• 主题
高级搜索  >
历史搜索 清空历史
首页> 外文会议>International Conference on Advanced Design and Manufacturing Engineering >Research for RBF Neural Networks Modeling Accuracy of Determining the Basis Function Center Based on Clustering Methods
【24h】

Research for RBF Neural Networks Modeling Accuracy of Determining the Basis Function Center Based on Clustering Methods

机译：基于聚类方法确定基础函数中心的RBF神经网络建模准确性

获取原文
页面导航
  • 摘要
  • 著录项
  • 相似文献
  • 相关主题

摘要

The radial basis function (RBF) neural network is superior to other neural network on the aspects of approximation ability, classification ability, learning speed and global optimization etc., it has been widely applied as feedforward networks, its performance critically rely on the choice of RBF centers of network hidden layer node. K-means clustering, as a commonly method used on determining RBF center, has low neural network generalization ability, due to its clustering results are not sensitive to initial conditions and ignoring the influence of dependent variable. In view of this problem, fuzzy clustering and grey relational clustering methods are proposed to substitute K-means clustering, RBF center is determined by the results of fuzzy clustering or grey relational clustering, and some researches of RBF neural networks modeling accuracy are done. Practical modeling cases demonstrate that the modeling accuracy of fuzzy clustering RBF neural networks and grey relational clustering RBF neural networks are significantly better than K-means clustering RBF neural networks, applying of fuzzy clustering or grey relational clustering to determine the basis function center of RBF neural networks hidden layer node is feasible and effective.
机译：径向基函数（RBF）神经网络优于近似能力，分类能力，学习速度和全局优化等的其他神经网络，它已被广泛应用于前馈网络，其性能批准依赖于选择RBF网络隐藏层节点的中心。 K-Means Clustering作为在确定RBF中心的常用方法具有低的神经网络泛化能力，由于其聚类结果对初始条件并不敏感并忽略所属变量的影响。鉴于此问题，提出模糊聚类和灰色关系聚类方法以替代K-means聚类，RBF中心由模糊聚类或灰色关系聚类的结果确定，并完成了RBF神经网络建模精度的一些研究。实用建模案例表明，模糊聚类RBF神经网络和灰色关系聚类RBF神经网络的建模精度明显优于K-means聚类RBF神经网络，应用模糊聚类或灰色关系聚类来确定RBF神经网络的基函数中心网络隐藏层节点是可行和有效的。

著录项

相似文献

  • 外文文献
  • 中文文献
  • 专利
获取原文

客服邮箱：kefu@zhangqiaokeyan.com

京公网安备：11010802029741号 ICP备案号：京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有

  • 客服微信

  • 服务号