Detection of Malicious Android Application Based on HTM Algorithm
QIU Shenjian;ZHANG Shibin;LIU Pingguang;College of Information Security Engineering,Chengdu University of Information Technology;College of Communication Engineering,Chengdu University of Information Technology;
With the Internet users' conversion from traditional PC to mobile terminal,the mobile security has been more and more concerned. In order to improve the detection efficiency of unknown malicious mobile application,aiming at the poor detection ability problem of traditional detection in detecting the malicious applications that introduces polymorphic and deformation techniques,a method to detect malicious Android mobile applications based on HTM algorithm is proposed,the application detection contains the feature extraction that aims at Android application Dalvik instructions characteristic and the feature selection and integration by using the information gain method,and the sequence mode is trained and deduced by HTM algorithm,then the feature extraction and fusion result of test sample is input into the HTM network that completes training,therefore,the purpose of detecting malicious applications is achieved. The experiment simulation show: the detection rate of designed malicious application detection method is nearly 100%,and has high detection efficiency; the false positive rate is 0. 08%. Compared to other algorithms,the detection rate,false positive rate and classification accuracy rate of proposed malicious detection method are better,and it can be applied to different types of malicious applications,but the training and testing time is longer.