Genetic algorithms for feature selection in an intrusion detection application

This research explores the applicability of genetic algorithms to the problem of selection of optimized feature subsets to reduce the error caused by using hand-selected features. Chapter I introduces the intrusion detection system that integrates fuzzy logic and data mining methods. Chapter II briefly reviews the knowledge of fuzzy logic, data mining, genetic algorithms and feature selection techniques. Chapter III introduces the problem that will be tackled in this thesis. Chapter IV presents the design issue of a general GA system for feature selection and fuzzy membership function optimization for a fuzzy system, including data structures for genes and chromosomes, and genetic operators including crossover, mutation. Chapter V presents the implementation of the GA system for this research. Chapter VI presents the experimental results, which demonstrate that genetic algorithms are effective for simultaneous feature selection and optimization of fuzzy membership functions for fuzzy control system. Chapter VII presents the summary.