ABSTRACT
Nowadays, cyber security seems to be an interesting topic that presents many complicated problems in terms of intrusion detection, data confidentiality, proactive identification of attacks, etc. With the exponential development of adversarial approaches, emerging challenges have grown progressively complex and sophisticated. For this reason, deep learning (DL) approaches have proved to be helpful to the defense sector and have become a core tool toward cyber security. Through pattern recognition, real-time cyberwarfare visualization, and comprehensive vulnerability assessment, DL preventively stamps out cyberattacks and promotes secure network infrastructure. The use of the DL method in cyber security is going to increase more than ever. DL promising responses which can be impactful toward zero-day attacks, like starting from IP traffic classification and filtering unwanted activity for intrusion detection. The latest study is now being undertaken using predictive traffic features, and DL approaches. A comprehensive analysis of DL approaches for cyber security applications, including numerous security and defensive privacy approaches, is mentioned in this chapter. For DL techniques with their implementation, a short beginner guide summary is given. A wide variety of attacks are addressed in this chapter, including poisoning attacks, backdoor, adversarial examples, and model stealing.
