Gaussian process regression is a non-parametric, Bayesian approach to regression. In contrast to parametric regression techniques, the complexity of the Gaussian process models scales with the number of training points which allows highly flexible function reconstruction. As Bayesian approach, the regression does not only provide a prediction but also an uncertainty measure for the quality of the regression.
This app allows to get hands-on experience with Gaussian process regression for educational purpose. The behavior of the Gaussian Process model with respect to new training data, different kernel functions and hyperparameters can be explored in an interactive way.
More information about Gaussian Processes and its application in control can be found on www.tbeckers.com
The app requires Google Chrome.