Binary Arithmetic Optimization Algorithm with Machine Learning based Intrusion Detection System

Abstract

Intrusion Detection Systems (IDS) are significant for preventing and identifying malicious actions in computer networks. Machine Learning (ML) approaches are extremely executed for recognizing intrusion since it is investigating huge volumes of network traffic data and recognize designs indicative of intrusions. But, the performance of these ML approaches is greatly dependent upon the choice of relevant features which efficiently represent the network traffic data. Feature Selection (FS) is the procedure of recognizing the most informative and discriminative aspects in a given database. As part of Intrusion Detection (ID) utilizing ML, FS purposes for identifying the subset of features that are efficiently differentiated between normal network behaviour and malicious activities. This article proposes a Binary Arithmetic Optimization Algorithm with Machine Learning based Intrusion Detection System (BAOA-MLIDS) technique. The BAOA-MLIDS technique employs FS with an optimal ML classifier for the ID process. To accomplish this, the BAOA-MLIDS technique performs data preprocessing to scale the input data. Besides, the BAOA-MLIDS technique comprises BAOA based FS approach to choose optimal features. Moreover, Extreme Learning Machine (ELM) approach is utilized for the identification of the intrusions. Furthermore, Hunger Games Search Optimization (HGSO) approach was employed for the hyperparameter optimization of the ELM approach. The performance assessment of the BAOA-MLIDS model was examined on a standard dataset and the outputs outperformed the advancement of the BAOA-MLIDS model in the ID process.