Jove
Visualize
Contact Us
JoVE
x logofacebook logolinkedin logoyoutube logo
ABOUT JoVE
OverviewLeadershipBlogJoVE Help Center
AUTHORS
Publishing ProcessEditorial BoardScope & PoliciesPeer ReviewFAQSubmit
LIBRARIANS
TestimonialsSubscriptionsAccessResourcesLibrary Advisory BoardFAQ
RESEARCH
JoVE JournalMethods CollectionsJoVE Encyclopedia of ExperimentsArchive
EDUCATION
JoVE CoreJoVE BusinessJoVE Science EducationJoVE Lab ManualFaculty Resource CenterFaculty Site
Terms & Conditions of Use
Privacy Policy
Policies

Related Experiment Videos

Unsupervised feature selection via anomaly-aware fuzzy graph fusion and diffusion multi-centroid learning.

Zhouqing Yan1, Ziping Ma2, Jinlin Ma3

  • 1School of Mathematics and Information Science, North Minzu University, Yinchuan, 750030, Ningxia, China.

Neural Networks : the Official Journal of the International Neural Network Society
|May 19, 2026
PubMed
Summary

Related Concept Videos

You might also read

Related Articles

Articles linked to this work by shared authors, journal, and citation graph.

Sort by
Same author

A peripheral subpopulation of retinal pigment epithelium resists oxidative damage through SERPINE3-mediated Caspase-1 inhibition.

The Journal of clinical investigation·2026
Same author

Unveiling microbial communities and biogeochemical cycles in Antarctic colored snow.

BMC microbiology·2026
Same author

Cisplatin-induced primary ovarian insufficiency-related injury in mice is mediated through defects in cholesterol synthesis and subsequent activation of p53 signaling.

Cellular signalling·2026
Same author

Identification of a pyrazolo[3,4-b]pyridine chemotype for dual focal adhesion kinase 1/2 (FAK1/2) targeting in triple-negative breast cancer.

Bioorganic chemistry·2026
Same author

Synthesis of fluorene-indole/benzofuran bicyclic molecules through palladium-catalyzed carboheterofunctionalization of alkynes.

Chemical communications (Cambridge, England)·2026
Same author

Clinical characteristics of <i>Mycoplasma pneumoniae</i> pneumonia in children and construction of a severe case prediction model: a retrospective study from Yan'an, China.

Frontiers in pediatrics·2026
This summary is machine-generated.

This study introduces a new unsupervised feature selection method (AGFMCFS) that improves data representation by considering fuzzy relationships and using multiple centroids. It effectively handles noisy data, leading to better clustering performance on benchmark datasets.

Area of Science:

  • Machine Learning
  • Data Science
  • Computer Vision

Background:

  • Unsupervised feature selection is crucial for dimensionality reduction in high-dimensional unlabeled data.
  • Existing methods struggle with fuzzy relationships and single-centroid representations, limiting discriminative power.
  • Anomalous samples can negatively impact graph construction in feature selection.

Purpose of the Study:

  • To propose a novel unsupervised feature selection method (AGFMCFS) addressing limitations of existing techniques.
  • To enhance the reliability of graph structures by incorporating anomaly detection.
  • To improve the discriminative power of learned representations using multi-centroid learning.

Main Methods:

  • Developed an anomaly-aware mechanism using the local outlier factor to quantify sample credibility and suppress outliers.
Keywords:
Adaptive diffusion bipartite graphFuzzy graph fusionLocal outlier factorMulti-centroid learningUnsupervised feature selection

Related Experiment Videos

  • Integrated multiple fuzzy graphs via a weighting mechanism for robust manifold structure learning.
  • Employed a diffusion bipartite graph model for propagative interaction between samples and multiple centroids.

    • Main Results:

    • The proposed AGFMCFS method demonstrated superior performance compared to existing approaches.
    • Experiments on twelve benchmark datasets validated the effectiveness and robustness of the method.
    • The approach yielded more discriminative feature representations for clustering tasks.

    Conclusions:

    • AGFMCFS effectively characterizes uncertain neighborhood relationships and enhances manifold structure robustness.
    • The multi-centroid learning approach significantly improves the discriminative power of feature representations.
    • The method offers a robust and effective solution for unsupervised feature selection and clustering.