Smooth on-line learning algorithms for hidden Markov models
作者： Pierre Baldi , Yves Chauvin
DOI: 10.1162/NECO.1994.6.2.307
关键词: Algorithm 、 Mathematics 、 Path (graph theory) 、 Viterbi algorithm 、 Markov model 、 Approximation algorithm 、 Probabilistic analysis of algorithms 、 Representation (mathematics) 、 Hidden Markov model 、 Weighted Majority Algorithm
摘要: A simple learning algorithm for Hidden Markov Models (HMMs) is presented together with a number of variations. Unlike other classical algorithms such as the Baum-Welch algorithm, described are smooth and can be used on-line (after each example presentation) or in batch mode, without usual Viterbi most likely path approximation. The have expressions that result from using normalized-exponential representation HMM parameters. All proved to exact approximate gradient optimization respect likelihood, log-likelihood, cross-entropy functions, usually convergent. These also casted more general EM (Expectation-Maximization) framework where they viewed GEM (Generalized Expectation-Maximization) algorithms. mathematical properties derived appendix.
mitpressjournals.org
core.ac.uk参考文章(16)
L. Baum, An inequality and associated maximization technique in statistical estimation of probabilistic functions of a Markov process Inequalities. ,vol. 3, pp. 1- 8 ,(1972)
P. Baldi, Y. Chauvin, T. Hunkapiller, M. A. McClure, Hidden Markov models of biological primary sequence information Proceedings of the National Academy of Sciences of the United States of America. ,vol. 91, pp. 1059- 1063 ,(1994) , 10.1073/PNAS.91.3.1059
Shun-ichi Amari, Information geometry of the EM and em algorithms for neural networks Neural Networks. ,vol. 8, pp. 1379- 1408 ,(1995) , 10.1016/0893-6080(95)00003-8
A. P. Dempster, N. M. Laird, D. B. Rubin, Maximum Likelihood from Incomplete Data Via theEMAlgorithm Journal of the Royal Statistical Society: Series B (Methodological). ,vol. 39, pp. 1- 22 ,(1977) , 10.1111/J.2517-6161.1977.TB01600.X
PIERRE BALDI, YVES CHAUVIN, Hidden Markov Models of the G-protein-coupled receptor family. Journal of Computational Biology. ,vol. 1, pp. 311- 336 ,(1994) , 10.1089/CMB.1994.1.311
R. M. Neal, A new view of the EM algorithm that justifies incremental and other variants Biometrika. ,(1993)
Lon R Cardon, Gary D Stormo, None, Expectation maximization algorithm for identifying protein-binding sites with variable lengths from unaligned DNA fragments. Journal of Molecular Biology. ,vol. 223, pp. 159- 170 ,(1992) , 10.1016/0022-2836(92)90723-W
Frank G. Ball, John A. Rice, Stochastic models for ion channels: Introduction and bibliography Mathematical Biosciences. ,vol. 112, pp. 189- 206 ,(1992) , 10.1016/0025-5564(92)90023-P
D. Haussler, A. Krogh, I.S. Mian, K. Sjolander, Protein modeling using hidden Markov models: analysis of globins hawaii international conference on system sciences. pp. 792- 802 ,(1993) , 10.1109/HICSS.1993.270611
Anders Krogh, Michael Brown, I.Saira Mian, Kimmen Sjölander, David Haussler, HIDDEN MARKOV MODELS IN COMPUTATIONAL BIOLOGY: APPLICATIONS TO PROTEIN MODELING Journal of Molecular Biology. ,vol. 235, pp. 1501- 1531 ,(1993) , 10.1006/JMBI.1994.1104