||Topic: Beginner Questions on SNN Charge Change
||Posted: 01-Jul-11 05:41
Im 16 years old and new to ANNs. I want to program a Spiking Neural Network but i have a problem.
I have a Neuron N1 and it is connected to the Neurons H1, H2 and H3.
The weight between N1 and H1 is 0.5, so if N1 shoots a spike to H1, H1's weight will be increased by 0.5 ( without sigmoid function ), am I right?
How is the charge of N1 decreased then (i have a function that continuosly decreases the charge slowly, but im sure thats not enough)? Or is it decreased at all?
I think it would be logic if the charge of N1 is decreased by 0.5V, the same value H1 is increased by.
But here is my problem:
If all weights leading away from N1 are 0.5 and N1's charge is 1V, which is over the threshold value how does the Neuron decide at which child it will shoot?
If it will shoot at all child Neurons and i subtract the weight of the Volt from N1 it will end up with a charge of -0.5V.
If the Neuron doesnt shoot at all of its childs does it shoot randomly at 2 of them?
Im sure this is not correct, but what else should happen?