Accelerating convergence of the EM algorithm via the vector ε algorithm
スポンサーリンク
概要
- 論文の詳細を見る
The EM algorithm of Dempster, Laird and Rubin (1977) is most popular iterative computational algorithm to find maximum likelihood estimates from incomplete data. The EM algorithm is broadly used to all most statistical analysis with missing data, because of its stability, flexibility and simplicity. However, it is criticized that the convergence of the EM algorithm is slow. The various algorithms based on the Newton-type algorithm are proposed to speed up the convergence of the EM algorithm. In this paper, we present an alternative accelerating EM algorithm utilizing the vector epsilon algorithm that is an non-linear technique to accelerate the convergence of slow converging sequences.
- 日本計算機統計学会の論文
- 2004-05-19
著者
-
KURODA Masahiro
Department of Radiology, Graduate School of Medicine and Dentistry, Okayama University Graduate Scho
-
Kuroda Masahiro
Department Of Radiology Okayama University Medical School
-
SAKAKIHARA Michio
Department of Information Science, Okayama University of Science
-
Sakakihara Michio
Department Of Applied Mathematics Okayama University Of Science
-
Sakakihara Michio
Department Of Information Science Okayama University Of Science
-
Kuroda Masahiro
Department Of Computer Science And Mathematics Kurashiki University Of Science And The Arts
関連論文
- The usefulness of serum thyroglobulin levels and Tl-201 scintigraphy in differentiating between benign and malignant thyroid follicular lesions
- Positive gallium-67 and thallium-201 scans in thymic rebound after chemotherapy for lymphoma
- Hepatectomy simulation discrepancy between radionuclide receptor imaging and CT volumetry : influence of decreased unilateral portal venous flow
- Nontumotous decrease in Tc-99m GSA accumulation
- MR Appearance of Endometrial Carcinoma and Mucinous Cystadenoma in a Postmenopausal Patient Treated with Tamoxifen for Breast Cancer
- Preoperative hyperthermochemoradiotherapy for soft tissue sarcoma
- Cervical Carcinoma: Dynamic MR Imaging with a Turbo-FLASH Technique
- BAYESIAN SEQUENTIAL LEARNING FROM INCOMPLETE DATA ON DECOMPOSABLE GRAPHICAL MODELS
- Accelerating convergence of the EM algorithm via the vector ε algorithm
- The Gauss-Seidel Iterative Method with the Preconditioning Matrix (I + S + S_m)
- L$^2(\Omega)$-ERROR ESTIMATE OF GALERKIN BOUNDARY ELEMENT METHOD WITH SINGLE LAYER POTENTIAL