Hey all,
I was hoping to follow up on this. If anything is unclear or there is a better place to post this please let me know. It seems unlikely that the ICA/SVM is actually consistently less accurate than chance trained on random data. That would mean I could guess better than random by choosing the opposite of the SVM. I could try running permutation tests on ICA and SVM implementations from other libraries but don't know where to start with that.
I've seen papers that use support vector classification on ICA decomposed data. If this is consistent (across implementation) I wonder if those accuracies are under-reported.
Best,
Jeff
########################################################################
To unsubscribe from the FSL list, click the following link:
https://www.jiscmail.ac.uk/cgi-bin/webadmin?SUBED1=FSL&A=1
|