Deep Learning Techniques for Intrusion Detection Systems: A Survey and Comparative Study

Nowadays cyber threats become increasingly sophisticated and prevalent. Intrusion Detection Systems (IDS) have been widely used, to achieve the necessary security requirements in computer networks because of their ability to detect network attacks. Recently, utilizing machine learning (ML) and deep learning (DL) models in IDS have demonstrated substantial improvements in identifying unknown attacks. This study conducts a comprehensive analysis of DL approaches for intrusion detection focusing on the recent research in the last five years, and explores the most used datasets in the field to highlight their characteristics and suitability for evaluating IDS performance. Finally, we present insights into the limitations, strengths, and future prospects of DL based IDS.