Network Intrusion Detection System using Deep Learning Paradigm

Abstract

Network Intrusion Detection Systems (NIDS) are essential for mitigating the pervasive threats posed by hackers in computer networks. With the rise in network traffic, applying deep learning techniques to enhance IDS performance has become crucial. Deep learning (DL), a subset of machine learning, involves multiple layers of information processing for learning. This paper compares the effectiveness of Deep Belief Networks (DBN), Convolutional Neural Networks (CNN), and Long Short-Term Memory Recurrent Neural Networks (LSTM-RNN) in developing a high-performance IDS. The deep learning techniques were validated using standard intrusion datasets that include novel attack patterns. The experimental results demonstrated accuracies of 80.61% for LSTM-RNN, 55% for DBN, and 81.07% for CNN, indicating that CNN outperformed the other techniques.