Skip to main content

EM for mixtures

Initialization requires special care


Maximum likelihood through the EM algorithm is widely used to estimate the parameters in hidden structure models such as Gaussian mixture models. But the EM algorithm has well-documented drawbacks: its solution could be highly dependent from its initial position and it may fail as a result of degeneracies. We stress the practical dangers of theses limitations and how carefully they should be dealt with. Our main conclusion is that no method enables to address them satisfactory in all situations. But improvements are introduced, first, using a penalized log-likelihood of Gaussian mixture models in a Bayesian regularization perspective and, second, choosing the best among several relevant initialisation strategies. In this perspective, we also propose new recursive initialization strategies which prove helpful. They are compared with standard initialization procedures through numerical experiments and their effects on model selection criteria are analyzed.

This is a preview of subscription content, access via your institution.

Fig. 1
Fig. 2
Fig. 3
Fig. 4
Fig. 5
Fig. 6
Fig. 7
Fig. 8
Fig. 9
Fig. 10
Fig. 11
Fig. 12


  1. Banfield, J.D., Raftery, A.E.: Model-based Gaussian and non-Gaussian clustering. Biometrics 49, 803–821 (1993)

    MATH  MathSciNet  Article  Google Scholar 

  2. Baudry, J.-P.: Sélection de modèle pour la classification non supervisée. Choix du nombre de classes. PhD thesis, Université Paris-Sud (2009)

  3. Baudry, J.-P., Maugis, C., Michel, B.: Slope heuristics: overview and implementation. Stat. Comput. 22, 455–470 (2011)

    MathSciNet  Article  Google Scholar 

  4. Berchtold, A.: Optimisation of mixture models: comparison of different strategies. Comput. Stat. 19, 385–406 (2004)

    MATH  MathSciNet  Google Scholar 

  5. Biernacki, C., Celeux, G., Govaert, G.: Assessing a mixture model for clustering with the integrated completed likelihood. IEEE Trans. Pattern Anal. Mach. Intell. 22, 719–725 (2000)

    Article  Google Scholar 

  6. Biernacki, C., Celeux, G., Govaert, G.: Choosing starting values for the EM algorithm for getting the highest likelihood in multivariate gaussian mixture models. Comput. Stat. Data Anal. 41, 561–575 (2003)

    MATH  MathSciNet  Article  Google Scholar 

  7. Birgé, L., Massart, P.: Minimal penalties for Gaussian model selection. Probab. Theory Relat. Fields 138, 33–73 (2007)

    MATH  Article  Google Scholar 

  8. Celeux, G., Govaert, G.: A classification EM algorithm for clustering and two stochastic versions. Comput. Stat. Data Anal. 14, 315–332 (1992)

    MATH  MathSciNet  Article  Google Scholar 

  9. Celeux, G., Govaert, G.: Parsimonious Gaussian models in cluster analysis. Pattern Recognit. 28, 781–793 (1995)

    Article  Google Scholar 

  10. Ciuperca, G., Ridolfi, A., Idier, J.: Penalized maximum likelihood estimator for normal mixtures. Scand. J. Stat. 30, 45–59 (2003)

    MATH  MathSciNet  Article  Google Scholar 

  11. Dempster, A.P., Laird, N.M., Rubin, D.B.: Maximum likelihood from incomplete data via the EM algorithm. J. R. Stat. Soc. Ser. B (Methodological) 39(1), 1–38 (1977)

    MATH  MathSciNet  Google Scholar 

  12. Fraley, C., Raftery, A., Wehrens, R.J.: Incremental model-based clustering for large datasets with small clusters. J. Comput. Graph. Stat. 14, 529–546 (2005)

    MathSciNet  Article  Google Scholar 

  13. Fraley, C., Raftery, A.E.: Model-based clustering, discriminant analysis and density estimation. J. Am. Stat. Assoc. 97, 611–631 (2002)

    MATH  MathSciNet  Article  Google Scholar 

  14. Fraley, C., Raftery, A.E.: Bayesian regularization for normal mixture estimation and model-based clustering. J. Classif. 24, 155–181 (2007)

    MATH  MathSciNet  Article  Google Scholar 

  15. Frazee, A.C., et al.: ReCount: a multi-experiment resource of analysis-ready RNA-seq gene count datasets. BMC Bioinform. 12, 449 (2011)

    Article  Google Scholar 

  16. Graveley, B.R., et al.: The development transcriptome of Drosophila melanogaster. Nature 471, 473–479 (2011)

    Article  Google Scholar 

  17. Keribin, C.: Consistent estimation of the order of mixture models. Sankhya A 62(1), 49–66 (2000)

    MATH  MathSciNet  Google Scholar 

  18. McLachlan, G., Krishnan, T.: The EM Algorithm and Extensions, 2nd edn. Wiley, Hoboken (2008)

    MATH  Book  Google Scholar 

  19. McLachlan, G.J., Peel, D.: Finite Mixture Models. Wiley, New York (2000)

    MATH  Book  Google Scholar 

  20. Papastamoulis, P., Martin-Magniette, M.-L., Maugis-Rabusseau, C.: On the estimation of mixtures of poisson regression models with large numbers of components. Computat. Stat. Data Anal. (to appear) (2014)

  21. Pelleg, D., Moore, A.W.: X-means: Extending k-means with efficient estimation of the number of clusters. In: Langley, P. (ed.) ICML, pp. 727–734. Morgan Kaufmann (2000)

  22. Rau, A., Maugis-Rabusseau, C., Martin-Magniette, M.-L., Celeux, G.: Co-expression analysis of high-throughput transcriptome sequencing data with Poisson mixture models. Bioinformatics. (to appear) (2015)

  23. Roeder, K., Wasserman, L.: Practical Bayesian density estimation using mixtures of normals. J. Am. Stat. Assoc. 92, 894–902 (1997)

    MATH  MathSciNet  Article  Google Scholar 

  24. Schwarz, G.: Estimating the dimension of a model. Ann. Stat. 6, 461–464 (1978)

    MATH  Article  Google Scholar 

Download references

Author information



Corresponding author

Correspondence to Gilles Celeux.

Rights and permissions

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

Baudry, JP., Celeux, G. EM for mixtures. Stat Comput 25, 713–726 (2015).

Download citation


  • Gaussian mixture models
  • EM algorithm
  • Initialization strategies
  • Recursive initialization
  • Regularized likelihood
  • Model selection criteria