Multimodal information gain in Bayesian design of experiments
From MaRDI portal
Publication:2135895
Abstract: One of the well-known challenges in optimal experimental design is how to efficiently estimate the nested integrations of the expected information gain. The Gaussian approximation and associated importance sampling have been shown to be effective at reducing the numerical costs. However, they may fail due to the non-negligible biases and the numerical instabilities. A new approach is developed to compute the expected information gain, when the posterior distribution is multimodal - a situation previously ignored by the methods aiming at accelerating the nested numerical integrations. Specifically, the posterior distribution is approximated using a mixture distribution constructed by multiple runs of global search for the modes and weighted local Laplace approximations. Under any given probability of capturing all the modes, we provide an estimation of the number of runs of searches, which is dimension independent. It is shown that the novel global-local multimodal approach can be significantly more accurate and more efficient than the other existing approaches, especially when the number of modes is large. The methods can be applied to the designs of experiments with both calibrated and uncalibrated observation noises.
Recommendations
- Informational energy in the sequential design of experiments in a Bayesian context
- Bayesian Experimental Design for Multiple Hypothesis Testing
- Integrated Bayesian Experimental Design
- Properties of Fisher information gain for Bayesian design of experiments
- Bayesian experimental design for models with intractable likelihoods
- Bayesian optimal designs for multi-factor nonlinear models
- Efficient Bayesian experimentation using an expected information gain lower bound
Cites work
- A Laplace method for under-determined Bayesian optimal experimental designs
- A Laplace-based algorithm for Bayesian adaptive design
- A Repelling–Attracting Metropolis Algorithm for Multimodality
- A-optimal design of experiments for infinite-dimensional Bayesian linear inverse problems with regularized _0-sparsification
- Accurate Approximations for Posterior Moments and Marginal Densities
- Bayesian Experimental Design: A Review
- Efficient D-optimal design of experiments for infinite-dimensional Bayesian linear inverse problems
- Fast Bayesian experimental design: Laplace-based importance sampling for the expected information gain
- Fast Bayesian optimal experimental design for seismic source inversion
- Fast estimation of expected information gains for Bayesian experimental designs based on Laplace approximations
- Fully Exponential Laplace Approximations to Expectations and Variances of Nonpositive Functions
- scientific article; zbMATH DE number 54240 (Why is no real title available?)
- Introduction to Bayesian Scientific Computing
- Large deviations techniques and applications.
- Large Sample Properties of Simulations Using Latin Hypercube Sampling
- Miscellanea. A note on Bayesian design for the normal linear model with unknown error variance
- Multilevel Monte Carlo estimation of expected information gains
- Nesterov-aided stochastic gradient methods using Laplace approximation for Bayesian design optimization
- On Information and Sufficiency
- On the convergence of the Laplace approximation and noise-level-robustness of Laplace-based Monte Carlo methods for Bayesian inverse problems
- Optimal Bayesian experimental design for subsurface flow problems
- Optimal design: Some geometrical aspects of D-optimality
- Optimum experimental designs, with SAS
- Taylor approximation and variance reduction for PDE-constrained optimal control under uncertainty
Cited in
(9)- Properties of Fisher information gain for Bayesian design of experiments
- Approximate Laplace importance sampling for the estimation of expected Shannon information gain in high-dimensional Bayesian design for nonlinear models
- Fast estimation of expected information gains for Bayesian experimental designs based on Laplace approximations
- Small-noise approximation for Bayesian optimal experimental design with nuisance uncertainty
- Modern Bayesian experimental design
- Accurate, scalable, and efficient Bayesian optimal experimental design with derivative-informed neural operators
- Laplace-based strategies for Bayesian optimal experimental design with nuisance uncertainty
- Gradient-free sequential Bayesian experimental design via interacting particle systems
- Natural variational annealing for multimodal optimization
This page was built for publication: Multimodal information gain in Bayesian design of experiments
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q2135895)
