| 前馈神经网络结构优化新算法 |
| |
| 引用本文： | 万晓玲,徐晋,李瑞海. 前馈神经网络结构优化新算法[J]. 成都理工大学学报(自然科学版), 2004, 31(4): 417-421 |
| |
| 作者姓名： | 万晓玲 徐晋 李瑞海 |
| |
| 作者单位： | 上海交通大学国际与公共事务学院,上海,200030;上海交通大学管理学院 |
| |
| 摘 要： | 前馈神经网络的结构直接影响网络的性能.首先基于拟牛顿NL2SOL法构造前馈神经网络模型,为了优化神经网络结构,尝试引入重置算法(Early Restart Algorithm),构建基于重置的NL2SOL动态前馈神经网络.最后通过对比实验表明,重置算法的引入有效地解决了神经网络的结构优化问题,优化后的神经网络具有良好的收敛性与稳定性.
|
| 关 键 词： | 重置算法 神经网络 结构优化 |
| 文章编号： | 1671-9727(2004)04-0417-05 |
| 修稿时间： | 2003-07-21 |
New algorithm for revising feed forward neural network structure |
| |
| WAN Xiao-ling,XU Jin,LI Rui-hai. New algorithm for revising feed forward neural network structure[J]. Journal of Chengdu University of Technology: Sci & Technol Ed, 2004, 31(4): 417-421 |
| |
| Authors: | WAN Xiao-ling XU Jin LI Rui-hai |
| |
| Affiliation: | WAN Xiao-ling~1,XU Jin~2,LI Rui-hai~2 |
| |
| Abstract: | The structure of feed forward neural network will affect its performance directly. The feed (forward) neural network based on NL2SOL algorithm is proposed firstly. Then, in order to optimize the neural network structure, the early restart algorithm is introduced and applied to the NL2SOL feed (forward) neural network. The comparison of experiment results demonstrates that the early restart (algorithm) can solve the structure optimization problem of neural network effectively, and the revised neural network performs well in convergence and stability. |
| |
| Keywords: | early restart algorithm neural network structure optimization |
| 本文献已被 CNKI 维普 万方数据 等数据库收录！ |
