An Ensemble-Based Classifier Network Intrusion Detection

Abstract

An Intrusion Detection System monitors the flow of data on a network through the computer for the search of malicious activities which are majorly known as threats and viruses. There are two types of intrusion detection, one, Signature-based detection where the intrusion detection collects the information, analyses it, and then compares them to the attack signatures stored in the database. The second one is an Anomaly-based intrusion detection system that learns normal and anomalous behaviour by analysis in various benchmark datasets. Common challenges for Intrusion Detection Systems are large amounts of data to process, low detection rates, and high rates of false alarms. Considering anomaly pattern as detecting a point in time where the behaviour of the system is unusual and significantly different from past behaviour. In such context anomaly detection mean detecting the behaviours that deviate from normal behaviours. An ensemble based classifier method is considered using Naïve Bayes and Multivariate Linear Regression algorithms. To get a better accuracy rate of the intrusion for real time data packets are received in the system. On the experimental results achieved, we are proposing the Naïve Bayes and Multivariate techniques as an efficient method for network intrusion detection.