EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

Addressing Non-IID with Data Quantity Skew in Federated Learning

Narisu Cha, Long Chang

Information · 04.10.2025 · 16 (10) · 861

DOI: 10.3390/info16100861 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Information, online erschienen am 04.10.2025. Verfasst von Narisu Cha und 1 weiteren Autor:innen. Die Arbeit wurde bisher 5-mal zitiert (Crossref) und nennt 46 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: Program for High-Quality Research Achievement of Inner Mongolia University of Finance and Economics, Natural Science Foundation of Inner Mongolia.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Non-IID is one of the key challenges in federated learning. Data heterogeneity may lead to slower convergence, reduced accuracy, and more training rounds. To address the common Non-IID data distribution problem in federated learning, we propose a comprehensive dynamic optimization approach based on existing methods. It leverages MAP estimation of the Dirichlet parameter β to dynamically adjust the regularization coefficient μ and introduces orthogonal gradient coefficients Δi to mitigate gradient interference among different classes. The approach is compatible with existing federated learning frameworks and can be easily integrated. Achieves significant accuracy improvements in both mildly and severely Non-IID scenarios while maintaining a strong performance lower bound.

Förderung

  • Program for High-Quality Research Achievement of Inner Mongolia University of Finance and Economics
  • Natural Science Foundation of Inner Mongolia

Zitieren

APA

Cha, N., & Chang, L. (2025). Addressing Non-IID with Data Quantity Skew in Federated Learning. Information, 16(10), 861. https://doi.org/10.3390/info16100861

Vancouver

Cha N, Chang L. Addressing Non-IID with Data Quantity Skew in Federated Learning. Information. 2025;16(10):861. doi:10.3390/info16100861

BibTeX
@article{Cha2025,
  author = {Cha, Narisu and Chang, Long},
  title = {Addressing Non-IID with Data Quantity Skew in Federated Learning},
  journal = {Information},
  year = {2025},
  volume = {16},
  number = {10},
  pages = {861},
  doi = {10.3390/info16100861}
}
Kostenlos starten →