EUVIMEDEUVIMED entdecken
Crossref

Communication-Efficient Decentralized Contextual $\mathcal{X}$ -Armed Bandit Learning in Multi-Agent Stochastic Networks

Youqi Li, Fan Li, Pan Zhou, Yu Wang

IEEE Transactions on Networking · 2026 · 34 · 4042-4057

DOI: 10.1109/ton.2026.3669909 ↗

Kurz gefasst

Fachartikel in der Zeitschrift IEEE Transactions on Networking, 2026. Verfasst von Youqi Li und 3 weiteren Autor:innen. Die Arbeit nennt 58 Literaturangaben. Gefördert von: Beijing Natural Science Foundation, Beijing Institute of Technology Research Fund Program for Young Scholars, National Natural Science Foundation of China.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Bandit with infinitely many arms (i.e., $\mathcal {X}$ -armed bandit) is an important variant of multi-armed stochastic bandits, which is useful to model different networking problems under both wired and wireless settings, e.g., online caching, dynamic channel/power allocation, rate adaption. However, the problem becomes challenging when the characteristic of the networking setting is affected by the side information (i.e., context) and distributed behavior. In this paper, we identify and study a novel problem, decentralized contextual $\mathcal {X}$ -armed bandit, where $N$ agents collaboratively solve the problem within time span $T$ . The problem is nontrivial because the infinite arms challenge and statistical information consensus issue make our setting go beyond a simple combination of $\mathcal {X}$ -armed bandits and multi-agent bandits. We develop a decentralized arm selection algorithm, called MACXUCB , by elaborating the contextual covering tree technique with a novel gossip communication protocol, which allows each agent to communicate efficiently with her neighbors. We prove that MACXUCB achieves a sublinear regret upper bound $\widetilde {\mathcal {O}}\left ({{D^{\frac {1}{2d_{\mathcal {X}}+2d_{\mathcal {Y}}+4}} N^{\frac {d_{\mathcal {X}}+d_{\mathcal {Y}}+1.5}{d_{\mathcal {X}}+d_{\mathcal {Y}}+2}} T^{\frac {d_{\mathcal {X}}+d_{\mathcal {Y}}+1}{d_{\mathcal {X}}+d_{\mathcal {Y}}+2}}}}\right)$ given aggregation period $D$ , and the covering dimensions $d_{\mathcal {X}}$ and $d_{\mathcal {Y}}$ of arm and context spaces, which asymptomatically matches the lower bound $\Omega \left ({{(DN)^{\frac {1}{d_{\mathcal {X}}+d_{\mathcal {Y}}+2}}T^{\frac {d_{\mathcal {X}}+d_{\mathcal {Y}}+1}{d_{\mathcal {X}}+d_{\mathcal {Y}}+2}}}}\right)$ up to a time-dependent factor. Moreover, MACXUCB enjoys a sublinear communication complexity $\widetilde {\mathcal {O}}(NDT^{0.5/(1-\rho)})$ when tuning parameter $\rho \in \left ({{0,\frac {1}{2}}}\right)$ . Finally, we carry out experiments to verify the performance of our MACXUCB . The results show the effectiveness and efficiency of our MACXUCB .

Förderung

  • Beijing Natural Science Foundation
  • Beijing Institute of Technology Research Fund Program for Young Scholars
  • National Natural Science Foundation of China
  • NSFC

Zitieren

APA

Li, Y., Li, F., Zhou, P., & Wang, Y. (2026). Communication-Efficient Decentralized Contextual $\mathcal{X}$ -Armed Bandit Learning in Multi-Agent Stochastic Networks. IEEE Transactions on Networking, 34, 4042-4057. https://doi.org/10.1109/ton.2026.3669909

Vancouver

Li Y, Li F, Zhou P, Wang Y. Communication-Efficient Decentralized Contextual $\mathcal{X}$ -Armed Bandit Learning in Multi-Agent Stochastic Networks. IEEE Transactions on Networking. 2026;34:4042-4057. doi:10.1109/ton.2026.3669909

BibTeX
@article{Li2026,
  author = {Li, Youqi and Li, Fan and Zhou, Pan and Wang, Yu},
  title = {Communication-Efficient Decentralized Contextual $\mathcalX$ -Armed Bandit Learning in Multi-Agent Stochastic Networks},
  journal = {IEEE Transactions on Networking},
  year = {2026},
  volume = {34},
  pages = {4042--4057},
  doi = {10.1109/ton.2026.3669909}
}
Kostenlos starten →