With robust statistics we mean methods that work well, if a chosen model is true and that are acceptable if the model is only an approximation. But if the model is far from the true one robust methods may be very bad (Huber 1980, Hampel & al 1986). Thus robust statistics should be used whenever we know that the chosen model is only an approximation to the true model.
KeywordsPosterior Distribution Covariance Function Gaussian Process True Distribution Choose Model
Unable to display preview. Download preview PDF.