First of all, it is nice to post here after such a long time of absence.

How should propagate error in feedforawrd network with backpropagation learning.

Should i backpropagate after each input vector data.

or maybe

Calculate all errors from my input vectors and than do something with this errors and backpropagate once for an epoch??
Posted on 2005-12-01 18:15:02 by AceEmbler
I rewrote Thomas Bleeker's NNet demo for OA32 some time ago.
You should check out his work first.
Posted on 2005-12-11 02:35:56 by Homer