Welcome to the Workshop on Uncertainty Quantification and Data-Driven Modeling

The purpose of this workshop is to bring together leading experts in uncertainty quantification, statistics, computer science, and computational science to discuss new research ideas in data-driving modeling.  With UQ now established as a core area of computational science and engineering, data science is quickly emerging as a critical accompanying technology area for enabling validated, predictive simulations.  However, data science is still in its infancy and new ideas are needed, especially in the context of UQ for multi-scale, multi-physics science and engineering applications.

Wide-ranging physical phenomena, from tubulent flows through materials physics, exhibit emergent behavior that can only be understood and predicted by detailed multi-scale models.  Existing multi-scale models based on first principles equations, and physical approximations thereof, tend to be either prohibitively expensive or inaccurate.  However, recent advances at the intersection of data and computational science have created new opportunities to meet the long-standing challenge of delivering quantitative predictivity in multi-scale computational physics.  This has led to a new paradigm of data-driven model development.  In addition to classical methods from inverse problems, this new framework brings to bear Bayesian inference, stochastic simulation, reduced order modelling, compressed sensing as well as advances in machine learning.  Of particular interest are approaches by which the above and related techniques can enhance the predictivity of large scale computational physics.

Through this workshop, we hope to highlight new advances in UQ and data-driven modeling that will lead to breakthroughs in our ability to develop predictive models for realistic applications.  Areas of interest include, but are not limited to

• Deterministic and stochastic inverse problems
• Data assimilation methodologies, particularly in the context of datasets with a high degree of noise and/or poorly characterized uncertainties
• Machine learning and data mining methods to inform model development
• Inference of models, including parameters and functional forms
• Validation and data-informed predictions for science and engineering applications
• Uncertainty quantification for large-scale, high-dimensional problems

The organization of this workshop is coordinated through the USACM Technical Thrust Area on Uncertainty Quantification and Probabilistic Analysis.