Unsure what to think about Stacked Regression Machines http://arxiv.org/abs/1608.04062 Did anyone repro, try on more datasets?
-
Layer-local greedy training doing that well is pretty surprising, yes.
the reason DL works so well is the joint training of the stacked reprs. Greedy approaches fundamentally flawed
-
-
Maybe. It doesn't seem logical that joint training should ever lose to greedy training, but if it does?
-
<5% chance that this paper upends backprop. The one that does will have a theoretical build-up that this paper lacks
- Show replies
New conversation -
Loading seems to be taking a while.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.