An intrusion detection scheme based on the ensemble of discriminant classifiers

The cyber-physical system and modern technology are useful in various applications of the network. However, various types of bugs and vulnerabilities are also brought along with modern technologies. The attacks caused by vulnerabilities create huge losses, necessitating the need to detect these attacks. Although, considerable work has been done by the researchers so far, but novel attacks are yet to be detected. Existing schemes are suitable for denial of service (DOS) type of attack class. However, these schemes are not efficiently detecting other types of attack classes such as Probing, Remote to User (R2L) and User to Root (U2R). In view of this, a new intrusion detection scheme based on ensemble of discriminant classifiers is proposed in this paper. In ensemble of discriminant classifier method, weak learners are converted into strong learners. KDDcup99 dataset has been used in the proposed scheme for empirical evaluation. The results show that the proposed scheme is superior in detecting all types of attack classes by achieving 98.9% overall accuracy. Network security, Cyber-physical system, Intrusion detection, Ensemble methods, Discriminant classifier, Denial of service (DOS) (C) 2020 Elsevier Ltd. All rights reserved.