Abstract: Time series from a large number of sources are ubiquitous, and may incur structural changes during data acquisition. For example, in fMRI analysis, brain regions associated with task-related stimuli or in a resting state become active. An activated time series can comprise readings from an activated region. Of interest is to control the uncertainty of discovering time series in activation (viz., activated regions in fMRI analysis) by using the false discovery rate (FDR) tool. We propose a simple, yet effective method that incorporates unknown asynchronous change patterns and spatial dependence. We justify the validity of our method in controlling the FDR using an asymptotic analysis. The results of our numerical experiments indicate that the proposed method is both accurate and powerful. An implementation is provided in the R package SLIP.

Key words and phrases: Change-point analysis, data splitting, false discovery rate, fMRI, regions of interest.