Topic：Some Recent Results for P-value Free FDR Controls

Speaker：Jun S LIU, Professor, Department of Statistics, Harvard University and Department of Statistics and Data Science, Tsinghua University

Host：Chien-Fu Jeff WU, X.Q. Deng Presidential Professor, School of Data Science, CUHK-Shenzhen

Date：5 September (Thursday), 2024

Time：11:00 AM - 12:00 PM, Beijing Time

Format：Onsite

Venue：103 Meeting Room, Daoyuan Building

Language：English

Abstract：

There has been significant interest among researchers in false discovery rate (FDR) control methods partially due to the strong desire from the scientific community for reproducibility and replicability of scientific discoveries. I will discuss our recent efforts trying to go beyond the recently popular p-value-free FDR control methods such as the knockoff filter (KF), data splitting (DS), and Gaussian mirror (GM). We present some power analysis of these methods under the weak-and-rare signal framework and discuss its implications under different correlation structures of the design matrix. We then focus on the DS procedure and its variant. In particular, we reformulate the DS method into a two-step procedure: using part of the data for estimation and feature ranking (in regression setting) and using the other part as checking/validation. FDR control can be achieved by monitoring how well the validation goes along the feature ranking. Under this setup, we may utilize external information and apply any procedure, such as a Bayesian method with spike-and-slab priors, to work on the first part of the data. We show that substantial power gain can be achieved in this way. The presentation is based on joint work with Buyu Lin, Tracy Ke, and Yuanchuan Guo.

Biography：

Jun Liu has been a tenured professor in the Department of Statistics at Harvard University since 2000. He has served as a Professor of Biostatistics at Harvard University, a Tenured Professor of Statistics at Stanford University, a Cheung Kong Chair Professor of Mathematics at Peking University, and a recipient of the National Outstanding Youth Fund (Category B). He has been Co-Editor-in-Chief of the Journal of the American Statistical Association and an Associate Editor of several leading international statistical journals. At Tsinghua University, he led the founding of the Center for Statistics in 2015 and the Department of Statistics and Data Science in 2024.

Jun Liu received the NSF CAREER Award (U.S.) in 1995 and was selected as a Terman Fellow by Stanford University in the same year. He received the Mitchell Best Paper Award from the International Bayesian Society in 2000, the COPSS Presidents' Award in 2002, an IMS Medallion Lecture in 2002, a Bernoulli Lecture in 2004, the Pao-Lu Hsu Award of the Pan-Chinese Statistical Association in 2016, and the Jerome Sacks Award for Outstanding Cross-Disciplinary Contributions in 2017.

Prof. Jun Liu has significantly contributed in several areas, including Markov Chain Monte Carlo methodology and applications, bioinformatics, statistical learning theory and methods, high-dimensional model selection, and prediction. Prof. Jun Liu has published more than 300 research papers and one monograph, and he has been cited more than 80,000 times (Google Scholar).