Online algorithm for time series prediction based on support vector machine philosophy

In this paper we prove the analytic connection between Support Vector Machines (SVM) and Regularization, Theory (RT) and show, based on this prove, a new on-line parametric model for time series forecasting based on Vapnik-Chervonenkis (VC) theory. Using the latter strong connection, we propose a regularization operator in order to obtain a suitable expansion of radial basis functions (RBFs) and expressions for updating neural parameters. This operator seeks for the "flattest" function in a feature space, minimizing the risk functional. Finally we mention some modifications and extensions that can be applied to control neural resources and select relevant input space.