Qi Xu

prof_pic.jpg

I am currently a postdoctoral researcher in the Department of Statistics & Data Science at Carnegie Mellon University, working with Kathryn Roeder and Jing Lei.

Previously, I obtained my PhD degree from University of California, Irvine, under supervision of Annie Qu. Prior to that, I got my Master and Bachelor degree from University of Illinois at Urbana Champaign and Tongji University.

Recent research interests

  • Direction 1: Integrating heterogeneous datasets for prediction, estimation and inference
  • Direction 2: Integrating AI (predictive, generative) models in the loop of statistical analysis
  • Tools: Semiparametric theory, representation learning, high-dimensional statistics

news

Sep 30, 2025 New Preprint Available: Blockwise Missingness meets AI [Arxiv]. Our new work tackles a long-standing challenge in statistics: semiparametric inference for data with non-monotone missing patterns. For over 30 years, the theoretically optimal estimator has been known but considered computationally intractable for practical use. We introduce an elegant RAY approximation to the optimal estimating equation, striking a crucial balance between statistical efficiency and computational feasibility. This approach can be seamlessly integrated into both classical semiparametric and modern prediction-powered inference frameworks, offering a powerful new tool for researchers.

selected publications

  1. Blockwise Missingness meets AI: A Tractable Solution for Semiparametric Inference
    Qi Xu, Lorenzo Testa, Jing Lei, and 1 more author
    arXiv preprint arXiv:2509.24158, 2025
  2. Representation retrieval learning for heterogeneous data integration
    Qi Xu and Annie Qu
    arXiv preprint arXiv:2503.09494, 2025
  3. Optimal individualized treatment rule for combination treatments under budget constraints
    Qi Xu, Haoda Fu, and Annie Qu
    Journal of the Royal Statistical Society Series B: Statistical Methodology, 2024
  4. Crowdsourcing utilizing subgroup structure of latent factor modeling
    Qi Xu, Yubai Yuan, Junhui Wang, and 1 more author
    Journal of the American Statistical Association, 2024
  5. Multi-label residual weighted learning for individualized combination treatment rule
    Qi Xu, Xiaoke Cao, Geping Chen, and 3 more authors
    Electronic Journal of Statistics, 2024