fatiguing Leaky Integrate and Fire Neurons
- fatiguing Leaky Integrate and Fire (fLIF) neurons are the model
we use.
- They're not as accurate as compartmental models, but they
are quite accurate and we can simulate 100,000 of them in
real-time on my PC.
- IF neurons are McCullouch Pitts Neurons (1943); neurons collect
activation; if they surpass a threshold they fire and send out
activation.
- LIF neurons are commonly used. If in a given cycle (or gradually
for continuous models) a neuron does not fire, some of its activation
leaks away.
- Fatigue means the more a neuron fires the harder it becomes
to fire. We do this by raising the threshold when a neuron fires
and decrease it (down to a base) when it doesn't fire.
- We're pretty much unique in using this model.
- Our model is discrete and ignores synaptic delays and
refractory periods.
- That means that each cycle is roughly equivalent to 10 ms
of biological time.
