3 ms·
I think the GP poster meant to contrast this paper not with other papers, but with the junk explanations for "how deep learning works" that science journalism c
by derefr 9y ago
I think the GP poster meant to contrast this paper not with other papers, but with the junk explanations for "how deep learning works" that science journalism comes up with.
- mathinpens 9y agotrust me some of the junk explanations for how deep learning works originate from researchers. we really just don't have a convincing picture about why these networks work or don't work in a theoretical way rather than empirical/heuristic handwaves
- crb002 9y agoBollocks see my comment. We know worst case bounds for Kolmogorov random transformations of a finite set, the most complex learning task there is.
- mathinpens 9y agohuh? not even wrong......