2025
Student Seminar: Yuyang Jiang
3:00–3:30 pm Jones 111
Tuesday, May 6, 2025, at 3:00 PM, in Jones 111, 5747 S. Ellis Avenue
Master’s Thesis l Presentation
Yuyang Jiang, Department of Statistics, The University of Chicago
“GPT-4V Cannot Generate Radiology Reports Yet”
Student Seminar: Evan Levine
2:30–3:00 pm Jones 111
Tuesday, May 6, 2025, at 2:30 PM, in Jones 111, 5747 S. Ellis Avenue
Master’s Thesis l Presentation
Evan Levine, Department of Statistics, The University of Chicago
“Bias-Adjusted Maximum Likelihood Estimators for LIMMA”
Student Seminar: Sparsho De
2:00–2:30 pm Jones 111
Tuesday, May 6, 2025, at 2:00 PM, in Jones 111, 5747 S. Ellis Avenue
Master’s Thesis l Presentation
Sparsho De, Department of Statistics, The University of Chicago
“Empirical bayes-powered skill model for darts players”
Student Seminar: Yuchen Qiu
2:00–2:30 pm Jones 111
Monday, May 5, 2025, at 2:00 PM, in Jones 111, 5747 S. Ellis Avenue
Master’s Thesis l Presentation
Yuchen Qiu, Department of Statistics, The University of Chicago
“Pre- and Post- Error Behavior in a Sequential Learning Task”
Student Seminars: Mark Lee
12:30–1:00 pm Jones 111
Monday, May 5, 2025, at 12:30 PM, in Jones 111, 5747 S. Ellis Avenue
Master’s Thesis l Presentation
Mark Lee, Department of Statistics, The University of Chicago
“TBA”
Statistics Colloquium: Jiaying Gu
11:30 am–12:30 pm Jones 303
Jiaying Gu, Department of Economics, University of Toronto
Title: Empirical Bayes for Compound Adaptive Experiments
Abstract:
We investigate Empirical Bayes analysis in the context of compound adaptive experiments, where the arm distribution in each experiment follows a normal distribution with an unknown mean parameter that we aim to estimate. There are two primary approaches to EB estimation: g-modeling, which estimates the prior by maximizing the marginal likelihood, and f-modeling, which directly computes posterior means from the sample distribution of observations. We establish that g-modeling remains a valid EB procedure even when it incorrectly assumes exogenous data collection; it holds regardless of the sampling algorithm used and the endogeneity of sample sizes. One can apply standard g-modeling methods by treating the data as if it were exogenously sampled, and restrict attention to only the sample mean of the data. Remarkably, we show that the risk guarantees established for g-modeling with i.i.d data remain valid for adaptively generated data, with no need for prior knowledge of the sampling algorithm, even when it varies across experiments. In contrast, the f-modeling approach results in biased estimates. We validate the robustness of the g-modeling approach through simulations involving commonly used adaptive algorithms and illustrate its applicability using a real-world dataset comprising multiple sequential experiments.
Student Seminar: YingTing Lu
10:30–11:00 am Jones 111
Monday, May 5, 2025, at 10:30 AM, in Jones 111, 5747 S. Ellis Avenue
Master’s Thesis l Presentation
YingTing Lu, Department of Statistics, The University of Chicago
“Evaluating Hybrid Machine Learning Models for Time Series Forecasting”
Student Seminar: Tony Yang
10:00–10:30 am Jones 111
Monday, May 5, 2025, at 10:00 AM, in Jones 111, 5747 S. Ellis Avenue
Master’s Thesis l Presentation
Tony Yang, Department of Statistics, The University of Chicago
“Welfare Dynamics in Human-AI Content Creation Competition”
Student Seminar: Mingkun Che
11:30 am–12:00 pm Jones 111
Friday, May 2, 2025, at 11:30 AM, in Jones 111, 5747 S. Ellis Avenue
Master’s Thesis l Presentation
Mingkun Che, Department of Statistics, The University of Chicago
“Randomization inference for quantiles of individual treatment effects under interference”
Student Seminar: Jay Lee
11:00–11:30 am Jones 111
Friday, May 2, 2025, at 11:00 AM, in Jones 111, 5747 S. Ellis Avenue
Master’s Thesis l Presentation
Jay Lee, Department of Statistics, The University of Chicago
“Discretization Schemes of Total Variation for Image Inpainting and Denoising”