Exclusive Content & Downloads from ASQ

A Computationally Stable Approach to Gaussian Process Interpolation of Deterministic Computer Simulation Data

Summary: Gaussian process (GP) models are commonly used to model the output of deterministic simulators with no replication error. This requires the computation of certain values for a number of correlation matrices which are sometimes approaching singularity. The usual method of dealing with near-singularity is the introduction of a nugget parameter, which can often over-smooth the data. The authors propose setting a minimum limit on the nugget and creating a predictor using an iterative regularization approach, minimizing over-smoothing and increasing the accuracy of the interpolation. This proposed predictor is shown to converge with the GP interpolator.

Anyone with a subscription, including Site and Enterprise members, can access this article.

Other Ways to Access content:

Join ASQ

Join ASQ as a Full member. Enjoy all the ASQ member benefits including access to many online articles.

  • Topics: Statistics
  • Keywords: Approximation, Computers, Data smoothing, Gaussian curve, Parameters
  • Author: Ranjan, Pritam; Haynes, Ronald; Karsten, Richard
  • Journal: Technometrics