Ensemble inference methods for models with noisy and expensive likelihoods
From MaRDI portal
Publication:5090110
Abstract: The increasing availability of data presents an opportunity to calibrate unknown parameters which appear in complex models of phenomena in the biomedical, physical and social sciences. However, model complexity often leads to parameter-to-data maps which are expensive to evaluate and are only available through noisy approximations. This paper is concerned with the use of interacting particle systems for the solution of the resulting inverse problems for parameters. Of particular interest is the case where the available forward model evaluations are subject to rapid fluctuations, in parameter space, superimposed on the smoothly varying large scale parametric structure of interest. {A motivating example from climate science is presented, and ensemble Kalman methods (which do not use the derivative of the parameter-to-data map) are shown, empirically, to perform well. Multiscale analysis is then used to analyze the behaviour of interacting particle system algorithms when rapid fluctuations, which we refer to as noise, pollute the large scale parametric dependence of the parameter-to-data map. Ensemble Kalman methods and Langevin-based methods} (the latter use the derivative of the parameter-to-data map) are compared in this light. The ensemble Kalman methods are shown to behave favourably in the presence of noise in the parameter-to-data map, whereas Langevin methods are adversely affected. On the other hand, Langevin methods have the correct equilibrium distribution in the setting of noise-free forward models, whilst ensemble Kalman methods only provide an uncontrolled approximation, except in the linear case. Therefore a new class of algorithms, ensemble Gaussian process samplers, which combine the benefits of both ensemble Kalman and Langevin methods, are introduced and shown to perform favourably.
Recommendations
- Derivative-Free Bayesian Inversion Using Multiscale Dynamics
- Interacting Langevin diffusions: gradient structure and ensemble Kalman sampler
- Ensemble Kalman filter based sequential Monte Carlo sampler for sequential Bayesian inference
- Ensemble Kalman inversion for general likelihoods
- Analysis of the ensemble Kalman filter for inverse problems
Cites work
- Accelerating MCMC via Kriging-based adaptive independent proposals and delayed rejection
- Affine invariant interacting Langevin dynamics for Bayesian inference
- An adaptive surrogate modeling based on deep neural networks for large-scale Bayesian inverse problems
- Asymptotic analysis for periodic structures
- Bayesian calibration of computer models. (With discussion)
- Calibrate, emulate, sample
- Combining Field Data and Computer Simulations for Calibration and Prediction
- Comparing dynamics: deep neural networks versus glassy systems
- Data assimilation: the Schrödinger perspective
- Delayed rejection in reversible jump Metropolis-Hastings.
- Derivative-Free Bayesian Inversion Using Multiscale Dynamics
- Design and analysis of computer experiments. With comments and a rejoinder by the authors
- Deterministic Nonperiodic Flow
- Discrete gradients for computational Bayesian inference
- Emulation of higher-order tensors in manifold Monte Carlo methods for Bayesian inverse problems
- Ensemble Kalman inversion: a derivative-free technique for machine learning tasks
- Ensemble Kalman methods for inverse problems
- Ensemble preconditioning for Markov chain Monte Carlo simulation
- Ensemble samplers with affine invariance
- Exponential convergence of Langevin distributions and their discrete approximations
- Extracting macroscopic dynamics: model problems and algorithms
- Fokker-Planck particle systems for Bayesian inference: computational approaches
- Interacting Langevin diffusions: gradient structure and ensemble Kalman sampler
- MCMC methods for functions: modifying old algorithms to make them faster
- Mean field limits for interacting diffusions in a two-scale potential
- Mean field limits for interacting diffusions with colored noise: phase transitions and spectral numerical methods
- Monte Carlo strategies in scientific computing.
- Multiscale Methods
- Numerical methods for multiscale inverse problems
- Parameter Estimation for Macroscopic Pedestrian Dynamics Models from Microscopic Data
- Pathwise accuracy and ergodicity of metropolized integrators for SDEs
- Probabilistic Sensitivity Analysis of Complex Models: A Bayesian Approach
- Rapid mixing for the Lorenz attractor and statistical limit laws for their time-1 maps
- Riemann manifold Langevin and Hamiltonian Monte Carlo methods. With discussion and authors' reply
- Sampling from rough energy landscapes
- Statistical and computational inverse problems.
- The pseudo-marginal approach for efficient Monte Carlo computations
- Wasserstein stability estimates for covariance-preconditioned Fokker–Planck equations
Cited in
(15)- Ensemble Kalman inversion for sparse learning of dynamical systems from time-averaged data
- Calibrate, emulate, sample
- Computational methods for parameter estimation in climate models
- scientific article; zbMATH DE number 3884009 (Why is no real title available?)
- Chapter 7: Ensemble variational methods
- Ensemble-Based Gradient Inference for Particle Methods in Optimization and Sampling
- Learning about structural errors in models of complex dynamical systems
- Less interaction with forward models in Langevin dynamics: enrichment and homotopy
- Sharp propagation of chaos for the ensemble Langevin sampler
- Ensemble Kalman methods: a mean-field perspective
- A survey of Monte Carlo methods for noisy and costly densities with application to reinforcement learning and ABC
- Hyperparameter optimization for randomized algorithms: a case study on random features
- Batch Bayesian optimization via particle gradient flows
- A localized consensus-based sampling algorithm
- Improved forecasting under imperfect posterior sampling
This page was built for publication: Ensemble inference methods for models with noisy and expensive likelihoods
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q5090110)
