Times Series Prediction Using ICA Algorithms and VC Theory

摘要

In this paper we propose a new method for volatile time series forecasting using Independent Component Analysis (ICA) algorithms and filtering as preprocessing tools. The endogenous learning machine, consisting of an Artificial Neural Network (ANN) based on radial basis functions (RBF), uses the preprocessed data from theses algorithms obtaining improvements in prediction results. The endogenous learning machine is a new on-line parametric model 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. Finally we mention some modifications and extensions that can be applied to control neural resources and select relevant input space.