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

On weighting clustering.

Richard Nock1, Frank Nielsen

  • 1Département Scientifique Inter-facultaire/GRIMAAG Lab., Université Antilles-Guyane, B.P. 7209, 97278 Schoelcher, Martinique, France. rnock@martinique.univ-ag.fr

IEEE Transactions on Pattern Analysis and Machine Intelligence
|August 5, 2006
PubMed
Summary
This summary is machine-generated.

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

Generalized Legendre Transforms Have Roots in Information Geometry.

Entropy (Basel, Switzerland)·2026
Same author

Mirror Descent and Exponentiated Gradient Algorithms Using Trace-Form Entropies.

Entropy (Basel, Switzerland)·2025
Same author

Two Types of Geometric Jensen-Shannon Divergences.

Entropy (Basel, Switzerland)·2025
Same author

Fast Proxy Centers for the Jeffreys Centroid: The Jeffreys-Fisher-Rao Center and the Gauss-Bregman Inductive Center.

Entropy (Basel, Switzerland)·2025
Same author

Symplectic Bregman Divergences.

Entropy (Basel, Switzerland)·2025
Same author

Divergences Induced by the Cumulant and Partition Functions of Exponential Families and Their Deformations Induced by Comparative Convexity.

Entropy (Basel, Switzerland)·2024
Same journal

Relation DETR+: Exploring Explicit Position Relation Prior for Dense Prediction.

IEEE transactions on pattern analysis and machine intelligence·2026
Same journal

RBF++: Quantifying and Optimizing Reasoning Boundaries across Measurable and Unmeasurable Capabilities for Chain-of-Thought Reasoning.

IEEE transactions on pattern analysis and machine intelligence·2026
Same journal

CAFE: Cross-View Adaptive Fusion and Cluster Center Enhancement for Robust Multi-View Clustering.

IEEE transactions on pattern analysis and machine intelligence·2026
Same journal

DIVER: Reinforced Diffusion Breaks Imitation Bottlenecks in End-to-End Autonomous Driving.

IEEE transactions on pattern analysis and machine intelligence·2026
Same journal

Ethics-Aware Safe Reinforcement Learning for Rare-Event Risk Control in Interactive Urban Driving.

IEEE transactions on pattern analysis and machine intelligence·2026
Same journal

Learning Shape Anchors for Holistic Indoor Scene Understanding.

IEEE transactions on pattern analysis and machine intelligence·2026
See all related articles

This study formalizes a new trend in unsupervised clustering, using instance point weights to focus on difficult data points, inspired by boosting algorithms. This data reweighting approach enhances clustering performance across various algorithms.

Area of Science:

  • Machine Learning
  • Data Mining
  • Artificial Intelligence

Background:

  • Recent trends in iterative unsupervised learning highlight a novel clustering approach.
  • This approach penalizes solutions using instance point weights, guiding clustering towards harder-to-cluster data.
  • Motivations stem from an analogy with supervised boosting algorithms, though previously lacking formalization.

Purpose of the Study:

  • To formally introduce and analyze a weighted approach for iterative unsupervised clustering.
  • To establish theoretical underpinnings for data reweighting in clustering, drawing parallels with boosting.
  • To demonstrate the practical benefits of this formalization through experimental validation.

Main Methods:

  • Clustering is framed as a constrained minimization of Bregman divergence.

Related Experiment Videos

  • Instance point weights are modified based on local variations in expected complete log-likelihoods.
  • Theoretical analysis is conducted to understand the impact of these weight modifications.

    • Main Results:

    • Theoretical results demonstrate benefits analogous to those observed in boosting algorithms.
    • Modified, weighted versions of established clustering algorithms (k-means, fuzzy c-means, EM, k-harmonic means) are derived.
    • Experimental results confirm the advantages of subtle data reweighting for improved clustering outcomes.

    Conclusions:

    • This paper provides the first formalization of a boosting-inspired weighted clustering approach.
    • The proposed method offers theoretical advantages and practical improvements across multiple clustering algorithms.
    • Data reweighting is shown to be a powerful technique for enhancing unsupervised learning performance.