摘要
为了提高网络流量的预测精度,针对极端学习机的训练样本选择问题,提出一种改进极端学习机的网络流量预测模型(IELM)。根据最优延迟时间和嵌入维数对网络流量重构,建立网络学习样本,将学习样本输入到改进极端学习机进行训练,随新样本加入而逐步求解网络的权值,以提高学习速度,引入cholesky分解方法提高模型的泛化能力,采用具体网络流量数据进行了仿真测试。结果表明,IELM不仅可以获得较传统网络流量预测模型更高的精度,并且大幅度减少了计算时间,提高了建模效率,可以较好地满足网络流量预测要求。
In order to improve prediction accuracy of traffic network, a novel network traffic prediction model is proposed based on improved extreme learning machine. The learning samples of network traffic data are obtained by the optimal delay time and embedding dimension, and then the samples are input into improved extreme learning machine to train, and the network weights are gradually updated by adding new samples to fasten the speed and cholesky method is introduced to improve generalization performance. The simulation experiments are carried out to test the model performance based on network traffic data. The results show that the proposed model has improved prediction accuracy of traffic network compared with traditional network traffic prediction models and greatly reduced the computing time to improve the effi- ciency, so it can meet the requirements of the online prediction for network traffic.
出处
《计算机工程与应用》
CSCD
2014年第21期91-95,146,共6页
Computer Engineering and Applications
关键词
网络流量
相空间重构
极端学习机
输出权值
network traffic
phase space reconstruction
extreme learning machine
output weight value