The "deep learning" revolution consists in just one thing --the ability to learn multi-layered representations jointly rather than greedily.
The best 1st layer in a 2-layer model isn't the same as in 20-layer model. Greedy representation learning is fundamentally broken.
-
-
And deep learning changed everything because it offered a computationally practical way to learn all layers at the same time.
-
At last --if you believe that SGD via backprop of a loss function is the only way to achieve this, or the best way, you are sorely mistaken.
- Show replies
New conversation -
-
-
@fchollet do you also think that layer wise pre- training is not useful?Thanks. Twitter will use this to make your timeline better. UndoUndo
-
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.