Categorisation
- The simulation work that I've done do far is really
on categorisation.
- You can find the simulator and links to parameter settings
for these experiments on
http://www.cwa.mdx.ac.uk/cant20/CANT20.html
- I started out with a grid of neurons, and most of the work
has been done with grids of 400, 900 or even 10,000 neurons.
- The first thing I did was I presented neurons to see if I
could get CAs to persist after external stimuli ceased.
That was pretty easy, though I find that they either persist
for a short number of cycles (up to 100), or for ever.
- The next thing I did was to present either neurons from the
top half or the bottom half.
- This allowed 2 CAs to form, and when neurons from one half were
presented, a lot more neurons became active and then persisted
in activation.
- One problem to be avoided was to have all of the neurons active
(simulated epilepsy).
- An ambiguous stimuli was presentation of some neurons from the
top and some from the bottom.
- Inhibitory neurons were needed, but this also worked, with
both CAs starting to ignite, then one supressing the other.
