电子学报 ›› 2008, Vol. 36 ›› Issue (7): 1396-1400.

• 论文 • 上一篇    下一篇

一种基于免疫系统的RBF网络在线训练方法

臧小刚, 宫新保, 常 成, 凌小峰, 唐 斌   

  1. 上海交通大学电子工程系,上海 200240
  • 收稿日期:2007-08-06 修回日期:2007-11-05 出版日期:2008-07-25 发布日期:2008-07-25

An Online Training RBF Network Based on Immune System

ZANG Xiao-gang, GONG Xin-bao, CHANG Cheng, LING Xiao-feng, TANG Bin   

  1. Department of Electronic Engineering,Shanghai Jiaotong University,Shanghai 200240,China
  • Received:2007-08-06 Revised:2007-11-05 Online:2008-07-25 Published:2008-07-25

摘要: 针对径向基函数(RBF)网络和免疫系统的相似性,本文提出了一种基于免疫模型的RBF网络在线学习方法以解决动态问题.该方法借鉴了免疫系统动态调整以对抗不断入侵的抗原的机制,通过免疫初步覆盖、免疫交叉响应和疫苗注射等免疫操作,加速算法效率、提高算法精度和动态性能.通过以上操作使得RBF网络能够根据样本的变化迅速地调整网络结构与参数.计算机仿真研究表明,采用这种方法设计的RBF网络在动态环境下具有优良的精度和泛化能力.

关键词: 径向基函数网络, 生物免疫系统, 动态系统, 免疫操作

Abstract: Inspired by the similarity between radial basis function (RBF) network and immune system,a RBF online training algorithm based on immune model is proposed to solve dynamic problems.The algorithm takes its inspiration from the dynamic adjustment mechanism of natural immune system to combat an ever-changing cast of antigens.Immune operations such as initial immune coverage,cross-reactive response and immune vaccination are implemented in the algorithm to improve the efficiency,precision and dynamic performance.Through these immune operations the RBF network can quickly adjust its structure and parameters according to changes in training set.Computer simulations demonstrate that the RBF network designed by the algorithm has good precision and generalization ability in dynamic environments.

Key words: radial basis function network, natural immune system, dynamic system, immune operation

中图分类号: