Neural Systems
- I've included neural systems here because they are my
current interest.
- I define neural systems as sytems based on simulated
neurons.
- A lot of connectionist architectures (e.g. MLPs) don't do
things like neurons.
- I work with a model called fatiguing Leaky Integrate and Fire (fLIF)
neurons.
- They're a pretty good model of neural behaviour but other models
(e.g. Hodgkin and Huxley neurons) are better.
- I've used the fLIF neurons to categorise data.
- The long term hope is that we can get effective cognitive systems
by simulating people at the neural level.
- For example, in my current agent we're trying to do vision
the way people do it.
- There is not a lot of industrial work in using neuropsychological
models for data mining.
