J4 ›› 2011, Vol. 29 ›› Issue (6): 585-589.

• 论文 • 上一篇    下一篇

基于模拟退火算法的Hopfield神经网络参数优化

齐小刚|王云鹤   

  1. 西安电子科技大学 数学与应用数学系|西安 710126
  • 出版日期:2011-11-24 发布日期:2011-12-06
  • 作者简介:齐小刚(1973—)|男|陕西宝鸡人|西安电子科技大学教授|博士|主要从事网络优化方法与应用研究|(Tel)86-29-88202060(E-mail) qixiaogang@gmail.com。

Hopfield Neural Network Parameter Optimization Based on Simulated Annealing Algorithm

QI Xiao-gang, WANG Yun-he   

  1. Department of Mathematics, Xidian University, Xi'an 710126,China
  • Online:2011-11-24 Published:2011-12-06

摘要:

为解决Hopfield神经网络应用过程中参数设置的问题,在研究Hopfield神经网络的工作原理的基础上,分析了神经网络模型在求解TSP(Traveling Salesman Problem)问题过程中参数的选取,通过对输出数据进行归一化处理建立网络的评价函数,然后引入模拟退火算法对参数进行最优化选取。实验结果表明,经过参数优化过的Hopfield神经网络模型能更有效,更快速地得到TSP问题的最优解。

关键词: 参数优化, 神经网络, TSP问题, 模拟退火

Abstract:

In order to solve the parameter setting problem during the application process of Hopfield neural network.The working principle of Hopfield neural network is described, the neural network model parameter selection problem in the TSP(Traveling Salesman Problem) problems solving process is analyed On the basis established the evaluation function of network by using normalized on output data, and then use simulated annealing algorithm to select the optimal parameters. The results show that, after optimization of parameters, Hopfield neural network can obtain the optimal solution of TSP problems more effective and more quickly.

Key words: parameter optimize, neural network, traveling salesman problem(TSP), simulated annealing

中图分类号: 

  • TP301