Decentralized Expectation Maximization Algorithm

Honghe Jin, Xiaoxiao Sun, Liwen Xu

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

As a promising paradigm that does not require a central node, decentralized computing provides better network load balance and has advantages over centralized computing in terms of data protection. Although decentralized algorithms such as decentralized gradient descent algorithms have been extensively studied, there is no such research on the expectation maximization (EM) algorithm, which includes the expectation step (E-step) and the maximization step (M-step) and is a popular iterative method for missing data studies and latent variable models. In this paper, we propose decentralized EM algorithms under different communication and network topology settings such as synchronous communication and dynamic networks. Convergence analysis of the proposed algorithms is provided in the synchronous scenario. Empirical studies show that our proposed algorithms have numerical advantages over the EM algorithms based on local data or full data only, especially when there is no closed-form maximization in the M-step.

Original languageEnglish (US)
Title of host publicationAlgorithms and Architectures for Parallel Processing - 20th International Conference, ICA3PP 2020, Proceedings
EditorsMeikang Qiu
PublisherSpringer Science and Business Media Deutschland GmbH
Pages512-527
Number of pages16
ISBN (Print)9783030602444
DOIs
StatePublished - 2020
Event20th International Conference on Algorithms and Architectures for Parallel Processing, ICA3PP 2020 - New York, United States
Duration: Oct 2 2020Oct 4 2020

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume12452 LNCS
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Conference

Conference20th International Conference on Algorithms and Architectures for Parallel Processing, ICA3PP 2020
Country/TerritoryUnited States
CityNew York
Period10/2/2010/4/20

Keywords

  • Asynchronous optimization
  • Decentralized computing
  • Dynamic network
  • EM algorithm
  • Mixture model

ASJC Scopus subject areas

  • Theoretical Computer Science
  • Computer Science(all)

Fingerprint

Dive into the research topics of 'Decentralized Expectation Maximization Algorithm'. Together they form a unique fingerprint.

Cite this