Android malware detection model based on lightGBM

Abstract

Android malware detection is an important research area against Android malware apps. In this paper, we propose an Android malware detection model based on LightGBM. The model consists of a new feature selection method, which contains Chi2 and ExtraTrees, and a LightGBM classifier method. A corpus of 2000 malware and equal number of benign samples are prepared to verify the model. Finally, an experiment is designed to test the model accuracy and training time. The results show high model accuracy (about 96.4%) and a heavy reduction in training time as compared to existing models.