Topology Problems
- There is a mismatch between CAs and Hopfield nets on connectivity
- Hopfiled nets are well connected but
- each neuron in the brain has only 1000 (or so) connections
- Our CANT models have even fewer connections
- In Hopfield nets this is equivalent to synapses with weights of
0
- However, the very sparse matrix constrains how a Hopfield stable state
- Another problem is that neurons are either excitatory or inhibitory
- This is contradictory to the two directed connection replacement for a single
bidirectional connection.
- At this point, we can no longer use the energy based calculation to solve
the stable state problem.
- But the Hopfield energy calculations can still inform this net
