Our CANT model
- We have been developing a computational model for a few years now.
- It's called CANT and a version can be found at
http://www.cwa.mdx.ac.uk/cant1.8/start.html
- I've also got a version here that I'd be happy to show you.
- The goal is to have a system that takes the essential mechanisms of
biological neurons and implements them efficiently.
- CANT currently works on these basic premises
- Neurons are leaky thresholding integrators
- Neurons fatigue
- Some neurons are inhibitory, some excitatory
- Synapses are unidirectional
- Synaptic connections are distance biased
- All learning is Hebbian
- The system runs in discrete (10 ms.) cycles
- These are not set in stone, but we've been using them for a few years
now.
