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

PRSOM: a new visualization method by hybridizing multidimensional scaling and self-organizing map.

Sitao Wu1, Tommy W S Chow

  • 1Department of Electric Engineering, City University of Hong Kong, Hong Kong, China.

IEEE Transactions on Neural Networks
|December 14, 2005
PubMed
Summary

This study introduces Probabilistic Regularized Self-Organizing Maps (PRSOM) for improved data visualization. PRSOM enhances topological structure preservation and interneuron distance representation, outperforming existing dimension reduction techniques.

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

Case report: Deep sequencing and long-read genome sequencing refine prior genetic analyses in families with apparent gonadal mosaicism in <i>PIK3CD</i>-related activated PI3K delta syndrome.

Frontiers in immunology·2024
Same author

STAR-RL: Spatial-Temporal Hierarchical Reinforcement Learning for Interpretable Pathology Image Super-Resolution.

IEEE transactions on medical imaging·2024
Same author

Causal Disentanglement Domain Generalization for time-series signal fault diagnosis.

Neural networks : the official journal of the International Neural Network Society·2024
Same author

ExpGCN: Review-aware Graph Convolution Network for explainable recommendation.

Neural networks : the official journal of the International Neural Network Society·2022
Same author

Partial Sequence Labeling With Structured Gaussian Processes.

IEEE transactions on neural networks and learning systems·2022
Same author

Modeling Self-Representation Label Correlations for Textual Aspects and Emojis Recommendation.

IEEE transactions on neural networks and learning systems·2022

Area of Science:

  • Data Science
  • Machine Learning
  • Computational Statistics

Background:

  • Self-Organizing Maps (SOM) offer nonlinear dimensionality reduction and data visualization but struggle with preserving interneuron distances, potentially degrading visualization quality.
  • Existing methods like Visualization-induced SOM (ViSOM) lack a formal cost function and principled weight-updating rules, limiting their effectiveness.
  • Multidimensional Scaling (MDS) preserves interpoint distances but is not a neural network approach.

Purpose of the Study:

  • To propose a novel Probabilistic Regularized Self-Organizing Map (PRSOM) for enhanced data visualization.
  • To integrate the strengths of SOM and MDS by preserving both topological structures and interneuron distances.
  • To provide a principled framework for weight-updating with a defined cost function.

Related Experiment Videos

Main Methods:

  • Developed Probabilistic Regularized Self-Organizing Maps (PRSOM) by incorporating a cost function and soft assignment mechanism.
  • Integrated principles from Multidimensional Scaling (MDS) to maintain interneuron distances from input space to output space.
  • Utilized soft assignment in PRSOM, contrasting with the hard assignment in ViSOM, to improve visualization fidelity.

Main Results:

  • PRSOM demonstrates superior visualization effects compared to traditional SOM and ViSOM.
  • The method effectively preserves interneuron distances, leading to more accurate topological representation in the reduced dimension space.
  • Experimental results validate the effectiveness of PRSOM over other dimension reduction techniques.

Conclusions:

  • PRSOM offers a principled and effective approach to nonlinear dimensionality reduction and data visualization.
  • The integration of SOM and MDS principles in PRSOM leads to significant improvements in preserving data structure and distances.
  • PRSOM represents a valuable advancement for visualizing complex datasets, enhancing interpretability and analytical insights.