The Use of Hide in Learning the Value of a Function

摘要

In this paper, we briefly review some recent work on Hypercircle inequality for data error (Hide) measured with square loss. We provide it in the case that the unit ball B is replaced by SB where S is any positive number. We study the problem in learning the value of a function in reproducing kernel Hilbert space (RKHS) by using the available material from Hide with different values of δ. Moreover, we compare our numerical experiment to the method of regularization, which is the standard method for learning problem. We also discuss the effect of the values of δ on the learning task under consideration.