Sparse Inference of Dynamical Systems

This lecture by the instructor covers a unified approach for sparse inference of dynamical systems from temporal measurements. It introduces the problem statement, types of dynamical systems, and the learning algorithm. The lecture discusses ODE, SDE, and PDE systems, along with the USDL algorithm for structure and parameter estimation. It explains the weak formulation and sparse signal recovery, highlighting the challenges and open issues in learning from populations and dealing with latent confounders. The presentation concludes with the USDL algorithm's benefits and theoretical metrics for accuracy assessment.