Harvard AI, Math, and Statistics Seminar: Nina Dornemann

Our upcoming event for the Our upcoming event for the Harvard AI, Math, and Statistics Seminars is scheduled for Friday, October 16th from 10:30 – 11:30pm (ET) and will be an in-person presentation at Maxwell-Dworkin, Room 134 a/b. This week's speaker will be Nina Dornemann, Assistant Professor at Aarhus University.

Title: Real-Time Change-Point Detection in High-Dimensional Random Matrices

Abstract: Detecting structural changes as data arrive is important in applications such as environmental monitoring and dynamic networks. We discuss a general approach to real-time monitoring of high-dimensional random matrices and illustrate it through the spiked Wigner model, where we monitor the largest eigenvalue for a phase transition indicating the emergence of a signal. Combining random matrix theory with Gaussian approximations, we construct a self-normalized CUSUM detector that requires no additional parameter estimation. In the second part, we discuss how Hölder-type statistics can lead to shorter detection delays.

Speaker's Bio: Nina Dörnemann is an Assistant Professor at Aarhus University, Denmark. She received her PhD from Ruhr University Bochum, Germany, in 2022 under the supervision of Holger Dette. After a short postdoctoral position in Bochum, she joined the Department of Statistics at the University of California, Davis, first as a Visiting Scholar and then as a Visiting Assistant Professor. This semester, she is a Visiting Scholar in the Department of Statistics at Harvard University, working with Pragya Sur. Her research focuses on high-dimensional statistics and random matrix theory, with further interests in change-point problems, time series analysis and functional data analysis.