A basic but extensible implementation of a multilayer perceptron neural network using VB.NET.
There are a lot of people who dislike VB, but I'd argue that they've never worked substantially with it (at least not in .NET form).It's actually really quite readable and has all of the features you'd expect from other modern high-level languages. I decided to implement a neural network in it for fun and since it was one of the first languages I learned fairly well.
Often-times machine learning algorithms are optimized (for good reason) to the point of being unreadable and difficult to understand. Here I've taken an object oriented approach instead that is really quite readable. It's not the fastest implementation, nor the most fully featured. But, it is a good start to learning how neural networks work and provides some entry points for exploring the many combinations of factors that can adjust its performance.
For now I've only included the simple canonical "xor" scenario as a test case. This is a well-known instance of a non-linear relationship that a single layer of neurons cannot solve.