An Improved Intrusion System Design using Hybrid Classification Technique

Abstract

data mining is an essential tool for current age technology. It is very useful for various applications such as business intelligence, computational cloud and other research and science based projects. In this presented paper a new kind of intrusion system design is proposed and their implementation is presented using MATLAB. The proposed attack classification technique is based on the C4.5 decision tree and the supervised back propagation neural network. During the attack classification the proposed system first pre- processes the input dataset. Therefore, in order to reduce the dimensions of the dataset, the KPCA algorithm is applied and then the C4.5 data model is implemented for extracting the classification rules form the C4.5 classifier. These rules are in the form of IF THEN ELSE format. Now the defined rules are processed by the using the back propagation neural network. The proposed model is tested on the different size of KDD cup dataset and the performance is provided. According to the obtained results the proposed data model provides the improved classification rates.