Abstract:
|
Approximate Bayesian computation (ABC) methods perform inference on model specific parameters of mechanistically motivated parametric models when evaluating likelihoods is difficult. Central to the success of ABC methods is computationally inexpensive simulation of data sets from the parametric model of interest. However, when simulating data sets from a model is so computationally expensive that the posterior distribution of parameters cannot be adequately sampled by ABC, inference is not straightforward. We present a class of computationally fast inference methods that extends ABC to models in which simulating data is expensive. We show that under mild assumptions, the posterior distribution obtained by our method converges to the posterior distribution obtained by ABC, as the number of data sets simulated from the parametric model and the sample size of the observed data set increase. We demonstrate the performance on a model of the admixture history of hybrid populations. We illustrates how, in population genetics, our method is of particular utility in scenarios that rely on conceptually straightforward but potentially slow forward-in-time simulations.
|
ASA Meetings Department
732 North Washington Street, Alexandria, VA 22314
(703) 684-1221 • meetings@amstat.org
Copyright © American Statistical Association.