EUVIMEDEUVIMED entdecken
Crossref

Optimal Cut Layer Bounds for Split Learning

Matea Marinova, Marija Poposka, Zoran Hadzi-Velkov, Valentin Rakovic

IEEE Communications Letters · 2025-04 · 29 (4) · 749-753

DOI: 10.1109/lcomm.2025.3542541 ↗

Kurz gefasst

Fachartikel in der Zeitschrift IEEE Communications Letters, 2025-04. Verfasst von Matea Marinova und 3 weiteren Autor:innen. Die Arbeit wurde bisher 5-mal zitiert (Crossref) und nennt 17 Literaturangaben.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Split learning (SL) is a distributed learning method where a deep learning model is partitioned between the client and server, aiming to optimize the training process. A key challenge in split learning is selecting the cut layer to minimize energy consumption while considering both computational and communication overheads. In this letter, we address this challenge within the context of a wireless system with multiple clients and a central server. We introduce a pruning-based cut layer selection scheme that effectively reduces the energy consumption for each client. Our approach leverages analytical bounds for optimal cut layer location, which we derive and validate against state-of-the-art SL benchmark schemes, demonstrating the high efficiency of our proposed method.

Zitieren

APA

Marinova, M., Poposka, M., Hadzi-Velkov, Z., & Rakovic, V. (2025). Optimal Cut Layer Bounds for Split Learning. IEEE Communications Letters, 29(4), 749-753. https://doi.org/10.1109/lcomm.2025.3542541

Vancouver

Marinova M, Poposka M, Hadzi-Velkov Z, Rakovic V. Optimal Cut Layer Bounds for Split Learning. IEEE Communications Letters. 2025;29(4):749-753. doi:10.1109/lcomm.2025.3542541

BibTeX
@article{Marinova2025,
  author = {Marinova, Matea and Poposka, Marija and Hadzi-Velkov, Zoran and Rakovic, Valentin},
  title = {Optimal Cut Layer Bounds for Split Learning},
  journal = {IEEE Communications Letters},
  year = {2025},
  volume = {29},
  number = {4},
  pages = {749--753},
  doi = {10.1109/lcomm.2025.3542541}
}
Kostenlos starten →