Comparing Feature Selection Method For Neural Network Classification

Siti Khalijah, Mohd Yasin (2012) Comparing Feature Selection Method For Neural Network Classification. Project Report. UTeM, Melaka, Malaysia. (Submitted)

	PDF (24 Pages) Comparing_Feature_Selection_Method_For_Neural_Network_Classification_-_24_Pages.pdf - Submitted Version Download (403kB)
	PDF (Full Text) Comparing_Feature_Selection_Method_For_Neural_Network_Classification_-_Full_Text.pdf - Submitted Version Restricted to Registered users only Download (1MB)

Abstract

Feature selection plays an important part in classifying systems in Neural Networks. A set of attributes which are relevant, irrelevant or redundant is desirable. The purpose of this project is to compare two methods of feature selection (Principle Component Analysis and Support Vector Machine) in order to gain the best result. Sample of images that has been converted into numeric data, it will be extracted and select by any feature selection technique. Back propagation is used to recognize the dataset and classified into the correct group. The extracted image or output from the two techniques has been used as an input for classification purpose. Back propagation is an iterative process that can often take a great deal of time to complete since the input data is feed forward network. Finally, the output or result of the classification will be analyzed in term of classification accuracy.

Actions (login required)

View Item