Rao-SVM Machine Learning Algorithm for Intrusion Detection System

Abstract

Most of the intrusion detection systems are developed based on optimization algorithms as a result of the increase in audit data features; optimization algorithms are also considered for IDS due to the decline in the performance of the human-based methods in terms of their training time and classification accuracy. This article presents the development of an improved intrusion detection method for binary classification. In the proposed IDS, Rao Optimization Algorithm, Support Vector Machine (SVM), Extreme Learning Machine (ELM), and Logistic Regression (LR) (feature selection and weighting) were combined with NTLBO algorithm with supervised ML techniques (for feature subset selection (FSS). Being that feature subset selection is considered a multi-objective optimization problem, this study proposed the Rao-SVM as an FSS mechanism; its algorithm-specific and parameter-less concept was also explored. The prominent intrusion machine-learning dataset, UNSW-NB15, was used for the experiments and the results showed that Rao-SVM reached 92.5% accuracy on the UNSW-NB15 dataset.