Intelligent Intrusion Detection System Using Deep Learning Models

Abstract

Cyber attacks are a very common issue in the modern world, and since there is a growing array of challenges in accurately detecting intrusion, this results in damage to security services, i.e. confidentiality, integrity, and availability of data. The attackers found new types of attacks day by day, first of all the type of attack should be analyzed properly with the help of IDS for the prevention of these types of attacks to offer the correct answers. IDS that play an important role in network security have three major features: first collection of data, selection of feature last is decision of engine. As data are increasing day by day due to which the attacks on the data also increases because of these increasing numbers of attacks on the data, the existing security applications are insufficient. SIDS and AIDS intrusion detection systems are separate proposed methods of intrusion detection to manage security threats. This paper has reviewed numerous deep learning algorithms that have been proposed to detect intrusion, i.e., Convolutional Neural Network, Recurrent Neural Network, Restricted Boltzmann Machine, Deep Brief Network and Auto encoder. It is designed to use IDS approach depending on a deep learning (DL) algorithm by using literature work comparisons and by providing the expertise either in intrusion detection or deep learning algorithms.