Research in the field of artificial intelligence (AI) provides new information about deep neural network (DNN)-based methods, also known as deep learning. This allows computational models, composed of countless processing layers, to learn representations of data with different levels of abstraction. Using deep learning in the field of cybersecurity can help with identify flaws, risks, and abnormal behavior, to improve the detection efficiency and protection offered by cybersecurity systems.

The main contributions of this book relate to various cybersecurity applications, including intrusion, cyberattacks and malware detection, face recognition, network traffic analysis, and some other tasks. This solutions-based reference synthesizes foundational deep learning methods, cybersecurity, and mathematical content with real-world applications. Its intended audience includes interdisciplinary researchers, students, and those new to the field.

The book contains 11 chapters, each addressing in detail how deep learning methods can be used to achieve specific cybersecurity goals.

Chapter 1 teaches readers about the new attack and defense techniques used for deep learning frameworks. The results of the attack methods experienced in the applications presented here will be of major interest to readers, representing real challenges for identifying defense solutions but also for integrating them into the deep learning process.

Chapter 2 explains an intelligent security control system based on deep learning methods. The authors present results of the application of neural networks in an attack detection scheme, as well as an adaptive neuro-fuzzy method for assessing the degree of compromise of the system.

Chapter 3 is about one of the major problems for cyber-physical systems: person re-identification. For the task of re-identifying a person, which aims to compare images of the same subject obtained from different cameras, efficient methods are presented based on convolutional neural networks (CNNs).

Chapter 4 discusses the effective counteracting of cyberattacks against electricity theft. Models based on deep learning to detect these attacks are carefully analyzed and very well argued.

Chapter 5 presents deep learning techniques of CNNs, used for the analysis of network traffic and the detection and classification of malicious traffic. For cybersecurity, these techniques are part of the development of innovative and effective malware defense mechanisms.

Chapter 6 covers a deep learning framework to counter domain name system (DNS) poisoning attacks. Violating the availability of web resources when accessing the DNS name, intercepting almost any traffic on the network and possibly replacing requests, network responses, or a trusted object on the network, are some types of threats of DNS attacks.

Developed based on deep learning architectures, and implemented and tested by the authors, the framework is able to learn domain names, classify them, and extract information specific to DNS attacks. The experimental results obtained using the framework for counteracting DNS attacks, with a careful analysis of the types of neural networks used, are presented in detail in chapter 7.

The detection of anomalies in software defined networking (SDN) based on a recurring neural network is the main topic addressed in chapter 8. SDN is a new approach to networks, which fundamentally changes the way they operate and manage. An intrusion detection system (IDS) is proposed, made with deep learning solutions and tested and evaluated in the context of SDN.

Chapter 9 discusses the modeling of Android malware in order to identify them. Classification methods based on neural and convolutional networks are applied. With a more efficient classification process and shorter training times, the methods presented are useful for both the protection of Android applications and the environments in which they run.

Lastly, chapters 10 and 11 are about the development of CNN models and methods that use invariant features characteristic of a face image. Used in cybersecurity to detect and identify a face, as well as to estimate age, the proposed methods are highly accurate and qualify as useful tools for combating child pornography.

Deep learning applications for cyber security addresses interdisciplinary topics that make deep learning a tool of major interest for cybersecurity. The proposed solutions are very well argued and exemplified. The accuracy of the presented methods is carefully analyzed and compared with the accuracy obtained by other techniques. Many of the results obtained allow the issuance of recommendations based on the behavior of the applications presented. This is why the book is recommended for researchers and students, as well as for all those interested in applying deep learning as part of cybersecurity products or platforms.