A Effective Variance Change Detection Method under constantly Changing Mean

Zhenguo Gao, Yan Li

Primary Area: probabilistic methods (Bayesian methods, variational inference, sampling, UQ, etc.)

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.

Keywords: Dual control windows, Variance change point detection, Smoothly changing mean trend, Subsampling, Liver procurement

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

Abstract:

Effectively evaluating the viability of a procured organ in the transplant patient prior the procedure is of critical importance. Current viability assessment methods rely on evaluating the organ’s morphology and/or laboratory biopsy results with limited effectiveness. A recently proposed, well-designed noninvasive method evaluated the viability status of organs by detecting the variance change point of their surface temperature through exploring the entire data profile. However, most part of the data in a temperature profile barely contains the change information, which yields a waste of computational resources of their method. This paper proposes an accelerating algorithm with a well-designed dual control windows scheme that can be extended to online change detection. The proposed method significantly improves the computational speed and retains the same change detection power as the method Gao19 through the removal of redundant data. Simulation and application results demonstrate the robust performance of the proposed method.

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: 630