Please use this identifier to cite or link to this item: https://doi.org/10.1162/08997660151134343
Title: Predictive approaches for choosing hyperparameters in gaussian processes
Authors: Sundararajan, S.
Keerthi, S.S. 
Issue Date: May-2001
Citation: Sundararajan, S., Keerthi, S.S. (2001-05). Predictive approaches for choosing hyperparameters in gaussian processes. Neural Computation 13 (5) : 1103-1118. ScholarBank@NUS Repository. https://doi.org/10.1162/08997660151134343
Abstract: Gaussian processes are powerful regression models specified by parameterized mean and covariance functions. Standard approaches to choose these parameters (known by the name hyperparameters) are maximum likelihood and maximum a posteriori. In this article, we propose and investigate predictive approaches based on Geisser's predictive sample reuse (PSR) methodology and the related Stone's cross-validation (CV) methodology. More specifically, we derive results for Geisser's surrogate predictive probability (GPP), Geisser's predictive mean square error (GPE), and the standard CV error and make a comparative study. Within an approximation we arrive at the generalized cross-validation (GCV) and establish its relationship with the GPP and GPE approaches. These approaches are tested on a number of problems. Experimental results show that these approaches are strongly competitive with the existing approaches.
Source Title: Neural Computation
URI: http://scholarbank.nus.edu.sg/handle/10635/61140
ISSN: 08997667
DOI: 10.1162/08997660151134343
Appears in Collections:Staff Publications

Show full item record
Files in This Item:
There are no files associated with this item.

SCOPUSTM   
Citations

52
checked on Jul 14, 2018

WEB OF SCIENCETM
Citations

44
checked on Jun 18, 2018

Page view(s)

17
checked on Apr 21, 2018

Google ScholarTM

Check

Altmetric


Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.