Enhancing Network Security: An Intrusion Detection Approach Using Artificial Neural Networks and NSL-KDD Dataset
R. Saranya, S. Silvia Priscila · 2024
This research work presents an innovative intrusion detection system that utilizes an Artificial Neural Network (ANN) architecture with an attention mechanism. Intrusion detection systems are critical in defending against increasingly sophisticated cyber threats. The proposed system incorporates the attention mechanism into the ANN framework, enabling the model to effectively prioritize important features in network traffic data and filter out irrelevant noise. Utilizing the NSL-KDD dataset, a well-regarded benchmark for intrusion detection systems, the methodology includes comprehensive preprocessing steps to ensure data quality. The proposed model demonstrates a significant enhancement in intrusion detection performance, achieving an accuracy score of 99.928%. Comparative analysis with Logistic Regression, SVM, and Random Forest shows better performance, underscoring its efficacy. It highlights a significant advancement in enhancing intrusion detection systems, focusing on the integration of attention mechanisms into neural network architectures, and provides a robust defense against evolving cyber threats.