Indeed, once upon a time, Hinton was interested in the same sorts of unification that I believe to be so important.https://twitter.com/danbri/status/1075653592596377600 …
-
..."[Most connectionist researchers accept this], though they expect that this ability may be implemented in ways that have not been anticipated within the standard symbol processing tradition." Pretty clear, and consistent with modern efforts and attitudes.
My prediction here (on record since 1998, esp 2001) is that those novel approaches will turn out to map perfectly onto to symbol-manipulation mechanisms, such as bindings and operations over variables. If they don’t, I turn out to be wrong; if they do, I turn out to be right.
-
-
I think the talk of variables is a mistake since one can perfectly well do logic without variables using eg combinators in the Schönfinkel style.
-
And the resulting methods are syntactically much easier to deal with though they aren’t really human readable
End of conversation
New conversation -
-
-
IMO, symboks is about comnunication behavior. Of course they are part of inferred patterns at working memory level, so they can be used in reasoning paths. Symbols are not necessary for reasoning.
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
"perfectly" is brave, but maybe there is wiggle-room in "turn out to map"
-
Bindings to names are obviously correct. However, operations over variables is the where thesis falls apart. Humans can attend to not more than 5 items at a time and only 3 levels of recursion. This kind of bounded rationality is no where near 'symbolic processing'.
- 1 more reply
New conversation -
-
-
Symbol manipulation mechanisms meaning GOFAI? If that is your position then you are very wrong.
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.