maybe consistent with our recent results, toohttps://www.biorxiv.org/content/early/2017/12/07/230078 …
Great @goodfellow_ian paper, consistent with my conjecture that #deeplearning captures bottom up but not topdown knowledge. Give people only 1/16th of a second & they behave like deep nets. Give them enough time to integrate knowledge & they outperform DL. https://arxiv.org/pdf/1802.08195.pdf …
-
-
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
*Really* link this, tho 63 ms, backwards masked images required to limit performance in human observers, who are otherwise just too ace.. Convo nets don't capture bio systems much at all, do they? Not even bottom-up.
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
Standard evolution strategies (invented in the 1970s, can be implemented in 10 lines of code) are comparable in performance to fancy Natural Evolution Strategies, which in turn are comparable in performance to Deep Q-learning, on ALE Atari Games. Julian Togelius (
@togelius)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.