Hmm... I don't know, seriously... How about start from reading? The WikiPedia may help you a bit.
If you think you understood the theory but do not know how to apply on an application, that would be a completely different story. You just feel disconnected. I too was disconnected and I went to ask my professor. Unfortunately, my professor answered me I don't know and he did nothing to help me understand. I had to move on and received a bad grade (C). You may have a better professor than I have that he could explain to you better. Even now, I understood the theory and how to compute it, but I have no idea how to apply the thing to an application...
Neural network programming isn't trivial. What other programming experience and training (theory) do you have already? Myself, I had to take a course from MIT to get reasonably up to speed on the subject, and I already had about 15 years of professional software engineering experience.
For a fun, practical way to get started with neural networks (or other types of machine learning) you could have a look at NetLogo. It's a modelling environment in which you can create and run a ton of different kinds of simulations. It has a large sample library, which includes some ANN examples as well. It's based on a simple programming language where you can easily add/adjust the simulation, but also add GUI controls for easy adjustment of your variable values while running the model. You can also create graphs and plots for a good overview of what's happening in your simulation.
It won't drive a car for you, but it'll give you a way to expirement and perhaps gain some more insight in the matter.