Trusted and Interactive Clustering for Time-Series Data
Keywords: Time-series clustering, evidence theory, information fusion
Abstract: Time-series clustering has gained abundant popularity and has been used in diverse scientific areas. However, few researchers take an information fusion perspective to combine information from the time and frequency domains to accomplish clustering, although these two domains offer distinct and complementary characteristics of time-series. Motivated by this issue, we propose a trusted and interactive model, which leverages evidence theory to combine time- and frequency-based clustering results produced by the corresponding contrastive learning module. After mathematizing clustering results from the two domains as mass functions, the uncertainty contained in these results can be quantified at the sample-specific level. The combined result thus promotes clustering reliability, and is optimized based on the pseudo-labels generated by k-means in an interactive learning paradigm. Both theoretical analysis and experimental results on 136 benchmark datasets validate the effectiveness of the proposed model in clustering performance. Extensive ablation experiments demonstrate the contribution of combining information from the time and frequency domains and using the interactive learning paradigm. The embeddings learned are also experimentally shown to perform well in other downstream tasks.
Primary Area: unsupervised, self-supervised, semi-supervised, and supervised representation learning
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/2025/AuthorGuide.
Reciprocal Reviewing: I understand the reciprocal reviewing requirement as described on https://iclr.cc/Conferences/2025/CallForPapers. If none of the authors are registered as a reviewer, it may result in a desk rejection at the discretion of the program chairs. To request an exception, please complete this form at https://forms.gle/Huojr6VjkFxiQsUp6.
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: 5798
Loading