LBBNN: Latent Binary Bayesian Neural Networks Using 'torch'

Latent binary Bayesian neural networks (LBBNNs) are implemented using 'torch', an R interface to the LibTorch backend. Supports mean-field variational inference as well as flexible variational posteriors using normalizing flows. The standard LBBNN implementation follows Hubin and Storvik (2024) <doi:10.3390/math12060788>, using the local reparametrization trick as in Skaaret-Lund et al. (2024) <https://openreview.net/pdf?id=d6kqUKzG3V>. Input-skip connections are also supported, as described in Høyheim et al. (2025) <doi:10.48550/arXiv.2503.10496>.

Version: 0.1.1
Depends: R (≥ 3.5)
Imports: ggplot2, torch, igraph, coro, svglite
Suggests: testthat (≥ 3.0.0), knitr, rmarkdown
Published: 2025-12-01
DOI: 10.32614/CRAN.package.LBBNN (may not be active yet)
Author: Lars Skaaret-Lund [aut, cre], Aliaksandr Hubin [aut], Eirik Høyheim [aut]
Maintainer: Lars Skaaret-Lund <lars.skaaret-lund at nmbu.no>
License: MIT + file LICENSE
NeedsCompilation: no
Language: en-US
Materials: README, NEWS
CRAN checks: LBBNN results

Documentation:

Reference manual: LBBNN.html , LBBNN.pdf
Vignettes: Getting started with LBBNN (source, R code)

Downloads:

Package source: LBBNN_0.1.1.tar.gz
Windows binaries: r-devel: not available, r-release: not available, r-oldrel: LBBNN_0.1.1.zip
macOS binaries: r-release (arm64): LBBNN_0.1.1.tgz, r-oldrel (arm64): LBBNN_0.1.1.tgz, r-release (x86_64): LBBNN_0.1.1.tgz, r-oldrel (x86_64): LBBNN_0.1.1.tgz

Linking:

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