决策树主分量特征优化跟踪及暂态性异常提取

发布时间：2018-02-10 08:09

本文关键词： 决策树小波变换网络流量　出处：《科技通报》2014年06期 　论文类型：期刊论文

【摘要】：为有效定位识别和提取网络流量序列的暂态性异常特征,针对网络异常流量特征扰动性和暂态性特点,提出一种基于小波分解的二叉分类回归决策树主分量特征优化跟踪特征提取算法。利用训练集建立决策树模型,采用二叉分类回归决策树模型进行主分量特征优化跟踪建模,利用双正交提升小波分解得到的各层细节信号对暂态性扰动特征的敏感性,实现网络流量异常特征的定位提取和识别。仿真实验表明,改进算法的抗干扰能力和分辨率提高显著,暂态性异常特征谱图分辨能力提高,异常特征分布谱清晰可见,展示了较好的特征提取和状态识别性能。
[Abstract]:In order to identify and extract the transient anomaly characteristics of network traffic sequence effectively, the characteristic disturbance and transient characteristics of network abnormal traffic are analyzed. This paper presents a feature extraction algorithm based on wavelet decomposition for feature optimization of principal component of binary classification regression decision tree. A decision tree model is established by using training set, and a binary classification regression decision tree model is used to model the optimal tracking of principal component feature. Using biorthogonal lifting wavelet decomposition, the sensitivity of each layer of detail signal to transient disturbance feature is obtained, and the location extraction and recognition of network traffic anomaly feature are realized. The simulation results show that, The improved algorithm can improve the anti-jamming ability and resolution, improve the resolution of transient anomaly feature spectrum, and the spectrum of abnormal feature distribution is clearly visible, which shows the better performance of feature extraction and state recognition.
【作者单位】：平顶山学院计算机科学与技术学院;
【分类号】：TP393.06

【相似文献】