What we've done with CAs
- I just tuned FLIF parameters to biological data and got something
like 90% correct spike timing.
- We've used FLIF CAs for categorisation tasks.
- We recently showed that a net of CAs is Turing complete.
- We have done some work with Hebbian learning rules and spontaneous
neural activation.
- I came up with a novel form of Variable Binding, by STP, which really
makes symbolic reasoning work.
- I used this binding to develop a neurocognitive model of Natural
Language Parsing (NLP is my doctoral work).
- We've developed some nice Associative Memory systems (Kailash Nadh).
- Belavkin and I came up with a nice cogntive model of a two choice task
that shows the ability to weight rules based on environmental feedback.
- Our recently finished EPSRC grant was on Cell Assembly roBots (CABots).
- The CABots used NLP, vision, planning and spatial cognitive map
learning to work with a user in a simple 3D games environment.
