Optimized Intrusion Detection System for Attack Classification Using Machine Learning and Deep Learning Techniques

This study addresses the escalating challenges in designing practical Intrusion Detection Systems (IDS) due to network traffic’s growing intricacy and volume. A novel approach is proposed, employing a hybrid feature encoding method and utilizing Machine Learning (ML) and Deep Learning (DL) techniques for binary and multiclass network traffic classification. The system incorporates RMSPROP as an optimizer for binary classification and Adam for multiclassification. Evaluations conducted on the NSL-KDD dataset demonstrate impressive accuracy, reaching 99.28% for ML and 97.76% for DL in binary classification, 97.03% for DL and 90% for ML in detecting DoS attacks, 97.03% for DL and 90% for ML in Prop attacks, 97.03% for DL and 78% for ML in R2L attacks, and 97.03% for DL and 99.3% for ML in U2R attacks. The results underscore the effectiveness of the proposed optimized IDS, showcasing advancements in accuracy and performance through state-of-the-art ML and DL algorithms.