Once, a long time ago, I had a neural networks class similar to the one you discuss. The major project of the class was to implement a decision tree application. I spoke to the prof about implementing my own decision tree engine in perl instead of using the class default program, which was a win16 app running on NT4. (PAINFUL).

I implemented it in a rather short amount of time. I even added support to make it self-optimizing. My decision tree would re-balanced itself if you asked the same question in different locations to require the fewest number of questions of the user based on an average for all of the leafs. I implemented it 3 weeks ahead of schedule for everyone else to be done adding questions & answers to their trees. I got to show it off to the class, and she thought it was great.

Up until test time, I thought I was going to ace the class. However, I failed due to 90% of the questions on the midterm were not about neural networks, but how to use the tool that she had the rest of the class use.