Android Malware Detection using Chi-Square Feature Selection and Ensemble Learning Method

The wide use of mobile phones has become a significant driving force behind a severe increase in malware attacks. These malware applications are hidden in the normal applications which make their classification and detection challenging. The existing techniques are based on signature based approach and are unable to detect unknown malware. In this paper, we propose a technique based on static and dynamic features for the detection of Android malware. We applied a chi-square feature selection algorithm to choose the appropriate features that contribute for detecting malware. After that, we stacked the different base classifiers to improve the detection rate. Furthermore, we compared the proposed method with existing well known machine learning classifiers. The experimental results demonstrate that the proposed technique (K-NN_ RF) achieves better detection accuracy i.e. 98.02%.