Machine Learning Detection and Analysis on Obfuscated Android Malware

University of Technology Sydney. Faculty of Engineering and Information Technology.Android is the leading mobile operating system, accounting for more than 80% of the market share. Millions of Android apps exist in various app markets such as GooglePlay, AppChina, Huawei app store, etc. The success of Android has attracted the interest of hackers who build thousands and even millions of Android malware that are distributed over the app markets. Detecting such malware is thus of the utmost importance to preserve the security and privacy of billions of Android app users. In this thesis, the author focuses on three main research challenges. First, the author investigates the problem of Android malware labeling. This is a key challenge that needs to be overcome to build reference datasets, to then be able to use AI techniques. Second, he investigates the impact of obfuscation and deobfuscation on the detection of piggybacked apps. Finally, he studies the problem of generating relevant testing samples (i.e., adversarial samples) to assess antivirus products