Show simple item record

dc.contributor.authorLuo, Hengrui
dc.contributor.authorNattino, Giovanni
dc.contributor.authorPratola, Matthew T.
dc.date.accessioned2022-05-30T11:05:41Z
dc.date.available2022-05-30T11:05:41Z
dc.date.issued2022-01-01
dc.identifier.issn1533-7928
dc.identifier.issn1532-4435
dc.identifier.urihttp://hdl.handle.net/10754/678325
dc.description.abstractIn this paper we introduce a novel model for Gaussian process (GP) regression in the fully Bayesian setting. Motivated by the ideas of sparsification, localization and Bayesian addi- tive modeling, our model is built around a recursive partitioning (RP) scheme. Within each RP partition, a sparse GP (SGP) regression model is fitted. A Bayesian additive frame- work then combines multiple layers of partitioned SGPs, capturing both global trends and local refinements with efficient computations. The model addresses both the prob- lem of efficiency in fitting a full Gaussian process regression model and the problem of prediction performance associated with a single SGP. Our approach mitigates the issue of pseudo-input selection and avoids the need for complex inter-block correlations in existing methods. The crucial trade-off becomes choosing between many simpler local model com- ponents or fewer complex global model components, which the practitioner can sensibly tune. Implementation is via a Metropolis-Hasting Markov chain Monte-Carlo algorithm with Bayesian back-fitting. We compare our model against popular alternatives on simu- lated and real datasets, and find the performance is competitive, while the fully Bayesian procedure enables the quantification of model uncertainties.
dc.description.sponsorshipThe authors wish to thank the helpful feedback of the editor, associate editor and two anonymous reviewers, which helped to substantially improve the paper. The work of M.T.P. was supported in part by the National Science Foundation under Agreement DMS-1916231 and in part by the King Abdullah University of Science and Technology (KAUST) Office of Sponsored Research (OSR) under Award No. OSR-2018-CRG7-3800.3.
dc.publisherMicrotome Publishing
dc.titleSparse Additive Gaussian Process Regression
dc.typeArticle
dc.contributor.institutionDepartment of Statistics, The Ohio State University, Columbus, OH 43210, USA
dc.contributor.institutionDivision of Biostatistics, College of Public Health, The Ohio State University, Columbus, OH 43210, USA
dc.identifier.volume23
kaust.grant.numberOSR-2018-CRG7-3800.3
dc.identifier.eid2-s2.0-85130407360
kaust.acknowledged.supportUnitOffice of Sponsored Research (OSR)


This item appears in the following Collection(s)

Show simple item record