I just realized: if a single neuron in the brain can connect to 10000 others, it means that if you want to model it with a regular lattice, you'd need at least 17 dimensions. #kissingNumber
-
-
Replying to @Plinz
… such that all dendrites/axons have the length of the lattice spacing? Why is this important?
1 reply 0 retweets 0 likes -
Replying to @mere_mortise
Imagine you want to build a microchip that efficiently encodes the neuron level a human cortex using a general pattern: you would have to make sure that each node can talk to at least 900 neighbors.
1 reply 0 retweets 1 like -
Replying to @Plinz @mere_mortise
Would that require a breakthrough? i.e., is it a very hard engineering problem given what we know? Or more a matter of someone deciding to do it?
1 reply 0 retweets 0 likes -
Replying to @DKedmey @mere_mortise
No, it would not be a breakthrough for AI. Representing a connectome alone is not sufficient, and this would merely the paper on which the connectome could be written.
1 reply 0 retweets 0 likes
Btw., it sounds grandiose but is trivial in practice. Each node in your graph computes a simple function over the values of up to 900 other nodes. (You also need to be able to integrate over a few timesteps, though.)
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.