High-Dimensional Change Point Analysis for Temporally Dependent Data
Abstract
This paper develops adaptive procedures for detecting and locating mean changes in high-dimensional time series. Quadratic CUSUM statistics target dense changes, whereas coordinatewise maximum statistics target sparse changes. Two weighting schemes are considered to accommodate both interior and boundary changes. Under general non-Gaussian vector dependence, we establish the limiting distributions, validate the required centering and scaling estimators, and prove asymptotic independence between matched quadratic and maximum statistics. These results justify Cauchy combination tests. We further establish single-change localization and consistent multiple-change recovery using wild binary segmentation. Numerical results illustrate the effectiveness of the proposed methods.
Related
- Nonparametric Change-Point Detection and Inference for High-Dimensional Distributions
- Adaptable High-Dimensional Change Point Detection via Ridge Regularization
- Simultaneous Change-Point Inference for High-Dimensional Functional Time Series
- Multiple Change Point Detection in the Mean and Covariance of High-Dimensional Matrix Time Series
- Graph-Based Change-Point Detection for Partially Observed High-Dimensional Data
BibTeX
@misc{wang2026highdimensional,
title = {{High-Dimensional Change Point Analysis for Temporally Dependent Data}},
author = {Xiaoyi Wang and Le Zhou and Jixuan Liu and Long Feng},
howpublished = {arXiv},
year = {2026},
month = oct,
doi = {10.48550/arxiv.2607.20828},
eprint = {2607.20828},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2607.20828},
note = {CC BY 4.0},
}
RIS
TY - UNPB
TI - High-Dimensional Change Point Analysis for Temporally Dependent Data
AU - Xiaoyi Wang
AU - Le Zhou
AU - Jixuan Liu
AU - Long Feng
PY - 2026
DA - 2026/10/26
PB - arXiv
DO - 10.48550/arxiv.2607.20828
UR - https://arxiv.org/abs/2607.20828
AB - This paper develops adaptive procedures for detecting and locating mean changes in high-dimensional time series. Quadratic CUSUM statistics target dense changes, whereas coordinatewise maximum statistics target sparse changes. Two weighting schemes are considered to accommodate both interior and boundary changes. Under general non-Gaussian vector dependence, we establish the limiting distributions, validate the required centering and scaling estimators, and prove asymptotic independence between matched quadratic and maximum statistics. These results justify Cauchy combination tests. We further establish single-change localization and consistent multiple-change recovery using wild binary segmentation. Numerical results illustrate the effectiveness of the proposed methods.
ER -