计算机工程 ›› 2011, Vol. 37 ›› Issue (10): 152-153.doi: 10.3969/j.issn.1000-3428.2011.10.051

• 人工智能及识别技术 • 上一篇    下一篇

基于k均值和量子遗传算法的RBF网络优化

赵 磊 1,贾振红 1,覃锡忠 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)
  • 出版日期:2011-05-20 发布日期:2011-05-20
  • 作者简介:赵 磊(1987-),男,硕士研究生,主研方向:量子遗传算法,数字图像处理;贾振红,教授;覃锡忠,讲师;杨 杰、庞韶宁,博士
  • 基金项目:
    科技部国际科技合作基金资助项目(2009DFA12870)

RBF Network Optimization Based on k-means and Quantum Genetic Algorithm

ZHAO Lei 1, JIA Zhen-hong 1, QIN Xi-zhong 1, YANG Jie 2, PANG Shao-ning 3   

  1. (1. College of Information Science & Engineering, Xinjiang University, Urumqi 830046, China;2. Institute of Image Processing and Pattern Recognition, Shanghai Jiaotong University, Shanghai 200240, China;3. Knowledge Engineering & Research Discovery Institute, Auckland University of Technology, Auckland 1020, New Zealand)
  • Online:2011-05-20 Published:2011-05-20

摘要: 针对遗传算法容易出现早熟的问题,提出一种基于k均值和量子遗传算法的径向基函数(RBF)神经网络组合优化方法。通过k均值聚类求取网络的中心,用量子遗传算法训练网络的权值,利用量子染色体的表示方式以及量子染色体的更新提高算法的并行性,从而解决遗传算法早熟的问题,提高网络的适应度。相对于PSO-RBF和ACO-RBF,该方法提高网络的收敛速度,实现对RBF网络的优化。

关键词: 早熟, k均值, 量子遗传算法, 适应度

Abstract: According to the prematurity matter of the Genetic Algorithm(GA), this paper gives a complex optimization method based on k-means and Quantum Genetic Algorithm(QGA). It accounts the center of the RBF neural network through k-means, and trains the weights of RBF neural network by QGA. The express and renewal of quantum chromosome is used to improve the parallel of this procedure, so the problem of prematurity is solved, the fitness of network is improved, and convergence speed of the network is improved compared with the PSO-RBF and ACO-RBF. Optimization of the RBF neural network is implemented.

Key words: prematurity, k-means, Quantum Genetic Algorithm(QGA), fitness

中图分类号: