A Log-Likelihood Ratio based Generalized Belief Propagation

Alexandru Amaricai, Mohsem Bahrami, Bane Vasić

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

Abstract

In this paper, we propose a reduced complexity Generalized Belief Propagation (GBP) that propagates messages in Log-Likelihood Ratio (LLR) domain. The key novelties of the proposed LLR-GBP are: (i) reduced fixed point precision for messages instead of computational complex floating point format, (ii) operations performed in logarithm domain, thus eliminating the need for multiplications and divisions, (iii) usage of message ratios that leads to simple hard decision mechanisms. We demonstrated the validity of LLR-GBP on reconstruction of images passed through binary-input two-dimensional Gaussian channels with memory and affected by additive white Gaussian noise.

Original languageEnglish (US)
Title of host publicationEUROCON 2019 - 18th International Conference on Smart Technologies
EditorsBoris Dumnic, Marko Delimar, Cedomir Stefanovic
PublisherInstitute of Electrical and Electronics Engineers Inc.
ISBN (Electronic)9781538693018
DOIs
StatePublished - Jul 2019
Event18th International Conference on Smart Technologies, EUROCON 2019 - Novi Sad, Serbia
Duration: Jul 1 2019Jul 4 2019

Publication series

NameEUROCON 2019 - 18th International Conference on Smart Technologies

Conference

Conference18th International Conference on Smart Technologies, EUROCON 2019
CountrySerbia
CityNovi Sad
Period7/1/197/4/19

Keywords

  • Probabilistic inference
  • generalized belief propagation (GBP)
  • graphical models

ASJC Scopus subject areas

  • Artificial Intelligence
  • Signal Processing
  • Management Science and Operations Research
  • Energy Engineering and Power Technology
  • Electrical and Electronic Engineering
  • Computational Mathematics
  • Control and Optimization
  • Health Informatics

Fingerprint Dive into the research topics of 'A Log-Likelihood Ratio based Generalized Belief Propagation'. Together they form a unique fingerprint.

Cite this