Classification Trees with Logistic Regression Functions for Network Based Intrusion Detection System

Abstract

Intrusion Detection Systems considered as an indispensable field of network security to detect passive and anomaly activities in network traffics and packets. In this paper a framework of network based intrusion detection system has been implemented using Logistic Model Trees supervised machine learning algorithm."NSL-KDD" dataset which is an updated dataset from "KDDCup 1999" benchmark dataset for intrusion detection has been used for the experimental analysis using percent of 60% for training phase and the rest for testing phase. The testing and experimental results from the proposed structure shows that using two way functions which are classification with regression combined in Logistic Model Tree is very accurate in term of accuracy and minimum false-positive average with high true-positive average. Two classifications has been performed in the proposed model which are (Attack or Normal)