Active Learning via Heteroskedastic Rational Kriging

Shangkun Wang, V. Roshan Joseph

Active learning methods for emulating complex computer models that rely on stationary Gaussian processes tend to produce design points that uniformly fill the entire experimental r…