I'm using sklearn's SVC with a linear kernel to train and predict brain states from functional MRI data. Upon completion, I want to extract the feature weights to identify which of these contain the most discriminative informaiton. The dimentionality of these images, however, is large (~250,000 features pr. observation) by nature, and thus I'd prefer to pre-compute the kernel prior to my CV procedure to improve run-time.
So my question is: Can I somehow backtrack to the explicit feature space after training my classifier in the implicit (kernel) feature space?
I know the SVC.coef_ attribute gives you these weightings if you pass the SVC a kernel='linear' argument and data in explicit feature space, but is it possible when I have pre-computed the kernel myself?
