A New Systematic Network Intrusion Detection System Using Deep Belief Network
A Akshai, M. V. Anushri, P Sonu · 2023
This study discusses a Network Intrusion Detection System (NIDS) leveraging Deep Belief Networks to classify network traffic into multiple categories. By stacking Multiple Restricted Boltzmann Machines, the NIDS model fosters a Deep Belief Network-a generative graphical model. This implied approach is particularly good at identifying and identifying high-dimensional representations. Using Gibbs Sampling and the Contrastive Divergence method, the Deep Belief Network is first pre-trained unsupervised before being fine-tuned supervised. Our research study utilized the CICIDS2018 dataset to train and assess the effectiveness of our proposed DBN technique. Our empirical findings show that our two-phase training technique not only maintains impressive performance against other types of assaults, but also dramatically improves detection accuracy when comprehending samples of peculiar attacks.