A Novel Permission Filtering Approach for Android Malware Detection and Malware Family Classification

As the most popular mobile operating system in the world, Android has become the major target of attacks. To combat this serious malware campaign, we need an effective and efficient approach to thwart them. Although numerous malware detection approaches have been proposed. Malware detection and family classification for a large bundle of apps remain a challenge. Existing malware detection and family classification methods are not accurate enough and cannot handle large numbers of samples efficiently. In this paper, we propose a novel permission filtering approach for Android malware detection and malware family classification. We design three levels of feature filtering to remove unimportant permissions by mining the permission. The dimensionality of permission as features is drastically reduced while keeping good performance both in malware detection and malware family classification. Experimental results validate that our approach can perform well in both malware detection and malware family classification using the final 33 permissions and common machine learning algorithms.