Learning from Data

Much current CCE research lies at the intersection of physical modeling with data-driven methods. These efforts include developing new methods for inverse problems, data assimilation, and broader problems in computational statistics. Learning from data is also fundamental to creating predictive “digital twins” of physical systems. CCE researchers also work in machine learning (ML), exploiting important connections between modern ML approaches and scientific computing. For example, CCE research is linking large-scale optimization methods and numerical algorithms for differential equations to many topics in learning and high-dimensional statistics.