List of Figures

1. GP regression with RBF and polynomial kernels
2. A graphical illustration of the online learning
3. Projection in the feature space
4. The evolution of the KL-divergences for the online training
5. Parameter decomposition
6. Errors made by ignoring the covariance-term in the score
7. A graphical illustration of the expectation propagation algorithm
8. Regression results for the $\ensuremath{\mathrm{sinc}}$ function.
9. Regression results for the Friedman dataset.
10. Regression results for the Boston dataset.
11. Sparse EP applied to Crab data.
12. Sparse EP applied to Sonar data.
13. Results for the USPS dataset for binary and combined classification.
14. Multiclass classification with rejecting the uncertain predictions.
15. The kernel used for density estimation.
16. Results for the GP density estimation.
17. The local ``likelihoods'' in the wind field model.
18. Updates for the vectorial GP
19. NWP wind-field prediction and sparse GP predictions.
20. Computing the relative weight of a GP approximation
21. The predicted wind-fields based on the sparse GP.
22. Grouping of the GP parameters (Fig 3.3 repeated).