Correlatory Learning
- Learning is essential to this categorisation process.
- The type of learning that the system is now using is
called Correlatory Learning.
- The weight of a connection tends toward the likelihood
that the post-synaptic neuron fires with the pre-synaptic
neuron fires.
- So, if every time neuron A fires, neuron B fires, the connection
from A to B will approach one.
- If on the other hand A fires only half the time B fires, the connection
from B to A will approach .5
- That's not quite right. There is a parameter, Synaptic Strength Median.
This sets the range of the weight. It's .9 so the AB synapse
will be 1.8 and the BA .9.
- That's still not quite right. There are inhibitory neurons and
they have an inverse weight. AB = 0, BA = -.9
