Android Malware Detection via Deep learning Approach

With the increasing number of malicious applications nowadays, the traditional methods (such as signature-based methods) cannot cope with these rapid changes and the need for new and effective approaches arises. Therefore, we present an Android malware detection model that depends on static analysis and deep learning. In static analysis, the source code of an application has been analyzed. Moreover, the proposed model combines two deep learning approaches: graph convolutional network and reservoir computing network. Empirical results indicate good performance with accuracy of 82.73% and recall of 91.86%. Moreover, the results show that our model can cope with real-time requirements.