The CANT model
- Our focus is on working on simplified models of neurons to do
computationally interesting things
- Hopefully, we don't throw out the important properties
- Model Properties
- Neurons as leaky integrators
- Neurons fatigue
- Simulations run in discrete steps not continuously
- Neurons are connected in a distance-biased way
- Neurons can be excitatory or inhibitory
- Learning is done locally by axonal weight change
- With this model we can recognise different patterns and
model competition
- We are hoping to model rules, hierarchies and structures with
it
