A Comprehensive Study on Machine Learning Algorithms for Intrusion Detection System

Rapid advancements in areas of communication and the internet ensued a significant boost in data and capacity of the network. As a consequence, a plethora of novel threats are being built, creating it harder for network security to efficiently identify breaches. Furthermore, intruders also initiate multiple attacks inside this network must be taken into account. An intrusion Detection System (IDS) is a method that checks network activity for availability, consistency, and secrecy to defend the network from potential invasions. Although the greatest efforts of professionals, IDSs continue to struggle with enhancing detecting accurateness whilst lowering false alarm rates and detecting newer intrusions. IDS systems based on ML (Machine Learning), as well as DL (Deep Learning), have recently been deployed as feasible procedures for swiftly identifying network intrusions. The taxonomy presented in this article is based on well-known ML as well as DL approaches for building network-based IDS (NIDS) systems, as well as it first defines the concept of IDS. In this comprehensive review of existing NIDS-based research, the merits and limitations of the suggested solutions are thoroughly explored. This evaluation may act as a standard for researchers and industry in terms of the advancement and development of Network Intrusion Detection Systems in the future.