Re: Neural net implementation question

On Apr 1, 8:46 pm, castiro...@xxxxxxxxx wrote:
Question: In Neural and Bayesian nets, is every node updated on every
clock pulse? Or are selective ones? I am thinking that you could
(not a pro.) update weights O( f ) less and grow the number of
connections O( f ). What kind of advanced AI is going on?

There are times not all nodes require updating.

Start at the end, adjust the weights on the output layer, then work
back layer by layer adjusting only the nodes connected to the node(s)
you just updated. (for backprop and backprop related algorithms).

[ comp.ai is moderated ... your article may take a while to appear. ]
.