A0378
Title: Adaptive two way change points detection
Authors: Likai Chen - Washington University in Saint Louis (United States) [presenting]
Jiaqi Li - University of Chicago (United States)
Abstract: A new detection method is proposed for multiple change points in high-dimensional time series. The method aggregates moving sum (MOSUM) statistics cross-sectionally by an $\ell^2$-norm and maximizes them over time. A Two-Way MOSUM statistic is introduced with adaptive window size to account for different break sizes, which can substantially improve the estimation accuracy. The asymptotic theory has been established for the limiting distribution of an $\ell^2$-aggregated statistic with varying window sizes for testing the existence of breaks, and the core is to extend a high-dimensional Gaussian approximation theorem to non-stationary and spatial-temporally dependent data generating processes. Consistency results of estimated break numbers, time stamps and sizes of breaks are provided.
