"How do we integrate symbol manipulation with learning.." is exactly the question I'm asking too. We agree on that. But I think deep (or gradient-based) learning is part of the answer. One of your tweets seems to agree. But your above jab at deep nets seem to suggests otherwise
not committing to the precise answer, which remains unknown, but yes these are exactly the right sorts of questions: how do we integrate symbol-manipulation with learning from large-scale data?https://twitter.com/GaneshNatesh/status/1066050801871675392 …
-
-
-
@ylecun I’ve been urging synthesis between symbols & associationism for 26 years! starting w first publication (1992, SRCD), continuing w 1st book (2001, subtitle “integrating connectionism w cognitive science”) thru 2018 arXiv, major conclusion = add symbols to deep learning. - 1 more reply
New conversation -
-
-
Fully agree with this trajectory. The human mind has evolved the ability to carry out symbolic manipulation on top of the infrastructure of an exquisitely complex neural network. And, IMHO that makes all the difference in the world.
- 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.