Research On Android Malware Detection Model Based On Machine Learning

With the faster development of the mobile Internet,functional advantages of themobile terminal such as intelligence and convenience are highlighted,which makesthe quantity of mobile users increasing rapidly.As the most popular mobileoperating system,Android APP faces more and more security threats,such asprivacy leaks,malicious deductions,and system damage occur from time to time.Therefore,effective detection of Android malicious applications has importantresearch significance and practical application value.In view of the influencing factors such as multiple feature types and high dimensions,this paper proposes a detection model based on machine learning,which combines static detection and dynamic detection technology,and conducts research from three aspects: extract the characteristics of the application,reduce the number of features and machine learning algorithm improvement.The main contributions of the paper are as follows:First,in view of the different contribution of application features to classification,a feature extraction method combining dynamic and static features is proposed.This method combines static feature extraction and dynamic feature extraction technology,which can extract both static and dynamic features.Second,in view of the problem of fine-grained features and high-dimensional application features,the information gain algorithm is proposed to reduce the number of the application feature set and a more representative feature set is selected.The experimental results confirm the effectiveness of feature selection.Third,in response to the current low detection rate of Android malicious applications,we use the weighted voting principle to improve the random forest algorithm.The category with the most votes is selected as the final category of the sample.The experimental results show that the algorithm overcomes the shortcomings of the random forest effectively,and improves the ability of the classifier for Android apps.In summary,according to the characteristics of Android applications and the commonly used attack methods of malicious software,this paper proposes a targeted application detection method.The comparison experiment results show that the detection accuracy rate reaches 99.3%,which has certain practical application value.