Encyclopedia of Machine Learning

2010 Edition
| Editors: Claude Sammut, Geoffrey I. Webb

Tree Augmented Naive Bayes

  • Fei Zheng
  • Geoffrey I. Webb
Reference work entry
DOI: https://doi.org/10.1007/978-0-387-30164-8_850




Tree augmented  naive Bayes is a  semi-naive Bayesian Learning method. It relaxes the naive Bayes attribute independence assumption by employing a tree structure, in which each attribute only depends on the class and one other attribute. A maximum weighted spanning tree that maximizes the likelihood of the training data is used to perform classification.

Classification with TAN

Interdependencies between attributes can be addressed directly by allowing an attribute to depend on other non-class attributes. However, techniques for learning unrestricted Bayesian networks often fail to deliver lower zero-one loss than naive Bayes (Friedman, Geiger, & Goldszmidt, 1997). One possible reason for this is that full  Bayesian networksare oriented toward optimizing the likelihood of the training data rather than the conditional likelihood of the class attribute given a full set of other attributes. Another possible reason is that full Bayesian networks have high variance...
This is a preview of subscription content, log in to check access.


  1. Chow, C. K., & Liu, C. N. (1968). Approximating discrete probability distributions with dependence trees. IEEE Transactions on Information Theory, 14, 462–467.zbMATHCrossRefGoogle Scholar
  2. Friedman, N., Geiger, D., & Goldszmidt, M. (1997). Bayesian network classifiers. Machine Learning, 29(2), 131–163.zbMATHCrossRefGoogle Scholar

Copyright information

© Springer Science+Business Media, LLC 2011

Authors and Affiliations

  • Fei Zheng
  • Geoffrey I. Webb

There are no affiliations available