MS39 Surrogate Models for Uncertainty Quantification

Associate Prof. Jean-Marc Bourinet: bourinet@sigma-clermont.fr

Session Chairs:
Jean-Marc Bourinet, Associate Professor, Sigma-Clermont, France, bourinet@sigma-clermont.fr
Sankaran Mahadevan, Professor, Vanderbilt University, USA, sankaran.mahadevan@vanderbilt.edu
Bruno Sudret, Professor, ETH Zurich, Switzerland, sudret@ethz.ch
Alexandros Taflanidis, Associate Professor, University of Notre Dame, USA, a.taflanidis@nd.edu

Abstract of the special session:
The use of surrogate models in uncertainty quantification has become quite popular in recent years. Such techniques are applied for the safe and optimal design of engineering structures and systems, whose analysis involves costly-to-evaluate physics-based computational models (e.g., finite element models). The aim of this mini-symposium is to confront various types of surrogate-modeling techniques in the context of uncertainty quantification (e.g. polynomial chaos expansions, kriging, support vector regression, neural networks, ...).
Papers that present new methodology developments (e.g. adaptive enrichment strategies, dimension reduction, multi-fidelity models, surrogate-based sampling, ...) as well as applications to problems representative of industrial ones are welcome.
This mini-symposium will discuss the applicability of the proposed approaches to challenging problems (e.g. high dimensional input spaces, complex physical models) and their efficiency/robustness/limits with respect to other reference techniques.