RKHS-Based Sample Reconstruction Ability Modulated Robust Kernel Fuzzy Clustering
IEEE Signal Processing Letters · 2026 · 33 · 2345-2349
DOI: 10.1109/lsp.2026.3699592 ↗Kurz gefasst
Fachartikel in der Zeitschrift IEEE Signal Processing Letters, 2026. Verfasst von Qi Cui und 4 weiteren Autor:innen. Die Arbeit nennt 31 Literaturangaben.
Abstract (Originalsprache)
Traditional fuzzy clustering cannot effectively characterize the structure of high-dimensional nonlinear data; furthermore, it is highly susceptible to outlier interference and suffers from cluster prototype deviation. This letter proposes an RKHS-based S ample R econstruction A bility modulated R obust K ernel F uzzy C lustering (SRA-RKFC) algorithm. Instead of utilizing standard kernel mapping, SRA-RKFC constructs an outlier-resistant projection subspace within the Reproducing Kernel Hilbert Space (RKHS) to accurately capture complex nonlinear distributions. Specifically, a robust binary weighting mechanism is introduced to mitigate the interference of outliers, and projection reconstruction is employed to preserve the primary structure of data. The global optimal mean is embedded into the joint optimization process to eliminate the mean shift caused by outliers and improve the accuracy of projection representation. Experiments are conducted on a synthetic nested spherical shell dataset and eight benchmark datasets. Results reveal that our proposed SRA-RKFC algorithm achieves superior clustering performance in comparison with current benchmark models across both original and corrupted data settings, indicating the effectiveness of coupling non-linear representation and noise resilience in improving the model robustness and stability.
Zitieren
APA
Cui, Q., He, X., Liu, Z., Wu, C., & Peng, Y. (2026). RKHS-Based Sample Reconstruction Ability Modulated Robust Kernel Fuzzy Clustering. IEEE Signal Processing Letters, 33, 2345-2349. https://doi.org/10.1109/lsp.2026.3699592
Vancouver
Cui Q, He X, Liu Z, Wu C, Peng Y. RKHS-Based Sample Reconstruction Ability Modulated Robust Kernel Fuzzy Clustering. IEEE Signal Processing Letters. 2026;33:2345-2349. doi:10.1109/lsp.2026.3699592
BibTeX
@article{Cui2026,
author = {Cui, Qi and He, Xie and Liu, Zhaohu and Wu, Chang and Peng, Yong},
title = {RKHS-Based Sample Reconstruction Ability Modulated Robust Kernel Fuzzy Clustering},
journal = {IEEE Signal Processing Letters},
year = {2026},
volume = {33},
pages = {2345--2349},
doi = {10.1109/lsp.2026.3699592}
}
EUVIMED