Leo Duan
Statistician · Associate Professor, University of Florida
About
Hi there, thanks for stopping by! My name is Leo Duan. I am a professional statistician and an (amateur) photographer.
Education and Experience
- 2025–presentAssociate Professor, Department of Statistics, University of Florida
- PresentAffiliate Faculty, McKnight Brain Institute, University of Florida
- 2018–2025Assistant Professor, Department of Statistics, University of Florida
- 2016–2018Postdoctoral Associate in Statistics, Duke University (Mentor: David Dunson)
- 2011–2015Ph.D. in Mathematics, University of Cincinnati (Advisors: Rhonda Szczesniak, Xia Wang)
- 2009B.S., Sichuan University (with honors)
Research Interests
My research focuses on the interface between optimization and Bayes.
The goal of our research group is to develop new statistical and machine learning toolboxes that just work — simple implementation, fast computation, and principled uncertainty quantification.
Achieving this goal is not trivial, which is why we’ve been working to show how one can borrow strengths from two communities: Bayes, who are good at probabilistic modeling, and optimization, who are good at efficient computing. Check out our work on building optimization-based priors, optimization-based likelihoods, and optimization-based generative models, and feel free to reach out with any questions.
My statistical interest is largely motivated by ongoing collaborative work in neuroscience, engineering, forensics, and data privacy.
Stance on AI. Embrace AI for inside-the-box tasks + check against slop. Work AI-unplugged for creativity. Reject AI spoon-feeding. Enjoy stats and math on our own.
Recent Talks
- September 2026Statistical Modeling of Combinatorial Response Data: An Inspiration from Albert and Chib (1993)
Université Paris Dauphine-PSL, PariSanté Campus - June 2026Graph-event Modeling for Inference on Persistent Homology
ISBA World Meeting, Nagoya - June 2026Bayesian Distance-to-Set Models
Seminar, Department of Decision Sciences, Bocconi University, Milan - April 2026Bayesian Distance-to-Set Models: from Latent Variable to Latent Projection
Seminar, University of Texas at Austin