Ensemble Machine Learning Approach for Android Malware Classification Using Hybrid Features

摘要: Feature-based learning plays a crucial role at building and sustaining the security. Determination of software based on its extracted features whether benign or malign process, particularly classification into correct malware family improves security operating system protects critical user’s information. In this paper, we present novel hybrid feature-based for Android samples. Static such as permissions requested by mobile applications, hidden payload, dynamic API calls, installed services, network connections are classification. We apply machine evaluate level in accuracy different classifiers extracting using fairly large set 3339 samples belonging to 20 families. The evaluation study has been scalable with 5 guest machines took 8 days processing. testing is reached 92%.