Machine Learning for Scientific Discovery, with Examples in Fluid Mechanics

Explore the application of machine learning in developing accurate and efficient nonlinear dynamical systems models for complex natural and engineered systems in this 58-minute lecture by Steve Brunton from the University of Washington. Delve into the sparse identification of nonlinear dynamics (SINDy) algorithm, which creates minimal dynamical system models balancing complexity and accuracy while avoiding overfitting. Discover how this approach promotes interpretable and generalizable models that capture essential system "physics." Learn about the importance of effective coordinate systems for sparse dynamics and see demonstrations of this sparse modeling approach in challenging fluid dynamics problems. Understand how to incorporate these models into existing model-based control efforts, with emphasis on interpretable, explainable, and generalizable machine learning solutions that respect known physics, particularly in fluid dynamics applications central to transportation, health, and defense systems.