A sensitivity test for network Intrusion detection based on Bayesian Network using a wrapper approach

Abstract

Anomalous traffic detection on internet is a major issue of security as per the growth of smart devices and this technology. Several attacks are affecting the systems and deteriorate its computing performance. Intrusion detection system is one of the techniques, which helps to determine the system security, by alarming when intrusion is detected. This Thesis presents a sensitivity test on intrusion detection based on Bayesian network using a wrapper approach. The results obtained were analyzed based on KDD, CIDD-005 data set and WEKA machine learning tool. Various performance measures and better accuracy was found with varying population size at 93.394% and 99.0235% respectively for population size 2 and 10. 98.7774%, 98.6821%, 98.5988%, 98.7695% respectively for population size of 30, 40 50, 60. The performance was compared with existing research and it was relatively good since accuracy was 90% and above.