北京航空航天大学学报 ›› 2000, Vol. 26 ›› Issue (5): 596-599.

• 论文 • 上一篇    下一篇

前馈神经网络的一种简单共轭梯度学习算法

梁久祯1, 何新贵2, 黄德双2   

  1. 1. 北京航空航天大学 计算机系;
    2. 北京系统工程研究所
  • 收稿日期:1999-03-25 出版日期:2000-05-31 发布日期:2010-09-27
  • 作者简介:梁久祯(1968-),男,山东东平人,博士生,100083,北京.
  • 基金资助:

    国家自然科学基金资助项目(69705001)

Simple Conjugation-Gradient BP Algorithm for Feedforward Neural Networks

LIANG Jiu-zhen1, HE Xin-gui2, HUANG De-shuang2   

  1. 1. Beijing University of Aeronautics and Astronautics, Dept. of Computer Science and Engineering;
    2. Beijing Institute of System Engineering
  • Received:1999-03-25 Online:2000-05-31 Published:2010-09-27

摘要: 针对前馈神经网络学习误差函数维数高、计算复杂度大的特点,对梯度下降BP算法加以改进从而构造出一种简单共轭梯度下降算法(MPARTAN算法).该算法计算复杂度不高于动量BP算法, 与FR共轭梯度法相比,该算法的稳定性好,又具有共轭梯度法的优点,收敛速度快.文中给出了该算法的收敛定理,并用2个实验例子比较了动量BP算法、FR共轭梯度法和MPARTAN算法的计算结果.

Abstract: The high dimension of the learning error function for BP networks and the difficult computation complexity are incestigated. A simple modified conjugation-gradient decent algorithm (MPARTAN) is proposed based on improving the gradient BP algorithm. That the computation complexity of this algorithm is not higher than that of the BP momentum algorithm. Compared with FR conjugation algorithm, this algorithm has better stability and fast speed quality of convergence. It is also investigated that the convergence theorems for this algorithm and comparison of the computing results by two computing examples for the promoted three algorithms: BP momentum algorithm, FR conjugation-gradient algorithm and the novel MPARTAN algorithm.

中图分类号: 


版权所有 © 《北京航空航天大学学报》编辑部
通讯地址:北京市海淀区学院路37号 北京航空航天大学学报编辑部 邮编:100191 E-mail:jbuaa@buaa.edu.cn
本系统由北京玛格泰克科技发展有限公司设计开发