Classifier Algorithm with Attribute Selection and Optimization for Intrusion Detection System.

Abstract

The purpose of this study is to determine how computer network security needed in line with the increasing number of interconnected networks. At present the attacks on computer networks continue to increase, so an efficient network intrusion detection mechanism is needed. Data mining methods over the past few years have been very popular for use in detecting network intrusion. In this paper, we propose reduce the dimensions of the datasets in the pre-processing step by using different state-of-the-art dimension reduction techniques, Principal Component Analysis (PCA) and Information Gain (IG). Particle swarm optimization (PSO) and Genetic Algorithm (GA), both used to find more appropriate set of attributes for classifying intrusion, and κ-Nearest Neighbors (κ-NN) algorithm is used as a classifier. The results of the experiments we conducted used the KDD99Cup dataset standard, showing a comparative level of accuracy from the use of dimension reduction and classification optimization. The use of reducing the IG dimension in the KDD99Cup dataset with κ-NN based PSO optimization can be better at detecting intrusion than other methods.