Spring 2022

Ulisses M. Braga-Neto, Ph.D.

Physics-Informed Neural Networks

Abstract: Scientific Machine Learning (SciML) has emerged recently as a promising development in
deep learning. Unlike traditional machine learning, SciML methods can extrapolate beyond the
available data, by using physical laws encoded in partial differential equations (PDEs). Physics-Informed
Neural Networks (PINNs) is the best-known SciML algorithm, having been deployed in a variety of
scientific and engineering problems with remarkable success. In this talk, we will introduce PINNs and
describe our progress in training methods to improve the convergence of PINNs in the solution of
difficult "stiff" nonlinear PDEs. We also intend to describe the activities and ongoing multidisciplinary
collaborative efforts at the recently-established Scientific Machine Learning Lab at the Texas A&M
Institute of Data Science (TAMIDS).

Seminar Zoom Session