Workshop on Statistical Inverse Problems

Location:

This workshop took place at SAMSI in Research Triangle Park, NC.

Description:

This workshop is intended to bring together active participants of Working Group I, The Statistical Inverse Problems group, which is run as a part of the 2016-2017 Program on Optimization.The areas of particular interest include sampling techniques for parameter estimation for large-scale Bayesian inverse problems, quantification of uncertainty, and optimal design of experiments for Bayesian inverse problems governed by Partial Differential Equations (PDE). Target applications include a wide class of problems ranging from the geosciences to medical imaging.

There are two main themes of this workshop:

  1. Optimal experimental design, which seeks to control experimental parameters to maximize the information gain about the estimated parameters of interest, subject to budget or physical constraints.
  2. Novel sampling techniques and the use of reduced order models to effectively sample high-dimensional distributions.

Schedule and Supporting Media

Speaker Titles/Abstracts

Thursday, January 26, 2017
SAMSI

Description Speaker Slides Videos
Registration
Opening Remarks Ilse Ipsen, SAMSI Assoc. Director  PDF
Plenary:Computational Methodologies for Large Data Assimilation Problems Adrian Sandu, Virginia Tech  PDF
Stochastic Newton and Quasi-Newton Methods for Large Linear Least-squares Problems Mathias Chung, Virginia Tech
Goal-Oriented Optimal Experimental Design Ahmed Attia, SAMSI Post Doc  PDF
Optimal Experimental Design for Constrained Inverse Problems Lars Ruthotto, Emory University
Gaussian Scale Mixtures for Inverse Problems in Imaging Dirk Lorenz, Technische Universität Braunschweig PDF

Friday, January 27, 2017
SAMSI

Description Speaker Slides Videos
Plenary:Markov Chain Monte Carlo Algorithms for Linear Inverse Problems John Bardsley, U of Montana  PDF
Hybrid Iterative Methods for Large-Scale Bayesian Inverse Problems Julianne Chung, Virginia Tech
Computationally Efficient Markov Chain Monte Carlo Methods for Hierarchical Bayesian Inverse Problems Andrew Brown, Clemson  PDF
Mitigating the Influence of the Boundary on PDE-based Covariance Operators Georg Stadler, New York University
FINAL REMARKS Alen Alexanderian and Arvind Saibaba, N.C. State University

Questions: email opt@samsi.info