[Eeglablist] How to deal with ICA Weights Changes for the Same EEG Dataset

[Makoto Miyakoshi]

Dear Shouyi,

> I wonder how can we deal with the ICA weights changes?

This has not been regarded as a problem. Why don't you disable
randomization to see what happens? At least it's not a requirement for ICA
if I understand it correctly. You get reproducibility in ICA results, at
that cost of ?? I haven't tried it myself so I don't know.

Makoto


On Fri, Aug 8, 2014 at 1:19 AM, shouyi wang <shouyisxty at gmail.com> wrote:

> Hi All,
> I just started to use EEGLAB to performa ICA study. However, I notice
> that the ICA weights changed (many weights changed slightly and some
> weights changed significantly) if I run 'Run ICA' several times for
> the same EEG dataset. As a results, when I do artifacts removal with
> ADJUST plugin. The rejected components can also change due to the ICA
> weights changes.
>
> I checked the 'runica' codes, there are several places using random
> generating numbers. I think that may be the major reason for the ICA
> weights changes run each time. I wonder how can we deal with the ICA
> weights changes? or I did something wrong here? Hope to get some
> suggestions here? Thanks a lot!
>
> regards,
> Shouyi
> _______________________________________________
> 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/20140811/4c50e5a9/attachment.html>