[Eeglablist] AMICA runtime

Makoto Miyakoshi mmiyakoshi at ucsd.edu
Mon Dec 14 17:59:19 PST 2015

Dear Mina,

If you have 30 channel data, you want to have minimum of (30^2)*30 = 27000
data points for ICA. The *30 part should be increased to 40, 50,... as you
increase the number of channels. Depending on which paper you read, they
say the number should be 20 or 30. This is just a rule of thumb. Jason says
if you have 100 channels, 1,000,000 data points would be good.

> The algorithm is completed in just a few seconds.

AMICA must have failed. Check out.txt.


On Mon, Dec 14, 2015 at 4:02 AM, Mina Marmpena <mina.marmpena at gmail.com>

> Hi,
> I use AMICA to clear the artifacts from my EEG data.
> I have noticed that when applying it on the continuous signal which is 45
> min long I get very good detection of EOG artifacts, but it takes long run
> time, around 4 hours.
> I tried to do the same on epoched data (total length around 7 min ).
> The algorithm is completed in just a few seconds. I was surprised to see
> that and I 'm a little worried maybe something is wrong. Is it something to
> be expected with this length of data?
> Thank you,
> Mina
> _______________________________________________
> Eeglablist page: http://sccn.ucsd.edu/eeglab/eeglabmail.html
> To unsubscribe, send an empty email to
> eeglablist-unsubscribe at sccn.ucsd.edu
> For digest mode, send an email with the subject "set digest mime" to
> eeglablist-request at sccn.ucsd.edu

Makoto Miyakoshi
Swartz Center for Computational Neuroscience
Institute for Neural Computation, University of California San Diego
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://sccn.ucsd.edu/pipermail/eeglablist/attachments/20151214/44e90d22/attachment.html>

More information about the eeglablist mailing list