ISCA Archive Eurospeech 1997
ISCA Archive Eurospeech 1997

Using simulated annealing expectation maximization algorithm for hidden Markov model parameters estimation

Jacques Simonin, Chafic Mokbel

This paper presents the use of a simulated annealing technique during the parameters estimation of a Hidden Markov Model (HMM) in a speech recognition system. This technique allows to move out of a local optimum which characterizes a classical Expectation Maximization (EM) algorithm, and thus to achieve a better estimation with a limited amount of training data. We choose here the Simulated Annealing Expectation Maximization (SAEM) algorithm introducing a simulated annealing technique in the EM method. The SAEM algorithm is compared to the classical EM algorithm, for both task- independent and task-dependent Viterbi training. The evaluation leads to significant improvement of recognition performances.


doi: 10.21437/Eurospeech.1997-155

Cite as: Simonin, J., Mokbel, C. (1997) Using simulated annealing expectation maximization algorithm for hidden Markov model parameters estimation. Proc. 5th European Conference on Speech Communication and Technology (Eurospeech 1997), 449-452, doi: 10.21437/Eurospeech.1997-155

@inproceedings{simonin97_eurospeech,
  author={Jacques Simonin and Chafic Mokbel},
  title={{Using simulated annealing expectation maximization algorithm for hidden Markov model parameters estimation}},
  year=1997,
  booktitle={Proc. 5th European Conference on Speech Communication and Technology (Eurospeech 1997)},
  pages={449--452},
  doi={10.21437/Eurospeech.1997-155},
  issn={1018-4074}
}