Citation
Sonai Muthu Anbananthen, Kalaiarasi and Hossen, Md. Jakir and Sayeed, Md. Shohel (2011) Comparison of network pruning and tree pruning on artificial neural network tree. Australian Journal of Basic and Applied Sciences, 5 (9). pp. 1093-1098. ISSN 1991-8178
Text
Comparison of network pruning and tree pruning on artificial neural network tree.pdf Restricted to Repository staff only Download (121kB) |
Abstract
Artificial Neural Network (ANN) has not been effectively utilized in data mining because of its “black box” nature. This issue was resolved by using the Artificial Neural Network Tree (ANNT) approach in the authors’ earlier works. The ANNT approach derives symbolic knowledge (rules) to provide some explanation of how the classification or prediction of the ANN is obtained. To enhance extraction, pruning will be incorporate with this approach where two pruning techniques are evaluated to see which method is best to use with ANNT. The first technique is to prune the neural network and the second technique is to prune the tree. The first technique analytically measures the amount of information gained by each of the ANN links as a result learning (training). The one with the lowest information is pruned. Rules are extracted from the pruned network. The second pruning will prune the tree that is built from the neural network and then the tree is converted to rules. These two techniques are evaluated with the ANNT algorithm in the insurance domain to see which method of pruning is most suitable with ANNT in terms of comprehensibility and accuracy.
Item Type: | Article |
---|---|
Additional Information: | SHDL Call Number: AN MFS:69707435 |
Uncontrolled Keywords: | Data mining, Artificial neural network, Network pruning, Tree network, Rule extraction |
Subjects: | T Technology > T Technology (General) |
Divisions: | Faculty of Engineering and Technology (FET) Faculty of Information Science and Technology (FIST) |
Depositing User: | Ms Rosnani Abd Wahab |
Date Deposited: | 13 Jan 2014 06:39 |
Last Modified: | 12 Jan 2017 07:42 |
URII: | http://shdl.mmu.edu.my/id/eprint/4818 |
Downloads
Downloads per month over past year
Edit (login required) |