Research & PresentationsPublications and PreprintsM. Celentano, A. Montanari, Y. Wei. The Lasso with general Gaussian designs with applications to hypothesis testing. 2020+. (paper) M. Celentano, A. Montanari, Y. Wu. The estimation error of general first order methods. COLT, 2020. (paper) M. Celentano. Approximate separability of symmetrically penalized least squares in high dimensions: characterization and consequences. In submission, 2019+. (paper) M. Celentano, A. Montanari. Fundamental barriers to high-dimensional regression with convex penalties. In submission, 2019+. (paper) PresentationsIT Forum, Stanford University. Fundamental barriers to estimation in high-dimensions. April 3, 2020. (video) (slides) CS Theory Seminar, Stanford University. Fundamental barriers to tractable estimation in high-dimensions. April 22, 2020. (slides) (These slides are similar to the IT Forum slides, but contain several new slides on proof techniques) Online Open Probability School, Problem Session Leader. Mean field methods in high-dimensional statistics and nonconvex optimization. July 7 & 8, 2020. I led two problem solving sessions for a five lecture course given by Andrea Montanari. (Session 1 Handout) (Session 2 Handout) Conference on Learning Theory. The estimation error of general first order methods. July 2020. (video) (slides) Bernoulli-IMS One World Symposium. The Lasso with general Gaussian designs. August 2020. (video) (slides) International Seminar on Selective Inference. The Lasso with general Gaussian designs with applications to hypothesis testing. September 10, 2020. (video) (slides) |