Learning the Structure of Augmented Bayesian Classifiers (2002)
| Citations: | 9 - 1 self |
BibTeX
@MISC{Keogh02learningthe,
author = {Eamonn J. Keogh and Michael J. Pazzani},
title = {Learning the Structure of Augmented Bayesian Classifiers},
year = {2002}
}
Years of Citing Articles
Bookmark
 |
 |
 |
 |
 |
 |
OpenURL
Abstract
this paper we take a different approach to constructing augmented Bayesian networks. Rather than directly attempting to approximate the underlying probability distribution, we concentrate solely on using the same representation to improve classification accuracy. We show two methods for constructing augmented Bayesian networks, a hill climbing greedy search, and a novel, more efficient search algorithm, which we call SuperParent. We compare these to Friedman and Goldszmidt's approach and show that the best approximation of the underlying probability distribution does not necessarily result in the most accurate classifier
