EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

Decentralized Policy Learning with Communication-Efficient Reinforcement Strategies for Collaborative Agents

Pierre Dubois, Julien Martin, Nicolas Lefèvre

International Journal of Computational and Biological Sciences · 05.04.2026 · 3 (1) · 251-256

DOI: 10.66238/ijcbs66 ↗

Kurz gefasst

Fachartikel in der Zeitschrift International Journal of Computational and Biological Sciences, online erschienen am 05.04.2026. Verfasst von Pierre Dubois und 2 weiteren Autor:innen. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

In decentralized decision environments, excessive communication among agents often leads to inefficiency without proportional performance gains. This work focuses on reducing communication redundancy while preserving coordination quality through a communication-efficient reinforcement learning strategy. A decentralized multi-agent PPO framework is combined with a learned communication gating mechanism that selectively filters inter-agent messages based on estimated marginal utility. The approach is evaluated on 9,800 decision episodes involving distributed scheduling and cooperative control tasks. Results indicate that communication volume is reduced by 32.6% while task success rate improves from 74.9% to 83.7%. Additionally, training stability is enhanced, with variance in cumulative reward decreasing by 21.3%. These findings suggest that communication-aware policy learning can significantly improve efficiency in decentralized collaborative systems.

Zitieren

APA

Dubois, P., Martin, J., & Lefèvre, N. (2026). Decentralized Policy Learning with Communication-Efficient Reinforcement Strategies for Collaborative Agents. International Journal of Computational and Biological Sciences, 3(1), 251-256. https://doi.org/10.66238/ijcbs66

Vancouver

Dubois P, Martin J, Lefèvre N. Decentralized Policy Learning with Communication-Efficient Reinforcement Strategies for Collaborative Agents. International Journal of Computational and Biological Sciences. 2026;3(1):251-256. doi:10.66238/ijcbs66

BibTeX
@article{Dubois2026,
  author = {Dubois, Pierre and Martin, Julien and Lefèvre, Nicolas},
  title = {Decentralized Policy Learning with Communication-Efficient Reinforcement Strategies for Collaborative Agents},
  journal = {International Journal of Computational and Biological Sciences},
  year = {2026},
  volume = {3},
  number = {1},
  pages = {251--256},
  doi = {10.66238/ijcbs66}
}
Kostenlos starten →