Detecting Change Points in Time Series via Curvatures of Representation Trajectories

Yooju Shin, Jaehyun Park, Susik Yoon, Hwanjun Song, Byung Lee, Jae-Gil Lee

Keywords: time series, change point detection, curvature

TL;DR: We propose a novel change point detection method that exploits the curvature of a representation trajectory for time series.

Abstract: Change points are the timestamps at which a time series experiences meaningful changes. Recently, representation-based change point detection has gained popularity, but its emphasis on consecutive distance difference backfires, especially when the changes are gradual. In this paper, we propose a change point detection method, RECURVE, based on a novel change metric, the curvature of a representation trajectory, to accommodate both gradual and abrupt changes. Here, a sequence of representations in the representation space is interpreted as a trajectory, and a curvature at each timestamp can be computed. Using the theory of random walk, we formally show that the mean curvature is lower near change points than at other points. Extensive experiments using diverse real-world time-series datasets confirm the superiority of RECURVE over state-of-the-art methods.

Primary Area: representation learning for computer vision, audio, language, and other modalities

Code Of Ethics: I acknowledge that I and all co-authors of this work have read and commit to adhering to the ICLR Code of Ethics.

Submission Guidelines: I certify that this submission complies with the submission instructions as described on https://iclr.cc/Conferences/2024/AuthorGuide.

Anonymous Url: I certify that there is no URL (e.g., github page) that could be used to find authors' identity.

No Acknowledgement Section: I certify that there is no acknowledgement section in this submission for double blind review.

Submission Number: 1662