Strategies for Application of Directional Statistics in Probabilistic Machine Learning

Statistical models are playing an important role in probabilistic machine learning applications. In this paper, we focus on a special type of data that is normalized, so that its “direction” is more important than its magnitude. Directional Statistics is a branch of statistics that deals with this type of data. Specifically, we consider protein backbone angles, Text Documents, and fMRI time series that lie either on the surface of the unit hypersphere, torus or on the real projective plane. For such data, we briefly review common mathematical models prevalent in machine learning, while also describing some technical algorithms for model-based clustering algorithms and some other challenges.