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

Efficient classification for multiclass problems using modular neural networks.

R Anand1, K Mehrotra, C K Mohan

  • 1Sch. of Comput. and Inf. Sci., Syracuse Univ., NY.

IEEE Transactions on Neural Networks
|January 1, 1995
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

Early radiographic accuracy and intra-operative parameters of patient-specific instrumentation versus conventional technique in knee osteotomies for adolescent deformity correction: A prospective comparative study.

Journal of clinical orthopaedics and trauma·2026
Same author

Correction: Performance of polyvinyl alcohol graphene oxide membrane for microplastic removal in wastewater with an IoT based monitoring approach.

Scientific reports·2026
Same author

Correction: Prediction of waste generation forecast and emission potential on the Erode City solid waste dump yards based on machine learning approach.

Scientific reports·2026
Same author

Airway Fire Prevention and Management in the Operating Room: A Mixed Methods Study on the Effectiveness of an Interprofessional Simulation-based Training Module.

Annals of African medicine·2026
Same author

Comparative evaluation and simulation of blockchain consensus mechanisms for secure and scalable peer to peer energy trading in microgrids.

Scientific reports·2025
Same author

Prediction of waste generation forecast and emission potential on the Erode City solid waste dump yards based on machine learning approach.

Scientific reports·2025
Same journal

Universal perceptron and DNA-like learning algorithm for binary neural networks: LSBF and PBF implementations.

IEEE transactions on neural networks·2013
Same journal

Guest editorial: special section on white box nonlinear prediction models.

IEEE transactions on neural networks·2011
Same journal

Data-based fault-tolerant control of high-speed trains with traction/braking notch nonlinearities and actuator failures.

IEEE transactions on neural networks·2011
Same journal

Guest editorial: special section on data-based control, modeling, and optimization.

IEEE transactions on neural networks·2011
Same journal

Neural network-based multiple robot simultaneous localization and mapping.

IEEE transactions on neural networks·2011
Same journal

Data-driven model-free adaptive control for a class of MIMO nonlinear discrete-time systems.

IEEE transactions on neural networks·2011
See all related articles

Training feedforward neural networks for multiclass problems with backpropagation is slow. A modular network approach, breaking K-class problems into K two-class problems, significantly speeds up convergence and enables training where nonmodular networks fail.

Area of Science:

  • Computer Science
  • Artificial Intelligence
  • Machine Learning

Background:

  • Backpropagation algorithm exhibits slow convergence for multiclass problems.
  • Initial iterations can increase output vector component differences, requiring many subsequent iterations for correction.
  • Small weight changes in standard backpropagation lead to prolonged training times.

Purpose of the Study:

  • To investigate a modular network architecture for improving the training speed of feedforward neural networks on multiclass problems.
  • To address the slow convergence rate associated with the standard backpropagation algorithm.
  • To explore an alternative approach that overcomes limitations of nonmodular networks.

Main Methods:

  • Reduced a K-class problem into K independent two-class problems.

Related Experiment Videos

  • Employed a modular network architecture with a dedicated network for each two-class subproblem.
  • Trained each subnetwork separately.

    • Main Results:

    • Achieved experimental speedups of one order of magnitude compared to nonmodular networks.
    • Demonstrated successful convergence in cases where nonmodular networks failed to converge.
    • Modular approach effectively mitigated initial error increases and accelerated overall training.

    Conclusions:

    • A modular network architecture offers a significant advantage in training speed for multiclass problems.
    • This approach provides a viable solution for problems intractable with standard backpropagation.
    • Modular networks enhance the efficiency and applicability of feedforward neural networks in complex classification tasks.