A1083
Title: Higher-criticism for multi-sensor change-point detection
Authors: Yao Xie - Georgia Institute of Technology (United States) [presenting]
Alon Kipnis - Reichman University (Israel)
Tingnan Gong - Georgia Institute of Technology (United States)
Abstract: The focus is on the distribution-free procedure based on higher criticism to address the sparse multi-stream sequential change-point detection problem. Namely, detecting a change point co-occurring is needed in a few data streams out of potentially many, while those affected streams are unknown in advance. The procedure involves testing for a change point in individual streams and combining p-values using higher criticism. As a by-product, the procedure also indicates a set of streams suspected to be affected by the change. It is shown that the procedure attains the information-theoretic optimal detection performance under a sparse Gaussian mean shift when individual tests are based on the LR or GLR. The effectiveness of the method is compared to other procedures using numerical evaluations.
