Offline Change Detection under Contamination

Sujay Bhatt, Guanhua Fang, Ping Li

Keywords: change detection, offline, contamination, robust, influence function, consistency

TL;DR: We propose a simple offline change detection algorithm having a robust scan statistic fashioned using influence functions, that can consistently detect changes in non-i.i.d and contaminated dataset.

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 de-signed 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.