摘要
采用主成分分析法 (PCA)来改善径向基函数网络的泛化性能 ,理论上可以根据PCA方法中的主成分累积贡献率 ηK 决定RBF网络的输入层节点数 .实例研究证明 。
The principal component analysis (PCA)method is investigated for improving generalization performance of radial basis function(RBF)network. It becomes theoretically available to determining the number of input neurons in RBF network based on the cumulative contribution rate η\-K which is the result of PCA.Case study shows that the RBF network with introducing PCA method consistently outperforms conventional RBF network in root mean square error and model efficiency R\-y.
出处
《武汉水利电力大学学报》
EI
CSCD
2000年第2期85-89,共5页
Engineering Journal of Wuhan University
关键词
人工神经网络
径向基函数
泛化性能
主成分
artificial neural network
radial basis function
generalization performance
principle component analysis method
