I’m going to go out on a philosophical limb and observe that there is evidence that complex neural networks - namely, humans - have invented symbolic representations, perhaps because they are more efficient mechanisms for computation for certain classes of problems.https://twitter.com/plinz/status/955300914117644288 …
-
-
Replying to @Grady_Booch
Also, our neural networks are not much like ANNs. Neuronal activation is roughly binary valued, while the dendritic interface does various kinds of conditional (even a bit of temporal) integration over its binary inputs. Links are mostly lateral, and only sparse between layers.
1 reply 0 retweets 1 like -
Replying to @Plinz
High five on that one. Our contemporary ANNs are akin to organic neurons as the first semiconductor by Bardeen, Brattain, and Shockley are to modern integrated circuits.
1 reply 0 retweets 0 likes -
Replying to @Grady_Booch
But do you think biological neurons are superior? Random access, long range connections, weight adjustments, real-valued computation, reliable determinism are all very expensive. At the end of the day, implementation details don't matter, the complexity of computed functions does
1 reply 0 retweets 0 likes -
Replying to @Plinz
Depends on how you measure superiority. I can still kick my computer when I’m annoyed at it, but it has not yet learned how to kick back. Still, I don’t make a habit of kicking my computer. It might have a long memory and sneak up on me when I least expect it.
1 reply 0 retweets 1 like
I am scared that my brain may one day kick back at me to retaliate for all the times I kicked it.
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.