西华大学学报（自然科学版）2018，Vol.37Issue(1)：46-50,5.DOI:10.3969/j.issn.1673-159X.2018.01.007
高斯核的一般误差估计
The Estimation of Generalization Error for Gauss Kernel
摘要
Abstract
The sample error and the approximation error have been investigated deeply and obtained different as ymptotic expressions.These two expressions can characterize the general error indirectly.In this paper,by adding conditions,the estimates of sample error and approximation error are unified into one estimate expression of general error:by fixing the hypothesis space,the general error has a decay of logarithmic order with the hypothesis space radius;by fixing number of sampling points,the general error has a decay of exponential order with sampling number.关键词
机器学习/最小二乘误差/高斯核/样本误差/逼近误差/误差估计Key words
machine learning/least squares error/Gaussian kernel/sample error/approximation error/error estimation分类
数理科学引用本文复制引用
韩永杰,陈洁..高斯核的一般误差估计[J].西华大学学报（自然科学版）,2018,37(1):46-50,5.基金项目
西华大学校重点基金资助项目(z1412621) （z1412621）
国家自然科学基金资助项目(15233593) （15233593）
四川省教育厅基金资助项目(15233448). （15233448）
