Solution Topology
- In earlier learning work we used a one subnet topology with distance
biased connectivity.
- More recently we've gone to multiple subnets, so that
there is more flexibility and it helps usability and debugging.
- Each square represents a neuron, with the circles firing in a
particular cycle.
- All neurons are excitatory.
- There are no internal connections in the Input Subnet, and
each of those has 10 connections to the SOM subnet.
- Each neuron in the SOM subnet has 20 connections to other
neurons in the subnet.
- All connections are random.
