Offline change detection under contamination

Sujay Bhatt, Guanhua Fang, Ping Li
Proceedings of the Thirty-Eighth Conference on Uncertainty in Artificial Intelligence, PMLR 180:191-201, 2022.

Abstract

In this work, we propose a non-parametric and robust change detection algorithm to detect multiple change points in time series data under non-adversarial contamination. The algorithm is designed for the offline setting, where the objective is to detect changes when all data are received. We only make weak moment assumptions on the inliers (uncorrupted data) to handle a large class of distributions. The robust scan statistic in the change detection algorithm is fashioned using mean estimators based on influence functions. We establish the consistency of the estimated change point indexes as the number of samples increases, and provide empirical evidence to support the consistency results.

Cite this Paper

BibTeX
@InProceedings{pmlr-v180-bhatt22a, title = {Offline change detection under contamination}, author = {Bhatt, Sujay and Fang, Guanhua and Li, Ping}, booktitle = {Proceedings of the Thirty-Eighth Conference on Uncertainty in Artificial Intelligence}, pages = {191--201}, year = {2022}, editor = {Cussens, James and Zhang, Kun}, volume = {180}, series = {Proceedings of Machine Learning Research}, month = {01--05 Aug}, publisher = {PMLR}, pdf = {https://proceedings.mlr.press/v180/bhatt22a/bhatt22a.pdf}, url = {https://proceedings.mlr.press/v180/bhatt22a.html}, abstract = {In this work, we propose a non-parametric and robust change detection algorithm to detect multiple change points in time series data under non-adversarial contamination. The algorithm is designed for the offline setting, where the objective is to detect changes when all data are received. We only make weak moment assumptions on the inliers (uncorrupted data) to handle a large class of distributions. The robust scan statistic in the change detection algorithm is fashioned using mean estimators based on influence functions. We establish the consistency of the estimated change point indexes as the number of samples increases, and provide empirical evidence to support the consistency results.} }
Endnote
%0 Conference Paper %T Offline change detection under contamination %A Sujay Bhatt %A Guanhua Fang %A Ping Li %B Proceedings of the Thirty-Eighth Conference on Uncertainty in Artificial Intelligence %C Proceedings of Machine Learning Research %D 2022 %E James Cussens %E Kun Zhang %F pmlr-v180-bhatt22a %I PMLR %P 191--201 %U https://proceedings.mlr.press/v180/bhatt22a.html %V 180 %X In this work, we propose a non-parametric and robust change detection algorithm to detect multiple change points in time series data under non-adversarial contamination. The algorithm is designed for the offline setting, where the objective is to detect changes when all data are received. We only make weak moment assumptions on the inliers (uncorrupted data) to handle a large class of distributions. The robust scan statistic in the change detection algorithm is fashioned using mean estimators based on influence functions. We establish the consistency of the estimated change point indexes as the number of samples increases, and provide empirical evidence to support the consistency results.
APA
Bhatt, S., Fang, G. & Li, P.. (2022). Offline change detection under contamination. Proceedings of the Thirty-Eighth Conference on Uncertainty in Artificial Intelligence, in Proceedings of Machine Learning Research 180:191-201 Available from https://proceedings.mlr.press/v180/bhatt22a.html.

Related Material