Toward robust learning of the Gaussian mixture state emission densities for hidden Markov models

Hao Tang, Mark Allan Hasegawa-Johnson, Thomas S Huang

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

One important class of state emission densities of the hidden Markov model (HMM) is the Gaussian mixture densities. The classical Baum-Welch algorithm often fails to reliably learn the Gaussian mixture densities when there is insufficient training data, due to the large number of free parameters present in the model. In this paper, we propose a novel strategy for robustly and accurately learning the Gaussian mixture state emission densities of the HMM. The strategy is based on an ensemble framework for probability density estimation in which the learning of the Gaussian mixture densities is formulated as a gradient descent search in a function space. The resulting learning algorithm is called "the boosting Baum-Welch algorithm." Our preliminary experiment results on emotion recognition from speech show that the proposed algorithm outperforms the original Baum-Welch algorithm on this task.

Original languageEnglish (US)
Title of host publication2010 IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2010 - Proceedings
Pages5242-5245
Number of pages4
DOIs
StatePublished - Nov 8 2010
Event2010 IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2010 - Dallas, TX, United States
Duration: Mar 14 2010Mar 19 2010

Publication series

NameICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing - Proceedings
ISSN (Print)1520-6149

Other

Other2010 IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2010
CountryUnited States
CityDallas, TX
Period3/14/103/19/10

Keywords

  • Baum-Welch algorithm
  • Boosting
  • Gaussian mixture density
  • Hidden Markov model

ASJC Scopus subject areas

  • Software
  • Signal Processing
  • Electrical and Electronic Engineering

Fingerprint Dive into the research topics of 'Toward robust learning of the Gaussian mixture state emission densities for hidden Markov models'. Together they form a unique fingerprint.

Cite this