Non-Negative Matrix Factorization for Binary Data

Factorize binary matrices into rank-k components using the logistic function in the updating process. See e.g. Tomé et al (2015) .

nmfbin: Non-Negative Matrix Factorization for Binary Data

The nmfbin R package provides a simple Non-Negative Matrix Factorization (NMF) implementation tailored for binary data matrices. It offers a choice of initialization methods, loss functions and updating algorithms.

NMF is typically used for reducing high-dimensional matrices into lower (k-) rank ones where k is chosen by the user. Given a non-negative matrix X of size $m \times n$, NMF looks for two non-negative matrices W ($m \times k$) and H ($k \times n$), such that:

$$X \approx W \times H$$

In topic modelling, W is interpreted as the document-topic matrix and H as the topic-feature matrix.

Unlike most other NMF packages, nmfbin is focused on binary (Boolean) data, while keeping the number of dependencies to a minimum. For more information see the website.

Installation

You can install the development version of nmfbin from GitHub with:

# install.packages("remotes")
remotes::install_github("michalovadek/nmfbin")

Usage

The input matrix can only contain 0s and 1s.

# load
library(nmfbin)

# Create a binary matrix for demonstration
X <- matrix(sample(c(0, 1), 100, replace = TRUE), ncol = 10)

# Perform Logistic NMF
results <- nmfbin(X, k = 3, optimizer = "mur", init = "nndsvd", max_iter = 1000)

Citation

@Manual{,
  title = {nmfbin: Non-Negative Matrix Factorization for Binary Data},
  author = {Michal Ovadek},
  year = {2023},
  note = {R package version 0.2.1},
  url = {https://michalovadek.github.io/nmfbin/},
}

Contributions

Contributions to the nmfbin package are more than welcome. Please submit pull requests or open an issue for discussion.