[EEG] Meet today! We have a cap! And I'm obsessed with ICA!

Jonathan Foote jtfoote at ieee.org
Thu Apr 16 16:30:46 PDT 2009


Cool!

ICA is bringing back grad school flashbacks (as well as flashbacks of
other kinds: the first time I met Tony Bell [ICA quasi-inventor] he
was wearing nothing but a coat of red paint. This was at Burning Man
naturally.)

To refresh people's memories, here's a good ICA demo:
http://www.cis.hut.fi/projects/ica/icademo/

I'm a little skeptical of some of the ICA claims, and it makes some
serious assumptions about signal forms  (non-Gaussian signals, linear
mixing, known number of components) but when it works (like toy
examples) its' super-cool. And you can get a lot of the ICA goodness
from PCA or similar approaches that don't have such rigid assumptions.

There's also a nice-looking Python library that's new to me:
http://mdp-toolkit.sourceforge.net/

See you tonight.

On Thu, Apr 16, 2009 at 3:46 PM, Kelly <hurtstotouchfire at gmail.com> wrote:
> So there's Open EEG on the calendar today from 6-8 for those of you that can
> make it.  We're on for our "regular" Sunday meeting as well, but since we've
> got a new toy it seems worth meeting today for everyone who was busy being
> religious this Sunday instead of talking about brains and hardware.  So come
> down and see the cap!
>
> Also, I had an exciting week -- met one of the field leaders in EEG
> research: http://httf.livejournal.com/79092.html and now I really really
> want to play with EEGLab and ICA.  Scott (The EEG Guy) told me that EEGlab
> comes with some sample data that will let you run visuals.  And this is
> particularly exciting because ICA is probably the best method for
> localization.  I had a brainstorming talk with Steen earlier this week about
> localizing sources for EEG signals and about what a mindfuck that is, so I'm
> hoping that ICA will make it easier.
>
> In general, I really want to get a leg up on EEG analysis, because I have
> only the vaguest idea how to do it, and even if we don't have operational
> hardware yet, I'd like to get started on learning.  Maybe I'll make a post
> to noisebridge discuss at some point soon to recruit some programmers.
> Seems like we lost a couple since the first meeting.
>
> Ok, I'm rambling.  So come tonight so I can ramble in meatspace.
>
> -Kelly
>
> PS. Tomorrow morning I'm going to be helping to run an EEG experiment!  I
> think some of it's ECOG actually, which is like... electrodes on the bare
> cortex.  I promise to report on that Sunday.  I'm planning to ask about
> noise and shielding, but we can discuss tonight if there's anything else in
> particular I should ask about.
>
> _______________________________________________
> EEG mailing list
> EEG at lists.noisebridge.net
> https://www.noisebridge.net/mailman/listinfo/eeg
>
>



More information about the EEG mailing list