A Comparative Study of Machine Learning Algorithms on Intrusion Detection System

Abstract

To detect malicious activity, an intrusion detection system (IDS) automates the procedure of observing and reasoning events that take place in the computer network. The existing intrusion detection system is confined to particular sorts of malicious activity, and it may not be able to identify new types of malicious activity, thus ML techniques were employed to implement the detection system at a faster rate. The intrusion detection system employs ML technologies such as random forest and support vector machines. This system has three main modules: data preparation, feature mapping, modelling and accuracy analyser. In this paper accuracy and sensitivity of both the support vector and random forest algorithms will be compared, with the results verified at a faster rate. The results show that machine learning approaches can aid intrusion detection using a dataset (KDD '99) that also highlights the findings of the prediction model which can differentiate between intrusions and normal connections.