Please use this identifier to cite or link to this item:
Title: Mixtures of experts for understanding model discrepancy in dynamic computer models
Authors: Nott, D.J. 
Marshall, L.
Fielding, M.
Liong, S.-Y. 
Keywords: Bayesian inference
Computer model
Mixtures of experts
Issue Date: Mar-2014
Citation: Nott, D.J., Marshall, L., Fielding, M., Liong, S.-Y. (2014-03). Mixtures of experts for understanding model discrepancy in dynamic computer models. Computational Statistics and Data Analysis 71 : 491-505. ScholarBank@NUS Repository.
Abstract: There are many areas of science and engineering where research and decision making are performed using computer models. These computer models are usually deterministic and may take minutes, hours or days to produce an output for a single value of the model inputs. Fitting mixtures of experts of computer models where the expert components use different values of the computer model parameters is considered. The efficient calibration of such models using emulators, which are fast statistical surrogates for the computer model, is discussed. It is argued that mixtures of experts are often insightful for describing model discrepancy and ways in which the computer model can be improved. This is not a strength of standard approaches to the statistical analysis of computer models where a certain "best input" assumption is usually made and model discrepancy is often described through a stationary Gaussian process prior on the discrepancy function. Application of the framework is presented for a dynamic hydrological rainfall-runoff model in which the mixture approach is helpful for highlighting model deficiencies. © 2013 Elsevier B.V. All rights reserved.
Source Title: Computational Statistics and Data Analysis
ISSN: 01679473
DOI: 10.1016/j.csda.2013.04.020
Appears in Collections:Staff Publications

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


checked on Nov 21, 2022


checked on Nov 21, 2022

Page view(s)

checked on Nov 24, 2022

Google ScholarTM



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