最小二乘算法在神经网络函数逼近方面的研究  

Research of OLS Algorithm in Neural Network Function Approximation

作　　者：朱艺[1] 孙丽英[1] 葛超[1] 张景春[1] 

机构地区：[1]河北理工大学信息学院,河北唐山063009

出　　处：《山西电子技术》2007年第5期13-14,共2页Shanxi Electronic Technology

摘　　要：介绍了人工神经网络的一个重要的模型径向基函数网络,并且对径向基函数中心的选取问题进行了研究,最后给出了应用实例,结果证明,由于计算过程中应用了最小二乘算法的正交化性质,所以网络调整时对已有模式的扰动可以达到最小。这说明最小二乘算法不仅简单有效,而且性能优越,并有较强的实用性,在许多领域有广泛应用。This paper introduces an important model named radial basis function（RBF）in artificial neural network（ANN）.The problem of selecting to network center is researched in making radial basic function network.Because the orthogonal of OLS algorithm is used in the process of independence calculate,it makes fewer disturbances to the former model.At the end of the paper an example of the application approximation of functions is given to explain the arithmetic.The example indicates that the OLS is an excellent algorithm because its orthogonal quality,so it has wide application.

关 键 词：最小二乘算法 径向基函数 函数逼近 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]