摘要
提出了一种基于鲁棒学习的鲁棒RBF神经网络。使用聚类方法选择RBF网络隐结点的中心,以鲁棒代价函数为目标函数,采用梯度下降法调整隐层结点的宽度和网络权值,从而使RBF网络的学习过程不受离群点的影响,并且能够快速收敛。仿真实验结果表明了RBF神经网络的鲁棒优越性。
In this paper,A Robust RBF Network Based on Robust Learning Algorithm is presented. The algorithm uses the k-means clustering method to select hidden node centers of RBF network ,and the gradient descent method with the robust loss function (RLF) as the objective function to adjust hidden node widths and the connection weight s of the network. Therefore,the learning of RBF network has robustness on dealing with outliers and fast rate of convergence. The simulation results show the advantages of the learning algorithm over traditional learning algorithms for RBF network.
出处
《科技信息》
2010年第07X期42-43,184,共3页
Science & Technology Information
关键词
RBF神经网络
聚类
鲁棒损失函数
离群点
RBF neural network
K-means clustering
Robust loss function
Outlier