CAs are Categorizers
- A neural network is a network of simulated neurons.
- A neural network is an attractor network.
- Another type of attractor net is a Hopfield net.
- A Hopfield net does not simulate neurons, but is
instead a connectionist net. (Neural nets are
also connectionist nets.)
- Attractor nets take input, and, following an algorithm,
go to a stable or pseudo-stable state.
- Neural nets go to pseudo-stable states
- The neural net is given input from the environment (neurons
fire).
- Activation is passed around as neurons fires.
- If a CA ignites, it will remain active.
- The neurons in the CA fire intermittently.
- If all the neurons in the CA fired each step, it would
be a stable state.
- As they do not fire each step, it is a psuedo stable state.
- CA ignition is categorisation.
- The CA that ignites is the one that represents the
category of the stimuli.
- So, if a stimuli is presented, and the dog CA ignites, then
the stimuli is a dog.