Adware detection and privacy control in mobile devices
Ianir Ideses, Assaf Neuberger · 2014
In this paper we propose a system and algorithms for detection of Adware in mobile devices that are based on machine learning algorithms and are capable of adapting to the ongoing transformation of Adware and Malware. The system is based on static and dynamic analysis of mobile applications, extraction of useful features and real-time classification. This classification is based on supervised machine learning algorithms with emphasis on fast, linear operations and efficient implementation on mobile devices. The system presented in this paper enables identification of relevant features that are salient in Adware and Malware, useful for further analysis by security researchers. The proposed system exhibits a detection rate of 97%. The system has been tested and verified on known, industry standard, datasets and is superior to state of the art solutions available in the market. These result have been verified by 3rdparty evaluators.