| 共轭梯度神经网络的研究 |
| |
| 引用本文： | 蔡正国,屈梁生.共轭梯度神经网络的研究[J].西安交通大学学报,1995,29(8):72-76. |
| |
| 作者姓名： | 蔡正国 屈梁生 |
| |
| 作者单位： | 湖北荆门石化总厂 |
| |
| 摘 要： | 针对ＢＰ人工神经网络存在的两个缺点：（１）局部收敛问题不能很好解决；（２）收敛速度慢，提出应用改进共轭梯度算法建立人工神经网络，以获得共轭梯度网络的全局收敛特性和快收敛速度。计算机仿真结果表明：改进巫轭梯度网络优于ＢＰ网络。
|
| 关 键 词： | 神经网络 共轭梯度算法 全局收敛性 |
RESEARCH ON CONJUGATE GRADIENT NEURAL NETWORKS |
| |
| Cai Zhengguo,Qu Liangsheng.RESEARCH ON CONJUGATE GRADIENT NEURAL NETWORKS[J].Journal of Xi'an Jiaotong University,1995,29(8):72-76. |
| |
| Authors: | Cai Zhengguo Qu Liangsheng |
| |
| Institution: | School of Mechanical Engineering |
| |
| Abstract: | Artificial neural networks based on back-propagation algorithm have two disadvantages(1)weakly local characteristics;(2)slow convergence speed.In this paper,an advanced conju-gate gradient algorithm is applied to architect artificial neural networks so as to get its globalminimum and relatively quick convergence speed.The computer simulation results declarethat the advanced conjugate gradient ANN is superior to BP networks. |
| |
| Keywords: | neural networks optimizing algorithm |
| 本文献已被 CNKI 维普 等数据库收录！ |
