Deep Learning in Malware Identification and Classification

Abstract

Albeit the cyber world has become an essential part and the lifeline of the present day, there are threats associated with it. People access the cyber world for various services like networking, banking, communication, shopping, and for other uses. Malware is one of the primary and perilous threats among malevolent software for the decades in the cyber and the computing world. Due to its magnification in volume and in complexity, malware and its variant identification and classification are the most central and severe problems nowadays. Since malware inception, more and more malware is engendered and designed, as time passes; more intricate malware is designed enormously. Researchers and analysts are perpetually probing for a solution that is the most efficacious to fight back with malware. The most-famed methods utilized for malware analysis is signature-based detection, static, and dynamic analysis. In recent years, signature-based detection has been proven ineffective against the escalation of malware and its variants. Malware classification is attracting widespread interest due to its vast proliferation. In this chapter, we have chosen to discuss and explore another method of malware analysis that is image-based malware analysis utilizing deep learning. We are specifically discussing malware classification utilizing malware visualization and deep learning, one of the most widely implemented techniques in many real-world applications. To better understand the concept from a practical perspective, we additionally discussed and implemented a fundamental level malware classifier, for the reader’s further research and study purpose. The main objective of this chapter is to avail readers a better and in-depth understanding of malware classification, visualization, deep learning algorithms and emerging challenges, open issues.