一种多层前馈神经网络的快速学习算法  

A Fast Multilayer Feedforward Neural Networks Training Algorithm

作　　者：陈亚军[1] 

机构地区：[1]四川师范学院物理与电子信息系,四川南充637002

出　　处：《河北师范大学学报（自然科学版）》2002年第6期582-587,共6页Journal of Hebei Normal University：Natural Science

摘　　要：对多层前馈神经网络的学习算法及其特点做了较为详细的讨论 ,提出了一种基于层内优化的快速学习算法 .在该算法中 ,输出层的连接权矩阵 (V)和前一层的输出矢量 (B)被作为 2个变量集合 ,通过最小化该层样本的总平方误差函数可求得一组它们的优化解 (V* ,B*p ) ;并将 B*p 作为前一层 (隐层 )的期望输出 ,用类似的方法同样可以求出隐层的连接权矩阵和输出矢量 .Discussed the training algorithms and its features of multilayer feedforward neural networks,and derived a fast training algorithm based on the layer by layer optimizing procedures.In this new algorithm,the weight matrix( V ) of the output layer and the output vector( B ) of the previous layer are treated as two variable sets.An optimal solution pair (V *,B * p )is found to minimize the sum square error of the patterns. B * p is then used as the desired output of the previous layer,the optimal weight matrix and output vector of the hidden layer is found with the same method as that used for the output layer.Computer simulation shows that the new algorithm is effective.

关 键 词：多层前馈神经网络 快速学习算法 权值矩阵 层内优化 计算机仿真 连接权矩阵 

分 类 号：TP183[自动化与计算机技术—控制理论与控制工程]