Finding clusters and components by unsupervised learning

Abstract

We present a tutorial survey on some recent approaches to unsupervised machine learning in the context of statistical pattern recognition. In statistical PR, there are two classical categories for unsupervised learning methods and models: first, variations of Principal Component Analysis and Factor Analysis, and second, learning vector coding or clustering methods. These are the starting-point in this article. The more recent trend in unsupervised learning is to consider this problem in the framework of probabilistic generative models. If it is possible to build and estimate a model that explains the data in terms of some latent variables, key insights may be obtained into the true nature and structure of the data. This approach is also reviewed, with examples such as linear and nonlinear independent component analysis and topological maps. © Springer-Verlag 2004.