文献类型：期刊文献

英文题名：CONVERGENCE ANALYSIS FOR KERNEL-REGULARIZED ONLINE REGRESSION ASSOCIATED WITH AN RRKHS

作者：Pan, Xiaoling[1];Liu, Lin[1];Sheng, Baohuai[1,2]

机构：[1]Shaoxing Univ, Sch Math Phys & Informat, Shaoxing 312000, Peoples R China;[2]Zhejiang Yuexiu Univ, Sch Int Business, Dept Econ Stat, Shaoxing 312000, Peoples R China

年份：2023

外文期刊名：COMMUNICATIONS ON PURE AND APPLIED ANALYSIS

收录：SCI-EXPANDED(收录号：WOS:000970732100001)、、Scopus(收录号：2-s2.0-85163324496)、WOS

基金：This work was supported by the NSFC/RGC Joint Research Scheme (Project No. 12061160462 and N-CityU102/20) and the National Natural Science Foundation of China under Grant No. 61877039.

语种：英文

外文关键词：Online learning; regression algorithm; parameterized loss; radon re-producing kernel Hilbert space; error analysis

外文摘要：It is known that the functional reproducing kernel Hilbert space (FRKHS) theory lays the functional analysis foundation for learning non-point evaluation functional data with kernel-regularized learning. In the present paper, we investigate the convergence of regression learning associated with Radon reproducing kernel Hilbert spaces (RRKHSs) and a parameterized loss. We provide a kind of online learning algorithm and establish an upper bound for the learning rate, and it shows that the learning rate may be improved by adjusting the parameter in the loss function. As incidental conclusions, we provide a method of constructing FRKHSs with Fourier analysis, some RRKHSs are defined for the classes of periodic functions defined on [-pi, pi](s); the functions defined on the interval [-1, 1]; the functions defined on R+ = [0, +infinity) and the functions defined on the unit sphere.