D. PoltemK. KhompurngsonB. NovaprateepBurapha UniversityPERDOUniversity of PhayaoMahidol University2018-06-112018-06-112012-12-01International Journal of Mathematics and Computers in Simulation. Vol.6, No.6 (2012), 521-528199801592-s2.0-84875711610https://repository.li.mahidol.ac.th/handle/20.500.14594/14026Abstract-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.Mahidol UniversityComputer ScienceMathematicsArticleSCOPUS