1 2 3 | nmf_mmle_mcem(V, initW, maxiters = 100, nsamples = 50, algo = "gibbs",
mincrease = "constant", alpha = 1, beta = 1, mstep = "C",
mult_updates = 1, saem = FALSE, L = 10, epsilon = 0.5, burnin = 0.5)
|
V |
matrix to be factorized |
maxiters |
maximum number of iterations |
nsamples |
number of samples per iteration in the E-step |
K |
number of latent factors (or dimensions) |
W |
initial matrix W |
mc |
mcem or saem. |
The number of iterations is set to maxiters. In the future, there should be a convergence check in case the KL converges before maxiters.
If saem is used, then the number of samples gradually increases until reaching nsamples for the last iterations
W last estimator of W traces$H: a K x N x samples tensor traces$W: a K x samples - matrix vector with the evolution of row norms
Add the following code to your website.
For more information on customizing the embed code, read Embedding Snippets.