just a quick question... haven't worked with NNs for a year now so im a bit rusty :wacko:
When the network propagates the errors backward through the network, are the bias weights updated along with all the other network weights?
For those who don't know, the bias weights are updated!
I'm glad you figured it out because I had forgotten too, but it seems like they would have to be trained. Recently, I've been playing around with FANN instead of building my own ANNs.
cool, thanks - this will save me time!