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

An On-line agglomerative clustering method for nonstationary data

I D Guedalia1, M London, M Werman

  • 1Institute of Computer Science, Hebrew University, Jerusalem 91904, Israel.idavidg@pacbell.net

Neural Computation
|February 9, 1999
PubMed
Summary

This study introduces an online agglomerative clustering algorithm that effectively handles nonstationary data. It efficiently determines the natural number of clusters and minimizes local distortion, outperforming existing methods for dynamic datasets.

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

Safety climate and workplace violence prevention in state-run residential addiction treatment centers.

Work (Reading, Mass.)·2012
Same author

Min-max operators in texture analysis.

IEEE transactions on pattern analysis and machine intelligence·2011
Same author

A curvelet-based patient-specific prior for accurate multi-modal brain image rigid registration.

Medical image analysis·2010
Same author

Characterizing spatio-temporal patterns for disease discrimination in cardiac echo videos.

Medical image computing and computer-assisted intervention : MICCAI ... International Conference on Medical Image Computing and Computer-Assisted Intervention·2007
Same author

Oral toxicity of the cyanobacterial toxin cylindrospermopsin in mice: long-term exposure to low doses.

Environmental toxicology·2006
Same author

Abnormal frontal activations related to decision-making in current and former amphetamine and opiate dependent individuals.

Psychopharmacology·2005

Area of Science:

  • Computer Science
  • Data Mining
  • Machine Learning

Background:

  • Existing online clustering algorithms often assume data stationarity, leading to poor performance with nonstationary datasets.
  • Determining the optimal number of clusters and handling clusters of varying sizes (mass) are significant challenges in data clustering.
  • Many current algorithms are computationally intensive, especially when identifying the correct number of clusters.

Purpose of the Study:

  • To develop an on-line agglomerative clustering algorithm capable of handling nonstationary data.
  • To address the temporal aspects of data clustering and determine the natural number of clusters.
  • To minimize local distortion, ensuring accurate representation of clusters with small mass.

Main Methods:

  • An on-line agglomerative clustering algorithm is proposed.

Related Experiment Videos

  • The algorithm's performance is evaluated against popular batch and on-line algorithms for stationary data.
  • The framework efficiently determines the natural number of clusters based on problem scale.

    • Main Results:

    • The proposed algorithm demonstrates comparable clustering performance to existing methods on stationary data.
    • It provides an efficient method for determining the natural number of clusters.
    • The algorithm effectively minimizes local distortion, improving representation of small-mass clusters.

    Conclusions:

    • The developed on-line agglomerative clustering algorithm offers a robust solution for nonstationary data.
    • It overcomes limitations of existing methods by efficiently determining cluster numbers and preserving small clusters.
    • This approach enhances data representation accuracy in dynamic environments.