Neural Sim was an eight month project that I participated in that involved neural networks. The application enables the user to design a neural network, specify all of its variables and import some training data. The network will take the training data and learn the pattern of the data. Simply put, it interpolates the in between points from the training data.
This is useful for modeling the response of just about anything. In this case the primary focus was to model the response of an electronic circuit operating at high frequencies. Circuits operating at high frequencies involve complex math in order to model. This is why it is a good idea to use something like Neural Sim as opposed to doing calculations by hand.
I was in charge of design and architecture of the software. I also took part in implementing the algorithms into the code. The program was built with Java.