Catalogo Articoli (Spogli Riviste)

OPAC HELP

Titolo:
SPACE-ALTERNATING GENERALIZED EXPECTATION-MAXIMIZATION ALGORITHM
Autore:
FESSLER JA; HERO AO;
Indirizzi:
UNIV MICHIGAN,DIV NUCL MED ANN ARBOR MI 48109 UNIV MICHIGAN,DEPT ELECT ENGN & COMP SCI ANN ARBOR MI 48109
Titolo Testata:
IEEE transactions on signal processing
fascicolo: 10, volume: 42, anno: 1994,
pagine: 2664 - 2677
SICI:
1053-587X(1994)42:10<2664:SGEA>2.0.ZU;2-C
Fonte:
ISI
Lingua:
ENG
Soggetto:
POSITRON-EMISSION TOMOGRAPHY; EM ALGORITHM; MAXIMUM-LIKELIHOOD; RECONSTRUCTION ALGORITHMS; IMAGE-RECONSTRUCTION; ITERATIVE RECONSTRUCTION; CONVERGENCE; PET;
Tipo documento:
Article
Natura:
Periodico
Settore Disciplinare:
Science Citation Index Expanded
Citazioni:
43
Recensione:
Indirizzi per estratti:
Citazione:
J.A. Fessler e A.O. Hero, "SPACE-ALTERNATING GENERALIZED EXPECTATION-MAXIMIZATION ALGORITHM", IEEE transactions on signal processing, 42(10), 1994, pp. 2664-2677

Abstract

The expectation-maximization (EM) method can facilitate maximizing likelihood functions that arise in statistical estimation problems. In the classical EM paradigm, one iteratively maximizes the conditional log-likelihood of a single unobservable complete data space, rather thanmaximizing the intractable likelihood function for the measured or incomplete data. EM algorithms update all parameters simultaneously, which has two drawbacks: 1) slow convergence, and 2) difficult maximization steps due to coupling when smoothness penalties are used. This paper describes the space-alternating generalized EM (SAGE) method, which updates the parameters sequentially by alternating between several small hidden-data spaces defined by the algorithm designer. We prove thatthe sequence of estimates monotonically increases the penalized-likelihood objective, we derive asymptotic convergence rates, and we provide sufficient conditions for monotone convergence in norm. Two signal processing applications illustrate the method: estimation of superimposed signals in Gaussian noise, and image reconstruction from Poisson measurements. In both applications, our SAGE algorithms easily accommodate smoothness penalties and converge faster than the EM algorithms.

ASDD Area Sistemi Dipartimentali e Documentali, Università di Bologna, Catalogo delle riviste ed altri periodici
Documento generato il 04/04/20 alle ore 08:18:31