AdversarialDroid: A Deep Learning based Malware Detection Approach for Android System Against Adversarial Example Attacks
Giuseppe D'Ambrosio, Wenjia Li · 2021
As the predominant mobile operating system world-wide, Android suffers from various types of malware, which could cause severe security and privacy issues. To cope with them, many research efforts have been made to develop effective mal ware detectors. However, malware authors tend to evade detection by launching adversarial example attacks, in which Android applications could be mutated and thus causing confusion to malware detectors. In this paper, we propose a deep learning based approach to identify malware for the Android system in the presence of adversarial example attacks. To validate the proposed approach, we have conducted experimental study on real-world datasets.