Sequential adaptive design for emulating costly computer codes

Dr Hossein Mohammadi had this to say about the paper’s findings:

“Approximating computationally expensive computer models is a common challenge in many real-world applications. This is often addressed using surrogate models such as Gaussian process emulators. Adaptive design of experiments provides a framework to efficiently sample from these expensive models—such as those used in landscape decision problems—to build accurate emulators.

In this paper, we propose a novel adaptive sampling criterion called VIGF (Variance of Improvement for Global Fit) for Gaussian processes. We assess the applicability of VIGF using a range of test functions and compare its performance against several sampling strategies. The results suggest that our method outperforms existing approaches in most cases when predicting benchmark functions”.