On Unifying Randomized Methods For Inverse Problems

Abstract

This work unifies the analysis of various randomized methods for solving linear and nonlinear inverse problems by framing the problem in a stochastic optimization setting. By doing so, we show that many randomized methods are variants of a sample average approximation. More importantly, we are able to prove a single theoretical result that guarantees the asymptotic convergence for a variety of randomized methods. Additionally, viewing randomized methods as a sample average approximation enables us to prove, for the first time, a single non-asymptotic error result that holds for randomized methods under consideration. Another important consequence of our unified framework is that it allows us to discover new randomization methods. We present various numerical results for linear, nonlinear, algebraic, and PDE-constrained inverse problems that verify the theoretical convergence results and provide a discussion on the apparently different convergence rates and the behavior for various randomized methods.

References in corpus (14)

• The Ensemble Kalman Filter for Inverse Problems
• Stein Variational Gradient Descent: A General Purpose Bayesian Inference Algorithm
• Dimension-independent likelihood-informed MCMC
• Covariance Estimation: The GLM and Regularization Perspectives
• Likelihood-informed dimension reduction for nonlinear inverse problems
• On the well-posedness of Bayesian inverse problems
• Solving Large-Scale PDE-constrained Bayesian Inverse Problems with Riemann Manifold Hamiltonian Monte Carlo
• Sub-Weibull distributions: generalizing sub-Gaussian and sub-Exponential properties to heavier-tailed distributions
• Stein Variational Gradient Descent Without Gradient
• Projected Stein Variational Gradient Descent
• Γ-convergence of Onsager-Machlup functionals. Part I: With applications to maximum a posteriori estimation in Bayesian inverse problems
• Sharper Sub-Weibull Concentrations
• A Data-Scalable Randomized Misfit Approach for Solving Large-Scale PDE-Constrained Inverse Problems
• Simultaneous shot inversion for nonuniform geometries using fast data interpolation