Machine Learning and Deep Learning approaches for Network Intrusion Detection: A Systematic Study

Abstract: Network dimensions and its related data is increasing at a rapid rate due to the huge improvement in the area of internet and broadcasting. This has widely imposed a hindrance in accurate detection of intrusions since more and more innovative tactics are introduced. Existence of invaders with a goal to harm the network cannot be disregarded. Inspection of network traffic is done in order to fend off a system from intrusion with the help of a tool known as intrusion detection system (IDS) so that the system’s confidentiality and integrity is maintained. IDS encounters various issues in achieving high detection accuracy rate and reduction of false alarm rate although several attempts have been made to improve it. Many IDS build on basis of ML and DL are being efficiently deployed as a possible remedy to detect intrusions in a system. This paper showcases the advancements of IDS with the help of a mechanism, selection of dataset and evaluation metrics. It also depicts the shortcomings of the existing IDS and highlights the scope of research for betterment of IDS.

Keywords: Network Intrusion Detection System, Firewall, Machine Learning, Cyberspace.