{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,9,12]],"date-time":"2026-09-12T00:15:50Z","timestamp":1789172150288,"version":"build-2803163510"},"reference-count":45,"publisher":"Oxford University Press (OUP)","issue":"3","license":[{"start":{"date-parts":[[2026,1,1]],"date-time":"2026-01-01T00:00:00Z","timestamp":1767225600000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/academic.oup.com\/pages\/standard-publication-reuse-rights"}],"content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":[],"published-print":{"date-parts":[[2026,7,6]]},"abstract":"<jats:title>Summary<\/jats:title>\n                  <jats:p>Gaussian mixture models are fundamental statistical tools for modelling heterogeneous data. Due to the nonconcavity of the likelihood function, the expectation-maximization (EM) algorithm is widely used for parameter estimation of each Gaussian component. Existing analyses of the EM algorithm\u2019s convergence to the true parameter focus on either the two-component case or multi-component settings with known mixing probabilities and isotropic covariance matrices. In this work, we study the convergence of the EM algorithm for multi-component Gaussian mixture models in full generality. The population-level EM algorithm converges to the true parameters provided that the minimum pairwise separation between Gaussian components exceeds a logarithmic factor of the maximum separation and the inverse of the smallest mixing weight. At the sample level, the EM algorithm is further shown to be minimax rate-optimal. We develop two novel analytical approaches, each tailored to a different separation regime, reflecting two complementary perspectives on the use of EM: parameter estimation and clustering. As a byproduct, our analysis reveals that the EM algorithm, when used for community detection, also achieves the minimax-optimal rate of misclustering error, an interesting result in its own right. Our results allow the number of Gaussian components, the minimum mixing weight, the component separation and the dimension to grow with the sample size. Simulation studies corroborate our theoretical findings.<\/jats:p>","DOI":"10.1093\/biomet\/asag047","type":"journal-article","created":{"date-parts":[[2026,7,8]],"date-time":"2026-07-08T11:56:48Z","timestamp":1783511808000},"source":"Crossref","is-referenced-by-count":0,"title":["Convergence and optimality of the EM algorithm under multi-component Gaussian mixture models"],"prefix":"10.1093","volume":"113","author":[{"ORCID":"https:\/\/orcid.org\/0000-0001-7462-9360","authenticated-orcid":false,"given":"Xin","family":"Bing","sequence":"first","affiliation":[{"name":"Department of Statistical Sciences, University of Toronto , 700 University Avenue , Toronto, Ontario M5G 1X6,","place":["Canada"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Dehan","family":"Kong","sequence":"additional","affiliation":[{"name":"Department of Statistical Sciences, University of Toronto , 700 University Avenue , Toronto, Ontario M5G 1X6, \u00a0 dehan.kong@utoronto.ca","place":["Canada"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Bingqing","family":"Li","sequence":"additional","affiliation":[{"name":"Department of Statistical Sciences, University of Toronto , 700 University Avenue , Toronto, Ontario M5G 1X6, \u00a0 bbingqing.li@mail.utoronto.ca","place":["Canada"]}],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"286","published-online":{"date-parts":[[2026,7,9]]},"reference":[{"key":"2026091119565668100_asag047-B1","doi-asserted-by":"crossref","first-page":"458","DOI":"10.1007\/11503415_31","volume-title":"Learning Theory","author":"Achlioptas","year":"2005"},{"key":"2026091119565668100_asag047-B2","doi-asserted-by":"crossref","first-page":"3099","DOI":"10.1214\/09-AOS689","article-title":"Identifiability of parameters in latent structure models with many observed variables","volume":"37","author":"Allman","year":"2009","journal-title":"Ann. Statist."},{"key":"2026091119565668100_asag047-B3","doi-asserted-by":"crossref","first-page":"37","DOI":"10.1007\/978-3-642-32512-0_4","volume-title":"Approximation, Randomization, and Combinatorial Optimization. Algorithms and Techniques","author":"Awasthi","year":"2012"},{"key":"2026091119565668100_asag047-B4","doi-asserted-by":"crossref","first-page":"77","DOI":"10.1214\/16-AOS1435","article-title":"Statistical guarantees for the EM algorithm: from population to sample-based analysis","volume":"45","author":"Balakrishnan","year":"2017","journal-title":"Ann. Statist."},{"key":"2026091119565668100_asag047-B5","doi-asserted-by":"crossref","first-page":"1234","DOI":"10.1214\/18-AOS1711","article-title":"CHIME: clustering of high-dimensional Gaussian mixtures with EM algorithm and its optimality","volume":"47","author":"Cai","year":"2019","journal-title":"Ann. Statist."