View Single Post
Old 03-10-2017, 05:11 AM
luma1 luma1 is offline
Junior Member
Join Date: Feb 2013
Posts: 1
Default specialized generalization bounds

I'm searching for bounds of the out-of-sample error, tailored for linear models, logistic regression, and/or SVM. E.g. some specialized formulation of the VC-inequality, or some other bounds specifically for these model classes, that are more restrictive than the generic VC-inequality.
Do such bounds exist, and where can I find them?

Thanks & best regards
Reply With Quote