Measures on Hidden Markov Models

Abstract

Hidden Markov models were introduced in the beginning of<br />the 1970's as a tool in speech recognition. During the last decade<br />they have been found useful in addressing problems in computational<br />biology such as characterising sequence families, gene finding,<br />structure prediction and phylogenetic analysis. In this paper<br />we propose several measures between hidden Markov models. We<br />give an efficient algorithm that computes the measures for leftright<br />models, e.g. profile hidden Markov models, and discuss how<br />to extend the algorithm to other types of models. We present an<br />experiment using the measures to compare hidden Markov models<br />for three classes of signal peptides.