},{"key":"2026091119565668100_asag047-B6","doi-asserted-by":"crossref","first-page":"221","DOI":"10.1214\/aos\/1176324464","article-title":"Optimal rate of convergence for finite mixture models","volume":"23","author":"Chen","year":"1995","journal-title":"Ann. Statist."},{"key":"2026091119565668100_asag047-B7","author":"Chen","year":"2024"},{"key":"2026091119565668100_asag047-B8","first-page":"203","article-title":"A probabilistic analysis of EM for mixtures of separated, spherical Gaussians","volume":"8","author":"Dasgupta","year":"2007","journal-title":"J. Mach. Learn. Res"},{"key":"2026091119565668100_asag047-B9","author":"Daskalakis","year":"2017"},{"key":"2026091119565668100_asag047-B10","doi-asserted-by":"crossref","first-page":"1","DOI":"10.1111\/j.2517-6161.1977.tb01600.x","article-title":"Maximum likelihood from incomplete data via the EM algorithm","volume":"39","author":"Dempster","year":"1977","journal-title":"J. R. Statist. Soc. B"},{"key":"2026091119565668100_asag047-B11","doi-asserted-by":"crossref","first-page":"62","DOI":"10.1214\/22-AOS2207","article-title":"Optimal estimation of high-dimensional Gaussian location mixtures","volume":"51","author":"Doss","year":"2023","journal-title":"Ann. Statist."},{"key":"2026091119565668100_asag047-B12","doi-asserted-by":"crossref","first-page":"3161","DOI":"10.1214\/19-AOS1924","article-title":"Singularity, misspecification and the convergence rate of EM","volume":"48","author":"Dwivedi","year":"2020","journal-title":"Ann. Statist."},{"key":"2026091119565668100_asag047-B13","doi-asserted-by":"crossref","first-page":"1066","DOI":"10.1214\/21-AOS2140","article-title":"Iterative algorithm for discrete structure recovery","volume":"50","author":"Gao","year":"2022","journal-title":"Ann. Statist."},{"key":"2026091119565668100_asag047-B14","doi-asserted-by":"crossref","first-page":"1233","DOI":"10.1214\/aos\/1013203452","article-title":"Entropies and rates of convergence for maximum likelihood and Bayes estimation for mixtures of normal densities","volume":"29","author":"Ghosal","year":"2001","journal-title":"Ann. Statist."},{"key":"2026091119565668100_asag047-B15","doi-asserted-by":"crossref","first-page":"2726","DOI":"10.1214\/16-AOS1444","article-title":"Convergence rates of parameter estimation for some weakly identifiable finite mixtures","volume":"44","author":"Ho","year":"2016","journal-title":"Ann. Statist."},{"key":"2026091119565668100_asag047-B16","doi-asserted-by":"crossref","first-page":"271","DOI":"10.1214\/16-EJS1105","article-title":"On strong identifiability and convergence rates of parameter estimation in finite mixture models","volume":"10","author":"Ho","year":"2016","journal-title":"Electron. J. Statist."},{"key":"2026091119565668100_asag047-B17","author":"Hsu","year":"2013"},{"key":"2026091119565668100_asag047-B18","author":"Jin","year":"2016"},{"key":"2026091119565668100_asag047-B19","author":"Kalai","year":"2010"},{"key":"2026091119565668100_asag047-B20","doi-asserted-by":"crossref","first-page":"1141","DOI":"10.1137\/S0097539704445925","article-title":"The spectral method for general mixture models","volume":"38","author":"Kannan","year":"2008","journal-title":"SIAM J. Comp"},{"key":"2026091119565668100_asag047-B21","author":"Kwon","year":"2020"},{"key":"2026091119565668100_asag047-B22","volume-title":"Mixture Models: Theory, Geometry, and Applications","author":"Lindsay","year":"1995"},{"key":"2026091119565668100_asag047-B23","doi-asserted-by":"crossref","first-page":"468","DOI":"10.1080\/01621459.1993.10476297","article-title":"Multivariate normal mixtures: a fast consistent method of moments","volume":"88","author":"Lindsay","year":"1993","journal-title":"J. Am. Statist. Assoc"},{"key":"2026091119565668100_asag047-B24","author":"Lu","year":"2016"},{"key":"2026091119565668100_asag047-B25","volume-title":"Mixture Models: Inference and Applications to Clustering","author":"McLachlan","year":"1988"},{"key":"2026091119565668100_asag047-B26","author":"Moitra","year":"2010"},{"key":"2026091119565668100_asag047-B27","doi-asserted-by":"crossref","first-page":"1114","DOI":"10.3390\/f16071114","article-title":"The pyrogeography of the Gran Chaco\u2019s dry forest: a comparison of clustering algorithms and the scale of analysis","volume":"16","author":"Naval-Fern\u00e1ndez","year":"2025","journal-title":"Forests"},{"key":"2026091119565668100_asag047-B28","doi-asserted-by":"crossref","first-page":"2096","DOI":"10.1214\/22-AOS2178","article-title":"Sharp optimal recovery in the two component Gaussian mixture model","volume":"50","author":"Ndaoud","year":"2022","journal-title":"Ann. Statist."