Convergence properties of the EM algorithm in constrained parameter spaces
نویسنده
چکیده
The established general results on convergence properties of the EM algorithm require the sequence of EM parameter estimates to fall in the interior of the parameter space over which the likelihood is being maximized. This paper presents convergence properties of the EM sequence of likelihood values and parameter estimates in constrained parameter spaces for which the sequence of EM parameter estimates may converge to the boundary of the constrained parameter space contained in the interior of the unconstrained parameter space. Examples of the behavior of the EM algorithm applied to such parameter spaces are presented. 1 EM CONVERGENCE PROPERTIES 2 R ESUM E Les r esultats de convergence existants concernant l'algorithme EM pr esupposent que les estimations successives du param etre appartiennent a l'int erieur de l'espace param etrique sur lequel la vraisemblance est maximis ee. Cet article examine la convergence de cette suite d'estimations et celle des valeurs de la vraisemblance qui leur sont associ ees dans le cas o u la suite des estimations du param etre peut converger vers un point qui se situe a la fronti ere de l'espace param etrique contraint tout en appartenant a l'int erieur de l'espace param etrique complet. Le comportement de l'algorithme EM est illustr e dans de tels cas.
منابع مشابه
A note on constrained EM algorithms for mixtures of elliptical distributions
The existence of a global maximizer of the likelihood on constrained parameter spaces has been proved here, extending a previous result due to Hathaway (1986) to the multivariate setting and to the general case of mixtures of elliptical distributions. Then, focusing on a particular data set which motivates our methodology, a new definition of weak homoscedasticity is introduced. Successively, a...
متن کاملW-convergence of the proximal point algorithm in complete CAT(0) metric spaces
In this paper, we generalize the proximal point algorithm to complete CAT(0) spaces and show that the sequence generated by the proximal point algorithm $w$-converges to a zero of the maximal monotone operator. Also, we prove that if $f: Xrightarrow ]-infty, +infty]$ is a proper, convex and lower semicontinuous function on the complete CAT(0) space $X$, then the proximal...
متن کاملConvergence theorems of multi-step iterative algorithm with errors for generalized asymptotically quasi-nonexpansive mappings in Banach spaces
The purpose of this paper is to study and give the necessary andsufficient condition of strong convergence of the multi-step iterative algorithmwith errors for a finite family of generalized asymptotically quasi-nonexpansivemappings to converge to common fixed points in Banach spaces. Our resultsextend and improve some recent results in the literature (see, e.g. [2, 3, 5, 6, 7, 8,11, 14, 19]).
متن کاملOn some fixed points properties and convergence theorems for a Banach operator in hyperbolic spaces
In this paper, we prove some fixed points properties and demiclosedness principle for a Banach operator in uniformly convex hyperbolic spaces. We further propose an iterative scheme for approximating a fixed point of a Banach operator and establish some strong and $Delta$-convergence theorems for such operator in the frame work of uniformly convex hyperbolic spaces. The results obtained in this...
متن کاملNew Complete-Data Spaces and Faster Algorithms for Penalized- Likelihood Emission Tomography
The classical expectation-maximization (EM) algorithm for image reconstruction suffers from particularly slow convergence when additive background effects such as accidental coincidences and scatter are included. In addition, when smoothness penalties are included in the objective function, the M-step of the EM algorithm becomes intractable due to parameter coupling. Thie paper describes the sp...
متن کامل