AI-Driven Intrusion Detection: A Machine Learning-Based Approach

As cyber threats continue to rise and intrusion techniques rapidly evolve, a more powerful security system is needed to detect intrusions. This paper applied machine learning to develop an intrusion detection system capable of analyzing attack signatures and classifying them as either malicious or benign. Traditional intrusion detection systems(IDS) rely on signature-based methods, which struggle to detect novel or evolving cyber threats, making them ineffective against newer, more sophisticated attacks. An AI-powered approach fixes this problem by leveraging its ability to recognize complex patterns. Supervised learning was used to build five machine learning-based classification models, K-Nearest Neighbor, Random Forest, Logistic Regression, Support Vector Machine, and Extreme Gradient Boosting. The models were applied and trained on the CSE-CIC-IDS2018 dataset after preprocessing. The best performing model was Extreme Gradient Boosting with a classification accuracy of 99.96%. The worst performing model was Logistic Regression with a classification accuracy of 97.43%.