Discussion
- CAs are Turing Complete, and we've implmented some cool toys.
What's the point.
- Firstly, the PP attachment stuff shows that we've got a technology
that is at or near the leading edge for an industrially viable
problem.
- It's good for mining associations.
- Secondly, the parser and two-choice learning show that CAs can
lead to good neuro-cognitive models.
- These are open to the criticism of being ad-hoc, but they can
also be linked to neural data.
- They can also be used to explore that data.
- Thirdly, simulated neurons and CAs provide a mechanism for
combining systems sensibly, a neuro-cognitive architecture.
- We've used subnets to separate systems and that's a nice engineering
approach.
- However, there is a lot to explore about how brain areas
work together.
