Learning Linear Dynamical Systems with Hankel Nuclear Norm Regularization

Explore the intricacies of learning linear dynamical systems from input-output data in this 35-minute lecture by Maryam Fazel from the University of Washington. Delve into the challenges of system identification with limited output samples, a crucial step in control and policy decision problems across various domains. Examine recent developments in finite-sample statistical analysis using least-squares regression, and investigate the benefits of adding a Hankel nuclear norm regularizer when seeking low-order systems. Gain new insights into the sample complexity of this regularized scheme and its practical applications in engineering. Part of the Mini-symposium on Low-Rank Models and Applications at the Fields Institute, this talk provides a comprehensive overview of cutting-edge research in linear dynamical systems and their identification.

Learning Linear Dynamical Systems with Hankel Nuclear Norm Regularization