I wrote an article “Neural Network Back-Propagation using C#” which appears in Visual Studio Magazine. See http://visualstudiomagazine.com/articles/2013/08/01/neural-network-back-propagation-using-c.aspx. A neural network is a software system that loosely models biological neurons and synapses. Neural networks can be used to make predictions such as determining the political party affiliation (Democrat, Republican, Other) of a person based on factors such as the person’s age, annual income, and sex. To make such predictions, a neural network has to be trained using existing data with known values. This essentially boils down to determining the values for a set of weights so that, for the set of known inputs, the neural network generates computed outputs which best match the known outputs.

There are several approaches for training a neural network, but the most common is called the back-propagation algorithm. In the Visual Studio Magazine article, I explain how the back-propagation algorithm works and present a complete C# demo program. Although the back-propagation algorithm itself is quite complex, the code is quite a bit shorter than you might expect.

It seems that in the software developer community there has been (based on my observations anyway) greatly increased interest in neural networks over the past 18-24 months. In addition to this article, I’ll be giving a talk on neural networks at the upcoming DevConnections conference, September 30 – October 4, 2013, in Las Vegas. See http://www.devconnections.com.