Using confidence interval of a regularization network

In this paper we establish new bounds on for the actual risk functional of a new on-line parametric machine for time series forecasting based on Vapnik-Chervonenkis (VC) theory. Using the strong connection between support vector machines (SVM) and Regularization theory (RT), we propose a regularization operator in order to obtain a suitable expansion of radial basis functions (RBFs) with the corresponding expressions for updating neural parameters. This operator seeks for the "flattest" function in a feature space, minimizing the risk functional and controlling the capacity of the learning machine. Finally, we mention some modifications and extensions that can be applied to control neural resources (complexity control) and select relevant input space (suitable expression) to avoid high computational effort.