The source code can be downloaded here. It depends on QTXlib, so
you will have to compile that prior to compiling the NN code.
A am using GCC on a linux-ix86 box;
you may need to do some tweaking to get it compiled. See the head of
the source code file for the gcc call I am using.
You should symlink Qt to the Qt include directory.
Random run with above program using a 10x10 output layer, eta=0.3 and
sigma=sqrt(3). (Gauss-only neighbourhood). The t= entry displays
the number of training cycles (one cycle is one input pattern feed).
There are 16 different input patterns. Note the complete separation of
all input patterns after roughly 400 cycles.