Model Parameter Estimation

作者: William Turin

DOI: 10.1007/978-1-4419-9070-9_4

关键词: Hidden Markov modelForward algorithmAlgorithmComputer scienceParameter identification problemStochastic processFree parameterExpectation–maximization algorithmMaximizationCurve fitting

摘要: In this chapter, we develop methods for approximating a stochastic process with an HMM and, in particular, fitting HMMs to experimental data. We present the iterativeexpectation maximization(EM) algorithm approximation which generalizes statistical EM theBaum-Welch algorithm(BWA) HMM. This generalized can be also applied curve and finding maximum of nonnegative function several variables. The is iterative converges slowly. Therefore, it important select good initial model. describe choosing question speeding up convergence addressed.

参考文章(49)
Christopher Chatfield, Statistical Inference Regarding Markov Chain Models Journal of The Royal Statistical Society Series C-applied Statistics. ,vol. 22, pp. 7- 20 ,(1973) , 10.2307/2346299
J. Durbin, The fitting of time series models Revue de l'Institut International de Statistique / Review of the International Statistical Institute , 28 (3) pp. 233-244. (1960). ,vol. 28, pp. 233- ,(1960) , 10.2307/1401322
Elwyn R. Berlekamp, Algebraic Coding Theory ,(1984)
S. Sivaprakasam, K. Sam Shanmugan, A forward-only recursion based HMM for modeling burst errors in digital channels global communications conference. ,vol. 2, pp. 1054- 1058 ,(1995) , 10.1109/GLOCOM.1995.502566
J. Makhoul, Linear prediction: A tutorial review Proceedings of the IEEE. ,vol. 63, pp. 561- 580 ,(1975) , 10.1109/PROC.1975.9792
William Turin, Fitting probabilistic automata via the em algorithm Communications in Statistics. Stochastic Models. ,vol. 12, pp. 405- 424 ,(1996) , 10.1080/15326349608807392
Frederick W. Leysieffer, Functions of Finite Markov Chains Annals of Mathematical Statistics. ,vol. 38, pp. 206- 212 ,(1967) , 10.1214/AOMS/1177699072