Hands-On Artificial Intelligence for Cybersecurity

From images to malware

In the description that follows, we will show an alternative approach to malware detection that takes advantage of the typical skills of CNNs in image recognition. But in order to do this, it is first necessary to represent the executable code of the malware in the form of an image to be fed to the CNN.

This approach was described in the paper entitled Towards Building an Intelligent Anti-Malware System: A Deep Learning Approach using Support Vector Machine (SVM) for Malware Classification by Abien Fred M. Agarap, in which each executable malware is treated as a binary sequence of zeros and ones, which is then translated into a gray-scale image.

In this way, it is possible to recognize the malware families based on ...