Show simple item record

dc.contributor.authorWilson, Andrew Gordon
dc.contributor.authorAdams, Ryan Prescott
dc.date.accessioned2013-11-25T20:46:27Z
dc.date.issued2013
dc.identifierQuick submit: 2013-08-08T11:43:14-04:00
dc.identifier.citationAndrew Gordon Wilson; Ryan Prescott Adams. 2013. Gaussian process kernels for pattern discovery and extrapolation. Journal of Machine Learning Research 28, no. 3: 1067-1075.en_US
dc.identifier.issn1532-4435en_US
dc.identifier.urihttp://nrs.harvard.edu/urn-3:HUL.InstRepos:11337457
dc.description.abstractGaussian processes are rich distributions over functions, which provide a Bayesian nonparametric approach to smoothing and interpolation. We introduce simple closed form kernels that can be used with Gaussian processes to discover patterns and enable extrapolation. These kernels are derived by modeling a spectral density – the Fourier transform of a kernel – with a Gaussian mixture. The proposed kernels support a broad class of stationary covariances, but Gaussian process inference remains simple and analytic. We demonstrate the proposed kernels by discovering patterns and performing long range extrapolation on synthetic examples, as well as atmospheric CO2 trends and airline passenger data. We also show that it is possible to reconstruct several popular standard covariances within our framework.en_US
dc.description.sponsorshipEngineering and Applied Sciencesen_US
dc.language.isoen_USen_US
dc.publisherMicrotome Publishingen_US
dash.licenseOAP
dc.titleGaussian Process Kernels for Pattern Discovery and Extrapolationen_US
dc.typeConference Paperen_US
dc.date.updated2013-08-08T15:43:44Z
dc.description.versionAccepted Manuscripten_US
dc.rights.holderAndrew Gordon Wilson; Ryan Prescott Adams
dc.relation.journalJournal of Machine Learning Researchen_US
dash.depositing.authorAdams, Ryan Prescott
dc.date.available2013-11-26T08:30:34Z
workflow.legacycommentsIn QSDB.en_US
dash.contributor.affiliatedAdams, Ryan Prescott


Files in this item

Thumbnail

This item appears in the following Collection(s)

Show simple item record