EUVIMEDEUVIMED entdecken
Open Access · CC BY-NC-ND 4.0

FedZMG: Efficient Client-Side Optimization in Federated Learning

Fotios Zantalis, Evangelos Zervas, Grigorios Koulouras

IEEE Open Journal of the Computer Society · 2026 · 7 · 1690-1701

DOI: 10.1109/ojcs.2026.3748315 ↗

Kurz gefasst

Fachartikel in der Zeitschrift IEEE Open Journal of the Computer Society, 2026. Verfasst von Fotios Zantalis und 2 weiteren Autor:innen. Die Arbeit nennt 34 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY-NC-ND 4.0).

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Federated Learning (FL) enables distributed model training on edge devices while preserving data privacy. However, clients tend to have non-Independent and Identically Distributed (non-IID) data, which often leads to client-drift, and therefore diminishing convergence speed and model performance. While adaptive optimizers have been proposed to mitigate these effects, they frequently introduce computational complexity or communication overhead unsuitable for resource-constrained IoT environments. This paper introduces Federated Zero Mean Gradients (FedZMG), a novel, parameter-free, client-side optimization algorithm designed to tackle client-drift by structurally regularizing the optimization space. Advancing the idea of Gradient Centralization, FedZMG projects local gradients onto a zero-mean hyperplane, effectively neutralizing the “intensity” or “bias” shifts inherent in heterogeneous data distributions without requiring additional communication or hyperparameter tuning. A theoretical analysis is provided, proving that FedZMG reduces the effective gradient variance and converges at an $\mathcal {O}(1/T)$ rate to the optimum of the constrained parameter space induced by the zero-mean projection. Extensive empirical evaluations on EMNIST, CIFAR100, and Shakespeare datasets demonstrate that FedZMG achieves higher final validation accuracy than the baseline FedAvg, the adaptive optimizer FedAdam, and the popular FedProx, and faster convergence in the highly non-IID CIFAR100 and Shakespeare tasks.

Zitieren

APA

Zantalis, F., Zervas, E., & Koulouras, G. (2026). FedZMG: Efficient Client-Side Optimization in Federated Learning. IEEE Open Journal of the Computer Society, 7, 1690-1701. https://doi.org/10.1109/ojcs.2026.3748315

Vancouver

Zantalis F, Zervas E, Koulouras G. FedZMG: Efficient Client-Side Optimization in Federated Learning. IEEE Open Journal of the Computer Society. 2026;7:1690-1701. doi:10.1109/ojcs.2026.3748315

BibTeX
@article{Zantalis2026,
  author = {Zantalis, Fotios and Zervas, Evangelos and Koulouras, Grigorios},
  title = {FedZMG: Efficient Client-Side Optimization in Federated Learning},
  journal = {IEEE Open Journal of the Computer Society},
  year = {2026},
  volume = {7},
  pages = {1690--1701},
  doi = {10.1109/ojcs.2026.3748315}
}
Kostenlos starten →