The 80s backprop claims seemed much too good to be true. I replicated key experiments around 1990, found they were nonsense. Rinse & repeat?
Nothing in “deep learning” is new. Impressive hype machine—like backprop in 80s. 80s results collapsed. This time?https://twitter.com/lukeprog/status/630784807182557185 …
-
-
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
@Meaningness The deep learning people themselves say the new results mostly due to more compute power and more data rather than new ideas. -
@mtraven Yes… Bugs me because wasn’t just scale that made backprop not work. I have guesses about what’s going on. Wonder if worth checking - 3 more 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.