Network Intrusion Detection System using Machine Learning and Deep Learning Approach

Rohan Chauhan

doi:10.37591/jons.v10i1.899

Abstract

Networks play a significant part in today’s world; fast internet and communication industries result in vast network size and data expansion. Furthermore, attackers aiming to launch various cyberattacks inside the system cannot be neglected. An IDS keeps track of the network’s software and hardware security to preserve its privacy, integrity, and accessibility. Despite the significant efforts of the researchers, current IDS continue to confront challenges in terms of accuracy rate, reducing false alerts, and detecting new attacks. To overcome the challenges listed above, IDS-based on ML and DL systems have recently been developed as practical approaches to identifying intrusions across the network effectively. As machine learning approaches generally apply, they can also discover unknown risks. DL is a high-performance ML subfield that becomes a significant study subject. This study first clarifies the IDS idea and taxonomy. Then there are regularly used machine and deep learning algorithms and assessment criteria and datasets utilized in network-based IDS design. In this comprehensive study of existing IDS-based studies, the approach employed in suggested solutions is explored. Finally, we emphasized a variety of research problems and suggested future research topics to improve ML and DL based NIDS by using current approaches limitations.

Abstract

Keywords