameter Fusion and Prototype Alignment},
  journal = {Sensors},
  year = {2025},
  volume = {25},
  number = {16},
  pages = {5076},
  doi = {10.3390/s25165076}
}
Kostenlos starten →