A2026
Title: Online change-point detection using foundation probabilistic forecasting models
Authors: Alan Moore - Iowa State University (United States)
Zhengyuan Zhu - Iowa State University (United States) [presenting]
Lynna Chu - Iowa State University (United States)
Abstract: Online change-point detection is a fundamental problem in sequential monitoring, where the goal is to detect distributional changes in a time series as quickly as possible while controlling false alarms. A general framework for online change-point detection is proposed that leverages foundation probabilistic forecasters, such as Chronos-2, which can capture complex temporal structure including trend, seasonality, heteroskedasticity, and nonlinear dependence. The approach transforms each incoming observation relative to its predictive quantile forecast into an approximately standard Gaussian monitoring statistic, which can then be used in classical sequential detection procedures such as CUSUM and MOSUM. To reduce adaptation of the forecaster to post-change observations, a buffer is introduced between the forecasting context window and the forecast target. Because this design can induce serial dependence in the monitoring statistics, a parametric bootstrap procedure is developed to calibrate critical thresholds and control false alarms in practice. Extensive simulations across data-generating processes involving non-normality, trend, seasonality, heteroskedasticity, and autoregressive dependence show that the proposed method achieves competitive and often superior detection performance relative to benchmark procedures, while requiring no explicit specification of a parametric pre-change model.
