Multifidelity Approximate Bayesian Computation with Sequential Monte Carlo Parameter Sampling

Abstract

Multifidelity approximate Bayesian computation (MF-ABC) is a likelihood-free technique for parameter inference that exploits model approximations to significantly increase the speed of ABC algorithms (Prescott and Baker, 2020). Previous work has considered MF-ABC only in the context of rejection sampling, which does not explore parameter space particularly efficiently. In this work, we integrate the multifidelity approach with the ABC sequential Monte Carlo (ABC-SMC) algorithm into a new MF-ABC-SMC algorithm. We show that the improvements generated by each of ABC-SMC and MF-ABC to the efficiency of generating Monte Carlo samples and estimates from the ABC posterior are amplified when the two techniques are used together.

References in corpus (7)

• Approximate Bayesian computation scheme for parameter inference and model selection in dynamical systems
• Low Dimensional Behavior of Large Systems of Globally Coupled Oscillators
• Approximation and inference methods for stochastic biochemical kinetics - a tutorial review
• Rethinking the Effective Sample Size
• Multifidelity Approximate Bayesian Computation
• Optimal proposals for Approximate Bayesian Computation
• Multilevel Monte Carlo in Approximate Bayesian Computation