Issue No.05 - May (2004 vol.26)
pp: 651-656
Zoran Zivkovic , IEEE Computer Society
<p><b>Abstract</b>—There are two open problems when finite mixture densities are used to model multivariate data: the selection of the number of components and the initialization. In this paper, we propose an online (recursive) algorithm that estimates the parameters of the mixture and that simultaneously selects the number of components. The new algorithm starts with a large number of randomly initialized components. A prior is used as a bias for maximally structured models. A stochastic approximation recursive learning algorithm is proposed to search for the maximum a posteriori (MAP) solution and to discard the irrelevant components.</p>
Online (recursive) estimation, unsupervised learning, finite mixtures, model selection, EM-algorithm.
Zoran Zivkovic, "Recursive Unsupervised Learning of Finite Mixture Models", IEEE Transactions on Pattern Analysis & Machine Intelligence, vol.26, no. 5, pp. 651-656, May 2004, doi:10.1109/TPAMI.2004.1273970