Forensic Identification of Android Trojans Using Stacked Ensemble of Deep Neural Networks.

As the user base of Android operating system grows steadily, the ecosystem became a growing target for malicious actors. With Trojans representing over 93% of all Android malware, this type of malicious code becomes a serious threat to Android users. In this paper, we present a forensic identification system to identify Android trojan families based on dynamic features extracted from malicious applications. Our proposed system is based on a stacked ensemble of deep neural networks. The proposed system was tested using CIC-AndMal-2020 dataset, and have shown accuracy and $$F_1$$ score exceeding 0.98 in identifying trojan families effectively.