摘要
传统抗恶意代码攻击算法无法区分恶意代码类别,抗攻击行为不具有针对性,导致传统算法的抗攻击成功率偏低。为此提出基于改进深度森林的抗恶意代码攻击算法。采用Gamma校正法对缩放处理后的灰度图像完成色彩空间规范性处理,确定灰度图像梯度后构建其梯度方向直方图,在块内归一化梯度直方图后采集方向梯度直方图特征。以方向梯度直方图特征为输入,采用深度森林算法划分恶意代码类别。计算深度森林各级级联结构内的不同类别恶意代码概率均值,构建四维增强特征向量,将其作为下级输入,改进深度森林算法,提升算法收敛速度,完成抗恶意代码攻击算法的设计。仿真结果验证了上述算法针对不同恶意代码类型抗攻击成功率均高于70%,具有更强的应用性。
The traditional anti-malicious code attack algorithms lack the identification of malicious code categories, resulting in non-targeted anti-attack behavior and a low anti-attack success rate. Therefore, we report an anti-malicious code attack algorithm based on improved deep forest. The Gamma correction method was used to process the scaled gray image. The gradient direction histogram was constructed after determining the gradient of the gray image for standardizing the color space. Meanwhile, after normalizing the gradient histogram in the block, the characteristics of the direction gradient histogram were collected. Taking the histogram feature of directional gradient as the input, the deep forest algorithm was applied to classify malicious code. The probability mean of different types of malicious code in each cascade structure of deep forest was calculated to establish a four-dimensional enhanced feature vector, and the established feature vector was used as the lower-level input, improving the deep forest algorithm and the convergence speed of the algorithm. Finally, the design of the anti-malicious code attack algorithm was achieved. The simulation results show that the algorithm has excellent applicability and a high anti-attack success rate(> 70%).
作者
李辉
王欢
王冬秀
LI Hui;WANG Huan;WANG Dong-xiu(College of Computer Science and Communication Engineering,Guangxi University of Science and Technology,Liuzhou Guangxi 545006,China)
出处
《计算机仿真》
北大核心
2022年第1期353-357,共5页
Computer Simulation
基金
广西自然科学基金项目(2020GXNSFBA159042)
广西自然科学基金项目(2018GXNSFAA294085)。
关键词
改进深度森林
恶意代码
抗攻击
代码转换
直方图特征
Improved depth forest
Malicious code
Anti attack
Code conversion
Histogram feature
