Feature Selection Technique-Based Network Intrusion System Using Machine Learning

Abstract

Internet is a global public network, and as internet traffic has grown, so has the demand for security mechanisms. There are both harmful and harmless users on the Internet, and both have access to the same information. Malicious users get access to any organization's systems and cause significant damage. As a result, the necessity for the organization's private resources security has increased dramatically. Firewalls were installed by every corporation to protect their networks, although no network can be completely secure. Firewalls are topped with intrusion detection systems (IDS). The firewall defends the company against malicious attacks, and the IDS detects and generates an alert if someone attempts to intrude the firewall and has access to the system. In this paper, an IDS based on Machine Learning (ML) is proposed. The K-Nearest Neighbour (KNN), Naive Bayes (NB), Random Farest (RF), and Decision Tree (DT) ML technique are applied for NSL-KDD dataset. Besides, a Recursive Feature Elimination (RFE) is used for feature selection technique to enhance the performance, accuracy, and processing time of the model.