An Algorithm for Incremental Tree-augmented Naive Bayesian Classifier Learning

A new algorithm for incremental Bayesian classifier is proposed in this paper. The algorithm has two advantages. First, because our algorithm is based on the score function, it can make full use of the prior knowledge. Meanwhile, it can take the degree of fit between the added instances and the original classifier into consideration. Second, our algorithm expand incremental model from the naïve Bayesian classifier to the tree-augmented naïve Bayesian classifier. Our experimental results on a variety of UCI data sets illustrate the efficiency of this new algorithm.