Workshops

Detecting simultaneous change-points from multiple sources

103
reads

David Siegmund

2011-12-20
10:05:00 - 10:55:00

101 , Mathematics Research Center Building (ori. New Math. Bldg.)

We consider sequences of aligned independent observations from multiple sources. The sequences are subject to change-points assumed to occur simultaneously in a fraction of the sources. The sources could be sensors in a network, which may have a geometric structure so that a change-point affecting one sensor also affects nearby sensors, or the relationships of the sources may be completely unstructured. The change-point(s) to be detected could represent the disruption of a steady state that must be detected sequentially as soon as possible, or they may represent local signals occurring at different places in long sequences of observations. I will describe a number of examples and discuss methods that seem useful for both the sequential and the fixed sample problems, with particular attention to unstructured problems. This is joint research with Nancy Zhang, Benjamin Yakir, and Yao Xie.