Survey on intrusion detection systems based on deep learning

Abstract

Intrusion Detection Systems (IDSs) have a significant role in all networks and information systems in the world to earn the required security guarantee. IDS is one of the solutions used to reduce malicious attacks. As attackers always changing their techniques of attack and find alternative attack methods, IDS must also evolve in response by adopting more sophisticated methods of detection. The huge growth in the data and the significant advances in computer hardware technologies resulted in the new studies existence in the deep learning field, including intrusion detection. Deep learning is sub-field of Machine Learning (ML) methods that are based on learning data representations. In this paper, a detailed survey of various deep learning methods applied in IDSs is given first. Then, a deep learning classification scheme is presented and the main works that have been reported in the deep learning works is summarized. Utilizing this approach, we have provided a taxonomy survey on the available deep architectures and algorithms in these works and classify those algorithms to three classes, which are: discriminative, hybrid and generative. After that, chosen deep learning applications are reviewed in a wide range of fields of intrusion detection. Finally, popular types of datasets and frameworks are discussed.

Application of Deep Learning for Quality of Service Enhancement in Internet of Things: A Review

“…In this review paper, DL techniques for collision avoidance and the detection of obstacles autonomously are presented together with various datasets for IoT DL-UAV systems. Lateef et al [29], carried out a survey of DL-based Intrusion Detection Systems (IDS) for IoT-based systems. They note that unsupervised DL models, such as AutoEncoders (AE), are more suitable for implementing IDS, but there is still a challenge of lack of training data.…”

“…This review paper's objective is to address this gap. (2) Various research papers recommend future research for the application of DL-based techniques for intrusion detection [29,30] and resource allocation and management [31], which are the main factors that determine the QoS of IoT networks and systems. Therefore, this review takes up this recommendation to provide researchers with the application of DL to QoS enhancement in IoTs.…”

EmSM: Ensemble Mixed Sampling Method for Classifying Imbalanced Intrusion Detection Data

“…To address this problem, machine learning techniques have been used in several areas to improve cyber security [3]. Based on current advancements, supervised machine learning techniques that learn and classify network behaviors have achieved higher true-positive and lower false-positive rates than existing signature-based approaches [4,5].…”

An Overview of Neural Net Models for Intrusion Detection System

“…Аналіз останніх досліджень і публікацій. В оглядових статтях [4][5][6][7][8][9][10] наводиться класифікація найбільш поширених методів роботи IDS: підходи на основі виявлення сигнатур потенційних атак і на основі пошуку аномалій у даних. У роботі [6] наводиться детальний аналіз переваг і недоліків цих двох методів.…”

“…У роботі [4] проведено класифікацію наявних методів глибинного навчання на генеративні, описові та гібридні. Генеративні методи, наприклад, нейронні мережі автокодувальники, використовуються для абстрактного представлення ознак мережевого запиту або розширення навчальної вибірки.…”