Abstract 

Scientific Machine Learning (SciML) is an emerging research area that applies machine learning techniques to solve scientific computing problems, such as numerically solving forward and inverse problems associated with partial differential equations. In this accessible talk, we will present the main SciML techniques and discuss how they can be applied not only to problems in numerical analysis but also to statistical learning based on prior information. In particular, we will introduce physics-informed neural networks which, unlike traditional deep neural networks, have greater interpretability and can be verified and validated for use in critical domains. Additionally, we will present examples of how SciML can be applied in epidemiology to forecast disease spread. With this talk, we aim to demonstrate the relevance of emerging SciML techniques for a modern training in Applied Mathematics, Statistics, and Computer Science.

Zoom

Recording