[Eeglablist] ICA does not start running properly

[Philip Michael Zeman]

Maarten,

I did much the same thing as you described.  When I removed the 'noise'
component on the first iteration of ICA, I reconstructed the data but
reprojecting the remaining components back to the scalp.

The second iteration did exactly as you described when I attempt to run ICA
on the full dimensionality of the data.  The runica algorithm learning rate
is adjusted a number of times and converges *very slowly*.  It does seem to
converge in most cases (or aborts).  I suspect this is because we reduced
the dimensionality of the data when removed the 'noise' component in the
first iteration of ICA.  You should find that if you reduce the
dimensionality of the data before running ICA (by using the PCA parameter),
the algo will converge more quickly.   For example, if you have 124
electrodes:

(1) Initial iteration of ICA:   runica (use full dimensions, no PCA
parameter required)

(2) Remove unwanted 'noise' components (say-- 3 components)

(3) Projected the remaining components to the scalp and recombine

(4) Run ICA on the reconstructed data:  use the PCA parameter for runica.
Reduce the dimensionality by a minimum of 3.  ie,   runica 'PCA',120 should
work for you.


Cheers!

~Phil Zeman




----- Original Message ----- 
From: "Maarten Mennes" <Maarten.Mennes at med.kuleuven.ac.be>
To: <eeglablist at sccn.ucsd.edu>
Sent: Tuesday, April 19, 2005 7:25 AM
Subject: [Eeglablist] ICA does not start running properly


> hi,
>
> after successfully running ICA on epoched data to remove eyemovement
> artifacts, I want to run ICA again on the filtered epochs to remove
> other artifact components. I do not want to work with the remaining
> components of the first ICA as these epochs can differ a lot from the
> task related epochs I want to use.
>
> However, for some unclear reason the ICA procedure starts stalling at
> the "Beginning ICA training..." print. Sometimes it gets to lowering its
> learning rate and starting again, but I tend to abort the thing after
> half an hour.
>
> Running the ICA for removing the eyemovements was no problem at all. It
> never took more than one ore two minutes. Even when I used it on more
> epochs than what I'm trying to now.
>
> What could be the problem?
>
> thanks,
> maarten
>
>
> _______________________________________________
> eeglablist mailing list eeglablist at sccn.ucsd.edu
> http://sccn.ucsd.edu/mailman/listinfo/eeglablist
> Eeglablist page: http://sccn.ucsd.edu/eeglab/eeglabmail.html
> To unsubscribe, send an empty email to
eeglablist-unsubscribe at sccn.ucsd.edu