Performs model fitting and sequential design for deep Gaussian processes using MCMC and elliptical slice sampling. Models extend up to three layers deep; a one layer model is equivalent to typical Gaussian process regression. Sequential design criteria include integrated mean square prediction error (IMSPE), active learning Cohn (ALC), and expected improvement (EI). Covariance structure is based on inverse exponentiated squared euclidean distance. Applicable to noisy and deterministic functions. Incorporates SNOW parallelization and utilizes C under the hood. Manuscript forthcoming; see Damianou and Lawrence (2013) <arXiv:1211.0358> for deep Gaussian process models and Murray, Adams, and MacKay (2010) <arXiv:1001.0175> for elliptical slice sampling.
| Version: | 0.1.0 |
| Depends: | R (≥ 3.6) |
| Imports: | grDevices, graphics, stats, doParallel, foreach, parallel |
| Suggests: | akima, knitr |
| Published: | 2020-10-29 |
| Author: | Annie Sauer |
| Maintainer: | Annie Sauer <anniees at vt.edu> |
| License: | LGPL-2 | LGPL-2.1 | LGPL-3 [expanded from: LGPL] |
| NeedsCompilation: | yes |
| Materials: | README |
| CRAN checks: | deepgp results |
| Reference manual: | deepgp.pdf |
| Package source: | deepgp_0.1.0.tar.gz |
| Windows binaries: | r-devel: deepgp_0.1.0.zip, r-release: deepgp_0.1.0.zip, r-oldrel: deepgp_0.1.0.zip |
| macOS binaries: | r-release: deepgp_0.1.0.tgz, r-oldrel: deepgp_0.1.0.tgz |
Please use the canonical form https://CRAN.R-project.org/package=deepgp to link to this page.