Johnny (Quanhan) Xi

profile.jpg

I am a Ph. D. Student in the Department of Statistics at the University of British Columbia, supervised by Ben Bloem-Reddy. I’m broadly interested in causal machine learning and/with generative models. Recently, I have been interested in how dynamical systems can be used to model data both with and without explicit temporal components. In the past, I was an intern with Jason Hartford at Valence Labs, where I worked on causal representation learning from multiple modalities.

news

May 1, 2025 Two papers accepted to ICML 2025: Causal Velocity Models and Identifying metric structures in LVMs. Looking forward to it!
Nov 1, 2024 My internship project on matching data from unpaired modalities using propensity scores will appear at NeurIPS 2024! In biology, you can in theory have multimodal measurements of the same cell in biology, but each measurement process is destructive, putting us naturally in a “potential outcomes” setting. Paper here: arXiv!
Dec 4, 2023 I will be presenting a poster at the CRL workshop on triangular monotonic maps for causal discovery. See the paper.

selected publications

  1. ICML 2025
    Distinguishing Cause from Effect with Causal Velocity Models
    Johnny XiHugh DancePeter Orbanz, and 1 more author
    In ICML 2025, 2025
  2. NeurIPS 2024
    Propensity Score Alignment of Unpaired Multimodal Data
    Johnny Xi, Jana Osea, Zuheng Xu, and 1 more author
    In NeurIPS 2024, 2024
  3. AISTATS 2023
    Indeterminacy in Generative Models: Characterization and Strong Identifiability
    Quanhan Xi, and Benjamin Bloem-Reddy
    In AISTATS, 2023