New semi-supervised fuzzy C-means clustering with asymmetric deviation constraints and fast algorithm

[Display omitted] •A new semi-supervised fuzzy clustering model is proposed.•A semi-supervised fuzzy clustering algorithm’s convergence is analyzed.•A fast semi-supervised fuzzy clustering algorithm is presented.•Experimental results show the proposed algorithm’s superiority and efficiency. Semi-sup...

Full description

Saved in:
Bibliographic Details
Published in:Expert systems with applications Vol. 298; p. 129648
Main Authors: Wu, Chengmao, Hou, Jun
Format: Journal Article
Language:English
Published: Elsevier Ltd 01.03.2026
Subjects:
Accelerated algorithm
Affinity filtering
Asymmetric deviation constraint
Fuzzy clustering
Membership scaling
Fuzzy clustering
Accelerated algorithm
Membership scaling
Asymmetric deviation constraint
Affinity filtering
ISSN:0957-4174
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:[Display omitted] •A new semi-supervised fuzzy clustering model is proposed.•A semi-supervised fuzzy clustering algorithm’s convergence is analyzed.•A fast semi-supervised fuzzy clustering algorithm is presented.•Experimental results show the proposed algorithm’s superiority and efficiency. Semi-supervised clustering leverages prior information to improve algorithm performance and is widely valued by researchers. This paper analyzes the traditional semi-supervised fuzzy C-means (SFCM) objective function, noting that as a labeled sample’s membership degree aligns with its prior information, the impact of this information on the deviation constraint weakens. This reduces its supervisory effect on optimizing the membership partition matrix, especially with a large regularization factor. To overcome this, we propose a novel semi-supervised fuzzy C-means method based on an asymmetric deviation constraint and develop a two-level alternating iterative optimization algorithm, supported by theoretical convergence analysis using Zangwill’s theorem and the bordered Hessian matrix. To address the slow convergence and high computational cost typical of semi-supervised fuzzy clustering, we further enhance the algorithm with affinity filtering and a membership scaling scheme for improved efficiency. Experimental results demonstrate that our methods significantly outperform existing state-of-the-art techniques, advancing semi-supervised fuzzy C-means clustering.
ISSN:0957-4174
DOI:10.1016/j.eswa.2025.129648