Yes. Neural networks work using weighted values which act somewhat like channels. I have a download on the UT. It is nnp.zip (n)eural (n)etwork (p)roject). Mike Yearwood and I are on Sedna with it as a project. Mike has oodles more experience with neural nets, so I am ecstatic about his decision to give me a hand. So, to answer your question - Yes I am using VFP to run neural networks. It works just fine and uses VFP tables in a local situation.
Y'know, with the new abilty of VFP 9 to rotate graphics, it might be possible to display a neural network's processing much like the seti@home screensaver does.
I ain't skeert of nuttin eh?
Yikes! What was that?