End-to-end malware detection for android IoT devices using deep learning

摘要: Abstract The Internet of Things (IoT) has grown rapidly in recent years and become one the most active areas global market. As an open source platform with a large number users, Android driving force behind rapid development IoT, also attracted malware attacks. Considering explosive growth years, there is urgent need to propose efficient methods for detection. Although existing detection based on machine learning achieved encouraging performance, these require lot time effort from analysts build dynamic or static features, so are difficult apply practice. Therefore, end-to-end without human expert intervention required. This paper proposes two deep learning. Compared methods, proposed have advantage their process. Our resample raw bytecodes classes.dex files applications as input models. These models trained evaluated dataset containing 8K benign malicious applications. Experiments show that can achieve 93.4% 95.8% accuracy respectively. our not limited by filesize, no manual feature engineering, low resource consumption, they more suitable application IoT devices.