An Approach to Malware Detection Using Error Back Propagation Networks

With new malware being created every day the onus is on the researchers to identify their unique signatures to detect them. This puts systems to risk against these unknown malware for a considerable amount of time. Also with the advent of polymorphic and metamorphic viruses the job of the researchers is even more arduous. Thus in this paper we propose the extraction of application programming interface (API) calls from malware subcategories. Also as each malware has its own infection mechanism API calls differ. We propose the use of Neural networks for classifying executables based on their relevant API calls.