摘要
恶意软件是当前互联网安全面临的最严重威胁之一。传统的恶意软件检测方法需要依赖大量的规则和模式匹配,这种方法需要耗费大量人力进行恶意软件的分析和标注,并且难以检测由于代码混淆而快速变种的恶意软件。提出了一种基于代码压缩和循环神经网络的恶意代码检测方法。该方法通过压缩源代码来降低维度,提高计算效率,并将压缩后的代码转化为灰度图像作为循环神经网络的输入。通过对恶意代码和良性代码的训练,循环神经网络能够自动学习恶意代码的特征,并用于检测新的未知恶意代码。实验结果表明,该方法在检测恶意代码时表现出了很好的性能,并且对于具有变形和混淆的恶意代码也有较好的识别能力。因此,该方法可以作为一种有效的恶意代码检测手段,具有广泛的应用前景。
Malware is one of the most serious threats to internet security today.Traditional malware detection methods rely on a large number of rules and pattern matching,which requires labor-intensive analysis and labeling of malware,and makes it difficult to detect malware that quickly mutates due to code obfuscation.A malicious code detection method is proposed in this paper based on code compression and recurrent neural network.This method reduces dimensionality by compressing the source code,improves computational efficiency,and converts the compressed code into a grayscale image as input to the recurrent neural network.By training on malicious and benign code,RNNs are able to automatically learn the characteristics of malicious code and use it to detect new unknown malicious code.Experimental results show that the proposed method has good performance in detecting malicious code,and also has good recognition ability for malicious code with deformation and confusion.Therefore,this method can be used as an effective malicious code detection method and has a wide application prospect.
作者
刘明珠
高丽婷
李倩芸
LIU Mingzhu;GAO Liting;LI Qianyun(Hebei University of Architecture,Zhangjiakou,Hebei 075000)
出处
《河北建筑工程学院学报》
CAS
2023年第4期246-251,共6页
Journal of Hebei Institute of Architecture and Civil Engineering
