sonia@singsing.berkeley.edu (Sonia Marx) (01/27/91)
I'm writing a program in C++ to structure a neural net based on prior knowledge about a dataset and then learn the dataset using backpropagation. I've been getting very discouraging results, but I'm not sure if they're due to faults in the theory or in my program. Does anyone out there have code to do backpropagation (so I can compare results), or code to display the changing weights in the network as it runs, preferably compatible with X windows? Also, if anyone is working on growing nets incrementally, I'd love to talk with you about it, since that will be the next stage of my project. Many thanks, and my apologies if this is the Nth request of this form... Sonia Marx sonia@guard.berkeley.edu