},{"key":"2026091119565668100_asag047-B30","first-page":"267","article-title":"Mixture models: inference and applications to clustering","volume":"82","author":"Redner","year":"1984","journal-title":"J. Am. Statist. Assoc"},{"key":"2026091119565668100_asag047-B31","author":"Regev","year":"2017"},{"key":"2026091119565668100_asag047-B32","author":"Sanjeev","year":"2001"},{"key":"2026091119565668100_asag047-B33","doi-asserted-by":"crossref","first-page":"339","DOI":"10.1007\/978-3-031-82484-5_25","volume-title":"Machine Learning, Optimization, and Data Science","author":"Schwab","year":"2025"},{"key":"2026091119565668100_asag047-B34","doi-asserted-by":"crossref","first-page":"4510","DOI":"10.1214\/21-EJS1905","article-title":"Improved convergence guarantees for learning Gaussian mixture models by EM and gradient EM","volume":"15","author":"Segol","year":"2021","journal-title":"Electron. J. Statist."},{"key":"2026091119565668100_asag047-B35","doi-asserted-by":"crossref","first-page":"1265","DOI":"10.1214\/aoms\/1177703862","article-title":"Identifiability of finite mixtures","volume":"34","author":"Teicher","year":"1963","journal-title":"Ann. Math. Statist."},{"key":"2026091119565668100_asag047-B36","volume-title":"Statistical Analysis of Finite Mixture Distributions","author":"Titterington","year":"1985"},{"key":"2026091119565668100_asag047-B37","author":"Wang","year":"2026"},{"key":"2026091119565668100_asag047-B38","first-page":"1","article-title":"The EM algorithm is adaptively-optimal for unbalanced symmetric Gaussian mixtures","volume":"23","author":"Weinberger","year":"2022","journal-title":"J. Mach. Learn. Res"},{"key":"2026091119565668100_asag047-B39","doi-asserted-by":"crossref","first-page":"95","DOI":"10.1214\/aos\/1176346060","article-title":"On the convergence properties of the EM algorithm","volume":"11","author":"Wu","year":"1983","journal-title":"Ann. Statist."},{"key":"2026091119565668100_asag047-B40","doi-asserted-by":"crossref","first-page":"1981","DOI":"10.1214\/19-AOS1873","article-title":"Optimal estimation of Gaussian mixtures via denoised method of moments","volume":"48","author":"Wu","year":"2020","journal-title":"Ann. Statist."},{"key":"2026091119565668100_asag047-B41","doi-asserted-by":"crossref","first-page":"143","DOI":"10.4171\/msl\/29","article-title":"Randomly initialized EM algorithm for two-component Gaussian mixture achieves near optimality in $ o(\\surd{n}) $ iterations","volume":"4","author":"Wu","year":"2021","journal-title":"Math. Statist. Learn"},{"key":"2026091119565668100_asag047-B42","author":"Xu","year":"2016"},{"key":"2026091119565668100_asag047-B43","doi-asserted-by":"crossref","first-page":"129","DOI":"10.1162\/neco.1996.8.1.129","article-title":"On convergence properties of the EM algorithm for Gaussian mixtures","volume":"8","author":"Xu","year":"1996","journal-title":"Neural Comp"},{"key":"2026091119565668100_asag047-B44","doi-asserted-by":"crossref","first-page":"209","DOI":"10.1214\/aoms\/1177698520","article-title":"On the identifiability of finite mixtures","volume":"39","author":"Yakowitz","year":"1968","journal-title":"Ann. Math. Statist."},{"key":"2026091119565668100_asag047-B45","author":"Yan","year":"2017"},{"key":"2026091119565668100_asag047-B46","doi-asserted-by":"crossref","first-page":"632","DOI":"10.1214\/19-EJS1660","article-title":"Statistical convergence of the EM algorithm on Gaussian mixture models","volume":"14","author":"Zhao","year":"2020","journal-title":"Electron. J. Statist."}],"container-title":["Biometrika"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/academic.oup.com\/biomet\/advance-article-pdf\/doi\/10.1093\/biomet\/asag047\/68738637\/asag047.pdf","content-type":"application\/pdf","content-version":"am","intended-application":"syndication"},{"URL":"https:\/\/academic.oup.com\/biomet\/article-pdf\/113\/3\/asag047\/68738637\/asag047.pdf","content-type":"application\/pdf","content-version":"vor","intended-application":"syndication"},{"URL":"https:\/\/academic.oup.com\/biomet\/article-pdf\/113\/3\/asag047\/68738637\/asag047.pdf","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,9,11]],"date-time":"2026-09-11T23:57:07Z","timestamp":1789171027000},"score":1,"resource":{"primary":{"URL":"https:\/\/academic.oup.com\/biomet\/article\/doi\/10.1093\/biomet\/asag047\/8729049"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2026]]},"references-count":45,"journal-issue":{"issue":"3","published-print":{"date-parts":[[2026,7,6]]}},"URL":"https:\/\/doi.org\/10.1093\/biomet\/asag047","relation":{},"ISSN":["0006-3444","1464-3510"],"issn-type":[{"value":"0006-3444","type":"print"},{"value":"1464-3510","type":"electronic"}],"subject":[],"published-other":{"date-parts":[[2026]]},"published":{"date-parts":[[2026]]},"article-number":"asag047"}}