Comparison of Two Learning Methods of the Tree Augmented Naïve Bayesian Network Classifier

Generative learning and discriminative learning are two different classifier learning methods. Bayesian network classifiers belong to in nature generative classifiers because the learners always attempt to find the Bayesian network that maximizes likelihood rather than classification accuracy. In order to improve the classification performance, many researchers is trying to train the generative classifier in a discriminative way. This paper introduces two learning approaches of a restricted Bayesian network classifier, tree augmented naive Bayesian network (TAN), and compares them from several different aspects through the experiments. The experimental results demonstrate that there are diversity between the generative learning and the discriminative learning of the TAN classifier