[Eeglablist] BINICA in Windows

[Scott Makeig]

But recall that using RUNICA/BINICA  in Extended Infomax mode (the default)
is advantageous, as simple Infomax should fail to locate independent
components whose probability distribution function (pdf) is subGaussian
(e.g., pure line noise).  For most accuracy, we showed in 2012 (Delorme et
al., 2012) that AMICA is the most powerful ICA approach because it fits its
component pdf model to each component as it learns.  Recently its author,
Jason Palmer, has made available an AMICA version on the UCSD supercomputer
that can be used for free via the Neuroscience Gateway (nsgportal.org) and
is many times faster through proper parallel computing configuration. This
is particularly important when using the 2nd major advantage of AMICA, its
ability to find multiple ICA models operating on different subsets of the
input data -- a sensitive measure of non-stationarity in the ICA source
solution (see SH Hsu et al., 2018, 2022).

Does CUDAICA implement Extended Infomax?  I do not recall ...

Scott

On Fri, Jan 17, 2025 at 2:06 PM Makoto Miyakoshi via eeglablist <
eeglablist at sccn.ucsd.edu> wrote:

> Hi Andrew,
>
> If you really need speed, try CUDAICA. It can be 15-25 times faster,
> depending on your environment.
>
> https://sccn.ucsd.edu/wiki/Makoto's_useful_EEGLAB_code#By_using_CUDAICA_.2805.2F09.2F2022_updated:_Thank_you_Ugo.2C_Yunhui.2C_John.2C_and_Ernie.21.29
>
> Makoto
>
> On Thu, Jan 16, 2025 at 1:24 PM Andrew Corcoran via eeglablist <
> eeglablist at sccn.ucsd.edu> wrote:
>
> > Hi list
> >
> > I'm rerunning some old preprocessing scripts and having some trouble
> > getting the binica implementation of runica to work with my version of
> > EEGlab (2024.2.1). The script runs fine with an archived version of
> EEGlab
> > from 2021 so I assume I have not set up the binary correctly.
> >
> > I have tried copying over the binica.exe file from my archived version
> and
> > updated icadefs.m to ensure the ICABINARY path is correct. Attempting to
> > run binica via pop_runica produces the following error:
> >
> > Output argument "wts" (and possibly others) not assigned a value in the
> > execution with "binica" function.
> > Error in pop_runica (line 532)
> >             [EEG.icaweights,EEG.icasphere] = binica( tmpdata, 'lrate',
> > 0.001, g.options{:} );
> >
> > I would like to continue using binica given its speed advantages but am
> > concerned that reproducibility of my analyses (by my future self and
> > possibly others) will be hampered if binica is difficult to set up...
> > Indeed,  I struggled to find a convenient download (for Windows) -- the
> > github code seems to require compilation, which I don't recall doing
> > previously. Is there a more straightforward alternative ?
> >
> > Thanks
> > Andrew
> >
> >
> > --
> > Dr. Andrew W. Corcoran
> > Post-doctoral Research Fellow
> > Monash Centre for Consciousness & Contemplative Studies
> > <
> >
> https://urldefense.com/v3/__http://www.monash.edu/consciousness-contemplative-studies__;!!Mih3wA!BKRCpLqNy1TXMHXE5WoaBYdNGJDn4TNyNuWKRMmybAZkKwVupR5uFEfh4l47c_3-RGLdiMeKTF0s9RLbs3O638jvtg3z3ui7kV8$
> > >
> >
> > Rm 422, 29 Ancora Imparo Way
> > Clayton, VIC 3800, Australia
> > +61 3 9905 4266
> >
> > Website
> > <
> >
> https://urldefense.com/v3/__https://www.monash.edu/arts/cognition-and-philosophy-lab/people/andrew-corcoran__;!!Mih3wA!BKRCpLqNy1TXMHXE5WoaBYdNGJDn4TNyNuWKRMmybAZkKwVupR5uFEfh4l47c_3-RGLdiMeKTF0s9RLbs3O638jvtg3zjTfBFD0$
> > >
> > | Google Scholar
> > <
> >
> https://urldefense.com/v3/__https://scholar.google.com/citations?user=yxKSfdsAAAAJ&hl=en__;!!Mih3wA!BKRCpLqNy1TXMHXE5WoaBYdNGJDn4TNyNuWKRMmybAZkKwVupR5uFEfh4l47c_3-RGLdiMeKTF0s9RLbs3O638jvtg3zvt-Dcdc$
> > > |
> > ResearchGate <
> >
> https://urldefense.com/v3/__https://www.researchgate.net/profile/Andrew-Corcoran-5__;!!Mih3wA!BKRCpLqNy1TXMHXE5WoaBYdNGJDn4TNyNuWKRMmybAZkKwVupR5uFEfh4l47c_3-RGLdiMeKTF0s9RLbs3O638jvtg3z1eQ4DBY$
> > > |
> > ORCID <
> >
> https://urldefense.com/v3/__https://orcid.org/0000-0002-0449-4883__;!!Mih3wA!BKRCpLqNy1TXMHXE5WoaBYdNGJDn4TNyNuWKRMmybAZkKwVupR5uFEfh4l47c_3-RGLdiMeKTF0s9RLbs3O638jvtg3zF4h0rQw$
> > > | Twitter
> > <
> >
> https://urldefense.com/v3/__https://twitter.com/mr_corcorana__;!!Mih3wA!BKRCpLqNy1TXMHXE5WoaBYdNGJDn4TNyNuWKRMmybAZkKwVupR5uFEfh4l47c_3-RGLdiMeKTF0s9RLbs3O638jvtg3zbVXNVaM$
> > >
> > _______________________________________________
> > 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
> >
> _______________________________________________
> 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



-- 
Scott Makeig, Research Scientist and Director, Swartz Center for
Computational Neuroscience, Institute for Neural Computation, University of
California San Diego, La Jolla CA 92093-0559, http://sccn.ucsd.edu/~scott