|DATE||July 31 (Fri), 2020|
|TITLE||Data-driven modeling for stochastic systems with machine learning|
Models of physical systems typically involve uncertainty in the input data such as those associated with coefficients initial or boundary conditions, geometry, etc. Estimating the propagation of this uncertainty into computational model output predictions is crucial to provide more instight to the true physics and produce predictions with high fidelity. This often leads to solve partial differential equations with many parameters. We discuss machine learning based algorithms to solve parametric PDEs and present numerical examples to demonstrate the effectiveness of the proposed methods.