Possible Solutions to CA3
- We need one solution to all of the problems.
- For memory based parsing we needed a solution for memory decay,
so I implmented it neurally using fatigue. However, it
won't work for even the reactivation problem
- Other mechanisms that we thought about include:
- Global inhibitory feedback (Chown and Kaplan)
- Refined neural model
- Refined learning model
- Multiple neuron types
- Refined Topology
- I expect that we'll have to combine several of these things
to get what we want.
