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

Temporally asymmetric learning supports sequence processing in multi-winner self-organizing maps.

Reiner Schulz1, James A Reggia

  • 1Departments of Computer Science and Neurology, University of Maryland, College Park, MD 20742, U.S.A. rschulz@cs.umd.edu

Neural Computation
|March 10, 2004
PubMed
Summary

Modified Kohonen self-organizing maps (SOMs) learn unique spatial representations for temporal sequences. This biologically inspired approach enhances sequence pattern recognition and visualization capabilities.

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

Performance and mental workload dynamics during in-person and virtually-based remote learning of action sequences.

Ergonomics·2026
Same author

Neurocognitive assessment under various human-robot teaming environments.

Annual International Conference of the IEEE Engineering in Medicine and Biology Society. IEEE Engineering in Medicine and Biology Society. Annual International Conference·2025
Same author

The DEAD box RNA helicase DDX42 is an intrinsic inhibitor of positive-strand RNA viruses.

EMBO reports·2022
Same author

Tunable Neural Encoding of a Symbolic Robotic Manipulation Algorithm.

Frontiers in neurorobotics·2021
Same author

NeuroLISP: High-level symbolic programming with attractor neural networks.

Neural networks : the official journal of the International Neural Network Society·2021
Same author

Extraction and high-throughput sequencing of oak heartwood DNA: Assessing the feasibility of genome-wide DNA methylation profiling.

PloS one·2021

Area of Science:

  • Computational neuroscience
  • Artificial intelligence
  • Machine learning

Background:

  • Traditional Kohonen self-organizing maps (SOMs) excel at feature mapping but struggle with temporal sequence representation.
  • Developing models that can process and represent temporal data is crucial for advancing pattern recognition.

Purpose of the Study:

  • To investigate the efficacy of modified self-organizing maps (SOMs) in learning unique representations of temporal sequences.
  • To explore biologically inspired extensions for SOMs to handle temporal data while maintaining map formation.

Main Methods:

  • Implemented two extensions to traditional SOMs: multiple simultaneous 'winner' selection and local intramap connections trained with a temporally asymmetric Hebbian learning rule.
  • Trained the extended SOM with variable-length temporal sequences of phoneme feature vectors representing phonetic transcriptions.

Related Experiment Videos

  • Evaluated the model's ability to transform input sequences into unique spatial representations on the map.

    • Main Results:

    • The modified SOM successfully learned unique spatial representations for distinct temporal sequences.
    • Training improved the uniqueness of spatial representations while preserving map formation based on input patterns.
    • A significant correlation was found between the spatial representation closeness and sequence similarity.

    Conclusions:

    • Modified SOMs can effectively learn unique representations of temporal sequences, supporting map formation.
    • These extended SOMs show potential as visualization tools for temporal sequences.
    • The model could serve as a valuable preprocessor for sequence pattern recognition systems.