A0308
Title: Adaptive block-based change-point detection for sparse spatially clustered data
Authors: Lynna Chu - Iowa State University (United States) [presenting]
Abstract: A non-parametric change-point detection approach is presented for detecting potentially sparse changes in a time series of high-dimensional observations or non-Euclidean data objects. A change in distribution is targeted over time that occurs in a smaller (unknown) subset of dimensions, where the dimensions may be spatially correlated. The motivation is the remote sensing application, where changes occur in small, spatially clustered regions over time. An adaptive block-based change-point detection approach is proposed that allows for complex spatial dependencies across dimensions and leverages these dependencies to boost detection power and estimation accuracy. It is demonstrated via simulation studies that the approach has superior performance in detecting sparse changes for datasets with spatial or local group structures. An application of detecting changes at the Natanz Nuclear facility in Iran using remote sensing images is carefully explored.
