Future Work
- Large Associative Memories
- Initial 'toy' experiments are in progress.
- We need AMR for a basis for expanding CA work.
- Large systems will bring new problems (e.g. simulated epilepsy)
- Do things that DBs can't do. (E.g. semantics for parsing my
original motivation)
- Sequences: early work was done on this; we should do some of our
own
- Hierarchies: intimately related to AM
- Cognitive Maps: build structure at a level above the CA.
- Variable Binding and Rules
- There seems to be a lot of thought about variable binding
particularly via synchrony.
- It is needed to deal with combined concepts.
- It's also needed for rules.
- Synchrony might be epiphenomenal from bound CAs: CAs are bound
via activation and this causes them to synchronise.
- Does anyone know of a working VB model from reasonable plausible
neurons?
- Understand CAs by themselves better
