A Novel Discriminative Naive Bayesian Network for Classification
Kaizhu Huang (Fujitsu Research and Development Centre Co. Ltd., China), Zenglin Xu (Chinese University of Hong Kong Shatin, Hong Kong), Irwin King (Chinese University of Hong Kong Shatin, Hong Kong), Michael R. Lyu (Chinese University of Hong Kong Shatin, Hong Kong) and Zhangbing Zhou (Zhou, Bell-Labs, Lucent Technologies, China)
Copyright: © 2007
Naive Bayesian network (NB) is a simple yet powerful Bayesian network. Even with a strong independency assumption among the features, it demonstrates competitive performance against other state-of-the-art classifiers, such as support vector machines (SVM). In this chapter, we propose a novel discriminative training approach originated from SVM for deriving the parameters of NB. This new model, called discriminative naive Bayesian network (DNB), combines both merits of discriminative methods (e.g., SVM) and Bayesian networks. We provide theoretic justifications, outline the algorithm, and perform a series of experiments on benchmark real-world datasets to demonstrate our model’s advantages. Its performance outperforms NB in classification tasks and outperforms SVM in handling missing information tasks.