An Analysis of K-means Algorithm Based Network Intrusion Detection System

In this modern age, information technology (IT) plays a role in a number of different fields. And therefore, the role of security is very important to control and assist the flow of activities over the network. Intrusion detection (ID) is a kind of security management system for computers and networks. There are many approaches and methods used in ID. Each approach has merits and demerits. Therefore this paper highlights the similar distribution of attacks nature by using K-means and also the effective accuracy of Random Forest algorithm in detecting intrusions. This paper describes full pattern recognition and machine learning algorithm performance for the four attack categories, such as Denial-of-Service (DoS) attacks (deny legitimate request to a system), Probing attacks (information gathering attacks), user-to-root (U2R) attacks (unauthorized access to local super-user), and remote-to-local (R2L) attacks (unauthorized local access from a remote machine) shown in the KDD 99 Cup intrusion detection dataset.