ebpmf_wbg3: Empirical Bayes Poisson Matrix Factorization (Background...

In stephenslab/ebpmf:

Description

Empirical Bayes Poisson Matrix Factorization (Background Model with weights)

Usage

ebpmf_wbg3(
  X,
  K,
  pm_func = list(f = ebpm::ebpm_gamma_mixture, l = ebpm::ebpm_gamma_mixture, f0 =
    ebpm::ebpm_gamma, l0 = ebpm::ebpm_gamma),
  init = NULL,
  pm_control = NULL,
  fix_option = list(l0 = FALSE, f0 = FALSE, gl = FALSE, ql = FALSE, gf = FALSE, qf =
    FALSE),
  maxiter = 100,
  tol = 1e-08,
  verbose = FALSE,
  seed = 123
)

Arguments

X	count matrix (dim(X) = c(n, p)).
pm_func	functions for solving the ebpm subproblem for L and F; It is a list list(l, f); For our purpose we use 'mle_pm' or 'ebpm_point_gammma'for L, and ebpm_gamma_mixture for F
init	Either NULL or list(qg, l0, f0, w)
pm_control	control parameters for pm_func function
maxiter	maximum number of iterations
tol	stopping tolerance for ELBO
seed	used when init is NULL
k	number of topics
fix_g	list(l, f) where l, f are either TRUE or FALSE

Value

A list containing elements:

l0

sample-wise mean

f0

feature-wise mean

qg

list(ql, gl,qf, gf)

ELBO

ELBO objective for this VEB algorithm

Examples

To add

stephenslab/ebpmf documentation built on Nov. 20, 2021, 10:55 a.m.