Jason Xu
Associate Professor of Biostatistics
University of California, Los Angeles
Email: jqxu at g dot ucla dot edu
© 2025 Jason Xu · All rights reserved ·
Research Interests
I’m broadly interested in stochastic modeling, statistical machine learning, and computational statistics. Much of my work focuses on contributing inferential tools in dynamic, dependent, and missing data settings, and I am particularly interested in likelihood-based formulations and the interface between optimization and Bayesian approaches.
Currently, I am developing latent variable approaches for inference in dynamic stochastic models, especially discrete valued Markov processes in continuous time such as branching processes and mechanistic compartmental models. Recently I am considering their extensions over spatial domains and networks, incorporating more flexible rules of behavior and interaction. We are also interested in casting these and related problems from an optimization viewpoint via EM and its generalization MM (majorization-minimization). These approaches admit fast, intuitive algorithms for tasks including clustering and estimation under general set-based constraints.
I work primarily on developing theory and methods, often driven by applications to systems biology and epidemiology. A list of publications along with software is available here.