my_Amari: The Amari Error
In hanspeter6/icaPlay: What the Package Does (One Line, Title Case)

Description Usage Arguments Details Value Author(s) References Examples

View source: R/my_Amari.R

Measures a distance between a mixing matrix inverse of A and an estimate of an unmixing matrix W.

1	my_Amari(A, W)

`A`	The inverse of the mixing matrix
`W`	The estimated unmixing matrix

This is an implementation of the algorithm found in Sorensen, J. 2008. An Assessment of Hermite Function-based Approximations of Mutual Information Applied to Independent Component Analysis. Entropy. Vol. 10. pp745-756

Sorensen, 2008 cites Amari et al. 1996 for the expression.

The Amari Error/Distance

Hans-Peter Bakker

See details

set.seed(1)
A <- matrix(runif(4), nrow = 2)
set.seed(2)
W <- matrix(runif(4), nrow = 2)

my_Amari(solve(A), W)
[1] 0.6755254


##---- Should be DIRECTLY executable !! ----
##-- ==>  Define data, use random,
##--	or do  help(data=index)  for the standard data sets.

## The function is currently defined as
function (A, W) 
{
    AInv <- A
    WInv <- solve(W)
    m <- nrow(A)
    AInvWInv <- AInv %*% WInv
    terms1 <- rowSums(abs(AInvWInv))/apply(abs(AInvWInv), 1, 
        max) - 1
    terms2 <- colSums(abs(AInvWInv))/apply(abs(AInvWInv), 2, 
        max) - 1
    return(1/(2 * m) * (sum(terms1) + sum(terms2)))
  }