| 一种新的基于目标反传的前馈式神经网络训练算法 |
| |
| 引用本文： | 张长胜,王雪峰,冯英浚. 一种新的基于目标反传的前馈式神经网络训练算法[J]. 哈尔滨商业大学学报(自然科学版), 2002, 18(2): 225-227 |
| |
| 作者姓名： | 张长胜 王雪峰 冯英浚 |
| |
| 作者单位： | 哈尔滨工业大学,数学系,黑龙江,哈尔滨,150001 |
| |
| 摘 要： | 提出了一种基于目标反传的前馈式神经网络训练算法,该算法将网络的目标输出信息反传到网络的每一个隐层上,于是将神经网络的训练问题转化为求解一系列线性方程组和线性不等式组的问题,数值实验结果表明本文提出的方法与传统的BP算法相比提高了网络的训练速度.
|
| 关 键 词： | 前馈式神经网络 目标反传 训练算法 |
| 文章编号： | 1004-1842(2002)02-0225-03 |
| 修稿时间： | 2001-10-16 |
New learning algorithm for feed-forward neural network based on objective back-propagation |
| |
| ZHANGChang-sheng,WANG Xue-feng,FENG Ying-jun. New learning algorithm for feed-forward neural network based on objective back-propagation[J]. Journal of Harbin University of Commerce :Natural Sciences Edition, 2002, 18(2): 225-227 |
| |
| Authors: | ZHANGChang-sheng WANG Xue-feng FENG Ying-jun |
| |
| Abstract: | A new algorithm is introduced based on obective back-propagation.The algorithm linearly back-propagates the network objective output to the hidden layer.So the training problem of neural network is translated to the solution of a series of linear equations and linear inequalities.The experimental results show that the new algorithm has better convergence results than BP algorithm. |
| |
| Keywords: | feed-forward neural network objective back-propagation training algorithm |
| 本文献已被 CNKI 维普 万方数据 等数据库收录！ |
