Prediction Error Adjusted Gaussian Process for Nonlinear Non-parametric System Identification

A variant of Gaussian Process is proposed in this study for nonlinear non-parametric system identification. Only local data is used to construct the estimate. Moreover, the hyperparameters are adjusted to minimize the local weighted prediction errors. The proposed scheme seems to have semi-global modeling properties of Gaussian Process for limited data sets and also possess local convergence properties if the data set is sufficient rich.