| In this talk, we shall discuss supervised classification of sequence data based on finite order Markov Models. A crucial step in the implementation of such classifiers is the identification of an appropriate order of the Markov Model. We shall discuss some criteria for determining the order of the Markov Model and discuss their performance in terms of resulting misclassification rates. We shall also consider sequence data arising from non-Markov Models and investigate the behavior of our classifiers based on Markov likelihood for such data. |