English

On the Arithmetic and Geometric Fusion of Beliefs for Distributed Inference

Signal Processing 2023-11-02 v2 Multiagent Systems Social and Information Networks

Abstract

We study the asymptotic learning rates under linear and log-linear combination rules of belief vectors in a distributed hypothesis testing problem. We show that under both combination strategies, agents are able to learn the truth exponentially fast, with a faster rate under log-linear fusion. We examine the gap between the rates in terms of network connectivity and information diversity. We also provide closed-form expressions for special cases involving federated architectures and exchangeable networks.

Keywords

Cite

@article{arxiv.2204.13741,
  title  = {On the Arithmetic and Geometric Fusion of Beliefs for Distributed Inference},
  author = {Mert Kayaalp and Yunus Inan and Emre Telatar and Ali H. Sayed},
  journal= {arXiv preprint arXiv:2204.13741},
  year   = {2023}
}

Comments

Accepted for publication in IEEE Transactions on Automatic Control