Low-Rank Models and Applications

Explore cutting-edge research in low-rank modeling through this comprehensive mini-symposium featuring 15 specialized talks on advanced mathematical and computational techniques. Delve into diverse applications including linear dynamical systems with Hankel nuclear norm regularization, over-parameterized tensor decomposition beyond lazy training frameworks, and generalized compressed sensing with novel measurement matrices. Examine dual principal component pursuit methods, multi-channel linear convolutional networks from a function space perspective, and data-driven approaches to dynamic interpolation and approximation. Investigate tomographic imaging under model uncertainty, rigidity theory for Gaussian graphical models, and non-separable relaxations of rank penalty classes. Learn about robust low-rank matrix completion using alternating manifold proximal gradient continuation methods, computational barriers in estimation from low-degree polynomials, and principal component analysis for high-dimensional heteroscedastic data. Discover connections between PCA, double descent phenomena, and Gaussian processes, while exploring sample optimal algorithms for low-rank approximation of positive semidefinite and distance matrices, concluding with techniques for imputing missing data using low-rank Gaussian copula models.