| 基于改进算法的多层神经网络的结构优化设计 |
| |
| 引用本文： | 雷铁安,吴作伟,杨周妮.基于改进算法的多层神经网络的结构优化设计[J].大连铁道学院学报,2004,25(3):49-52. |
| |
| 作者姓名： | 雷铁安 吴作伟 杨周妮 |
| |
| 作者单位： | 北京交通大学机电学院,北京交通大学机电学院,北京交通大学机电学院 北京100044,北京100044,北京100044 |
| |
| 摘 要： | 简要地介绍了工程中应用较多的多层前向神经网络(BP网络)的结构和网络的设计．通过采用一种改进的训练算法(参考隐层的输出),并结合科学的样本选择方法(“超立方体”法)设计网络,进行网络训练．最后通过数字仿真证明了这个网络模型用于结构优化设计的有效性．
|
| 关 键 词： | 多层神经网络 改进算法 多层前向神经网络 超立方体 训练算法 网络训练 BP网络 证明 科学 参考 |
Structural Optimization of Multilayer Neural Networks Based on An Enhanced Algorithm |
| |
| LEI Tie-an,WU Zuo-wei,YANG Zhou-ni.Structural Optimization of Multilayer Neural Networks Based on An Enhanced Algorithm[J].Journal of Dalian Railway Institute,2004,25(3):49-52. |
| |
| Authors: | LEI Tie-an WU Zuo-wei YANG Zhou-ni |
| |
| Abstract: | This paper introduces the structure and the design of MLP network ("back-propagarion networks") which was often applied to practical engineering. A network was designed and trained by a enhanced training algorithm(reference the output of the hidden layer) combined with a sample selection method ("hypercube"method). The network model is proved feasible by numerical simulation results. |
| |
| Keywords: | Multilayer neural networks Backpropagation hypercube method |
| 本文献已被 CNKI 维普 等数据库收录！ |
