Coursera Flash Sale
40% Off Coursera Plus for 3 Months!
Grab it
Explore an innovative approach to system identification through invariant measures in this 37-minute conference talk from IPAM's Scientific Machine Learning Workshop. Learn about an alternative methodology that moves beyond traditional trajectory-based system identification methods, particularly valuable when dealing with infrequently sampled trajectory data that makes time derivative estimation challenging or impossible. Discover how physical measures of dynamical systems can be leveraged for system identification, including PDE-based approximation methods for physical measures and regularity results for optimal-transportation-based fidelity functions essential for efficient gradient-based optimization. Gain insights into the mathematical foundations underlying this approach and understand its practical applications in scenarios where conventional methods fall short. The presentation concludes with discussion of future research directions and open questions in this emerging field at the intersection of dynamical systems theory and machine learning.
