Stable States
- Hopfield nets fall into stable states
- To some extent (asynchronous mode), these are attractor states (see
SOM)
- CAs are pseudo-stable states
- if fatigue is turned off, there will be a stable or osciallating state
- with fatigue on, a near osciallatory pattern is formed
- this oscillation is close to repeating
- there is physiological evidence for this [Abeles]
- Can the stable states of Hopfield nets help us form stable CAs?
- Clearly we have set up stable states that we want
- However, can we learn the states?
- Do the states act as appropriate attractors?
- Are they good CAs?
- How does fatigue effect this stable state?
- If DeVries is right, CAs should be able to persist in pseudo-stable
states indefinitely.
- These stable states must also act as
attractor states
