Well guys, I'm having a hard time learning enough about ANNs to apply them to a problem.

Firstly my general understanding is that the network adapts during the "training" phase to produce the correct output with the given set of inputs. This is obviously useful with image recognition, or handwriting recognition.

But it has these properties by working in a specific way, and I'm mostly having difficulty implementing an ANN for a (very) simple application.

So... If someone has some experience with them, a little insight into what they really do, and how they do it would be most helpful!