FEATURE SELECTION FOR THE FUZZY ARTMAP NEURAL NETWORK USING A HYBRID GENETIC ALGORITHM AND TABU SEARCH
Loading...
Date
2007-07
Authors
TANG, WENG CHIN
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
The performance of Neural-Network (NN)-based classifiers is strongly
dependent on the data set used for learning. In practice, a data set may contain noisy
or redundant data items. Thus, feature selection is an important step in building an
effective and efficient NN-based classifier. In this thesis, the research of a hybrid
algorithm of Genetic Algorithm (GA) and Tabu Search (TS) for feature selection in the
Fuzzy ARTMAP NN classifier is presented. The proposed GA-TS algorithm embeds
the recency and frequency memory structures of TS into the search process of the GA.
The recency memory structure helps induce an additional diversification mechanism in
the GA search process. On the other hand, the frequency memory structure provides
guidance to genetic operator and helps intensify the GA search process. A series of
empirical studies comprising benchmark and real-world problems is employed to
evaluate the effectiveness of the proposed hybrid GA-TS algorithm. A simulated noisy
feature injection method is devised to assess the capabilities of GA-TS in identifying
and removing noisy features that can degrade classification accuracy. Experimental
results demonstrate that proposed GA-TS performs better in terms of feature
compactness (the number of features reduced) and classification accuracy than the
ordinary GA.
Description
Keywords
NETWORK , SEARCH