The problem of hardware neurons/synapses: it seems we still miss significant stuff on biological neurons/synapses for instance biological neurons/synapse are much less "locality" constrained, and to give this level of freedom to hardware "neurons/synapses" (and with the right "rules") will waiste a lot of hardware real estate. And I don't even start on 3D space.
What I mean is you will need probably a lot more hardware neurons/synapses in order to break free from the hardware 2D locality limitations towards a 3D space with much less locality constraints.
I dare to presume that 3D + much less locality constraints (with the right "rules") would give more "efficient" cognitive functions. Yeah, that's probably way to much to presume.
What I mean is you will need probably a lot more hardware neurons/synapses in order to break free from the hardware 2D locality limitations towards a 3D space with much less locality constraints.
I dare to presume that 3D + much less locality constraints (with the right "rules") would give more "efficient" cognitive functions. Yeah, that's probably way to much to presume.