A minimum discrimination information approach for hidden Markov modeling
作者： Y. Ephraim , A. Dembo , L. Rabiner
DOI: 10.1109/ICASSP.1987.1169727
关键词:
摘要: A new iterative approach for hidden Markov modeling of information sources which aims at minimizing the discrimination (or cross-entropy) between source and model is proposed. This does not require commonly used assumption that to be modeled a process. The algorithm started from estimated by traditional maximum likelihood (ML) alternatively decreases over all probability distributions agree with given measurements models. proposed procedure generalizes Baum ML modeling. shown descent measure its local convergence proved.
ucsb.edu
sci-hub.se 参考文章(5)
L. Bahl, P. Brown, P. de Souza, R. Mercer, Maximum mutual information estimation of hidden Markov model parameters for speech recognition international conference on acoustics, speech, and signal processing. ,vol. 11, pp. 49- 52 ,(1986) , 10.1109/ICASSP.1986.1169179
Nguyen A., On the uniqueness of the maximum-likeliwood estimate of structured covariance matrices IEEE Transactions on Acoustics, Speech, and Signal Processing. ,vol. 32, pp. 1249- 1251 ,(1984) , 10.1109/TASSP.1984.1164445
Michael Sabin, Robert Gray, Global convergence and empirical consistency of the generalized Lloyd algorithm IEEE Transactions on Information Theory. ,vol. 32, pp. 148- 155 ,(1986) , 10.1109/TIT.1986.1057168
R. Gray, A. Gray, G. Rebolledo, J. Shore, Rate-distortion speech coding with a minimum discrimination information distortion measure IEEE Transactions on Information Theory. ,vol. 27, pp. 708- 721 ,(1981) , 10.1109/TIT.1981.1056410
I. Csiszar, $I$-Divergence Geometry of Probability Distributions and Minimization Problems Annals of Probability. ,vol. 3, pp. 146- 158 ,(1975) , 10.1214/AOP/1176996454