| 神经网络自适应学习步长研究 |
| |
| 引用本文： | 李众立,王成端.神经网络自适应学习步长研究[J].电子科技大学学报(社会科学版),1996(6). |
| |
| 作者姓名： | 李众立 王成端 |
| |
| 作者单位： | 西南工学院信息与控制工程系 |
| |
| 摘 要： | 论述了神经网络学习算法的研究现状和存在的问题，分析了ＢＰ算法中学习步长选取的局限性，建立了ＢＰ网络输出误差的非线性规划模型，并根据文献［６］的状态空间混合算法，推导出了一种自适应调整学习步长的公式。推得的学习步长能够利用环境改变的低息反馈不断自动改变。
|
| 关 键 词： | 神经网络，BP算法，学习步长，自适应 |
Research on Self-adaptive Study Step of Neural Networks |
| |
| Li Zhongli, Wang Chengduan.Research on Self-adaptive Study Step of Neural Networks[J].Journal of University of Electronic Science and Technology of China(Social Sciences Edition),1996(6). |
| |
| Authors: | Li Zhongli Wang Chengduan |
| |
| Abstract: | In this paper,the current state and the existing problems of the study algorithm of neural networks are reviewed, the selecting problems of the study step in BP algorithm are analysed, and the noalinear programming model of output error in BP networks is built. Based on the state space hybrid method in Ref.6],the formulae of the study step which is self-adaptive and adjustable are derived. The study step derived in this poper can Change automatically with the environonent signal feedback. |
| |
| Keywords: | neural network BP algorithm study step self-adaptive |
| 本文献已被 CNKI 等数据库收录！ |
