Incremental Estimation of Discrete Hidden Markov Models Based on a New Backward Procedure
Abstract
We address the problem of learning discrete hidden Markov models from very long sequences of observations. Incremental versions of the Baum-Welch algorithm that approximate the β-values used in the backward procedure are commonly used for this problem, since their memory complexity is independent of the sequence length. We introduce an improved incremental Baum-Welch algorithm with a new backward procedure that approximates the β-values based on a one-step lookahead in the training sequence. We justify the new approach analytically, and report empirical results that show it converges faster than previous incremental algorithms.
Cite
Text
Florez-Larrahondo et al. "Incremental Estimation of Discrete Hidden Markov Models Based on a New Backward Procedure." AAAI Conference on Artificial Intelligence, 2005.Markdown
[Florez-Larrahondo et al. "Incremental Estimation of Discrete Hidden Markov Models Based on a New Backward Procedure." AAAI Conference on Artificial Intelligence, 2005.](https://mlanthology.org/aaai/2005/florezlarrahondo2005aaai-incremental/)BibTeX
@inproceedings{florezlarrahondo2005aaai-incremental,
title = {{Incremental Estimation of Discrete Hidden Markov Models Based on a New Backward Procedure}},
author = {Florez-Larrahondo, German and Bridges, Susan and Hansen, Eric A.},
booktitle = {AAAI Conference on Artificial Intelligence},
year = {2005},
pages = {758-763},
url = {https://mlanthology.org/aaai/2005/florezlarrahondo2005aaai-incremental/}
}