DROIDANALYZER: EFFICIENT FRAMEWORK FOR ANDROID MALWARE DETECTION

Abstract

Mobile Application Market has evolved and is continuously expanding with over millions of applications. Many mobile operating systems are available, most popular among them is Android. Due to its popularity and reach, malware developers are targeting android markets for distributing malware. This has led to an increase in risk associated with Android devices. The growth of mobile malware is so huge that traditional techniques for malware detection are inefficient. Therefore, effective and robust malware detection techniques are required. Many researchers have proposed static and dynamic approaches for effective Android malware detection. In this research, we have proposed a fine-grained hybrid model for efficient android malware detection using multi-modal learning. We have extracted static and dynamic features from a set of 4000 applications. We have used multi-modal learning to better classify the samples. We have compared our implementation with other techniques. Our analysis suggest that multi-modal learning outperforms other state of the art techniques.