Love your enthusiasm! The composite images do sound really interesting for this application. Wonder how it would perform with larger noise levels in the network. Anyway I have no idea how these work, but I'll do my research, thanks!

Thanks for the advice, it sounds promising and I'll definitely give it a go, but I gotta learn how to walk first!

Thanks a lot, I'll look into it, unfortunately I only have one example each, I currently just need it to tell me which if which when I run one of the examples through. Edit: Also how do I implement logistic regression if I have 10 possible outputs instead of just yes/no?