This is pretty fascinating! Though the concept seems to work only for convex problems (in particular, problems which have strong duality; this excludes NNs in a

This is pretty fascinating! Though the concept seems to work only for convex problems (in particular, problems which have strong duality; this excludes NNs in almost their entirety, except 1 layer nets), but the application is nice and straightforward. I wonder if there is a similar lower-bound which can be constructed for non convex problems which retain enough properties for this method to be useful?