Logistic Non-negative Matrix Factorization

This function performs Logistic Non-negative Matrix Factorization (NMF) on a binary matrix.

Usage

nmfbin(
  X,
  k,
  optimizer = "mur",
  init = "nndsvd",
  max_iter = 1000,
  tol = 1e-06,
  learning_rate = 0.001,
  verbose = FALSE,
  loss_fun = "logloss",
  loss_normalize = TRUE,
  epsilon = 1e-10
)

Arguments

X

A binary matrix (m x n) to be factorized.

k

The number of factors (components, topics).

optimizer

Type of updating algorithm. mur for NMF multiplicative update rules, gradient for gradient descent, sgd for stochastic gradient descent.

init

Method for initializing the factorization. By default Nonnegative Double Singular Value Decomposition with average densification.

max_iter

Maximum number of iterations for optimization.

tol

Convergence tolerance. The optimization stops when the change in loss is less than this value.

learning_rate

Learning rate (step size) for the gradient descent optimization.

verbose

Print convergence if TRUE.

loss_fun

Choice of loss function: logloss (negative log-likelihood, also known as binary cross-entropy) or mse (mean squared error).

loss_normalize

Normalize loss by matrix dimensions if TRUE.

epsilon

Constant to avoid log(0).

Value

A list containing:

• W: The basis matrix (m x k). The document-topic matrix in topic modelling.

• H: The coefficient matrix (k x n). Contribution of features to factors (topics).

• c: The global threshold. A constant.

• convergence: Divergence (loss) from X at every iter until tol or max_iter is reached.

Examples

if (FALSE) {
# Generate a binary matrix
m <- 100
n <- 50
X <- matrix(sample(c(0, 1), m * n, replace = TRUE), m, n)

# Set the number of factors
k <- 4

# Factorize the matrix with default settings
result <- nmfbin(X, k)
}