10 Abr 2019
Salón de Grados, Departamental 2
Physics-based Deep Learning – Can Computers Learn Physics by Example?
Nils Thuerey is probably the most successful researcher combining animation and modern machine learning. In this talk he will focus on the possibilities that arise from recent advances in the area of deep learning for accelerating and improving physics simulations. He will focus on fluids, which encompass a large class of materials we encounter in our everyday lives. In addition to being ubiquitous, the underlying physical model, the Navier-Stokes equations, at the same time represent a challenging, non-linear advection-diffusion PDE that poses interesting challenges for deep learning methods.
He will explain and discuss several recent research projects from that focus on temporal predictions of physical functions, temporally coherent adversarial training, and predictions of steady-state turbulence solutions. Among other things, it turns out to be useful to make the learning process aware of the underlying physical principles. E.g., the transport component of the Navier-Stokes equations play a crucial role. I will also give an outlook about open challenges in the area of deep learning for general physics problems. Among others, trained models could server as priors for a variety of inverse problems and control tasks.