I'm trying to grasp this concept... I understand the basics of how it functions once it's set up. but I just don't really understand how to set one up? how do you know how many neurodes to make, and how many layers, and how to connect them? And how do you adjust weights and biases for learning? I've read over several introductions to neural networks, but maybe I just need to look at some sample code or something... I think there's something that I just keep missing and it's really bugging me. I should be able to understand this. Any help?

Have a look at ai-junkie if you haven't already. While I'm still learning the basics of neural nets, it's helped a lot, and I've learned loads about genetic algorithms from there too.

To try and answer some of your questions, the number of neurons depends on the problem. The first layer will always have the same amount of neurons as there are inputs into the neural net, and the output layer the same as the number of outputs obviously. As for the hidden layers, it just seems to me that you should experiment, and that there is no definite answer for a solution.

I know that the weights can be ajusted either by back-propogation, in which the output of the network is used to go back through each neuron and ajust the weight accordingly, or by using a genetic algorithm.

Hope some of that helps.

Thanks a lot mate!
That link was much more hepful than things I've been reading so far, at least for getting started.

you can use the perceptron as well. i find it easier than the backpropagation of the error....