摘要
当前的非授权代码的检测过程忽略了获取线性特征集,代码类型无法得以高精度匹配,导致传统方法出现检测准确率低、耗时长问题。为解决以上问题,提出基于线性特征集的非授权代码敏感路径检测方法。引入深度学习,设计非授权代码特征检测步骤。以非授权代码为目标样本,基于线性特征集判断非授权代码敏感路径判断,利用深度学习提取正常代码样本的函数图特征,并设置阈值,实现非授权代码敏感路径的检测。实验结果表明:与传统方法相比,所提非授权代码敏感路径检测方法具有更高的检测准确率,且耗时更短,为该领域的深入研究提供理论支持。
The current detection process of unauthorized codes ignores the acquisition of linear feature sets, and the code types cannot be matched with high precision, which leads to the problems of low detection accuracy and long time consumption of traditional methods. To solve the above problems, a sensitive path detection method for unauthorized code based on linear feature set is proposed. Deep learning is introduced to design the steps of unauthorized code feature detection. The unauthorized code was taken as the target sample, the sensitive path judgment of unauthorized code was judged based on the linear feature set, and the function graph features of normal code samples were extracted by deep learning, and the threshold value was set to realize the sensitive path detection of unauthorized code. Experimental results show that compared with traditional methods, the proposed method has higher detection accuracy and shorter time consumption, which provides theoretical support for further research in this field.
作者
梁涛
李毅成
段玉莹
LIANG Tao;LI Yi-cheng;DUAN Yu-ying(Fuzhou Medical College of Nanchang University,Jiangxi Fuzhou 344000,China;Jingdezhen Ceramic InstituteyJiangxi Jingdezhen 333403,China;Institute of Computer Science and Technology,Shandong University of Technology,Shandong Zibo 255000,China)
出处
《计算机仿真》
北大核心
2021年第6期373-377,共5页
Computer Simulation
关键词
非授权代码
敏感路径
线性特征集
深度学习
特征提取
函数图特征
Unauthorized code
Sensitive path
Linear feature set
Deep learning
Feature extraction
Function graph feature
