Stabilization of Hierarchical Supervised Adversarial Mechanism With Dynamic Compensation for Two-View Soft Stimulation
Rattachement africain : cn, hk. Niveau de preuve : code pays fourni par la source.
Le résumé fourni par la source
In this study, we propose a hierarchical two-view Takagi–Sugeno–Kang (TSK) fuzzy classifier based on a supervised adversarial mechanism (ADVML-FC) designed to address the limitations of traditional single- and two-view fuzzy classifiers, such as data underutilization, challenges in feature fusion, and poor generalization. The proposed model leverages the output of one view as perturbation information to challenge the other view, thereby promoting information exchange and fusion between the two views. Unlike conventional approaches that rely on random noise perturbations, the model uses signals derived from the opposing view outputs and dynamically adjusts the attack success rate (ρ) to prevent overfitting and enhance generalization. The model simplifies the objective function design and integrates the least-squares method to efficiently estimate the consequent parameters of fuzzy rules, thereby substantially reducing computational complexity. Moreover, the model retains a zero-order TSK fuzzy rule structure and integrates fuzzy C-means clustering with Gaussian membership functions to ensure semantic interpretability. Experimental results demonstrate that ADVML-FC achieves superior performance in terms of average training accuracy, average testing accuracy, F1-score, Kappa, and Matthews correlation coefficient across nine two-view datasets. The proposed model not only delivers exceptional classification performance in terms of interpretability, highlighting its strong potential for broad applicability.
Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.
Le contrôle bibliographique ouvert
DOI retrouvé dans Crossref DOI retrouvé ; titre concordant.
- Titre Crossref
- Stabilization of Hierarchical Supervised Adversarial Mechanism With Dynamic Compensation for Two-View Soft Stimulation
- Date Crossref
- 01/08/2026
- Éditeur
- Institute of Electrical and Electronics Engineers (IEEE)
- Type
- journal-article
Ce recoupement confirme des métadonnées liées au DOI. Il ne confirme ni la méthode ni les conclusions de l’étude, et il ne compte pas comme une seconde source scientifique indépendante.
Où se fait cette recherche
-
Jiangsu University of Science and Technology pays non établi dans la noticeUniversité ou école supérieure
-
Wuhan University of Technology pays non établi dans la noticeUniversité ou école supérieure
-
Wuhan Ship Development & Design Institute pays non établi dans la noticeStructure de recherche
-
Hong Kong Polytechnic University pays non établi dans la noticeUniversité ou école supérieure
-
China Ship Research and Development Center pays non établi dans la noticeInstitution
-
The Hong Kong Polytechnical University pays non établi dans la noticeUniversité ou école supérieure
Jiangsu University of Science and Technology, Wuhan University of Technology et Wuhan Ship Development & Design Institute, avec 3 autres affiliations.
Une affiliation ne permet pas de déduire la nationalité d’un auteur.