Detection of Android malicious applications based on APIs

Abstract

Machine learning technology is widely used to detect Android malware, among which the selection of features is particularly important. Compared with the permission applied, the APIs invoked by applications can better reflect the behavior of the application. Therefore, many researches choose APIs as the features of machine learning classification algorithm to detect Android malware. Generally, researchers simply select APIs as features, rather than focus on how to select features more beneficial for detection. This paper proposes an Android malware detection method based on APIs frequent pattern selection and use Relief and particle swarm optimization algorithm to jointly select feature subset for detection.