Bayesian M-T clustering for reduced parameterisation of Markov chains used for non-linear adaptive elements

0410587 - UTIA-B 20010056 RIV GB eng J - Článek v odborném periodiku
Valečková, Markéta - Kárný, Miroslav - Sutanto, E. L.
Bayesian M-T clustering for reduced parameterisation of Markov chains used for non-linear adaptive elements.
Automatica. Roč. 37, č. 6 (2001), s. 1071-1078. ISSN 0005-1098. E-ISSN 1873-2836
Grant CEP: GA ČR GA102/99/1564
Grant ostatní: IST(XE) 1999/12058
Výzkumný záměr: AV0Z1075907
Klíčová slova: Markov chain * clustering * Bayesian mixture estimation
Kód oboru RIV: BC - Teorie a systémy řízení
Impakt faktor: 1.449, rok: 2001

Markov chains are black box models ideal for describing stochastic digitised systems. Although the identification of their parameters can be a relatively easy task to perform, the dimensionality involved become undesirable large. This significant drawback can be overcome by exploiting smoothness of the underlying system. The paper present a novel hybrid off-line algorithm to locate areas which merit detailed model description. It comprises Bayesian parameter estimation and Mean tracking algorithm.
Trvalý link: http://hdl.handle.net/11104/0130676