"A New Feature Selection Technique For Intrusion Detection Systems"

Network-based technologies continue to grow in use among individuals, professionals, and businesses worldwide. However, most network-based systems have been discovered to be significantly vulnerable to attacks. A malicious attack on network-based systems can have severe and devastating consequences. For example, an attack on a power utility network could result in the loss of electricity for millions of individuals and offices. Computer scientists have been trying to develop advanced Intrusion Detection Systems (IDSs) capable of identifying and counteracting malicious attacks in order to strengthen network security. Machine Learning (ML) algorithms have been proven to be promising in the automatic detection of attacks and intrusions on a network. The practice of selecting data features that a model can use when making predictions is important in developing and training ML-based IDSs. Researchers from Canadian University Dubai in the UAE created a new feature selection method that could help allow the development of more effective ML-based IDSs. Their feature selection method, MICorr, addresses some of the limitations of existing feature selection techniques. The researchers tested their method on the CSE-CIC-IDS2018 dataset, which contains 10,000 benign and malicious network intrusion instances. The new feature method is said to address the challenge of considering continuous input features and discrete target values. They showed that their proposed method performs well against the benchmark selection methods. This article continues to discuss the vulnerability of network-based systems to attacks, ML-based IDSs, and the new feature selection technique developed for IDSs. 

NewsUpdate UK reports "A New Feature Selection Technique For Intrusion Detection Systems"