Lazy Bayesian Rules: A Lazy Semi-Naive Bayesian Learning Technique Competitive to Boosting Decision Trees

Lbr is a lazy semi-naive Bayesian classi er learning technique, designed to alleviate the attribute interdependence problem of naive Bayesian classi cation. To classify a test example, it creates a conjunctive rule that selects a most appropriate subset of training examples and induces a local naive Bayesian classi er using this subset. Lbr can signi cantly improve the performance of the naive Bayesian classi er. A bias and variance analysis of Lbr reveals that it signi cantly reduces the bias of naive Bayesian classi cation at a cost of a slight increase in variance. It is interesting to compare this lazy technique with boosting and bagging, two well-known state-of-the-art non-lazy learning techniques. Empirical comparison of Lbr with boosting decision trees on discrete valued data shows that Lbr has, on average, signi cantly lower variance and higher bias. As a result of the interaction of these e ects, the average prediction error of Lbr over a range of learning tasks is at a level...

• Type: Conference
• Year: 1999
• Where: ICML
• Authors: Zijian Zheng, Geoffrey I. Webb, Kai Ming Ting
• Keywords: Bayesian Classi Cation, Bayesian Classi Er, ICML 1999, Machine Learning, Naive Bayesian Classi