Improved Sampling for Diagnostic Reasoning in Bayesian Networks
Bayesian networks offer great potential for use in automating large scale diagnostic reasoning tasks. Gibbs sampling is the main technique used to perform diagnostic reasoning in large richly interconnected Bayesian networks. Unfortunately Gibbs sampling can take an excessive time to generate a representative sample. In this paper we describe and test a number of heuristic strategies for improving sampling in noisy-or Bayesian networks. The strategies include Monte Carlo Markov chain sampling techniques other than Gibbs sampling. Emphasis is put on strategies that can be implemented in distributed systems.
Keywords: Diagnostic reasoning, MCMC Sampling.
PS Link: http://www.cs.rmit.edu.au/~mrh/uai95.ps
PDF Link: /papers/95/p315-hulme.pdf
AUTHOR = "Mark Hulme
TITLE = "Improved Sampling for Diagnostic Reasoning in Bayesian Networks",
BOOKTITLE = "Proceedings of the Eleventh Conference Annual Conference on Uncertainty in Artificial Intelligence (UAI-95)",
PUBLISHER = "Morgan Kaufmann",
ADDRESS = "San Francisco, CA",
YEAR = "1995",
PAGES = "315--322"