> I think there needs to be a rethinking of the Grand Theory of Deep Learning Architecture TM. strong agree. Although perhaps not so much a rethinking as a the

> I think there needs to be a rethinking of the Grand Theory of Deep Learning Architecture TM. strong agree. Although perhaps not so much a rethinking as a theory of all. Huge dearth of theory in the field. Daily practition involves regular use of black magic intuition for arch, problem posing and debugging. Weird times.