Android malware classification using feature fusion and AdaBoost learning

The monetary benefits associated with the Android operating system have attracted the cybercriminals to generate malware application. Malware creators tend to leak sensitive information to their servers for financial gains. They can turn mobile devices into bots to perform malicious actions. To speed the process of malware classification, Android malware can be converted into digital images using the visualization technique. The malicious content of the malware application is visualized as an image. In this paper, we have employed two different types of image descriptors namely Gray Level Co-occurrence Matrix-based (GLCM) and GIST for feature extraction. To evaluate the classification results, the extracted features are classified using the AdaBoost classifier. The classification performance is also evaluated using the fusion of GLCM and GIST image descriptors. The experiment results show that feature fusion enhances the classification performance than the standalone image descriptors. Using feature fusion and AdaBoost learning, the performance measures such as accuracy, precision, recall, and F1 score are achieved to be 92.75%, 90%, 95.32%, and 92.32% respectively.