[pymvpa] How to do hyperalignment with the unit of trial

Yaroslav Halchenko debian at onerussian.com
Wed Nov 18 13:36:03 UTC 2015

On Wed, 18 Nov 2015, 孔令军 wrote:
> Sorry for not describe that completionly .
> There are 27 subjects , every subject do two runs , each run consisted
> of 8 trials  , 
> each trial consisted of 12 stimulate events .
> When do the hyperalignment , the unit is run , but my data only have two runs ,
> so the training data is lack relativity , the common space is only half of it ,
> of course the accuracy rate is low .
> I wanna know if and how can we do the hyperalighnment with the unit of trial 
> to get more training data

I guess I would need more detail to actually understand what is going on
;)  but the point is -- for hyperalignment ideally you use all of your
fMRI volumes (not just trials, not just events you extract) and have
aligned in time across subjects events.  If you have your dataset ready
may be you could just provide output of

print your_dataset.summary()

But also given that you have only 2 runs, it is unlikely that
volumes would be corresponding across subjects in time if you randomized
your sequences of stimuli... if you didn't randomize -- you have another
problem ;)

Yaroslav O. Halchenko
Center for Open Neuroscience     http://centerforopenneuroscience.org
Dartmouth College, 419 Moore Hall, Hinman Box 6207, Hanover, NH 03755
Phone: +1 (603) 646-9834                       Fax: +1 (603) 646-1419
WWW:   http://www.linkedin.com/in/yarik        

More information about the Pkg-ExpPsy-PyMVPA mailing list