Machine Learning Through Probabilistic Programming

In this article we give an introduction to the Probabilistic Programming (PP) paradigm for .NET engineers. We start by explaining the differences between PP and traditional approaches and show a “Hallo World” equivalent in PP. We then move on to building a sample probabilistic model from scratch using Infer.NET. The example chosen is the TrueSkill rating system, which we design and implement step by step, as well as demonstrate how to query for training and prediction. At the end we discuss the benefits of PP and close up with pointers to further reading.