Abstract

This paper addresses the problem of state estimation in the case where the prior distribution of the states is not perfectly known but instead is parameterized by some unknown parameter. Thus in order to support the state estimator with prior information on the states and improve the quality of the state estimates, it is necessary to learn this unknown parameter first. Here we assume a parameterized Gaussian Markov random field to model the prior distribution of the states and propose an algorithm that is able to learn its parameters from given observations on these states. The effectiveness of this approach is proven experimentally by simulations.

Original languageEnglish (US)
Title of host publication2010 IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2010 - Proceedings
Pages3070-3073
Number of pages4
DOIs
StatePublished - 2010
Event2010 IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2010 - Dallas, TX, United States
Duration: Mar 14 2010Mar 19 2010

Publication series

NameICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing - Proceedings
ISSN (Print)1520-6149

Other

Other2010 IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2010
Country/TerritoryUnited States
CityDallas, TX
Period3/14/103/19/10

Keywords

  • EM algorithm
  • Factor graph
  • Gaussian Markov random field
  • Sum-product algorithm
  • Unsupervised learning

ASJC Scopus subject areas

  • Software
  • Signal Processing
  • Electrical and Electronic Engineering

Fingerprint

Dive into the research topics of 'Learning in Gaussian Markov random fields'. Together they form a unique fingerprint.

Cite this