Results
- We have been doing work on this for a few years now, and have
made progress on Categorisation.
- One of the first things we did was have the system learn two
distinct CAs. We succeeded with Pearson's measurements for
reliable activation (~.22), uniqueness (~-.03) and persistence
(~.75).
- I'll explain some results on spontaneous neural activation
and how that effects the memory system.
- I'll then show some overlapping CA work, which uses
a compensatory learning mechanism and is used as
the basis of hierarchical categorisation.
- Results on two datamining tasks will be presented.
- Finally, I'll say something about the results of some
theoretical work.
