Detección de malware con modelo de lenguaje y su clasificación mediante SVM

Malware analysis is a more difficult task each time, due to malware developers are taking care of avoiding and detection techniques, from obfuscation to destroying, hard drive (if malware detect that has been analyzed). In this paper we release a malware dynamic analysis of six types of malware: Trojans, Worms, Virus, Trojan-Spys, Backdoors y Rootkits, moreover a set of Whiteware using an n-grams language model in Win API calls for every sample. Finally, we used a Polynomial Kernel SVM to malware classification prediction, where we have obtained 70% of performance and malware classification and 100% for Whiteware classification in which case determine if is a malicious sample and in this form we can do malware detection.