Statistics and Computing

, Volume 22, Issue 5, pp 1009–1020 | Cite as

An adaptive sequential Monte Carlo method for approximate Bayesian computation

Article

Abstract

Approximate Bayesian computation (ABC) is a popular approach to address inference problems where the likelihood function is intractable, or expensive to calculate. To improve over Markov chain Monte Carlo (MCMC) implementations of ABC, the use of sequential Monte Carlo (SMC) methods has recently been suggested. Most effective SMC algorithms that are currently available for ABC have a computational complexity that is quadratic in the number of Monte Carlo samples (Beaumont et al., Biometrika 86:983–990, 2009; Peters et al., Technical report, 2008; Toni et al., J. Roy. Soc. Interface 6:187–202, 2009) and require the careful choice of simulation parameters. In this article an adaptive SMC algorithm is proposed which admits a computational complexity that is linear in the number of samples and adaptively determines the simulation parameters. We demonstrate our algorithm on a toy example and on a birth-death-mutation model arising in epidemiology.

Keywords

Approximate Bayesian computation Markov chain Monte Carlo Sequential Monte Carlo 

Preview

Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.

References

  1. Andrieu, C., Johanes, T.: A tutorial on adaptive Markov chain Monte Carlo methods. Stat. Comput. 18, 343–373 (2008) MathSciNetCrossRefGoogle Scholar
  2. Andrieu, C., Berthelesen, K., Doucet, A., Roberts, G.O.: The expected auxiliary variable method for Monte Carlo simulation. Working paper (2008) Google Scholar
  3. Beaumont, M.A., Cornuet, J.M., Marin, J.M., Robert, C.P.: Adaptive approximate Bayesian computation. Biometrika 86, 983–990 (2009) MathSciNetCrossRefGoogle Scholar
  4. Beskos, A., Crisan, D., Jasra, A.: On the stability of sequential Monte Carlo methods in high dimensions. Technical Report, Imperial College London (2011) Google Scholar
  5. Bortot, P., Coles, S., Sisson, S.: Inference for stereological extremes. J. Am. Stat. Assoc. 102, 84–92 (2009) MathSciNetCrossRefGoogle Scholar
  6. Chopin, N.: A sequential particle filter for static models. Biometrika 89, 539–552 (2002) MathSciNetMATHCrossRefGoogle Scholar
  7. Del Moral, P., Doucet, A., Jasra, A.: Sequential Monte Carlo samplers. J. R. Stat. Soc. B 68, 411–436 (2006) MATHGoogle Scholar
  8. Del Moral, P., Doucet, A., Jasra, A.: On adaptive resampling strategies for sequential Monte Carlo methods. Bernoulli (2011, to appear) Google Scholar
  9. Drovandi, C.C., Pettit, A.N.: Estimation of parameters for macroparasite population evolution using approximate Bayesian computation. Biometrics 67, 225–233 (2011) MathSciNetMATHCrossRefGoogle Scholar
  10. Fearnhead, P., Prangle, D.: Semi-automatic approximate Bayesian computation. Technical Report, University of Lancaster (2010) Google Scholar
  11. Gilks, W.R., Berzuini, C.: Following a moving target—Monte Carlo inference for dynamic Bayesian models. J. R. Stat. Soc. B 63, 127–146 (2001) MathSciNetMATHGoogle Scholar
  12. Gramacy, R., Samworth, R., King, R.: Importance tempering. Stat. Comput. 20, 1–7 (2010) MathSciNetCrossRefGoogle Scholar
  13. Grelaud, A.: Méthodes sans vraisemblance appliquées à l’étude de la sélection naturelle et la prédiction de structure tridimensionnelle des protéines. Ph.D. Thesis, Université Paris-Dauphine (2009) Google Scholar
  14. Kitagawa, G.: Monte Carlo filter and smoother for non-Gaussian nonlinear state space models. J. Comput. Graph. Stat. 5, 1–25 (1996) MathSciNetGoogle Scholar
  15. Liu, J.S.: Monte Carlo Strategies in Scientific Computing. Springer, New York (2001) MATHGoogle Scholar
  16. Majoram, P., Molitor, J., Plagnol, V., Tavaré, S.: Markov chain Monte Carlo without likelihoods. Proc. Natl. Acad. Sci. 100, 15324–15328 (2003) CrossRefGoogle Scholar
  17. Marin, J.-M., Pudlo, P., Robert, C.P., Ryder, R.: Approximate Bayesian computational methods. Stat. Comput. (2011, to appear) Google Scholar
  18. Neal, R.M.: Annealed importance sampling. Stat. Comput. 11, 125–139 (2001) MathSciNetCrossRefGoogle Scholar
  19. Peters, G.W., Fan, Y., Sisson, S.A.: On sequential Monte Carlo, partial rejection control and approximate Bayesian computation. Technical report, University of New South Wales (2008) Google Scholar
  20. Sisson, S., Fan, Y., Tanaka, M.M.: Sequential Monte Carlo without likelihoods. Proc. Natl. Acad. Sci. 104, 1760–1765 (2007) MathSciNetMATHCrossRefGoogle Scholar
  21. Sisson, S., Fan, Y., Tanaka, M.M.: Correction: Sequential Monte Carlo without likelihoods. Proc. Natl. Acad. Sci. 106, 1760 (2009) CrossRefGoogle Scholar
  22. Small, P.M., Hopewell, P., Singh, P., Paz, A., Parsonnet, J., et al.: The epidemiology of tuberculosis in San Francisco: a population-based study using conventional and molecular methods. N. Engl. J. Med. 330, 1703–1709 (1994) CrossRefGoogle Scholar
  23. Tanaka, M.M., Francis, A.R., Luciani, F., Sisson, A.: Using approximate computation to estimate tuberculosis transmission parameters from genotype data. Genetics 173, 1511–1520 (2006) CrossRefGoogle Scholar
  24. Toni, T., Welch, D., Strelkowa, N., Ipsen, A., Stumpf, M.P.H.: Approximate Bayesian computation scheme for parameter inference and model selection in dynamical systems. J. R. Soc. Interface 6, 187–202 (2009) CrossRefGoogle Scholar

Copyright information

© Springer Science+Business Media, LLC 2011

Authors and Affiliations

  1. 1.Centre INRIA Bordeaux Sud-Ouest & Institut de MathématiquesUniversité Bordeaux ITalence cedexFrance
  2. 2.Department of StatisticsUniversity of OxfordOxfordUK
  3. 3.Department of Statistics and Applied ProbabilityNational University of SingaporeSingaporeSingapore

Personalised recommendations