Malware analysis based on smart agents and image classification

Windows-based systems and operating systems in general are significantly damaged, affecting infrastructures. At present, Malware analysis is performed in laboratories that use high costs and resources; so there are few methods of classification of Malware, based on artificial intelligence that consumes few resources. This article provides a system that was developed for the dynamic analysis of malware in Windows and classified using SIFT, SURF, and Bayesian networks. This involves the transformation of infected files into image files that allows the identification and classification of Malware. The samples of malicious software that allows generating a contingency plan were identified. The system was developed using intelligent agents. The analysis of Postal worm malware is presented as an example. When comparing with other malware detection and classification systems, it is observed that the multi-agent-based system is competitive.