Model-free Change-point Detection Using Modern Classifiers

In contemporary data analysis, it is increasingly common to work with non-stationary complex datasets. These datasets typically extend beyond the classical low-dimensional Euclidean space, making it challenging to detect shifts in their distribution without relying on strong structural assumptions. This paper introduces a novel offline change-point detection method that leverages modern classifiers developed in the machine-learning community. With suitable data splitting, the test statistic is constructed through sequential computation of the Area Under the Curve (AUC) of a classifier, which is trained on data segments on both ends of the sequence. It is shown that the resulting AUC process attains its maxima at the true change-point location, which facilitates the change-point estimation. The proposed method is characterized by its complete nonparametric nature, significant versatility, considerable flexibility, and absence of stringent assumptions pertaining to the underlying data or any distributional shifts. Theoretically, we derive the limiting pivotal distribution of the proposed test statistic under null, as well as the asymptotic behaviors under both local and fixed alternatives. The weak consistency of the change-point estimator is provided. Extensive simulation studies and the analysis of two real-world datasets illustrate the superior performance of our approach compared to existing model-free change-point detection methods.