advocates of #machinelearning, I am told that you all know that (current) #ML is limited. fair enough. but which limits are you willing to *publicly* acknowledge?https://twitter.com/NotSimplicio/status/1173373706674085888 …
-
The limitations you implicitly mention are limitations of supervised learning and reinforcement learning (deep or not), not of deep learning per se.
@GaryMarcus keeps making that confusion.
hopeful thinking @ylecun, at least for now. you are correct that these are limits of deep learning as used in rl and supervised contexts, but thus far there is to my knowledge no compelling evidence that other applications of deep learning evade the limits mentioned above.