A course of 21 hours on "Sparsity and Big Data in Control, Systems Identification and Machine Learning", taught by Mario Sznaier (Northeastern Univ., USA), is offered as part of the European Embedded Control Institute (EECI) International Graduate School on Control, will be held on april 29 and 30 and may 2nd and 3rd, 2024 in Toulouse, France.
One of the hardest challenges faced by the systems community stems from the exponential explosion of data, fueled by recent advances in sensing technology. During the past few years a large research effort has been devoted to developing computationally tractable methods that seek to mitigate the ``curse of dimensionality" by exploiting sparsity.
The goals of this course are:
1) provide a quick introduction to the subject for people in the systems community faced with ``big data" and scaling
problems ;
2) serve as a ``quick reference" guide for researchers, summarizing the state of the art.
Part I of the course covers the issue of handling large data sets and sparsity priors in systems identification, model (in)validation and control. presenting recently developed techniques that exploit a deep connection to semi-algebraic geometry, rank minimization and matrix completion.
Part II of the course focuses on applications, including control and filter design subject to information flow constraints, subspace clustering and classification on Riemannian manifolds, and time-series classification, including activity recognition and anomaly detection
