[FieldTrip] classification with 3 conditions?

[Irina Simanova]

Dear Akiko,

you can use DML functions for multiclass classification: dml.one_against_one or dml.one_against_rest

one_against_rest builds one SVM per class, trained to distinguish the trials   in a single class from the trials  in all remaining classes. Classification of a test trial is done according to the maximum output among all SVMs.

one_against_one builds one SVM for each pair of classes (a pairwise comparison). Classification of a test trial is done according to the maximum voting , where each SVM votes for one class. 

You can see for yourself which method is better for your research question/data. 
You call it like this: 
cfg.mva =  dml.one_against_rest('mva', {dml.standardizer() dml.svm()}) 
or
cfg.mva =  dml.one_against_one('mva', {dml.standardizer() dml.svm()}) 


best,
Irina



 
On Apr 10, 2013, at 12:32 AM 4/10/13, Akiko Ikkai wrote:

> Hi,
> 
> I have a 3 conditions, A, B, and C in my MEG data. Multivariate classification using selected sensors between conditions A and B, B and C, and A and C work beautifully with chance = 50%. I'm now curious whether I could run a classification of 3 conditions with chance = 33%. 
> 
> ft_freqstatistics could take more than 2 frequency datasets as inputs, but ft_statistics_crossvalidate using +dmlt/svm.m seems to require 2 datasets (because it's using a linear kernel?). 
> 
> Does anyone have suggestions how I might go about it...?
> Thanks in advance! Akiko
> 
> -- 
> Akiko Ikkai, Ph.D. 
> Postdoctoral Fellow
> Department of Psychological and Brain Sciences
> Johns Hopkins University
> Ames Hall, 3400 N. Charles St.
> Baltimore, MD 21218
> 
> _______________________________________________
> fieldtrip mailing list
> fieldtrip at donders.ru.nl
> http://mailman.science.ru.nl/mailman/listinfo/fieldtrip

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.science.ru.nl/pipermail/fieldtrip/attachments/20130410/93654435/attachment-0001.html>