ongoing

We developed a scalable approximate inference algorithm for robust Gaussian process regression with contaminated normal noise. Initial experiments with simulated and actual data are promising.