Deep Learning-Based Malware Detection via Meta-Information and Manifest Analysis in Android Systems

Day by day, the number of users of electronic devices is increasing around the world, as electronic-based systems have become a part of human life.The Android system is one of the largest and most popular systems used in different devices such as mobile devices.The ease of using Android system applications and their availability is main reason behind the widespread a vast number of applications.For example, gaming, social networking, and other sensitive applications such as banking application systems.However, as android is an open-source system, it becomes a valuable opportunity to increase malware and malicious applications to achieve cybercrime such as privacy violation, theft, extortion, and also other crimes that have become a major challenge for security information specialist and a important threat to the daily human life.Regardless of many developed protection systems and methods for detecting malware, the development and methods used by cybercriminals make some of these traditional methods useless in protecting devices and information.Hence, researchers have resorted to the use of artificial intelligence-based techniques to develop different protection systems against malwares.This study proposes deep learning-based technique for malware detection based on the meta information of applications; accuracy results for Various Deep Learning Techniques (RNNs, LSTMs, CNNs, and Bi-LSTMs) are 88.3, 88.3, 88.4 and 88.2 respectively .