JADE (version 1.1-0)

ComonGAP: Comon's Gap

Description

Comon's GAP criterion to evaluate the performance of an ICA algorithm.

Usage

ComonGAP(A, A.hat)

Arguments

A

The true square mixing matrix.

A.hat

The estimated square mixing matrix.

Value

• The value of the Comon's GAP. } references{ cite{Comon, P., (1994), Independent Component Analysis, A new concept?, emph{Signal Processing}, bold{36}, 287--314. } } author{Klaus Nordhausen, email{[e-mail]}} seealso{code{amari.error}, code{SIR}} examples{ S <- cbind(rt(1000, 4), rnorm(1000), runif(1000)) A <- matrix(rnorm(9), ncol = 3) X <- S %*% t(A) A.hat <- JADE(X, 3)$A ComonGAP(A, A.hat) } % Add one or more standard keywords, see file 'KEYWORDS' in the % R documentation directory. keyword{ multivariate }

Details

Comon's GAP criterion is permutation and scale invariant. It can take every positive value and 0 corresponds to an optimal separation. If A is however nearly singular the values of the criterion can be huge. Note that this function assumes the ICA model is $X = S A'$, as is assumed by JADE and ics. However fastICA and PearsonICA assume $X = S A$. Therefore matrices from those functions have to be transposed first.