Thursday, October 13 2022, 4pm 204 Caldwell Hall Event Flyer (1.64 MB) Georgios Fellouris Department of Statistics University of Illinois at Urbana-Champaign Speaker's Website Sequential multiple testing Abstract In this talk we will consider the multiple testing problem when theobservationsare collected in real time over multiple data streams and the goal isto decide as quickly as possible. In this context, a multiple testingprocedure consists of not only a decision rule but also a stopping rule.We will present such pairs that minimize asymptoticallythe expected time for decision, uniformly with respect to the unknownsignal configuration, as user-specified bounds on the false positive andthe false negative rate go to 0. We will also consider the case whereit is not possible to sample all data streamsat each time instance, and one needs to additionally specify anadaptive sampling rule. Finally, we will discuss the setup where thedecisions for the various binary testing problems can be taken atdifferent times and the goal is to minimize the total samplesize. Biography Dr. Georgios Fellouris is an Associate Professor in the Department of Statistics of University of Illinois at Urbana-Champaign. He received his PhD from Columbia University in 2010. His research interests include Sequential hypothesis testing, Quickest change detection, Sequential parameter estimation, Sequential design. He is also interested in decision making under communication constraints, and educational measurement and cognitive assessment.
