Numerical issues in maximum likelihood parameter estimation for Gaussian process interpolation
This work addresses a basic but widespread problem affecting many studies that rely on off-the-shelf Gaussian process implementations, though it is incremental in nature.
The paper investigates numerical issues in maximum likelihood parameter estimation for Gaussian process interpolation and proposes simple strategies to improve open-source software implementations, which are critical for reliable and reproducible studies in fields like Bayesian optimization.
This article investigates the origin of numerical issues in maximum likelihood parameter estimation for Gaussian process (GP) interpolation and investigates simple but effective strategies for improving commonly used open-source software implementations. This work targets a basic problem but a host of studies, particularly in the literature of Bayesian optimization, rely on off-the-shelf GP implementations. For the conclusions of these studies to be reliable and reproducible, robust GP implementations are critical.