Machine Learning Based Android Malware Detection: A comprehensive Survey

Abstract

Android Operating System’s open-source nature has attracted a broader range of developers which has resulted in an unprecedented proliferation of Android-based devices in various sectors of the economy. However, despite the fact that this growth has resulted in significant technological advances and the ease of conducting business (e-commerce) and social interactions, they have also become powerful platforms for unregulated cyber-attacks and espionage against business infrastructures and individual users of these mobile devices. Malicious application attacks, as opposed to other attack strategies such as social engineering, have taken the lead when it comes to cyber-attack tactics. Android malware has advanced in complexity and intelligence to the point where it is now highly resistant to existing detection systems, especially signature-based systems. Machine Learning techniques have emerged as a more capable choice for countering the complexity and innovation used by emerging Android malware. Machine Learning models operate by first learning existing patterns of malware behavior and then using that information to isolate or classify any related behavior from unknown sources. As found in recent literature, this paper presents a thorough analysis of Machine Learning techniques and their applications in Android malware detection.