We gratefully acknowledge support from
the Simons Foundation and member institutions.
Full-text links:


Current browse context:


Change to browse by:

References & Citations


(what is this?)
CiteULike logo BibSonomy logo Mendeley logo del.icio.us logo Digg logo Reddit logo ScienceWISE logo

Statistics > Methodology

Title: Potentially Predictive Variance Reducing Subsample Locations in Local Gaussian Process Regression

Abstract: Gaussian process models are commonly used as emulators for computer experiments. However, developing a Gaussian process emulator can be computationally prohibitive when the number of experimental samples is even moderately large. Local Gaussian process approximation (Gramacy and Apley, 2015) was proposed as an accurate and computationally feasible emulation alternative. However, constructing local sub-designs specific to predictions at a particular location of interest remains a substantial computational bottleneck to the technique. In this paper, two computationally efficient neighborhood search limiting techniques are proposed, a maximum distance method and a feature approximation method. Two examples demonstrate that the proposed methods indeed save substantial computation while retaining emulation accuracy.
Comments: 25 pages, 8 figures
Subjects: Methodology (stat.ME); Computation (stat.CO)
DOI: 10.5705/ss.202016.0138
Cite as: arXiv:1604.04980 [stat.ME]
  (or arXiv:1604.04980v2 [stat.ME] for this version)

Submission history

From: Chih-Li Sung [view email]
[v1] Mon, 18 Apr 2016 04:05:59 GMT (189kb,D)
[v2] Sat, 26 Nov 2016 20:05:41 GMT (209kb,D)

Link back to: arXiv, form interface, contact.