Android malware detection applying feature selection techniques and machine learning

Android operating system is known as one of the most popular mobile operating systems. The malware intrusion increases in the same pace as the production of applicable software. Propagation of new and transformed malware in seconds is a critical challenge in malware detection. Android software supplies thousands of features, providing assistance to identify malware applications. In this paper, a novel method based on a random forest algorithm, which applied three different feature selection techniques is proposed. This paper assesses the consequence of applying three different feature selection types including effective, high weight and effective group feature selection. Experiments conducted on Drebin dataset indicate applying the feature selection methods ameliorate the accuracy in terms of metrics and required time. In addition, comparison between the candidate feature selection model and a variety of algorithms as baselines proves the merit of applying feature selection on Random Forest, which outperforms other models based on several metrics.