Learning rates of least-square regularized regression with strongly mixing observation

Abstract

This paper considers the regularized learning algorithm associated with the least-square loss, strongly mixing observations and reproducing kernel Hilbert spaces. We first give the bound of the sample error with exponentially strongly mixing observations and the rate of approximation by Jackson-type theorem of approximation based on exponentially strongly mixing sequence. Then the generalization error of the least-square regularized regression is obtained by estimating sample error and regularization error.