计算机工程与应用 ›› 2011, Vol. 47 ›› Issue (28): 59-61.

• 研究、探讨 • 上一篇    下一篇

用蛙跳算法优化RBF神经网络参数的研究

薛升翔1,贾振红1,杨 杰2,庞韶宁3   

  1. 1.新疆大学 信息科学与工程学院,乌鲁木齐 830046
    2.上海交通大学 图像处理与模式识别研究所,上海 200240
    3.新西兰奥克兰理工大学 知识工程与开发研究所,新西兰 奥克兰 1020
  • 收稿日期:1900-01-01 修回日期:1900-01-01 出版日期:2011-10-01 发布日期:2011-10-01

Using shuffled frog leaping algorithm to optimize the parameters of radial basis function neural network

XUE Shengxiang1,JIA Zhenhong1,YANG Jie2,PANG Shaoning3   

  1. 1.School of Information Science and Engineering,Xinjiang University,Urumqi 830046,China
    2.Institute of Image Processing and Pattern Recognition,Shanghai Jiaotong University,Shanghai 200240,China
    3.Knowledge Engineering and Discovery Research Institute,Auckland University of Technology,Auckland 1020,New Zealand
  • Received:1900-01-01 Revised:1900-01-01 Online:2011-10-01 Published:2011-10-01

摘要: 针对径向基函数(Radial Basis Functions,RBF)神经网络结构参数确定问题,提出了一种基于蛙跳算法优化RBF神经网络参数的新方法。将RBF神经网络参数组成一个多维向量,作为蛙跳算法中的参数进行优化。以适应度函数为标准,在可行解空间中搜索最优解,并对蛙跳算法进行了改进。非线性函数逼近实验结果表明,该优化算法相对标准遗传优化算法、粒子群优化算法有较小的均方误差,具有更好的逼近能力。

关键词: 蛙跳算法, 径向基函数神经网络, 非线性函数逼近, 参数优化

Abstract: In allusion to being difficult to determine the parameters of Radial Basis Functions Neural Network(RBFNN),a new method on the parameters optimization of radial basis function neural network based on Shuffled Frog Leaping Algorithm(SFLA) is proposed.The parameters of the RBFNN compose a multidimensional vector which is regarded as parameters of SFLA to optimize.According to the fitness function,the feasible sampling space is searched for the global optima,further more,the SFLA has been improved.The simulation test on nonlinear function approximation shows that compared to Genetic Algorithm(GA) and Particle Swarm Optimization(PSO) the new method has less mean square error and better approximation ability.

Key words: shuffled frog leaping algorithm, Radial Basis Functions Neural Network(RBFNN), nonlinear function approximation, parameters optimization