Brett Larsen

Headshot.jpg

I am a Research Scientist at Databricks Mosaic Research and Guest Researcher at the Flatiron Institue and NYU. I completed my PhD in the Neural Dynamics and Computation Lab at Stanford University co-advised by Surya Ganguli and Shaul Druckmann and supported by the DOE Computational Science Graduate Fellowship. Through this program, I also worked in Sandia National Labs’ Extreme-Scale Data Science group as a visiting researcher, leading to frequent collaborations with Tammy Kolda.

My research combines theoretical and empirical approaches to understand and improve deep learning, low-rank models, and representations in the brain. This has led to the opportunity to work with many excellent collaborators, including Jonathan Frankle, Karolina Dziugaite, Mansheej Paul, Cody Blakeney, Stanislav Fort, and Alex Williams.

Prior to my PhD, I completed the MPhil in Scientific Computing (Machine Learning Focus) at the University of Cambridge working with Carola Bibiane Schönlieb and Martin Benning. I did my undergrad at Arizona State University and Barrett, the Honors College where I worked on statistical signal processing.

Contact: brettlarsen [at] flatironinstitute.org

Interests

  • Machine Learning and Data Science
  • High-Dimensional Loss Landscapes
  • Numerical Optimization
  • Randomized Algorithms
  • Low-Rank Models
  • Robust Statistics

Education

  • PhD in Physics, 2022

    MS in Statistics, 2020

    Stanford University

  • MPhil in Scientific Computing, 2016

    University of Cambridge

  • BSE in Electrical Engineering, 2015

    BS in Physics, 2015

    Arizona State University (ASU)