Comparing and evaluating HMM ensemble training algorithms using train and test and condition number criteria

Authors:
R. I. A. Davis;Brian C. Lovell
Affiliations:
The University of Queensland, The Intelligent Real-Time Imaging and Sensing (IRIS) Group, School of Information Technology and Electrical Engineering,, Australia;The University of Queensland, The Intelligent Real-Time Imaging and Sensing (IRIS) Group, School of Information Technology and Electrical Engineering,, Australia
Venue:
Pattern Analysis & Applications
Year:
2003

Citing 0
Cited 2

Evaluation of incremental learning algorithms for HMM in the recognition of alphanumeric characters

Pattern Recognition
Improved parameters estimating scheme for E-HMM with application to face recognition

ICB'06 Proceedings of the 2006 international conference on Advances in Biometrics

Quantified Score

Hi-index	0.00

Visualization

Abstract

Hidden Markov Models have many applications in signal processing and pattern recognition, but their convergence-based training algorithms are known to suffer from over-sensitivity to the initial random model choice. This paper describes the boundary between regions in which ensemble learning is superior to Rabiner’s multiple-sequence Baum-Welch training method, and proposes techniques for determining the best method in any arbitrary situation. It also studies the suitability of the training methods using the condition number, a recently proposed diagnostic tool for testing the quality of the model. A new method for training Hidden Markov Models called the Viterbi Path Counting algorithm is introduced and is found to produce significantly better performance than current methods in a range of trials.