| 基于李雅普诺夫函数的BP神经网络算法的收敛性分析 |
| |
| 引用本文： | 张茂元,卢正鼎. 基于李雅普诺夫函数的BP神经网络算法的收敛性分析[J]. 小型微型计算机系统, 2004, 25(1): 93-95 |
| |
| 作者姓名： | 张茂元 卢正鼎 |
| |
| 作者单位： | 华中科技大学,计算机科学与技术学院,湖北,武汉,430074 |
| |
| 摘 要： | 针对前馈神经网络应时变输入的自学习机制，采用李雅普诺夫函数来分析权值的收敛性，从而揭示BP神经网络算法朝最小误差方向调整权值的内在因素，并在分析单参数BP算法收敛性基础上，提出单参数变调整法则的离散型BP神经网络算法．
|
| 关 键 词： | 李雅普诺夫函数 前馈神经网络 BP算法 单参数 变调整法则 |
| 文章编号： | 1000-1220(2004)01-0093-03 |
Lyapunov-based Analyse of Weights'''' Convergence on Backpropagation Neural Networks Algorithm |
| |
| ZHANG Mao-yuan,LU Zheng-ding. Lyapunov-based Analyse of Weights'''' Convergence on Backpropagation Neural Networks Algorithm[J]. Mini-micro Systems, 2004, 25(1): 93-95 |
| |
| Authors: | ZHANG Mao-yuan LU Zheng-ding |
| |
| Abstract: | In this letter, with the use of a backpropagation algorithm, Lyapunov function method is adopted to analyze the convergence of weights for feedforward Neural Networks learning with time varying inputs. So the factor of convergence can be proposed to minima of the error function. With this factor, a backpropagation algorithm based on single parameter is sound and quick. Then a backpropagation algorithm named with single parameter and multi-regulations is proposed. |
| |
| Keywords: | lyapunov feedforward neural networks backpropagation algorithm single parameter multi-regulations |
| 本文献已被 CNKI 维普 万方数据 等数据库收录！ |
