Machine Learning-Based Network Intrusion Detection System

Abstract

The Internet is an extensive and interconnected network of computers and gadgets worldwide. The demand for security measures is more critical as usage and traffic increase. There are both benign and malignant users on the Internet, and they all have equal access to information. Unfortunately, malicious users, also known as hackers, use this access to infiltrate organizations' systems and inflict significant harm. These attacks can take many forms, including stealing sensitive data, disrupting operations, and spreading viruses or malware. Organizations must implement robust security mechanisms to counteract these threats to safeguard their networks, devices, and data. Robust security mechanisms include measures such as firewalls, encryption, antivirus software, and employee training on best practices for online security. By adopting a preventive security strategy, organizations can mitigate the risk of cyber-attacks and protect their critical data. The firewall defends the enterprise from malicious attacks, but no network can be completely secure. So, it is proposed to install IDS on top of the firewall; it monitors any attempt to breach it, gets access to systems on the trusted side, and issues an alert. Using machine learning (ML), an IDS is suggested in this paper. The NSL-KDD dataset is used to test the Logistic Regression (LR), Stacking CV Classifier, K-Nearest Neighbor (KNN), MLP classifier, Catboost Classifier, Linear Support Vector Machine (LSVM), Random Forest (RF), Decision Tree (DT), and XGBoost Classifier technique.