Tianyu Zhang will be giving a talk on "Winners with Confidence: Argmin Inference and Algorithmic Stability" on Wednesday, October 15, 2025 in HSSB 1173.
Wed October 15 PSTAT Seminar 3:30 pm - 4:30 pm, HSSB 1173 Speaker: Tianyu Zhang, UCSB
Title:
Winners with Confidence: Argmin Inference and Algorithmic Stability
Abstract:
In this talk, I will discuss the problem of identifying the index of the minimum value of a vector from noisy observations, which arises in policy comparison, discrete maximum likelihood, and model selection. Jointly with Hao Lee and Jing Lei, we developed an asymptotically normal test statistic for establishing an index set containing the population minimizer with high probability. The method is applicable in high-dimensional settings and with potentially many ties in the population mean vector. I will discuss the concept of algorithmic stability and how it implies a central limit theorem for globally dependent summations, which is a key technical ingredient of the proposed framework.
Bio:
Tianyu Zhang is currently an assistant professor in the Department of Statistics and Applied Probability at the University of California, Santa Barbara. He obtained his PhD from the Department of Biostatistics at the University of Washington and had his postdoctoral training at the Department of Statistics & Data Science at Carnegie Mellon University. His recent research interest includes the theory and application of nonparametric estimation, model selection, and transfer learning.