Open Access
June 2019 CHIME: Clustering of high-dimensional Gaussian mixtures with EM algorithm and its optimality
T. Tony Cai, Jing Ma, Linjun Zhang
Ann. Statist. 47(3): 1234-1267 (June 2019). DOI: 10.1214/18-AOS1711


Unsupervised learning is an important problem in statistics and machine learning with a wide range of applications. In this paper, we study clustering of high-dimensional Gaussian mixtures and propose a procedure, called CHIME, that is based on the EM algorithm and a direct estimation method for the sparse discriminant vector. Both theoretical and numerical properties of CHIME are investigated. We establish the optimal rate of convergence for the excess misclustering error and show that CHIME is minimax rate optimal. In addition, the optimality of the proposed estimator of the discriminant vector is also established. Simulation studies show that CHIME outperforms the existing methods under a variety of settings. The proposed CHIME procedure is also illustrated in an analysis of a glioblastoma gene expression data set and shown to have superior performance.

Clustering of Gaussian mixtures in the conventional low-dimensional setting is also considered. The technical tools developed for the high-dimensional setting are used to establish the optimality of the clustering procedure that is based on the classical EM algorithm.


Download Citation

T. Tony Cai. Jing Ma. Linjun Zhang. "CHIME: Clustering of high-dimensional Gaussian mixtures with EM algorithm and its optimality." Ann. Statist. 47 (3) 1234 - 1267, June 2019.


Received: 1 October 2017; Published: June 2019
First available in Project Euclid: 13 February 2019

zbMATH: 07053507
MathSciNet: MR3911111
Digital Object Identifier: 10.1214/18-AOS1711

Primary: 62G15
Secondary: 62C20 , 62H35

Keywords: EM algorithm , Gaussian mixture model , High-dimensional data , Minimax optimality , misclustering error , unsupervised learning

Rights: Copyright © 2019 Institute of Mathematical Statistics

Vol.47 • No. 3 • June 2019
Back to Top