Enhanced Intrusion Detection using Machine Learning Techniques

Over the last few years, increased cyber threat activity has highlighted the need for secure network systems. Intrusion Detection Systems (IDS) are key in detecting and mitigating malicious activity within network traffic. This work explores the application of several machine learning techniques Logistic Regression, Random Forest, XGBoost, and K-Nearest Neighbors (KNN) to enhance intrusion detection using the KDDTrain+ dataset. The dataset is an enhanced version of the original KDD'99, aiding in eliminating issues of redundancy as well as imbalance. Preprocessing of the dataset, normalization of features, and performance evaluation of the models in terms of accuracy, precision, recall, and F1-score are done. The results demonstrate that ensemble methods such as XGBoost and Random Forest are superior to the classic classifiers in terms of detection rate as well as computational cost. The proposed work proves that machine learning-based IDS is capable of intrusion detection and can be further optimized for real-time use.