Bruno Herencic wrote:
"How are those sinapse weights set in a human brain? I understand we can
simulate a simple neural net on a computer and change those weights, but how did
the nature set those weights? when? are they being changed over the time... how?
if a brain wants to remember something, memorize, or do
something, it must change those weights, doesn't it? this means input data can
have some inapct on those nets in the brain...?"
It sounds as though you are referring to multilayer perceptrons (MLP), the most
common artificial neural network. Most artificial neural networks (including
MLP) are very crude approximations of biological nervous systems, and I wouldn't
count many of them as 'simulations' of actual biological activity. More
biolobically accurate models have been built: see the work of Freeman or Alkon,
predictor at compuserve.com