Fixed points of the EM algorithm and nonnegative rank boundaries

Abstract

Mixtures of $r$ independent distributions for two discrete random variables can be represented by matrices of nonnegative rank $r$. Likelihood inference for the model of such joint distributions leads to problems in real algebraic geometry that are addressed here for the first time. We characterize the set of fixed points of the Expectation-Maximization algorithm, and we study the boundary of the space of matrices with nonnegative rank at most $3$. Both of these sets correspond to algebraic varieties with many irreducible components.

Published in at http://dx.doi.org/10.1214/14-AOS1282 the Annals of Statistics (http://www.imstat.org/aos/) by the Institute of Mathematical Statistics (http://www.imstat.org)

References in corpus (3)

• Identifiability of parameters in latent structure models with many observed variables
• Self-scaled bounds for atomic cone ranks: applications to nonnegative rank and cp-rank
• Algebraic boundary of matrices of nonnegative rank at most three

Cited by in corpus (11)

• Self-scaled bounds for atomic cone ranks: applications to nonnegative rank and cp-rank
• A universality theorem for nonnegative matrix factorizations
• Logarithmic Voronoi cells
• Singularity structures and impacts on parameter estimation in finite mixtures of distributions
• The maximum likelihood degree of mixtures of independence models
• Nonnegative rank depends on the field II
• New directions in algebraic statistics: Three challenges from 2023
• Mixtures of Discrete Decomposable Graphical Models
• Maximum likelihood geometry in the presence of data zeros
• Exact solutions in low-rank approximation with zeros
• On tropical and nonnegative factorization ranks of band matrices