Explainable Learning with Gaussian Processes