Appallingly little willingness from some people to consider the merits instead of defending their identities.
@carlesgelada, history seems likely to vindicate you!https://twitter.com/carlesgelada/status/1248347372259700736 …
-
-
Replying to @sama @carlesgelada
I suspect that the opposite of what Gelada says may be true. Neural networks have locked us into a local optimum of function approximation algorithms. They work relatively well on linear algebra tasks, but some other methods we discarded may turn out to be more general.
0 replies 1 retweet 9 likes
What is it that you want to approximate? Every computable function! While neural networks are Turing complete (they can express every computer program), not every computer program can be efficiently learned by the current NN learning paradigms.
7:28 PM - 9 Apr 2020
0 replies
0 retweets
1 like
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.