rnnmf: Regularized Non-Negative Matrix Factorization

A proof of concept implementation of regularized non-negative matrix factorization optimization. A non-negative matrix factorization factors non-negative matrix Y approximately as L R, for non-negative matrices L and R of reduced rank. This package supports such factorizations with weighted objective and regularization penalties. Allowable regularization penalties include L1 and L2 penalties on L and R, as well as non-orthogonality penalties. This package provides multiplicative update algorithms, which are a modification of the algorithm of Lee and Seung (2001) <http://papers.nips.cc/paper/1861-algorithms-for-non-negative-matrix-factorization.pdf>, as well as an additive update derived from that multiplicative update. See also Pav (2004) <doi:10.48550/arXiv.2410.22698>.

Version:	0.3.0
Depends:	R (≥ 3.0.2)
Imports:	Matrix
Suggests:	testthat, dplyr, ggplot2, scales, viridis, knitr
Published:	2024-11-04
DOI:	10.32614/CRAN.package.rnnmf
Author:	Steven E. Pav [aut, cre]
Maintainer:	Steven E. Pav <shabbychef at gmail.com>
BugReports:	https://github.com/shabbychef/rnnmf/issues
License:	LGPL-3
URL:	https://github.com/shabbychef/rnnmf
NeedsCompilation:	no
Citation:	rnnmf citation info
Materials:	README ChangeLog
CRAN checks:	rnnmf results

Documentation:

Reference manual:	rnnmf.pdf
Vignettes:	An Iterative Algorithm for Regularized Non-negative Matrix Factorizations (source, R code)

Downloads:

Package source:	rnnmf_0.3.0.tar.gz
Windows binaries:	r-devel: rnnmf_0.3.0.zip, r-release: rnnmf_0.3.0.zip, r-oldrel: rnnmf_0.3.0.zip
macOS binaries:	r-release (arm64): rnnmf_0.3.0.tgz, r-oldrel (arm64): rnnmf_0.3.0.tgz, r-release (x86_64): rnnmf_0.3.0.tgz, r-oldrel (x86_64): rnnmf_0.3.0.tgz

Linking:

Please use the canonical form https://CRAN.R-project.org/package=rnnmf to link to this page.