Program analysis and machine learning techniques for mobile security

• Main Author: Soh, Charlie Zhan Yi
• Other Authors: Chen Lihui
• Format: Theses and Dissertations
• Language: English
• Published: 2019
• Subjects: DRNTU::Engineering::Computer science and engineering::Software::Software engineering; DRNTU::Engineering::Computer science and engineering::Computing methodologies::Artificial intelligence
• Online Access: https://hdl.handle.net/10356/106079; http://hdl.handle.net/10220/47894
• Institution: Nanyang Technological University

Over the past few years, concerns have been raised with respect to the increasing number of malicious and clone apps infiltrating the Android markets. Android malware may perform a range of malicious activities (e.g., exfiltrating sensitive information and sending premium SMS) and clone apps steal revenue from the original developer. The detection of these adversary apps is non-trivial as in depth understanding of the apps is required. Furthermore, due to the arms race between the adversary apps and the detection algorithms, the adversary apps are constantly evolving and becoming more sophisticated. Hence, new and more effective algorithms are imperative. This thesis proposes three methods and one empirical study with suggested solutions for Android apps analysis. We address four specific issues that plague Android security, namely, clone detection, third-party library detection, malware detection and concept drift. We do so through leveraging on program analysis, Machine Learning and Deep Learning techniques.