Android Malware Detection Using Incremental Learning Approach

Android is the most popular Mobile operating system covering above 70% of all smartphone users. Due to its popularity, an increasing number of malware applications are targeting the Android platform, stealing sensitive information, accessing banking apps, etc. These malware applications can be detected with the help of machine learning algorithms by extracting features from those applications. However, previous attempts use traditional methods which require high memory overhead and computational power which is not suitable for continuous real-time data inflow. In this paper, we have pro-posed an incremental decision tree approach to detect mobile malware. Our model has achieved an accuracy of 93.33%, precision of 91.27%, recall of 95.83 %, and F1-Score of 93.50%. Our proposed model can help reduce model retraining time when new malware samples appear. It also requires less storage for historical data while retraining the model.