Study on Android Hybrid Malware Detection Based on Machine Learning
暂无分享,去创建一个
With the popularity of smart phones, many users are using the Android system. The major reason is that the Android system can download and install application function from third part market easily. Therefore, many malware attacks are proposed by the illegal hacker. How to detect these malware attacks accurately? Until now, many methods were proposed to improve the accuracy rate for malware detection. In this thesis, we will propose a malware detection system which combines the machine learning methods (SVM or Random Forest) and hybrid analysis model. Here, the major feature of hybrid analysis model is combination of the Permissions characteristic from the static analysis method and API from the dynamic analysis method. According to the experimental results, by using our proposed scheme, the accuracy rate and TP (true positive) rate are 88% and 89%, respectively. Comparing with Arshad et al. scheme, our proposed scheme is better than them.
[1] Hongnian Yu,et al. SAMADroid: A Novel 3-Level Hybrid Malware Detection Model for Android Operating System , 2018, IEEE Access.
[2] Wei-Ting Lin,et al. Mobile malware detection in sandbox with live event feeding and log pattern analysis , 2016, 2016 18th Asia-Pacific Network Operations and Management Symposium (APNOMS).
[3] Ali Feizollah,et al. AndroDialysis: Analysis of Android Intent Effectiveness in Malware Detection , 2017, Comput. Secur..