Personally I am very deeply concerned by the Symbol Grounding problem and it's not going away any time soon. However, recent work on integrating symbolic and sub-symbolic approaches is promising: https://homes.cs.washington.edu/~pedrod/papers/cacm19.pdf … and http://www.semantic-web-journal.net/content/neural-symbolic-integration-and-semantic-web …. Interested in your thoughts.
-
-
-
Symbol grounding is just a specific framing of model grounding. A model is fully ‘grounded’ if it is isomorphic to the system it models. Here, we are looking for a good-enough isomorphism to the universe, i.e. an interactive VR that predicts sensory data.
- 4 more replies
New conversation -
-
-
It's a continuous economic problem. It tries to maximize the usable resources it can gain (~minimize free energy) by optimizing the organization of its available resources.
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
I agree, this is just sufficient to specify the search space for organismic control in the most general case, not an actual implementation, which needs to be narrowed down.
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
The symbolic vs subsymbolic divide may be a red herring. There is a more general way to think about models as parameter sets constraining each other via computable functions, and principles of organization that will make efficient use of the available substrate computations.
-
I agree. Unfortunately Cognitive Science is still obsessing over this false dichotomy
End of conversation
New conversation -
-
-
One would assume that both principles/ hypotheses are interconnected by the Constructor Theory of Information (Deutsch&Marletto).
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.