GD Seo
Ph.D. Student in Statistics · University of Illinois Urbana-Champaign
I am GD Seo, a Ph.D. student in Statistics at the University of Illinois Urbana-Champaign, where I began my doctoral studies in August 2026.
My research focuses on causal inference in the presence of unmeasured confounding. In particular, I am interested in using proxy variables under proximal causal inference assumptions to address hidden confounding, as well as in settings with multiple environments where causal mechanisms may shift.
I received my M.S. in Statistics from Yonsei University in February 2026, where I conducted research under the supervision of Prof. Kyungwoo Song, and my B.A. in Applied Statistics from Yonsei University in August 2022. Before graduate school, I worked as a Data Analyst at NAVER FINANCIAL.
Research
- Unmeasured confounding: I study how proxy variables can be used, under proximal causal inference assumptions, to identify causal effects in the presence of hidden confounders.
- Causal mechanism shifts: I developed FANS (Function And Noise Separation), which uses normalizing flows to distinguish changes in causal functions from changes in noise distributions. This work appeared at ICML 2026.
- Robust machine learning: My collaborative work explores black-box visual prompting and causal-aware in-context learning for out-of-distribution generalization.
Browse my publications for more details.