Application of the tree augmented naive Bayes network to classification and forecasting

Tang, Adelina Lai Toh. (2005). Application of the tree augmented naive Bayes network to classification and forecasting PhD Thesis, School of Information Technology and Electrical Engineering, The University of Queensland.

       
Attached Files (Some files may be inaccessible until you login with your UQ eSpace credentials)
Name Description MIMEType Size Downloads
THE18487.pdf Full text application/pdf 5.84MB 1
Author Tang, Adelina Lai Toh.
Thesis Title Application of the tree augmented naive Bayes network to classification and forecasting
School, Centre or Institute School of Information Technology and Electrical Engineering
Institution The University of Queensland
Publication date 2005
Thesis type PhD Thesis
Supervisor Prof Tom Downs
Dr. Marcus Gallagher
Total pages 114
Language eng
Subjects 0906 Electrical and Electronic Engineering
0899 Other Information and Computing Sciences
L
Formatted abstract

The thesis commences with a review of basic Static Bayesian Networks (SBNs) and describes some of the methods of probabilistic inference from SBNs, including Pearl's causal tree methodology and the more general NP-complete clique tree methodology due to Lauritzen and Speigelhalter. The use of SBNs in pattern classification will then be described with special consideration given to the use of elementary forms of SBNs including the Naive Bayes and the Tree Augmented NaŃ—ve Bayes (TAN) classifiers. The possibility of applying boosting and other ensemble methods to these elementary forms in order to obtain superior performance will then be explored. The discussion will conclude by investigating the suitability of correlation measures in computing the TAN as an initial step in adapting it to solve the forecasting problem. Attention will then be turned to the use of Bayesian Networks (BN) in forecasting. Numerous techniques have been developed to create accurate forecasting models and the BN approach, with a time element incorporated, is among the more successful of these. The time element will be introduced by means of a series of SBNs, acting as time-slices to create a Dynamic Bayesian Network (DBN), to solve the forecasting problem. The suitably modified TAN combined with the Pearl causal tree, now called the TAN-Pearl Network (TPN), will form the basis of the DBN. The objective of the forecasting problem will be to minimize overall error, computed from the differences between the computed beliefs at each time-slice, and those of the actual events. The final major investigation concerns the application of boosting to regression. It draws upon the parallel between time-slices in a DBN and instances in regression analysis and it will be shown that the accuracy of the time-slices, and therefore that of the DBN as a whole, can be improved through boosting. 

Keyword Bayesian statistical decision theory
Neural networks (Computer science)

 
Citation counts: Google Scholar Search Google Scholar
Access Statistics: 131 Abstract Views, 1 File Downloads  -  Detailed Statistics
Created: Thu, 24 May 2012, 15:03:06 EST by Talha Alam on behalf of The University of Queensland Library