Android Malware Detection using Feature Selection With Random Forest and Ensemble Learning

As cell phones have become important tools in our today s life, the security challenges have become more serious. Malware detection is a set of techniques used to examine and understand how android applications work and to identify malwares. Malware detection using traditional methods is not reliable. So, in this research, the first step is seeking and creating a suitable and large dataset. Then using feature selection and ensemble method tries to provide better malware detection results compared to the other researches. Our opinion for feature selection is to use Random Forest. Random Forest reduces the number of features from 874 to 24 in two steps. The results show that the combination of classifiers into an ensemble model provides better accuracy than an individual classifier. The malware detection rate is up to 99.62% in our experimental evaluation.