Non-bayesian learning in the presence of byzantine agents

Lili Su, Nitin H. Vaidya

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


This paper addresses the problem of non-Bayesian learning over multi-agent networks, where agents repeatedly collect partially informative observations about an unknown state of the world, and try to collaboratively learn the true state. We focus on the impact of the Byzantine agents on the performance of consensus-based non-Bayesian learning. Our goal is to design an algorithm for the non-faulty agents to collaboratively learn the true state through local communication. We propose an update rule wherein each agent updates its local beliefs as (up to normalization) the product of (1) the likelihood of the cumulative private signals and (2) the weighted geometric average of the beliefs of its incoming neighbors and itself (using Byzantine consensus). Under mild assumptions on the underlying network structure and the global identifiability of the network, we show that all the non-faulty agents asymptotically agree on the true state almost surely.

Original languageEnglish (US)
Title of host publicationDistributed Computing - 30th International Symposium, DISC 2016, Proceedings
EditorsCyril Gavoille, David Ilcinkas
Number of pages14
ISBN (Print)9783662534250
StatePublished - 2016
Event30th International Symposium on Distributed Computing, DISC 2016 - Paris, France
Duration: Sep 27 2016Sep 29 2016

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume9888 LNCS
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349


Other30th International Symposium on Distributed Computing, DISC 2016


  • Adversary attacks
  • Byzantine agreement
  • Distributed learning
  • Faulttolerance
  • Security

ASJC Scopus subject areas

  • Theoretical Computer Science
  • Computer Science(all)


Dive into the research topics of 'Non-bayesian learning in the presence of byzantine agents'. Together they form a unique fingerprint.

Cite this