# Approximate Bayesian inference from noisy likelihoods with Gaussian process emulated MCMC

@inproceedings{Jarvenpaa2021ApproximateBI, title={Approximate Bayesian inference from noisy likelihoods with Gaussian process emulated MCMC}, author={Marko Jarvenpaa and Jukka Corander}, year={2021} }

We present an efficient approach for doing approximate Bayesian inference when only a limited number of noisy likelihood evaluations can be obtained due to computational constraints, which is becoming increasingly common for applications of complex models. Our main methodological innovation is to model the log-likelihood function using a Gaussian process (GP) in a local fashion and apply this model to emulate the progression that an exact Metropolis-Hastings (MH) algorithm would take if it was… Expand

#### Figures from this paper

#### One Citation

A survey of Monte Carlo methods for noisy and costly densities with application to reinforcement learning

- Computer Science, Mathematics
- ArXiv
- 2021

This survey gives an overview of Monte Carlo methodologies using surrogate models, for dealing with densities which are intractable, costly, and/or noisy, and a modular scheme which encompasses the considered methods is presented. Expand

#### References

SHOWING 1-10 OF 71 REFERENCES

Bayesian Synthetic Likelihood

- Mathematics
- 2017

ABSTRACT Having the ability to work with complex models can be highly beneficial. However, complex models often have intractable likelihoods, so methods that involve evaluation of the likelihood… Expand

The Bayesian choice

- Computer Science
- 1994

This paperback edition, a reprint of the 2001 edition, is a graduate-level textbook that introduces Bayesian statistics and decision theory and was awarded the 2004 DeGroot Prize for setting a new standard for modern textbooks dealing with Bayesian methods. Expand

A table of normal integrals

- Mathematics
- 1980

Integrals of functions of the univariate, bivariate, trivariate and multivariate normal densities are given. Both indefinite and definite integrals are included.

Batch simulations and uncertainty quantification in Gaussian process surrogate approximate Bayesian computation

- Mathematics, Computer Science
- UAI
- 2020

B batch-sequential Bayesian experimental design strategies to parallellise the expensive simulations and a numerical method to fully quantify the uncertainty in, for example, ABC posterior moments is proposed. Expand

Variational Bayesian Monte Carlo

- Computer Science, Mathematics
- NeurIPS
- 2018

A novel sample-efficient inference framework, Variational Bayesian Monte Carlo (VBMC), which combines variational inference with Gaussian-process based, active-sampling Bayesian quadrature, using the latter to efficiently approximate the intractable integral in the variational objective. Expand

On Markov chain Monte Carlo methods for tall data

- Computer Science, Mathematics
- J. Mach. Learn. Res.
- 2017

An original subsampling-based approach is proposed which samples from a distribution provably close to the posterior distribution of interest, yet can require less than $O(n)$ data point likelihood evaluations at each iteration for certain statistical models in favourable scenarios. Expand

A general framework for updating belief distributions

- Mathematics, Medicine
- Journal of the Royal Statistical Society. Series B, Statistical methodology
- 2016

It is argued that a valid update of a prior belief distribution to a posterior can be made for parameters which are connected to observations through a loss function rather than the traditional likelihood function, which is recovered as a special case. Expand

GPS-ABC: Gaussian Process Surrogate Approximate Bayesian Computation

- Computer Science, Biology
- UAI
- 2014

This work develops two new ABC sampling algorithms that significantly reduce the number of simulations necessary for posterior inference and stores the information obtained from every simulation in a Gaussian process which acts as a surrogate function for the simulated statistics. Expand

Likelihood ridges and multimodality in population growth rate models.

- Computer Science, Medicine
- Ecology
- 2009

Using the theta-Ricker model as a simple but flexible description of density dependence, theory and simulations are applied to show how multimodality and ridges in the likelihood surface can emerge even in the absence of model misspecification or observation error. Expand