Exploring Temporal Semantic for Incomplete Clustering
Keywords: Temporal semantic, incomplete clustering, human motion segmentation
Abstract: Clustering data with incomplete features has garnered considerable scholarly attention; however, the specific challenge of clustering sequential data with missing attributes remains largely under-explored. Conventional heuristic methods generally address this issue by first imputing the missing features, thereby making the clustering results heavily reliant on the quality of imputation. In this paper, we introduce a novel clustering framework, termed ETC-IC, which directly clusters incomplete data with rigorous theoretical guarantees, whilst concurrently leveraging temporal semantic consistency to enhance clustering performance. Empirical evaluations demonstrate that the proposed model consistently surpasses current state-of-the-art methods in clustering human motion data.
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.
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: 7038
Loading