Wu Lin

Research Machine Learning | Learn Math and Physics

Email: yorker.lin@gmail.com  ·  Google Scholar  ·  GitHub

About Me

I am working on computational aspects of differential, geometric, and algebraic structures (i.e., probability distributions and matrices). My research so far has mostly focused on geometric methods for numerical optimization and approximate inference in machine learning.

Working Experiences

• Institute of Artificial Intelligence, University of Central Florida, Orlando (2026 - Present)
  Assistant Professor in Computer Science

• Vector Institute for Artificial Intelligence, Toronto (2023 - 2026)
  Distinguished Postdoctoral Fellow (Hosts: Dr. Roger B. Grosse and Dr. Alireza Makhzani)

• Department of CS, University of British Columbia, Vancouver (2018 - 2023)
  Ph.D in Machine Learning (Advisors: Dr. Mark Schmidt and Dr. Emtiyaz Khan)

• Center for Advanced Intelligence Project, RIKEN, Tokyo (2017)
  Technical Staff (Advisor: Dr. Emtiyaz Khan)

Research Interests

I am interested in exploiting (hidden) structures and symmetries in machine learning with a focus on practical and numerical methods for optimization and statistical inference.

Selected Papers

• Improving optimizers Shampoo and SOAP via natural-gradient descent (ICLR 2026): Paper, Code
• Adaptive gradient methods as natural-gradient descent (ICML 2024): Paper, Code
• Natural-gradient descent for exponential-family mixtures (ICML 2019): Paper, Code

• Natural-gradient descent for Bayesian deep learning (ICML 2018): Paper, Code

• Natural-gradient variational inference for non-conjugate models (AI&Stats 2017): Paper, Code

For more publications, see my Google Scholar page. For an introduction to natural-gradient methods, see my Blog Posts.