Using machine learning to extract coarse-scale PDEs from fine-scale data

Abstract

Combining techniques such as Gaussian processes and neural networks with feature selection and manifold learning approaches yields broad possibilities for improved data-driven modeling.