I think there is nothing controversial in this post. But it’s unclear what exactly counts as “Deep Learning”, which makes the discussion about limits almost useless. Many recent advances are feature (=architecture) engineering and not fundamental advances, or “Neural Net + X”https://twitter.com/GaryMarcus/status/1068897657530138629 …
-
-
It stood out to me because you mentioned AlphaGo, but I don’t even see that as an advancement in pure Deep Learning. It just happened to use a CNN, bust most likely there exit many other function approximators (which we may not have discovered yet) that would work just as well.
-
indeed you could swap that part sooner than you could replace the monte carlo tree search. i was using as example of different point, viz hype & results being presented as having larger impact than has really been demonstrated
- 1 more reply
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.