Yale Institute for Foundations of Data Science, Kline Tower 13th Floor, Room 1327, New Haven, CT 06511
Speaker: Dr. Yuejie Chi
Sense of Wonder Group, Endowed Professor in AI Systems,
Department of Electrical and Computer Engineering,
Machine Learning Department and CyLab (by courtesy),
Carnegie Mellon University
Wednesday, February 21, 2024
Lunch: 11:30 am (Kitchen)
Talk: 12:00 pm (Seminar Room #1327)
at the Yale institute for Foundations of Data Science, Kline Tower, 13th Floor
Virtual link for this talk:
Title: Learning from Data with Low-rank Priors and Beyond
Abstract: Generative priors are effective countermeasures to combat the curse of dimensionality, and enable efficient learning and inversion that otherwise are ill-posed, in data science. This talk begins with the classical low-rank prior, and introduces scaled gradient descent (ScaledGD), a simple iterative approach to directly recover the low-rank factors for a wide range of matrix and tensor estimation tasks. ScaledGD provably converges linearly at a constant rate independent of the condition number at near-optimal sample complexities, while maintaining the low per-iteration cost of vanilla gradient descent. In addition, ScaledGD continues to admit fast global convergence from a small random initialization when the rank is over-specified, confirming the benign generalization of ScaledGD in learning overparameterized models. Going beyond low rank, the talk discusses diffusion models as a promising data prior in inverse problems, and highlights some ongoing efforts from algorithmic foundations to applications in materials science.
Bio: Dr. Yuejie Chi is the Sense of Wonder Group Endowed Professor of Electrical and Computer Engineering in AI Systems at Carnegie Mellon University, with courtesy appointments in the Machine Learning department and CyLab. She received her Ph.D. and M.A. from Princeton University, and B. Eng. (Hon.) from Tsinghua University, all in Electrical Engineering. Her research interests lie in the theoretical and algorithmic foundations of data science, signal processing, machine learning and inverse problems, with applications in sensing, imaging, decision making, and generative AI. Among others, Dr. Chi is a recipient of the Presidential Early Career Award for Scientists and Engineers (PECASE), the inaugural IEEE Signal Processing Society Early Career Technical Achievement Award for contributions to high-dimensional structured signal processing, and multiple paper awards including the SIAM Activity Group on Imaging Science Best Paper Prize and IEEE Signal Processing Society Young Author Best Paper Award. She is an IEEE Fellow (Class of 2023) for contributions to statistical signal processing with low-dimensional structures.