摘要
针对目前计算机网络业务流管理问题,提出了一种基于复合神经网络的网络业务分类方案.将复合神经网络用于网络业务源特征提取与分类的研究中,打包法和BP算法结合用于复合神经网络的训练.并分析比较与单个神经网络和模糊神经网络算法用于网络业务分类效果.计算机仿真结果表明,复合神经网络分类收敛快、误差小,比单个神经网络算法和模糊神经网络算法更优越;同时,研究结果为解决网络业务源特征提取与分类提供了一种有效的途径.
An algorithm of classifying network traffic based on ensemble neural networks is proposed in this paper. Ensemble neural networks (ENN) is used to study the feature extraction and classifying of the network traffic. Bagging and back-propagation algorithms are used for the training of ensemble neural networks. Compared with the non-ensemble neural networks (NN) and fuzzy neural networks ( FNN), the simulation result demonstrates that ENN not only can be well applicable for classifying NN and FNN. This paper supplies the fundamental research to solve the network traffic, but also is superior to the problem of classifying network traffic.
出处
《天津大学学报》
EI
CAS
CSCD
北大核心
2006年第B06期121-124,共4页
Journal of Tianjin University(Science and Technology)
基金
教育部博士学科点基金(20030056007)
中国博士后科学基金项目(2005037529)
天津市高等学校科技发展基金项目(20041325)
天津理工大学育苗项目(LG03018).
关键词
复合神经网络
BP算法
网络业务
特征提取与分类
ensemble neural networks
back-propagation algorithms
network traffic
feature extraction and classifying