Multi-source Learning

Summary (LLM read my papers; human bias-correction applied)

In multi-source learning, my work aims to make models reliable when data come from multiple studies, hospitals, or environments that do not exactly agree. Rather than assuming a single pooled distribution, I define the target through an explicit robust optimization principle—typically a minimax (distributionally robust) objective that guards against the worst-case mixture or shift across sources—so the estimand itself is stable to heterogeneity. A key contribution is then to make this target computable and statistically actionable: in some settings, robust objectives admit an interpretable dual characterization that yields a closed-form solution or a simple reweighting rule; more generally, I develop efficient primal–dual / saddle-point algorithms tailored to the minimax structure, with theoretical guarantees. Finally, I provide valid uncertainty quantification for these optimization-defined estimands, even in nonstandard/nonregular regimes where classical smooth asymptotics can fail (e.g., due to nonsmooth objectives, boundary solutions, or source-adaptive behavior).

Wang, Z., Liu, M., Lei, J., Bach, F.^✉, and Guo, Z^✉.(2025).
StablePCA: Learning Shared Representations across Multiple Sources via Minimax Optimization
Technical Report

Guo, Z., Wang, Z., Hu, Y., and Bach, F.(2025).
Statistical Inference for Conditional Group Distributionally Robust Optimization with Cross-Entropy Loss
Technical Report

Gu, Y., Fang, C., Xu, Y., Guo, Z., and Fan, J.(2025).
Fundamental Computational Limits in Pursuing Invariant Causal Prediction and Invariance-Guided Regularization
Technical Report

Wang, Z.^#, Hu, Y.^#, Bühlmann, P., and Guo, Z.(2024).
Causal Invariance Learning via Efficient Optimization of a Nonconvex Objective
Technical Report

Wang, Z., Si, N., Guo, Z., and Liu, M.(2024).
Multi-source stable variable importance measure via adversarial machine learning
Technical Report

Zhan, K., Xiong, X., Guo, Z., Cai, T., and Liu, M.(2024).
Transfer Learning Targeting Mixed Population: A Distributional Robust Perspective.
Technical report

Xiong, X.,Guo, Z.^✉, and Cai, T.^✉(2023).
Distributionally Robust Transfer Learning
Technical Report

Wang, Z., Bühlmann, P., and Guo, Z. (2025+).
Distributionally Robust Learning for Multi-source Unsupervised Domain Adaptation.
Annals of Statistics, to appear.

Guo, Z., Li, X., Han, L., and Cai Tianxi. (2025+).
Robust Inference for Federated Meta-Learning
Journal of the American Statistical Association, to appear. [Codes]

Guo, Z. (2024).
Statistical Inference for Maximin Effects: Identifying Stable Associations across Multiple Studies .
Journal of the American Statistical Association, 119(547), 1968-1984. [Codes]

Wang, X., Zhou, H., · · ·, 4CE,Avillach, P.^✉, Guo, Z.^✉, and Cai, T.^✉ (2022)
Surv-Maximin: Robust Federated Approach to Transporting Survival Risk Prediction Models.
Journal of Biomedical Informatics, 134 (2022): 104-176.

underline indicates supervised students ; # indicates equal contribution; * indicates alphabetical ordering ; ✉ indicates corresponding authorship.