Convergence of the Expectation Maximization algorithm

par Dominikus Noll (Institut de Mathématiques de Toulouse)

jeudi 29 oct. 2026, 11:00 → 12:15 Europe/Paris

Auditorium 3 (Toulouse School of Economics)

Abstract:  The EM-algorithm assures monotone increase of the incomplete data likelihood, but does not in general guarantee convergence of the parameter estimates. We take a fresh look at the situation and present sufficient conditions for convergence that are conveniently verifiable in practical situations. Illustrations with typical applications of the EM-algorithm are given.
Key words: Kullback-Leibler divergence, Fisher information, exponential family, proximal point method, definability, information geometry