A pure python implementation of a neural network.  

Original Author: Neil Schemenauer <nas@arctrix.com>
Modified Author: James Howard <james.w.howard@gmail.com>
Modified Author: Christopher J. Volny <cvolny@gmail.com>

1. Custom patterns entered from command line
2. Early Stopping
3. Configurable number of hidden nodes
4. Uses python logging to better configure verbose output
5. Refactored to better support multiple training methods
6. Uses numpy for matrix/array work


TODO:
	COMMENTS!
	Training:
		- GA
		- PSO
	Regression:
		- readd regression support that cvolny broke
	Basian weight decay:
		- cvolny to learn what this is!