EM_FCtemplateICA {templateICAr}R Documentation

EM Template ICA

Description

EM Algorithm for FC Template ICA Model

Usage

EM_FCtemplateICA(
  template_mean,
  template_var,
  template_FC,
  prior_params = c(0.001, 0.001),
  BOLD,
  AS_0,
  maxiter = 100,
  miniter = 3,
  epsilon = 0.001,
  Gibbs_nsamp = 1000,
  Gibbs_nburn = 50,
  Gibbs_nchain = 10,
  eps_inter = NULL,
  verbose
)

Arguments

template_mean

(V \times Q matrix) mean maps for each IC in template, where Q is the number of ICs, V=nvox is the number of data locations.

template_var

(V \times Q matrix) between-subject variance maps for each IC in template

template_FC

(list) Parameters of functional connectivity template

prior_params

Alpha and beta parameters of IG prior on tau^2 (error variance)

BOLD

(V \times T matrix) preprocessed fMRI data

AS_0

(list) initial guess at latent variables: A (TxQ mixing matrix), and S (QxV matrix of spatial ICs)

maxiter

Maximum number of EM iterations. Default: 100.

miniter

Minimum number of EM iterations. Default: 3.

epsilon

Smallest proportion change in log-posterior between iterations. Default: 0.001.

Gibbs_nsamp

the number of Gibbs posterior samples of A and S to output after burn-in

Gibbs_nburn

the number of Gibbs posterior samples of A and S to throw away before saving

Gibbs_nchain

the number of simultaneous Gibbs chains to run

eps_inter

Intermediate values of epsilon at which to save results (used to assess benefit of more stringent convergence rules). Default: NULL (do not save). These values should be in decreasing order (larger to smaller error) and all values should be between zero and epsilon.

verbose

If TRUE, display progress of algorithm. Default: FALSE.

Details

EM_FCtemplateICA implements the expectation-maximization (EM) algorithm for the functional connectivity (FC) template ICA model

Value

A list of computed values, including the final parameter estimates.


[Package templateICAr version 0.8.8 Index]