Optimization of HMM by a genetic algorithm

Authors:
  • SKS. Kwong
    S. Kwong
  • CDC.K. Diu
    C.K. Diu
  • WFW.R. Fahrner
    W.R. Fahrner
Publisher website
Google Scholar
Abstract
The hidden Markov model (HMM) is a natural and highly robust statistical methodology for automatic speech recognition. It is also being tested and proved considerably important in a wide range of applications. The model parameters of the HMM are essential in describing the behavior of the utterance of the speech segments. Many successful heuristic algorithms are developed to optimize the model parameters in order to best describe the trained observation sequences. However, all these methodologies explore for only one local maxima in practice. No one methodology can recover from the local maxima the global maxima or other more optimized local maxima. A stochastic search method called the genetic algorithm (GA) is presented for HMM training. The GA mimics natural evolution and perform global searching within the defined searching space. Experiments showed that using the GA for HMM training (GA-HMM training) result in a better performance than using other heuristic algorithms.