Tree-augmented Naive Bayes ensembles

Ensemble learning is an effective method of improving classification accuracy of the classifier. TAN, Tree-Augmented Naive Bayes, is a tree-like Bayesian network. The standard TAN learning algorithm is the stable, which is difficult to improve its accuracy by bagging technique. In this paper, a new TAN learning algorithm called RTAN is presented, and the diversity of the TAN classifiers generated by RTAN is investigated by K statistic. And then Bagging-MultiTAN algorithm generates a TAN ensemble classifier. Through the comparisons of this TAN ensemble classifier with the standard TAN classifier in the experiments, the TAN ensemble classifier shows higher classification accuracy than the standard TAN classifier on the most data.