A systematic and comprehensive survey of recent advances in intrusion detection systems using machine learning: deep learning, datasets, and attack taxonomy

Recently, intrusion detection systems (IDS) have become an essential part of most organisations’ security architecture due to the rise in frequency and severity of network attacks. To identify a security breach, the target machine or network must be watched and analysed for signs of an intrusion. It is defined as efforts to compromise the confidentiality, integrity, or availability of a computer or network or to circumvent its security mechanisms. Several IDS have been proposed in the literature to efficiently detect such attempts exploiting different characteristics of cyberattacks. These systems can provide with timely sensing the network intrusions and, subsequently, notifying the manager or the responsible person in an organisation. Important actions are then carried out to reduce the degree of damage caused by the intrusion. Organisations use such techniques to defend their systems from the network disconnectivity and increase reliance on the information systems by employing intrusion detection. This paper presents a detailed summary of recent advances in IDS from the literature. Nevertheless, a review of future research directions for detecting malicious operations and launching different attacks on systems is discussed and highlighted. Furthermore, this study presents detailed description of well-known publicly available datasets and a variety of strategies developed for dealing with intrusions.