| 一种前向神经网络自适应学习算法 |
| |
| 引用本文： | 高书江,李成峰.一种前向神经网络自适应学习算法[J].应用能源技术,2006,12(4):49-51. |
| |
| 作者姓名： | 高书江 李成峰 |
| |
| 作者单位： | 哈尔滨电机厂有限责任公司,黑龙江,哈尔滨,150040 |
| |
| 摘 要： | 针对时变和(或)非线性输入的前向神经网络提出了一种感知自适应算法,其本质是要求输出的实际值和期望值的误差满足一个渐稳定的差分方程,而不是用后向传播方法使误差函数极小化。通过适当排列扩张输出可以避免算法的奇异性。
|
| 关 键 词： | 感知自适应 前向神经网络 非线性 渐进稳定 |
| 文章编号： | 1009-3230(2006)04-0049-03 |
| 收稿时间： | 2006-03-30 |
| 修稿时间： | 2006-04-07 |
A Self- adaptive Learning Algorithm for Feedforward Neural Network |
| |
| GAO Shu-jiang,LI Cheng-feng.A Self- adaptive Learning Algorithm for Feedforward Neural Network[J].Applied Energy Technology,2006,12(4):49-51. |
| |
| Authors: | GAO Shu-jiang LI Cheng-feng |
| |
| Affiliation: | Harbin Electric Machinery Co.Ltd, Heilongjiang, Harbin 150040,China |
| |
| Abstract: | A perceptron adaptation algorithm is generalized for the feedforward neural networks with time varying and/or nonlinear inputs.The essence of the algorithm is to force the error between actual and desired outputs to satisfy an asymptotically stable difference equation,rather than to minimize an error function using backpropagation procedure.By appropriate arrangement of augmented inputs,the singularity of the algorithm can be avoided. |
| |
| Keywords: | perceptron adaptation feedforward neural network nonlinear asymptotically stability |
| 本文献已被 CNKI 维普 万方数据 等数据库收录！ |
