A Deep Learning-based Approach for Malware Classification using Machine Code to Image Conversion

Abstract

Malwares are the number one threat to computer security. The number is staggering, as some antivirus companies report over 5 million malware samples per day. The security team cannot deal with all this malware at the same time. Malware classification schemes are typically required to concentrate on these occurrences. The size, type, and complexity of malware are increasing. Hackers and attackers often develop systems that can automatically transfer and encrypt code to avoid detection. Methodologies or Classic machine learning techniques, in which classifiers learn based on handmade feature vectors are not suitable methods for classifying malware families. Deep convolutional neural networks have recently illustrated efficiency in malware classification and detection. This article proposes a system that classifies malware corresponding to its family. Introducing a new method for multi-class classification challenges. What I want to drop deeper into is not only the classification of Malware but also using Deep Learning Models for classifications. Transforms malwares binaries to grayscale images and applies CNN to classify their families.