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 Video

Updated: Jul 7, 2026

Large-scale Reconstructions and Independent, Unbiased Clustering Based on Morphological Metrics to Classify Neurons in Selective Populations
12:27

Large-scale Reconstructions and Independent, Unbiased Clustering Based on Morphological Metrics to Classify Neurons in Selective Populations

Published on: February 15, 2017

A new neural network for cluster-detection-and-labeling.

T Eltoft1, R P deFigueiredo

  • 1Department of Physics, Faculty of Science, University of Tromsø, N-9037 Tromsø, Norway.

IEEE Transactions on Neural Networks
|February 8, 2008
PubMed
Summary

A new unsupervised neural network, the cluster-detection-and-labeling (CDL) network, effectively clusters data of any shape without prior knowledge of cluster count. It outperforms the winner-take-all network in complex data clustering tasks.

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

Adaptive regularized constrained least squares image restoration.

IEEE transactions on image processing : a publication of the IEEE Signal Processing Society·2008
Same author

Illumination control as a means of enhancing image features in active vision systems.

IEEE transactions on image processing : a publication of the IEEE Signal Processing Society·1995
Same author

An evolution-oriented learning algorithm for the optimal interpolative net.

IEEE transactions on neural networks·1992
Same author

A Novel Surface Reconstruction and Display Method for Cardiac PET Imaging.

IEEE transactions on medical imaging·1984

Area of Science:

  • Machine Learning
  • Artificial Intelligence
  • Data Science

Background:

  • Unsupervised learning methods are crucial for data analysis when labels are absent.
  • Existing clustering algorithms like winner-take-all (WTA) struggle with arbitrary cluster shapes and unknown cluster numbers.

Purpose of the Study:

  • To introduce a novel unsupervised neural network, the cluster-detection-and-labeling (CDL) network.
  • To enable data clustering for arbitrary shapes and an unknown number of clusters using a generic interpoint similarity measure.

Main Methods:

  • The CDL network combines similarity and distance concepts, representing clusters via prototypes.
  • Inner products between input vectors and prototypes, adjusted by distance-dependent thresholds, determine similarity.
  • An iterative reclustering process refines clusters until a termination criterion is met.

More Related Videos

End-To-End Deep Neural Network for Salient Object Detection in Complex Environments
03:31

End-To-End Deep Neural Network for Salient Object Detection in Complex Environments

Published on: December 15, 2023

Related Experiment Videos

Last Updated: Jul 7, 2026

Large-scale Reconstructions and Independent, Unbiased Clustering Based on Morphological Metrics to Classify Neurons in Selective Populations
12:27

Large-scale Reconstructions and Independent, Unbiased Clustering Based on Morphological Metrics to Classify Neurons in Selective Populations

Published on: February 15, 2017

End-To-End Deep Neural Network for Salient Object Detection in Complex Environments
03:31

End-To-End Deep Neural Network for Salient Object Detection in Complex Environments

Published on: December 15, 2023

Main Results:

  • The CDL network successfully clusters data into arbitrary shapes without predefining the number of clusters.
  • Performance comparisons show the CDL network significantly outperforms the WTA network, especially on complex cluster structures.
  • The network effectively assigns appropriate cluster labels to new inputs.

Conclusions:

  • The CDL network offers a robust solution for unsupervised clustering of complex datasets.
  • It provides a flexible alternative to traditional methods like WTA, handling diverse cluster geometries effectively.