Third Workshop on Higher-Order Asymptotics and Post-Selection Inference

Washington University, Saint Louis MO

Investigators

Abstract

The Third Workshop on Higher-Order Asymptotics and Post-Selection Inference (WHOA-PSI) will be held on the campus of Washington University in St. Louis from September 8-10, 2018. Post-selection inference refers to the question of valid statistical inference performed after a model selection procedure has been utilized. Higher-order asymptotics provide the tools and insights needed to refine basic large-sample results for post-selection inference procedures and make them more accurate and powerful, even when the scientist only has access to small to moderate sample sizes. WHOA-PSI seeks to merge these fields, foster collaboration and discussion, and push forward the post-selection inference research frontier in the direction of higher-order accuracy and more powerful inference procedures. NSF funding will support junior researchers attending this conference. This workshop attempts to address the needs of applied statisticians utilizing model selection procedures who also need extremely accurate and powerful inference procedures with theoretical guarantees. Topics from post-selection inference include selective inference after variable selection, high-dimensional and post-regularized inference, inference after change-point estimation, and simultaneous inference. Some talks will focus on methods from higher-order asymptotics, which can yield insights and improvements to post-selection inference procedures, including series expansions, saddlepoint approximations, and bootstrap-based refinements. Additional talks will present important applied problems for which new post-selection methods are needed. Examples include statistical genetics, neuroscience, and finance. (See http://www.math.wustl.edu/~kuffner/WHOA-PSI-3.html) This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.