Unveiling Android Malware: A Deep Learning Perspective Using RNN and LSTM
Abdullah Alqammaz, Mohammad Rasmi Al-Mousa, Ali Abu Zaid, Mahmoud Rajallah Asassfeh, Mohammad Abdulrazzaq Rajab, Sherien Dajah, Abdulla M F Alali, Mohammed Khouj · 2024
The advent and increasing popularity of Android devices have brought about undesirable mobile malware and its attendant applications that constitutes a danger to society in terms of privacy and information leakage. Because the Android open-source OS is attractive to criminals, it seems impossible for old methods of virus detection to be effective as they used to be. The present study tackles this vital problem by offering an innovative enhancement that uses deep learning algorithms: recurrent neural networks and long short-term memory networks for android malware detection. New solid classification schema was created which focused on the meta-information of applications in order to differentiate entre benign and malicious apps. In our experiments, both RNN and LSTM models yielded improvements in performance metrics which suggests that they can be used to counter the current computer malware challenges. This study adds to the body of knowledge available on the detection of malware and lays the foundation for other studies designed to employ other couplings of deep learning algorithms in further boosting the security of Android applications.