An Optimized Decision Tree based Android Malware Detection Approach using Machine Learning
Mosaddek Hossain, S. Rafi, Md. Shohrab Hossain · 2020
The growing trend of attacking Android smart phones using malicious app has started posing significant threats for the users. Many approaches have been introduced for protecting the users against such malware. However, those solutions tend to use many features to get better accuracy in detecting Android malware which eventually results in higher complexity in creating machine learning based models. Hence, an effective model is required to find the most significant features resulting in a faster model yet having better accuracy. In this paper, we have proposed a robust approach to detect android malwares using only selective features that are extracted using Ranker search method and Gain Ratio attribute evaluator. We have used machine learning algorithms which include J48 Decision Tree, Random Forest and Random Tree to classify the preprocessed dataset into malware and benign. We have produced faster results using Random Tree algorithm and obtained higher accuracy using Random Forest algorithm. Further, we have measured and compared various performance metrics with respect to different numbers of attributes and different classifiers. Our proposed detection method can help users distinguish malicious applications from benign ones in a faster yet precise manner.