Machine Learning for Intrusion Detection in Cyber Security: Applications, Challenges, and Recommendations
Samreen Naeem, Aqib Ali, Sania Anam, Muhammad Munawar Ahmed · Innovative Computing Review · 2022
Modern life revolves around networks and cybersecurity has emerged as a critical study field. The health of the software and hardware running on a network is monitored by an Intrusion Detection System (IDS) which is a fundamental cybersecurity approach. After decades of research, the existing IDSs have developed the capability to confront hurdles in order to improve detection accuracy, reduce false alarm rates, and detect unexpected attacks. Many academics have concentrated on designing such IDSs that employ machine learning approaches to overcome the aforementioned difficulties. Machine learning approaches are capable to discover important distinctions that exist between normal and aberrant data with great accuracy. Moreover, these approaches are also very generalizable which allows them to detect unknown attacks. The survey conducted in the current study offers ataxonomy of IDS based on machine learning that uses data objects as the critical dimension to classify and summarize the IDS literature. This form of classification structure is appropriate for cyber security researchers.