Social time with refreshments: 10:30

Seminar begins: 11:00

Speaker: Yang Ning (Cornell University)

Title: Active Subsampling for Binary Response Models

Abstract: In the measurement-constrained problems, despite the availability of large datasets, we may be only affordable to observe the labels on a small portion of the large dataset. This poses a critical question that which data points are most beneficial to label given a budget constraint. In this paper, we focus on the estimation of the optimal individualized threshold in a measurement-constrained M-estimation framework. Our goal is to estimate a high-dimensional parameter in a linear threshold for a continuous variable X such that the discrepancy between whether X exceeds the threshold and a binary outcome is minimized. We propose a novel K-step active subsampling algorithm to estimate θ, which iteratively samples the most informative observations and solves a regularized M-estimator. We show that the two-step algorithm yields an estimator with the parametric convergence rate.  Furthermore, we formulate a minimax framework for the measurement-constrained M-estimation problem and prove that our estimator is minimax rate optimal up to a logarithmic factor. Finally, we demonstrate the performance of our method in simulation studies and apply the method to analyze a large diabetes dataset.

Biography: Yang Ning is an associate professor of statistics and data science. The theme of his research is to develop statistical methods and theory to quantify the uncertainty (confidence interval and hypothesis test) in modern data sets, which are characterized by high dimensionality, complexity, and heterogeneity. He enjoys working at the interface of mathematical statistics, machine learning, and stochastic optimization. He is also interested in applied projects in genomics, neuroscience, epidemiology, and clinical trials. 

He has been awarded the NSF CAREER award and the David Byar Young Investigator Award, and has published papers in Annals of Statistics, Journal of American Statistical Association, Biometrika, and Journal of the Royal Statistical Society.

Prior to joining Cornell, he completed a postdoctoral fellowship at Princeton University, where he developed statistical methods and theory for analyzing big and complex data. Ning obtained a bachelor's degree in mathematics from Fudan University, China, and a Ph.D in biostatistics from Johns Hopkins University.