×

zbMATH — the first resource for mathematics

A simulated annealing-based method for learning Bayesian networks from statistical data. (English) Zbl 1101.62015
Summary: The problem of learning Bayesian networks from statistical data is described and reformulated as a discrete optimization problem. For a solution we employ the stochastic algorithm that is known as simulated annealing and that is based on the Markov Chain Monte Carlo approach. Numerical examples are included to illustrate the efficiency of the method.

MSC:
62F15 Bayesian inference
65C40 Numerical analysis or methods applied to Markov chains
90C90 Applications of mathematical programming
65C05 Monte Carlo methods
94C99 Circuits, networks
PDF BibTeX XML Cite
Full Text: DOI
References:
[1] Buntine, IEEE Trans Knowl Data Eng 8 pp 195– (1996)
[2] , , . Bayesian data analysis. London: Chapman and Hall; 1995.
[3] Heckerman, Mach Learn 20 pp 197– (1995) · doi:10.1007/BF00994016
[4] editor. Learning in graphical models. Dordrecht: Kluwer; 1998. · doi:10.1007/978-94-011-5014-9
[5] Chickering, J Mach Learn Res 3 pp 507– (2002)
[6] . Finding optimal Bayesian networks. In: Proc 18th Conf on Uncertainty in Artificial Intelligence, August 1–4, 2002, Univerity of Alberta, Edmonton, Alberta, Canada; 2002. pp 94–102.
[7] Geman, IEEE Trans Patt Anal Mach Intell 6 pp 721– (1984)
[8] Lakshmanan, IEEE Trans Patt Anal Mach Intell 11 pp 799– (1989)
[9] Image analysis, random fields and dynamic Monte Carlo methods. Berlin: Springer-Verlag; 1995. · doi:10.1007/978-3-642-97522-6
[10] Introduction to Bayesian networks. London: UCL Press; 1996.
[11] Graphical models. Oxford: Oxford University Press; 1996.
[12] , , editors. Markov chain Monte Carlo in practice. London: Chapman and Hall; 1996. · Zbl 0832.00018
This reference list is based on information provided by the publisher or from digital mathematics libraries. Its items are heuristically matched to zbMATH identifiers and may contain data conversion errors. It attempts to reflect the references listed in the original paper as accurately as possible without claiming the completeness or perfect precision of the matching.