基于函数型数据的系数正则化回归的收敛速度
ON THE CONVERGENCE RATE OF COEFFICIENT-BASED REGULARIZED REGRESSION FOR FUNCTIONAL DATA

本文研究了基于函数型输入和l1-正则化的最小二乘回归问题的推广性能.利用基于Rademacher平均的分析技术,获得了学习速度的估计,推广了已有的欧式空间有限维输入结果.
This paper investigates the generalization performance of least square regression with functional data and l1-regularizer. The estimate of learning rate is established by Rademacher average technique. The theoretical result is a natural extension for coefficient-based regularized regression when input space is a subset of infinite-dimensional Euclidean space