Research And Implementation Of Android Malicious Application Detection Technology Based On Depth Learning Algorithm

Android is a Linux-based open source operating system, with its low cost of introduction, a good user experience, open source resource structure and so on, fit the development trend of China’s mobile Internet,a market share leader in mobile phone operation system. With the popularization of Android system in many mobile devices, a large number of malicious applications such as malicious chargeback, privacy disclosure and communication blocking appear in the application market,causing huge economic losses to the users, at the same time, domestic and foreign security field put forward Many malicious application detection methods, while achieving results, there are still a large number of constantly changing malicious applications can not be detected. A new and effective detection method has become an urgent problem to be solved.This paper is of great signifcance for the study of deep learning in the direction of Android malicious application.The common detection methods of Android system security and malicious applications was studies in this paper. analyzes The advantages and disadvantages of the mainstream detection methods were analyzes here.A malicious detection technology based on depth learning Android application are proposed, and the experiment basically meets the requirements, the main work of this paper is as follows:(1) The main methods of Android system security and Android application are researched, and the realization process and problems of static detection and dynamic monitoring are studied respectively. The feasibility of applying machine learning technology to Android malicious detection method is studied, significance.(2) A static detection algorithm for Android malicious application based on depth learning is designed, and a series of data processing flows such as feature extraction, feature selection and feature expression are designed, and a malicious application detection algorithm based on LSTM algorithm is designed.(3) The detection algorithm of Android malicious application based on depth learning is realized. The feature extraction, the selection of effective features and the rational expression of the feature are realized respectively, and a deep learning model with classification ability is built and trained.(4) It evaluates the discrimination ability and validity of the detection algorithm based on depth learning, and analyzes the problems in the experimental results, and gives a reasonable explanation.