摘要
针对传统的基于特征码的恶意应用检测技术,在应对新的恶意应用产生情况下处理速度上的不足,提出一种基于深度学习的安卓恶意应用检测方法。通过对包含应用静态信息的文件进行反编译处理,提取可表征应用是否为恶意应用的信息,经过数据预处理后生成特征信息输入矩阵,采用多层卷积神经网络进行训练,优化得到较优的参数。实验结果表明,所提方法能有效检测出恶意应用。
Aiming at the traditional signature application-based malicious application detection technology and dealing with the lack of processing speed in the case of new malicious applications,a deep learning-based Android malicious application detection method was proposed.Through decompilation processing of files containing application static information,information that could be used to represent whether the application was a malicious application was extracted from it.After data preprocessing,the feature information input matrix was generated,and the multi-layer convolutional neural network was used for training and optimized to get better parameters.Experimental verification shows that the proposed method can effectively detect malicious applications.
作者
王亚洲
王斌
WANG Ya-zhou;WANG Bin(Beijing Institute of Computer Technology and Applications,Second Academy of China Aerospace Science and Industry Corporation,Beijing 100854,China)
出处
《计算机工程与设计》
北大核心
2020年第10期2752-2757,共6页
Computer Engineering and Design
关键词
安卓恶意应用
静态检测
深度学习
卷积神经网络
反编译
Android malicious application
static detection
deep learning
convolutional neural network
decompilation