Android malware detection using deep learning classification approach / Nur Amirah Amri and Mohd Faris Mohd Fuzi

Android devices are becoming increasingly popular and there are more threats to Android users because malware writers are shifting their focus to exploiting vulnerabilities of Android devices for malicious behaviour. This paper will study Android malware detection using a deep learning classification approach. Deep learning is a thriving research area with many successful applications in different fields. Recently, these techniques have been applied to detect mobile malware and have once again shown their ability to remedy this type of problem. In this paper, Android software will be analysed by using malware analysis tools like APKTool and 010 Editor. Some selected features will be extracted from this process and compiled into a csv file. The selected features will be trained using the CNN and RNN model approach. The performance of Android malware detection using CNN and RNN model will be analysed by measuring its accuracy based on Metric Formula Definition Accuracy. According to the development process, CNN is performing better by detecting android malware with a 96 percent accuracy, while RNN delivers a 75 percent accuracy.