A0477
Title: Sequential change point detection via denoising score matching
Authors: Liyan Xie - University of Minnesota (United States) [presenting]
Abstract: Sequential change-point detection plays a critical role in numerous real-world applications, where timely identification of distributional shifts can greatly mitigate adverse outcomes. Classical methods commonly rely on parametric density assumptions of pre- and post-change distributions, limiting their effectiveness for high-dimensional, complex data streams. A score-based CUSUM change-point detection is proposed, in which the score functions of the data distribution are estimated by injecting noise and applying denoising score matching. Both offline and online versions of score estimation are considered. Through theoretical analysis, it is demonstrated that denoising score matching can enhance detection power by effectively controlling the injected noise scale. Finally, the practical efficacy of the method is validated through numerical experiments on two synthetic datasets and a real-world earthquake precursor detection task, demonstrating its effectiveness in real scenarios.
