Skip to main navigation Skip to search Skip to main content

Cooperative Training over Networks via Consensus-based Algorithms

  • Zhongguo Li
  • , Bo Liu
  • , Zhen Dong
  • , Zhengtao Ding
  • Loughborough University
  • University of Manchester

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

Abstract

In this paper, the training problem for a group of neural networks with private datasets is considered. Approximated gradients are employed to replace the true gradients in the proposed algorithms, due to the presence of gradient noises in the training problems. Consensus tools are used to achieve identical weights of the distributed neural networks trained using local dataset only. The convergence of the proposed algorithms is established by exploring the error dynamics of the connected agents, through which upper bounds for the learning rates are derived. Performances are analysed for the proposed algorithms with and without gradient noises. Simulation examples are provided to validate the effectiveness of the proposed algorithms.

Original languageEnglish
Title of host publicationProceedings of the 40th Chinese Control Conference, CCC 2021
EditorsChen Peng, Jian Sun
PublisherIEEE Computer Society
Pages5309-5314
Number of pages6
ISBN (Electronic)9789881563804
DOIs
StatePublished - 26 Jul 2021
Externally publishedYes
Event40th Chinese Control Conference, CCC 2021 - Shanghai, China
Duration: 26 Jul 202128 Jul 2021

Publication series

NameChinese Control Conference, CCC
Volume2021-July
ISSN (Print)1934-1768
ISSN (Electronic)2161-2927

Conference

Conference40th Chinese Control Conference, CCC 2021
Country/TerritoryChina
CityShanghai
Period26/07/2128/07/21

Keywords

  • Consensus
  • convergence analysis
  • distributed training
  • multi-agent systems
  • neural networks
  • optimisation

Fingerprint

Dive into the research topics of 'Cooperative Training over Networks via Consensus-based Algorithms'. Together they form a unique fingerprint.

Cite this