A rare event approach to high dimensional Approximate Bayesian computation

Dennis Prangle,Richard G. Everitt,Theodore Kypraios

DOI: https://doi.org/10.48550/arXiv.1611.02492

2017-04-04

Abstract:Approximate Bayesian computation (ABC) methods permit approximate inference for intractable likelihoods when it is possible to simulate from the model. However they perform poorly for high dimensional data, and in practice must usually be used in conjunction with dimension reduction methods, resulting in a loss of accuracy which is hard to quantify or control. We propose a new ABC method for high dimensional data based on rare event methods which we refer to as RE-ABC. This uses a latent variable representation of the model. For a given parameter value, we estimate the probability of the rare event that the latent variables correspond to data roughly consistent with the observations. This is performed using sequential Monte Carlo and slice sampling to systematically search the space of latent variables. In contrast standard ABC can be viewed as using a more naive Monte Carlo estimate. We use our rare event probability estimator as a likelihood estimate within the pseudo-marginal Metropolis-Hastings algorithm for parameter inference. We provide asymptotics showing that RE-ABC has a lower computational cost for high dimensional data than standard ABC methods. We also illustrate our approach empirically, on a Gaussian distribution and an application in infectious disease modelling.

Computation

What problem does this paper attempt to address?

The problem that this paper attempts to solve is the difficulties encountered when using the Approximate Bayesian Computation (ABC) method for statistical inference under high - dimensional data. Specifically, the standard ABC method performs poorly when dealing with high - dimensional data because high - dimensional data makes "nearest neighbor" matching extremely rare, which is known as the "curse of dimensionality". To overcome this challenge, the author proposes a new ABC method based on the rare - event method, called RE - ABC (Rare Event - ABC). This method estimates the probability that the latent variable corresponds to data that is approximately consistent with the observed data under a given parameter value by introducing latent variables and systematically searching the latent variable space using Sequential Monte Carlo (SMC) and slice sampling techniques. This probability estimate is used as the likelihood estimate in the pseudo - marginal Metropolis - Hastings algorithm for parameter inference.

In short, the core problem of the paper is to improve the efficiency and accuracy of the ABC method when dealing with high - dimensional data, and this goal is achieved by developing a new method based on rare events.