Incorporating Symmetry for Learning Spatiotemporal Dynamics - IPAM at UCLA

Explore the intersection of deep learning and physical principles in this lecture on incorporating symmetry for learning spatiotemporal dynamics. Delve into the challenge of integrating physical laws into deep learning models, focusing on Noether's Law and its connection between conserved quantities and symmetry groups. Discover techniques for building neural networks that respect given symmetries, enhancing physical consistency, sample efficiency, and generalization in spatiotemporal dynamics modeling. Gain insights from Rose Yu of the University of California, San Diego, as she presents at IPAM's Learning and Emergence in Molecular Systems Workshop, demonstrating how these approaches significantly improve the accuracy and reliability of predictions in scientific domains.