Andromaly: a behavioral malware detection framework for android devices

摘要: This article presents Andromaly--a framework for detecting malware on Android mobile devices. The proposed realizes a Host-based Malware Detection System that continuously monitors various features and events obtained from the device then applies Machine Learning anomaly detectors to classify collected data as normal (benign) or abnormal (malicious). Since no malicious applications are yet available Android, we developed four applications, evaluated Andromaly's ability detect new based samples of known malware. We several combinations detection algorithms, feature selection method number top in order find combination yields best performance Android. Empirical results suggest is effective devices general particular.