Optimal projection of observations in a Bayesian setting

DOI10.1016/J.CSDA.2018.03.002MaRDI QIDQ1662871zbMATH OpenOpenAlexWikidataFDO

Authors L. Giraldi, O. Le Maître, I. Hoteit, Omar M. Knio

Publication date 20 August 2018

Published in Computational Statistics and Data Analysis (Search for Journal in Brave)

Full work available at URL https://arxiv.org/abs/1709.06606, http://hdl.handle.net/10754/627354

zbMATH Keywords

information theoryGaussian linear modeloptimal data reductionoptimal dimensionality reduction

Mathematics Subject Classification ID

Computational methods for problems pertaining to statistics (62-08) Bayesian inference (62F15)

Abstract: Optimal dimensionality reduction methods are proposed for the Bayesian inference of a Gaussian linear model with additive noise in presence of overabundant data. Three different optimal projections of the observations are proposed based on information theory: the projection that minimizes the Kullback-Leibler divergence between the posterior distributions of the original and the projected models, the one that minimizes the expected Kullback-Leibler divergence between the same distributions, and the one that maximizes the mutual information between the parameter of interest and the projected observations. The first two optimization problems are formulated as the determination of an optimal subspace and therefore the solution is computed using Riemannian optimization algorithms on the Grassmann manifold. Regarding the maximization of the mutual information, it is shown that there exists an optimal subspace that minimizes the entropy of the posterior distribution of the reduced model; a basis of the subspace can be computed as the solution to a generalized eigenvalue problem; an a priori error estimate on the mutual information is available for this particular solution; and that the dimensionality of the subspace to exactly conserve the mutual information between the input and the output of the models is less than the number of parameters to be inferred. Numerical applications to linear and nonlinear models are used to assess the efficiency of the proposed approaches, and to highlight their advantages compared to standard approaches based on the principal component analysis of the observations.

Recommendations

• scientific article; zbMATH DE number 2119842
• Goal-oriented optimal approximations of Bayesian linear inverse problems
• Certified Dimension Reduction for Bayesian Updating with the Cross-Entropy Method
• Likelihood-informed dimension reduction for nonlinear inverse problems
• Model reduction of linear dynamical systems via balancing for Bayesian inference

Cites work

• scientific article; zbMATH DE number 486467 (Why is no real title available?)
• scientific article; zbMATH DE number 5223994 (Why is no real title available?)
• scientific article; zbMATH DE number 3340881 (Why is no real title available?)
• scientific article; zbMATH DE number 3046994 (Why is no real title available?)
• A Survey of Some Closed Methods for Inverting Matrices
• A weighted empirical interpolation method: a priori convergence analysis and applications
• Accurate Approximations for Posterior Moments and Marginal Densities
• Algorithm AS 136: A K-Means Clustering Algorithm
• An `empirical interpolation' method: Application to efficient reduced-basis discretization of partial differential equations
• Approximation of boundary element matrices
• Bayesian inference of earthquake parameters from buoy data using a polynomial chaos-based surrogate
• Characteristic Classes. (AM-76)
• Comparison of Some Reduced Representation Approximations
• Fast Monte Carlo Algorithms for Matrices I: Approximating Matrix Multiplication
• Goal-oriented optimal approximations of Bayesian linear inverse problems
• Likelihood-informed dimension reduction for nonlinear inverse problems
• Optimal low-rank approximations of Bayesian linear inverse problems
• Principal component analysis.
• Pymanopt: a Python toolbox for optimization on manifolds using automatic differentiation
• Random projections for Bayesian regression
• Riemannian geometry of Grassmann manifolds with a view on algorithmic computation
• Spectral Methods for Uncertainty Quantification
• Tensor spaces and numerical tensor calculus
• Trust-region methods on Riemannian manifolds

Cited in

(5)

• Toward Bayesian Data Compression
• Optimal experimental design: formulations and computations
• Model reduction of linear dynamical systems via balancing for Bayesian inference
• Parametric inference in the large data limit using maximally informative models
• scientific article; zbMATH DE number 2119842 (Why is no real title available?)

Describes a project that uses

Uses Software

• Autograd
• Pymanopt
• AS 136

This page was built for publication: Optimal projection of observations in a Bayesian setting

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q1662871)