Portrait of Yinyihong Liu
Hi, I'm Yinyihong Liu (刘殷仪泓).
I'm a Ph.D. candidate in Statistical Science at Duke University, advised by Eric B. Laber and Rebecca C. Steorts. My research focuses on scalable Bayesian and variational methods for entity resolution, and on tree embeddings for learning monotone functions under partial-order constraints. Before starting my Ph.D., I earned an M.Sc. from Duke in 2024 and a B.Sc. in Mathematics and Data Science from NYU Shanghai in 2022.
yinyihong.liu@duke.edu Durham, NC
Research
Bayesian & Variational Entity Resolution
Scalable Bayesian and variational methods for matching identities across noisy records — graphical and comparison-based entity resolution, de-duplication, and streaming record linkage. With R. C. Steorts.
"J. Smith" "Jon Smith" "J. Smth" 🧑 Jon Smith LATENT ENTITY
Monotone Tree Embeddings
A tree-embedding method for learning monotone functions under arbitrary partial-order constraints — discretizing the covariate space via trees and penalizing violations of a user-defined order. With E. B. Laber.
x x' x ⪯ x′ ⟹ μ̂(x) ≤ μ̂(x′) ARBITRARY PARTIAL ORDER
Distributional In-Context Reinforcement Learning
Distributional in-context reinforcement learning for calibrated exploration and information acquisition, with benchmarks for Thompson sampling and information-directed sampling. With A. Pacchiano and E. B. Laber.
s₁ a₁ r₁ s₂ a₂ r₂ ⋮ CONTEXT TRANSFORMER DISTRIBUTION OVER Q
Synthetic Data & Privacy
A review of how synthetic data acts as both a privacy-preserving release mechanism and a tool for empowering downstream models. With J. P. Reiter.
PRIVATE DATA [ MECHANISM ] SYNTHETIC DATA ANALYSIS / ML
Publications
2026 · JSSAM · Accepted
A Comparative Analysis of Bayesian Graphical Entity Resolution Models
Liu, Y., Aleshin-Guendel, S., Marchant, N. G., Steorts, R. C.
2026 · Annual Rev. Stat. · Accepted
Synthetic Data: A Tool for Privacy Protection and Model Empowerment
Liu, Y., Reiter, J. P.
2026 · Privacy in Stat. Databases · Accepted
Streaming Variational Inference for Fast Beta Linkage
Liu, Y., Aleshin-Guendel, S., Steorts, R. C.
2026 · Privacy in Stat. Databases · Accepted
Bayesian and Variational fastLink for Record Linkage
Eason, S., Liu, Y., Steorts, R. C.
2021 · IEEE · ICSPML
Airbnb Pricing Based on Statistical Machine Learning Models
Liu, Y.
2026 · Survey Methodology · Submitted
A Variational Approximation for Record Linkage
Liu, Y., Kundinger, B., Aleshin-Guendel, S., Steorts, R. C.
Teaching & honors
Teaching
2026 SpSTA 640 — Causal Inference · Duke
2024 FaSTA 240L — Probability for Statistical Inference, Modeling & Data Analysis · Duke
2023 SuBayesian Inference for Nuclear Physics — workshop TA · virtual
F19·S20MATH-SHU 235 — Probability & Statistics · NYUSH
Honors & awards
2024Master's Portfolio Award · Duke
2023Dean's Research Award for Master's Students · Duke
2022Major Honors in Mathematics — top mathematics major · NYUSH
2022NYU Shanghai Excellence Award — top 20% of class · NYUSH
2018–2022Dean's List, every semester · NYUSH
Beyond research

When I'm not working on research, I'm probably in the kitchen 🍳. I love baking, both Chinese and Western, and lately I've been trying to make the fluffiest, pull-apart milk toast 🍞 I can — tips are always welcome. I enjoy trying new recipes, playing around with flavors, and figuring out little ways to make things healthier.

Outside the kitchen, I like cooking, running, reading, and traveling. I also used to play the piano and Guzheng, along with several traditional Chinese wind instruments, including the Dizi, Xiao, Hulusi, and Bawu.

All models are wrong, but some are useful.

— G. E. P. Box