An Efficient Android Malware Detection Model using Convnets and Resnet Models

Android is the most widely used operating system in mobile environments. Users can install mobile applications from play store as well as from other sources. While installing the applications, it may be difficult to check whether the app is malware app or benign app. Identifying malicious mobile applications is essential as vast numbers of apps are available on the Google Play store. Much of the previous works relies on static or dynamic or hybrid (mixture of static and dynamic) feature analysis for android malware detection. Static analysis extract features without running an app, whereas dynamic analysis extract features after executing the app in virtual environments. This paper proposed an efficient model for android malware detection based on image-oriented representation using Convolutional Neural Networks and Resnet models. The image representations of android apks are generated in two different ways. In first method, entire android app is converted as image, where as in second method only dex part of android app is converted as image. After creating two datasets, convolutional neural networks and resent models are proposed on two datasets. This work investigated innovative methods to improve the detection of Android malware by using deep learning techniques. This research contributes to the continuous endeavors to protect mobile devices from harmful apps.