A Deep Learning-Based Framework for Android Malware Family Classification

The rising threat of Android malware infection causes malware detection a research concern. The archaic techniques used by antimalware systems are limited to signature-based detection and deficient for new variants of Android malware. Various approaches have been proposed by various researchers to stand up to the attacks of Android mobile world. Apart from detecting inhabitation of malware in Android applications (apps) family reorganization of that malware is important as well. In this paper, a Deep Neural Network (DNN)-based detection framework is proposed for family classification of Android malware. This framework uses permissions and intents as features from Android apps. The proposed framework detects the malware and also able to classify the family it belongs. The framework achieved malware detection accuracy of 96.12% and also able to classify Android malware family with accuracy 91.54%.