Fast Convergence Rates for Distributed Non-Bayesian Learning

被引：131

作者：

Nedic, Angelia ^{[1
]}

Olshevsky, Alex ^{[2
,3
]}

Uribe, Cesar A. ^{[4
,5
]}

机构：

[1] Arizona State Univ, ECEE Dept, Tempe, AZ 85287 USA

[2] Boston Univ, ECE Dept, Boston, MA 02215 USA

[3] Boston Univ, Div Syst Engn, Boston, MA 02215 USA

[4] Univ Illinois, ECE Dept, Urbana, IL 61820 USA

[5] Univ Illinois, Coordinated Sci Lab, Urbana, IL 61820 USA

来源：

IEEE TRANSACTIONS ON AUTOMATIC CONTROL | 2017年 / 62卷 / 11期

基金：

美国国家科学基金会;

关键词：

Algorithm design and analysis; Bayes methods; distributed algorithms; estimation; learning; SENSOR NETWORKS; ASYMPTOTIC AGREEMENT; LOCALIZATION; CONSENSUS;

D O I：

10.1109/TAC.2017.2690401

中图分类号：

TP [自动化技术、计算机技术];

学科分类号：

0812 ;

摘要：

We consider the problem of distributed learning, where a network of agents collectively aim to agree on a hypothesis that best explains a set of distributed observations of conditionally independent random processes. We propose a distributed algorithm and establish consistency, as well as a nonasymptotic, explicit, and geometric convergence rate for the concentration of the beliefs around the set of optimal hypotheses. Additionally, if the agents interact over static networks, we provide an improved learning protocol with better scalability with respect to the number of nodes in the network.

引用

页码：5538 / 5553

页数：16

共 60 条

[51]

Su L., 2016, ARXIV PREPRINT ARXIV

[52]

Su L., 2016, ARXIV160603418

[53] Multi-sensor optimal information fusion Kalman filter* [J].