DEEP LEARNING APPLICATIONS IN CYBER SECURITY: A COMPREHENSIVE REVIEW, CHALLENGES AND PROSPECTS

Abstract

— Deep Learning applications are starting to pervade every commercial and technological sector. This paper surveys deep learning (DL) methods for cyber security applications, highlights the security considerations when using deep learning networks and presents possible malicious uses of such models. Initially, a brief description of the architectures studied and presented in this paper is provided. In the next section, the security applications are indicated for different models including CNNs, RNNs and GANs. Lastly, security is-sues are divided into two sections: attacks on various deep learning net-works, and attacks on a network using Deep Learning models themselves.