On the structure of hidden Markov models

Karim Abou-Moustafa, Ching Suen, Mohamed Cheriet

Abstract: This paper investigates the effect of HMM structure on the performance of HMM-based classifiers. The investigation is based on the framework of graphical models, the diffusion of credits of HMMs and empirical experiments. Although some researchers have focused on determining the number of states, this study shows that the topology has a stronger influence on increasing the performance of HMM-based classifiers than the number of states.