×

Convergence of the Monte Carlo expectation maximization for curved exponential families. (English) Zbl 1043.62015

Summary: The Monte Carlo expectation maximization (MCEM) algorithm is a versatile tool for inference in incomplete data models, especially when used in combination with Markov chain Monte Carlo simulation methods. In this contribution, the almost-sure convergence of the MCEM algorithm is established. It is shown, using uniform versions of ergodic theorems for Markov chains, that MCEM converges under weak conditions on the simulation kernel. Practical illustrations are presented, using a hybrid random walk Metropolis Hastings sampler and an independence sampler. The rate of convergence is studied, showing the impact of the simulation schedule on the fluctuation of the parameter estimate at the convergence. A novel averaging procedure is then proposed to reduce the simulation variance and increase the rate of convergence.

MSC:

62F12 Asymptotic properties of parametric estimators
65C05 Monte Carlo methods
60J10 Markov chains (discrete-time Markov processes on discrete state spaces)
65C40 Numerical analysis or methods applied to Markov chains
65C60 Computational problems in statistics (MSC2010)
PDFBibTeX XMLCite
Full Text: DOI Euclid

References:

[1] BISCARAT, J. (1994). Almost sure convergence of a class of stochastic algorithms. Stochastic Process. Appl. 50 83-94. · Zbl 0819.60033 · doi:10.1016/0304-4149(94)90149-X
[2] BOOTH, J. and HOBERT, J. (1999). Maximizing generalized linear mixed model likelihoods with an automated Monte Carlo EM algorithm. J. R. Stat. Soc. Ser. B Stat. Methodol. 61 265-285. · Zbl 0917.62058 · doi:10.1111/1467-9868.00176
[3] BRANDIÈRE, O. (1998). The dy namic sy stem method and the traps. Adv. in Appl. Probab. 30 137-151. · Zbl 0980.62067
[4] BRÖCKER, T. (1975). Differentiable Germs and Catastrophes. Cambridge Univ. Press. · Zbl 0302.58006
[5] CAPPÉ, O., DOUCET, A., LAVIELLE, M. and MOULINES, E. (1999). Simulation-based methods for blind maximum-likelihood filter identification. Signal Processing 73 3-25. · Zbl 0926.93066 · doi:10.1016/S0165-1684(98)00182-0
[6] CELEUX, G. and DIEBOLT, J. (1992). A stochastic approximation ty pe EM algorithm for the mixture problem. Stochastics Stochastics Rep. 41 119-134. · Zbl 0766.62050
[7] CHAN, J. and KUK, A. (1997). Maximum likelihood estimation for probit-linear mixed models with correlated random effects. Biometrics 53 86-97. JSTOR: · Zbl 1065.62503 · doi:10.2307/2533099
[8] CHAN, K. and LEDOLTER, J. (1995). Monte Carlo EM estimation for time series models involving counts. J. Amer. Statist. Assoc. 90 242-252. JSTOR: · Zbl 0819.62069 · doi:10.2307/2291149
[9] CHEN, H., GUO, L. and GAO, A. (1988). Convergence and robustness of the Robbins-Monro algorithm truncated at randomly varying bounds. Stochastic Process. Appl. 27 217-231. · Zbl 0632.62082 · doi:10.1016/0304-4149(87)90039-1
[10] DALAL, S. and WEERAHANDI, S. (1992). Some approximations for the moments of a process used in diffusion of new products. Statist. Probab. Lett. 15 181-189. · Zbl 0850.62194 · doi:10.1016/0167-7152(92)90187-A
[11] DALAL, S. and WEERAHANDI, S. (1995). Estimation of innovation diffusion models with application to a consumer durable. Marketing Letters 6 123-136.
[12] DELy ON, B., LAVIELLE, M. and MOULINES, E. (1999). Convergence of a stochastic approximation version of the EM algorithm. Ann. Statist. 27 94-128. · Zbl 0932.62094 · doi:10.1214/aos/1018031103
[13] DEMPSTER, A. P., LAIRD, N. M. and RUBIN, D. B. (1977). Maximum likelihood from incomplete data via the EM algorithm (with discussion). J. Roy. Statist. Soc. Ser. B 39 1-38. JSTOR: · Zbl 0364.62022
[14] FORT, G., MOULINES, E., ROBERTS, G. O. and ROSENTHAL, J. S. (2003). On the geometric ergodicity of hy brid samplers. J. Appl. Probab. 40 123-146. · Zbl 1028.65002 · doi:10.1239/jap/1044476831
[15] GUO, S. and THOMPSON, E. (1991). Monte-Carlo estimation of variance component models for large complex pedigrees. IMA Journal of Mathematics Applied in Medicine and Biology 8 171-189. · Zbl 0739.62081
[16] HALL, P. and HEy DE, C. C. (1980). Martingale Limit Theory and Its Application. Academic Press, New York. · Zbl 0462.60045
[17] JAMSHIDIAN, M. and JENNRICH, R. (1997). Acceleration of the EM algorithm by using quasiNewton methods. J. R. Stat. Soc. Ser. B Stat. Methodol. 59 569-587. JSTOR: · Zbl 0889.62042 · doi:10.1111/1467-9868.00083
[18] MENG, X. and SCHILLING, S. (1996). Fitting full-information item factor models and an empirical investigation of bridge sampling. J. Amer. Statist. Assoc. 91 1254-1267. · Zbl 0925.62220 · doi:10.2307/2291744
[19] MEy N, S. P. and TWEEDIE, R. L. (1993). Markov Chains and Stochastic Stability. Springer, London. · Zbl 0925.60001
[20] PIERRE-LOTI-VIAUD, D. (1995). Random perturbations of recursive sequences with an application to an epidemic model. J. Appl. Probab. 32 559-578. JSTOR: · Zbl 0841.58058 · doi:10.2307/3215113
[21] PÓLy A, G. and SZEGÖ, G. (1976). Problems and Theorems in Analy sis 2. Springer, New York.
[22] POLy AK, B. (1990). New stochastic approximation ty pe procedures. Automat. Remote Control 51 937-946.
[23] SHAPIRO, A. and WARDI, Y. (1996). Convergence analysis of stochastic algorithms. Math. Oper. Res. 21 615-628. JSTOR: · Zbl 0873.93084 · doi:10.1007/BF02190104
[24] SHERMAN, R., HO, Y. and DALAL, S. (1999). Conditions for convergence of Monte Carlo EM sequences with an application to product diffusion modeling. Econom. J. 2 248-267. · Zbl 0955.91050 · doi:10.1111/1368-423X.00032
[25] TANNER, M. A. (1996). Tools for Statistical Inference, 3rd ed. Springer, New York. · Zbl 0846.62001
[26] WEI, G. and TANNER, M. (1990). A Monte-Carlo implementation of the EM algorithm and the poor man’s data augmentation algorithms. J. Amer. Statist. Assoc. 85 699-704.
[27] WU, C. (1983). On the convergence properties of the EM algorithm. Ann. Statist. 11 95-103. · Zbl 0517.62035 · doi:10.1214/aos/1176346060
[28] ZEGER, S. L. (1988). A regression model for time series of counts. Biometrika 75 621-629. JSTOR: · Zbl 0653.62064 · doi:10.1093/biomet/75.4.621
This reference list is based on information provided by the publisher or from digital mathematics libraries. Its items are heuristically matched to zbMATH identifiers and may contain data conversion errors. In some cases that data have been complemented/enhanced by data from zbMATH Open. This attempts to reflect the references listed in the original paper as accurately as possible without claiming completeness or a perfect matching.