Research On Android Malicious Application Detection Model Based On Deep Learning

With the widespread popularity of mobile terminals,it has become a necessity of people’s life.Because of the open source nature of Android system and the huge potential economic benefits,Android mobile devices have become the target of many malicious programs,and the security problems are also emerging one after another.The existing Android malicious application detection model and technology have low detection rate and low effectiveness,and cannot effectively detect the currently rapidly growing malicious applications.Based on the research and analysis of the behavior characteristics of malicious applications in Android system,this paper proposes a new detection model of malicious applications in Android system based on the combination of feature weighting and Bi-directional Long Short-Term Memory(Bi-LSTM)neural network algorithm.Main work of this paper:1.The deficiencies of Android architecture and existing detection methods are analyzed.This paper studies the security mechanism of Android system and application installation package,and through the analysis of Android application behavior characteristics,as well as the comparative analysis of the advantages and disadvantages of the model detection algorithm,it studies the shortcomings of the existing Android malicious application detection model.2.The feature vector model of malicious behavior is designed and constructed.This model uses the static analysis method to extract different types of behavior features from Android applications,selects permissions,intents and component information as the main behavior features,and uses the improved feature weighting method to construct the feature vectors of malicious and normal applications.3.The deep learning detection algorithm based on Bi-directional Long Short-Term Memory neural network is designed and implemented.Based on theanalysis and further optimization of the length of feature vector and the number of hidden layer elements,this algorithm designs a Bi-directional Long Short-Term Memory neural network algorithm,and constructs a detection model of Android malicious applications with high classification detection capability.4.The effectiveness of Android malicious application detection model based on deep learning is verified by experiments.By comparing with other detection models and algorithms,the experimental results show that the detection model has high detection accuracy and recall rate for malicious applications.