Videos

Many of my talks are available for viewing online.

Institute for Mathematical and Statistical Innovation – Workshop on Data Assimilation and Inverse Problems, Fall 2025.
“Efficient Low-dimensional compression for deep overparameterized learning and fine-tuning.”

Institute for Mathematical and Statistical Innovation – Workshop on Reduced Order and Surrogate Modeling, Fall 2025.
“Optimization for Low-dimensional Modeling”

One World Mathematics of Information, Data, and Signals (1W-MINDS) Seminar, April 2024
“Efficient Low-dimensional compression for deep overparameterized learning and fine-tuning.”

Fields Institute Summer 2021: Heteroscedastic PCA

Rutgers Spring 2021: Heteroscedastic PCA and Variety Completion

IAS Fall 2020: Variety Completion

IST Lisbon Summer 2020: Preference Modeling with Context-Dependent Salient Features

IAS Winter 2020: Preference Modeling with Context-Dependent Salient Features

Simons Fall 2018: Heteroscedastic PCA

Simons Fall 2018: Subspace Clustering with Ensembles of K-Subspaces

Simons Winter 2017: Active Subspace Clustering

Summer 2014: Missing Data (the Big Data Dilemma) – 3 min overview

UM Stats Fall 2013: Learning Subspaces when Data are Missing and Corrupted

Spring 2012: High-rank Matrix Completion

Congratulations Soo Min!

July 1, 2026 By Laura Balzano

Congratulations Dr. Soo Min Kwon for successfully defending his PhD in May and for winning the AIStats CauScale Workshop best paper award!

Soo Min’s dissertation was titled “Deep Learning through Low-Dimensional Representations: Theory and Algorithms” and included a range of exciting results on theory of low-dimensional structure in deep learning, as well as a novel algorithm for learning Grassmannian geodesic structure from compressive measurements.

Soo Min was co-advised by my colleague Qing Qu. Another of Qing’s students, Alec Xu, was lead co-author on the paper that won a workshop best paper award. That paper studies a mathematical model for out-of-distribution generalization we sometimes see empirically with in-context learning. You can read about the best paper award here.

CPAL Rising Star

May 15, 2026 By Laura Balzano

Congratulations Can Yaras for receiving the CPAL (Conference on Parsimony and Learning) Rising Star recognition! Very well deserved!

Congratulations Javier!

May 1, 2026 By Laura Balzano

Today I hooded Dr. Javier Salazar Cavazos, who defended his thesis in March and today walked the stage at commencement. His dissertation is entitled “Learning Representations from Noisy Data and Brain Imaging: Subspace Modeling for Heteroscedastic Data and Deep Learning for Functional MRI in Alzheimer’s Disease,” and included his paper on ALPCAH, an algorithm for heteroscedastic subspace learning. Next he’ll be going to KLA. Congratulations Javier!