| BP网络泛化能力改进研究 |
| |
| 引用本文： | 王晖,何新贵. BP网络泛化能力改进研究[J]. 系统工程与电子技术, 2001, 23(3): 85-87 |
| |
| 作者姓名： | 王晖 何新贵 |
| |
| 作者单位： | 1. 北京航空航天大学,
2. 北京系统工程研究所, |
| |
| 摘 要： | 详细阐述了防止BP网络过度训练的方法，总结了常用的增强网络泛化能力的方法。并在此基础上，提出了基于主要影响因素和基于修正误差函数来增强网络泛化能力的方法，取得了良好的效果，提高了BP神经网络的适应性。
|
| 关 键 词： | 算法 网络 训练 |
| 文章编号： | 1001-506X(2001)03-0085-03 |
| 修稿时间： | 2000-03-14 |
Study of ImprovingGeneralization of BP Network |
| |
| Wang Hui,He Xingui. Study of ImprovingGeneralization of BP Network[J]. System Engineering and Electronics, 2001, 23(3): 85-87 |
| |
| Authors: | Wang Hui He Xingui |
| |
| Affiliation: | Wang Hui He Xingui Beijing University of Aeronautics and Astronautics,100083 Beijing System Engineering Institute,100101 |
| |
| Abstract: | The methods to improve the generalization ability of neural networks are summarized. Two new methods are presented to improve the generalization ability, one is through finding the main affecting variables and the other is through modifying the error function. The results are very good, and set up a sound basis to apply the BP network in real projects. |
| |
| Keywords: | Algorithm Network Training |
| 本文献已被 CNKI 万方数据 等数据库收录！ |
