Theoretical and methodological aspects of MCMC computations with noisy likelihoods
Andrieu, C., Lee, A., & Vihola, M. (2018). Theoretical and methodological aspects of MCMC computations with noisy likelihoods. In S. A. Sisson, Y. Fan, & M. Beaumont (Eds.), Handbook of Approximate Bayesian Computation : Likelihood-Free Methods for Complex Model (pp. 243-268). Chapman and Hall/CRC.
Embargoed until: 2019-08-10Request copy from author
© the Authors & Chapman and Hall/CRC, 2018.
Approximate Bayesian computation (ABC) [11, 42] is a popular method for Bayesian inference involving an intractable, or expensive to evaluate, likelihood function but where simulation from the model is easy. The method consists of defining an alternative likelihood function which is also in general intractable but naturally lends itself to pseudo-marginal computations , hence making the approach of practical interest. The aim of this chapter is to show the connections of ABC Markov chain Monte Carlo with pseudo-marginal algorithms, review their existing theoretical results, and discuss how these can inform practice and hopefully lead to fruitful methodological developments.