Skip to main content
Article
Massively Parallel Approximate Gaussian Process Regression
SIAM/ASA Journal on Uncertainty Quantification
  • Robert B. Gramacy, University of Chicago
  • Jarad Niemi, Iowa State University
  • Robin M. Weiss, University of Chicago
Document Type
Article
Publication Version
Published Version
Publication Date
9-30-2014
DOI
10.1137/130941912
Abstract

We explore how the big-three computing paradigms---symmetric multiprocessor, graphical processing units (GPUs), and cluster computing---can together be brought to bear on large-data Gaussian processes (GP) regression problems via a careful implementation of a newly developed local approximation scheme. Our methodological contribution focuses primarily on GPU computation, as this requires the most care and also provides the largest performance boost. However, in our empirical work we study the relative merits of all three paradigms to determine how best to combine them. The paper concludes with two case studies. One is a real data fluid-dynamics computer experiment which benefits from the local nature of our approximation; the second is a synthetic example designed to find the largest data set for which (accurate) GP emulation can be performed on a commensurate predictive set in under an hour.

Comments

This is an article from SIAM/ASA Journal on Uncertainty Quantification 2 (2014): 564, doi: 10.1137/130941912. Posted with permission.

Copyright Owner
2014
Language
en
File Format
application/pdf
Citation Information
Robert B. Gramacy, Jarad Niemi and Robin M. Weiss. "Massively Parallel Approximate Gaussian Process Regression" SIAM/ASA Journal on Uncertainty Quantification Vol. 2 Iss. 1 (2014) p. 564 - 584
Available at: http://works.bepress.com/jarad_niemi/11/