Developing Neural Network architectures

I've been putting a lot of my outside efforts into studying neural networks. After following along with David Miller's great vimeo tutorial to produce my first neural network in C++, I've began reading primers not only on how neural networks are constructed, and what types of neural networks best fit certain problems. I've been through a few resources, but my best recommendation would be Neural Network Design by Martin Hagan et al. I recommend following along with the book slowly and digest each new piece of information slowly, and, instead of looking at the available files, create your own working examples. You'll see on my github that I've done just that, using Octave. It was my initial intention to follow along in Racket/LISP, but it quickly became a detriment trying to force lists of lists to behave like matrices, especially for the finer things like transposing matrices. I know there's an available library, but it falls back on "arrays", and it all just feels very un-lisp-like! Eventually I will meet this goal. However, for now, check out my github for updates on the ANN stuff I've developing. I'll try to get some solid documentation available so you can just clone and go.