Approximate Bayesian Computation

About the Project

Approximate Bayesian computation (ABC) is a paradigm which allows one to perform Bayesian statistical inference, when the associate probability is totally intractable in a certain manner. In order to conduct this afore-mentioned inference, one must resort to Monte Carlo estimation.

A/P Jasra and Dr Jo are currently developing a method which allows Monte Carlo estimation in the context of ABC, which permits a reduction in computational effort, for a given level of error in the Monte Carlo estimate. Their methodology relies upon the important multilevel Monte Carlo method coupled with advanced sequential Monte Carlo approximation techniques.

Team
Related Publications
Ajay Jasra, Seongil Jo, David Nott, Christine Shoemaker, Raul Tempone
In the following article we consider approximate Bayesian computation (ABC) inference. We introduce a method for numerically approximating ABC posteriors using the multilevel Monte Carlo (MLMC). A sequential Monte Carlo version of the approach is developed and it is shown under some assumptions that for a given level of…
Stochastic Analysis and Applications
Ajay Jasra, Seongil Jo, David Nott, Christine Shoemaker, Raul Tempone
In the following article we consider approximate Bayesian computation (ABC) inference. We introduce a method for numerically approximating ABC posteriors using the multilevel Monte Carlo (MLMC). A sequential Monte Carlo version of the approach is developed and it is shown under some assumptions that for a given level of…
Stochastic Analysis and Applications