Of course, we still have long ways to go before we get anywhere close to our animal cognitive architecture. Present NN are far from it. But Fodor's main claim that NN "have no combinatorial structure in mental representations" seems to be not entirely valid even for modern NN.
-
-
-
Here is for example a recent paper that shows one way to deal with combinatorial structure: "Artificial Neural Networks that Learn to Satisfy Logic Constraints" Gadi Pinkas, Shimon Cohen https://arxiv.org/abs/1712.03049v1 …pic.twitter.com/HxAWrDyyUF
End of conversation
New conversation -
-
-
You might like to delve into how a machine learning monstrosity won 28-0 against the best rule-based chess player ever:https://www.technologyreview.com/s/609736/alpha-zeros-alien-chess-shows-the-power-and-the-peculiarity-of-ai/ …
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
On a reading of the abstract it could be *more* charitable since it allows the underlying plumbing to be Connectionist while (apparently) supporting more general/classical cognition.pic.twitter.com/l4ITDYaSEv
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.