Deep Learning for Cybersecurity: A Review
Zhaolin Chen
Abstract
With the development of Internet technology, the scale of Internet has increased considerably, which brings with a large number of cyber-attacks. Traditional protection techniques are confronted with complex, advanced and ongoing evolvement adversarial situations, which have to be more adaptive and responsive in order to handle future security and privacy problems. Deep Learning (DL), as one of the most currently remarkable machine learning techniques, has a great potential in cybersecurity. In this paper, author is committed to analyze current cyber-attacks, to review recent state-of-the-art deep learning algorithms and Figure out pros and cons of them, and to discuss the feasibility of deep learning technology applied to cybersecurity to defend malware attacks, DDoS attacks and spoofing attacks. This article also analyzes some vulnerabilities of deep learning algorithms and potential security problems which might come out when DL is applied to cybersecurity. Finally, paper discusses some challenges a DL-based defense mechanism has to overcome, and status and future directions of DL-based defense technology.