Malicious Android Applications’ Classification using Machine Learning

Smartphones have been an integral part of everyday life for society since their development. Naturally, the popularity of these devices has brought an equal amount of malware designed specifically for these smartphone devices. The struggle to keep these devices secure and in turn the sensitive information stored on these devices from getting into the wrong hands has become an ever-evolving endeavor. With the sheer amount of malware produced coupled with the intelligent, polymorphic nature of the malicious software, it has become increasingly difficult to protect against them. In this paper, we propose a dynamic approach to classifying malicious Android applications that does not rely solely on the signatures of said applications. Instead, we analyze the Android Manifest of the dataset to classify whether an application should be considered malicious or benign.

Keywords:APK; repackaging; Android Manifest; Smali; Dex; Dalvic executable; Google play store