计算机工程与应用 ›› 2011, Vol. 47 ›› Issue (11): 37-38.

• 研究、探讨 • 上一篇    下一篇

一种粒子群优化的神经网络综合训练算法研究

徐 刚,黄先玖   

  1. 南昌大学 数学系,南昌 330031
  • 收稿日期:1900-01-01 修回日期:1900-01-01 出版日期:2011-04-11 发布日期:2011-04-11

Study on nueral network synthesis training algorithm based on particle swarm algorithm

XU Gang,HUANG Xianjiu   

  1. Department of Mathematics,Nanchang University,Nanchang 330031,China
  • Received:1900-01-01 Revised:1900-01-01 Online:2011-04-11 Published:2011-04-11

摘要: 基于粒子群优化算法,将灰色关联分析法应用到BP神经网络隐层结点数的确定,实现了BP神经网络结构的优化;然后将贝叶斯正则法应用于神经网络训练,进一步提高网络的泛化性能。仿真结果表明泛化能力明显优于其他改进的BP算法,拟合效果较好。

关键词: 粒子群优化, 灰色关联分析, 贝叶斯推理, 神经网络

Abstract: Based on Particle Swarm Optimization(PSO),by using the grey correlation analysis on the hidden node number’s determination of BP nueral network,this method realizes the optimal of BP nueral network.Bayesian inference methods are applied to the training of feed-forward neural networks in order to improve their generalization capabilities.The results show that the algorithm has better generalization capacity than other improved BP algorithms,and furthermore,it has better effects of approximation.

Key words: particle swarm optimization, grey correlation analaysis, Bayesian inference, nueral network