Skip to contents

Run ICA over a given no_comp with random initialisations of w_init

rs_ica_iters.Rd

[Experimental] This function implements a stabilised ICA like algorithm in Rust. Briefly, it generates random w_init matrices (total number being no_random_init) and runs ICA given the x_processed and k data over these. The function returns combined S from the different runs and a boolean vector indicating if this specific run converged.

Usage

rs_ica_iters(x1, k, no_comp, no_random_init, ica_type, random_seed, ica_params)

Arguments

x1

Numerical matrix. The processed matrix (but not yet whitened!)

k

Numerical matrix. The whitening matrix. Needs at least no_comp rows, only the first no_comp are used.

no_comp

Integer. Number of independent components to return.

no_random_init

Integer. Number of random initialisations to test.

ica_type

String. One of c("logcosh", "exp").

random_seed

Integer. Seed for randomisations. Run i uses random_seed + i.

ica_params

A list containing:

  • maxit - Integer. Maximum number of iterations for ICA.

  • alpha - Float. The alpha parameter for the logcosh version of ICA. Should be between 1 to 2.

  • max_tol - Maximum tolerance of the algorithm

  • verbose - Verbosity of the function, i.e., shall individual iters be shown.

If the list is empty or the expected elements are not found, default values are used.

Value

A list containing:

  • s_combined - The combined matrices for S. Dimensions are nrows = features; and ncols = ncomp * no_random_init.

  • converged - Boolean vector indicating if the respective run reached convergence. Length = no_random_init