Function representation using hypercircle inequality for data error

Abstract

Abstract-In this paper, the study of Hypercircle inequality for data error (Hide) is briefly reviewed. Within the framework of Hide and to find a function representation from inaccurate data, the midpoint algorithm is provided. We give a new result for a function representation that has the form of the representer theorem. We illustrate some important facts for a practical computation and study the problem in the learning value of a function for a learning kernel. We demonstrate the potential of this framework by comparing our result to the regularization method, which is the standard method in the learning value of a function. The present example compares the performance of the methods when the optimal values of regularization parameters are used.