Abstract: In this paper we investigate deep learning architectures combined with low-discrepancy sampling as surrogate models. The aim is to provide a quick estimate of the outcome of an expensive ...