摘要
恶意代码数量已经呈现爆炸式增长,对于恶意代码的检测防护显得尤为重要。近几年,基于深度学习的恶意代码检测方法开始出现,基于此,提出一种新的检测方法,将恶意代码二进制文件转化为十进制数组,并利用一维卷积神经网络(1 Dimention Convolutional Neural Networks,1D CNN)对数组进行分类和识别。针对代码家族之间数量不平衡的现象,该算法选择在分类预测上表现良好的XGBoost,并对Vision Research Lab中的25个不同恶意软件家族的9458个恶意软件样本进行了实验。实验结果表明,所提的方法分类预测精度达到了97%。
In the context of the current explosive eruption of malicious codes,the detection and protection of malicious codes is particularly important.In recent years,a new method of using deep learning to detect malicious codes has emerged.Based on this,this article proposes a new detection method that converts malicious code binary files into decimal arrays and uses 1 Dimention Convolutional Neural Networks(1 D CNN)to perform classification and recognition.Aiming at the imbalance in the number of code families,this article chooses XGBoost,which performs well in the classification prediction competition.We conducted experiments on 9458 malware samples from 25 different malware families in the Vision Research Lab.Experimental results show that the classification prediction accuracy of this algorithm reaches 97%.
作者
黄科
袁启平
董薇
孙沂昆
亢勇
王天翔
HUANG Ke;YUAN Qiping;DONG Wei;SUN Yikun;KANG Yong;WANG Tianxiang(Beijing Special Engineering Design and Research Institute,Beijing 100028,China)
出处
《电视技术》
2021年第10期129-135,共7页
Video Engineering
