[<<][math][>>][..]
Tue Jun 2 14:10:46 CEST 2009

Throwing away the right information

While the computation mechanism explained in the previous posts might
be interesting as a mathematical toy, I'm not sure whether building
something like that is practical.  It requires a huge amount of
connections.

  A human brain contains about 10^11 cells with on average 7x10^3
  connections.  In log10 that's 11 and 3.84 with a ratio of 2.6

The thing is though, the architecture has failure built-in.  So
manufacturing failure should probably not throw off things too much.
It can be built with low yield, might might drasticly reduce
fabrication costs, to a point where it might be feasible in a home
lab with something other than sillicon.



[Reply][About]
[<<][math][>>][..]