[Mne_analysis] CSP for multi classification

JR KING jeanremi.king at gmail.com
Thu Jan 10 03:03:15 EST 2019
Search archives:

        External Email - Use Caution        

Hi,

The object mne.decoding.CSP works with more than 2 classes, but will be
increasingly slow as the number of class increases.

Since CSP is a spatial filtering technique, the more classes you have, the
less you'll be filtering.

You should thus avoid spatial filtering for multi-class problems,

You can check
https://pyriemann.readthedocs.io/en/latest/generated/pyriemann.classification.TSclassifier.html#pyriemann.classification.TSclassifier

JR

On Thu, 10 Jan 2019 at 02:52, A S <eng.emetsasa at gmail.com> wrote:

>         External Email - Use Caution
>
> Hi all,
> Does Common Spatial Patterns algorithm (from mne.decoding import CSP)
> works well for multi classification? Or is it much more better by
> binary classificaiton?
>
> Thanks in Advance
>
> _______________________________________________
> Mne_analysis mailing list
> Mne_analysis at nmr.mgh.harvard.edu
> https://mail.nmr.mgh.harvard.edu/mailman/listinfo/mne_analysis
>

>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mail.nmr.mgh.harvard.edu/pipermail/mne_analysis/attachments/20190110/228180dd/attachment.html 


More information about the Mne_analysis mailing list