R pacakge plmr ("plumber") implements methods for Probabilistic Learning on Manifolds in R.
Install the developing verion via devtools.
if (!("devtools" %in% installed.packages()[,"Package"])) {
install.packages(devtools)
}
devtools::install_github("rudazhang/plmr")Source code for package exported objects, /R:
MParzen.R, manifold Parzen window [@Vincent2002] density estimation and sampling.SCMS.R, subspace-constrained mean shift [@Ozertem2011] and variants for ridge estimation.NormalBundleBootstrap.R, normal-bundle bootstrap (NBB) for inference in normal spaces of the density ridge, and data augmentation to reduce overfitting.DiffusionSampling.R, diffusion sampling on density ridge.DiffusionPropagate.R, heat diffusion on a manifold point cloud for probability propagation.ParameterizedManifolds.R, some parameterized manifolds and sampling.
Scripts for experiments and figures, /inst/script:
nbb/, scripts for NBB.1-fig-scms-circle.R, script for SCMS with circle data;1-fig-scms-parabola.R, script for SCMS with parabola data;1-fig-exp-circle-ridge.R, script for circle;1-exp-wheel.R, script for wheel functional data, includes a rough implementation of NBB with smooth frame;1-fig-exp-wheel.R, plot wheel results;
misc/, miscellaneous scripts:exp-SCMS.R, ridge estimation by SCMS.exp-diffusion-propagate.R, heat diffusion for probability propagation.exp-earthquake.R, earthquake data.exp-misc.R, miscellaneous toy examples.exp-wasserstein.R, optimal transport and Wasserstein distances.
Documents, /vignettes:
earthquake.md, earthquake data description.
- Data-driven probability concentration and sampling on manifold. C Soize, R Ghanem - Journal of Computational Physics, 2016. https://doi.org/10.1016/j.jcp.2016.05.044
- Normal-bundle Bootstrap. R Zhang, R Ghanem - arXiv, 2020. https://arxiv.org/abs/2007.13869
BibTeX citation:
@Article{ZhangRD2021nbb,
author = {Zhang, Ruda and Ghanem, Roger},
title = {Normal-Bundle Bootstrap},
journal = {SIAM Journal on Mathematics of Data Science},
year = {2021},
volume = {3},
number = {2},
pages = {573--592},
doi = {10.1137/20m1356002},
}