A Study on Machine Learning based Malware Detection System

Abstract. The rapid increase of malware variances has been a big security threat over the past years. Malware invade our systems over the network through mails, websites, external devices, etc. to steal data and information without the user knowing. The traditional ways to fight malware has lost its effectiveness due to the inability to counter new unknown malware variants. So, there has been a rise in use of machine learning in detection systems where the process involves feature extraction and malware classification. This has been reported to be highly effective to pace up with the advancement in malware variances. In this paper, some of the recent machine learning based malware detection approaches with focus on ransomware are studied. With the aim to aid new researchers, a preliminary work is also discussed using CICAndMal2017 as the dataset and Random Forest as classifier with a result of 79.79% accuracy. Keywords: Machine Learning, Malware Detection, Ransomware Detection, Classification.