基于梯度下降法选取高斯径向基函数形状参数  

Gaussian Radial Basis Function Shape Parameter Is Selected Based on Gradient Descent Method

作　　者：谢志超 王玲 龚佃选 孙建 田炜印 

机构地区：[1]华北理工大学理学院,河北 唐山

出　　处：《应用数学进展》2023年第4期1640-1647,共8页Advances in Applied Mathematics

摘　　要：高斯径向基函数是径向基函数中常用的函数,对于高斯径向基函数插值形状参数的取值,常规的方法一般是通过人工修改形状参数的取值,计算成本较高,插值精度较低。本文基于梯度下降算法,通过设定针对性的目标函数,通过迭代的方式得到形状参数,有更好的插值效果,为高斯径向基函数的应用提供了更加便捷的途径。Gaussian radial basis function is a commonly used function in radial basis function. For the value of shape parameters interpolated by Gaussian radial basis function, the conventional method is gen-erally to manually modify the value of shape parameters, which has high calculation cost and low interpolation accuracy. In this paper, based on the gradient descent algorithm, shape parameters are obtained through iteration by setting targeted objective functions, which has better interpola-tion effect and provides a more convenient way for the application of Gaussian radial basis function.

关 键 词：梯度下降法 高斯径向基函数 插值逼近 形状参数 

分 类 号：G63[文化科学—教育学]