Automaticlly Learning Featurs Of Android Apps Using CNN
Yuxin Ding, Wei-Guo Zhao, Zhipan Wang, Longfei Wang · 2018
With the popularity of intelligent mobile terminals, mobile malware has been spreading quickly and widely. Among them, Android-based mobile devices have surpassed the mobile devices of other platforms as the most popular tool. Android-based smartphone users can download free programs from Android Application Market, which may be malicious programs that wm threaten the security of user's information. In this paper, we use deep learning techniques to detect android malware and aim to enhance security of mobile users. We firstly extract android APIs from APK files, and represent API sequences as a two-dimensional image file. Then convolutional neural network is applied to automatically learn the features of API files and classify whether an application is goodware or malware. The experiments show that convolutional neural network can automatically extract features of API files and achieves a high performance on malware detection.