Deep Learning Methods applied to Intrusion Detection: Survey, Taxonomy and Challenges
Oumaima Lifandali, Noreddine Abghour · 2021 International Conference on Decision Aid Sciences and Application (DASA) · 2021
Because of the popularity of the Internet of Things (IoT), the rapid expansion of computer networks, and the vast number of important applications, cyber security has lately garnered a lot of attention in today's security issues. As a result, identifying different cyber-attacks or abnormalities in a network, as well as constructing an efficient intrusion detection system that plays a key part in today's security, is becoming increasingly critical. Such a data-driven intelligent intrusion detection system may be built using artificial intelligence, particularly machine learning techniques. This survey provides a thorough review of Machine Learning (ML) techniques for cybersecurity intrusion detection systems, with an emphasis on new Deep Learning-based approaches (DL). The study examines current techniques in terms of intrusion detection processes, performance outcomes, and limits, as well as whether or not they use benchmark datasets to provide a fair assessment. In addition, a thorough examination of cybersecurity benchmark datasets is provided. This article aims to offer a roadmap for readers interested in learning more about the potential of deep learning techniques for cybersecurity and intrusion detection systems, as well as a thorough examination of the benchmark datasets used to train DL models in the literature.