Statistics Colloquium : Dr. Aleksey Polunchenko

Abstract: Sequential (quickest) change-point detection is a branch of statistics concerned with the design and analysis of methods for rapid but reliable anomaly detection in “live” monitored random processes. The subject’s areas of application are virtually unlimited, and include quality and process control, anomaly and failure detection, surveillance and security, finance, seismology, navigation, intrusion detection, boundary tracking—to name a few. This talk is intended to provide a brief overview of the state-of-the-art in quickest change-point detection, with particular emphasis placed on the recently proposed Generalized Shiryaev–Roberts (GSR) detection procedure (it was proposed in 2008, but the paper didn’t come out until 2011). Notwithstanding its relatively “young age”, the GSR procedure has already been shown to have strong optimality properties, not exhibited by such mainstream detection methods as the Cumulative Sum (CUSUM) “inspection scheme” and the Exponentially Weighted Moving Average (EWMA) chart. Hence the interest in the GSR procedure.