李擎, 郑德玲. 混沌时间序列神经网络拓扑结构的选取方法[J]. 工程科学学报, 1999, 21(1): 90-93. DOI: 10.13374/j.issn1001-053x.1999.01.026
引用本文: 李擎, 郑德玲. 混沌时间序列神经网络拓扑结构的选取方法[J]. 工程科学学报, 1999, 21(1): 90-93. DOI: 10.13374/j.issn1001-053x.1999.01.026
Li Qing, Zheng Deling. Determining Topology Architecture for Chaotic Time Series Neural Network[J]. Chinese Journal of Engineering, 1999, 21(1): 90-93. DOI: 10.13374/j.issn1001-053x.1999.01.026
Citation: Li Qing, Zheng Deling. Determining Topology Architecture for Chaotic Time Series Neural Network[J]. Chinese Journal of Engineering, 1999, 21(1): 90-93. DOI: 10.13374/j.issn1001-053x.1999.01.026

混沌时间序列神经网络拓扑结构的选取方法

Determining Topology Architecture for Chaotic Time Series Neural Network

  • 摘要: 采用3层前向神经网络描述混沌时间序列的动力学模型,给出了该网络拓扑结构的确定方法.以及使网络泛化误差达到最小为依据确定网络的输入节点和隐含节点个数.仿真结果表明:该方法不仅优化了网络的结构,而且大大减少了网络的泛化误差.

     

    Abstract: Three-layered feed-forward neural network is used to establish the model for chaotic time series dynamic systems. Determination method of topology architecture for network is given.The size of the input node and hidden node is determined by the minimization the generalization error of the network. The simulation results show that the network architecture is optimized and the generalization error is dramatically decreased when this method is used.

     

/

返回文章
返回