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

Even on Finite Test Sets Smaller Nets may Perform Better.

THOMAS ELSKEN1

  • 1Universität Osnabrück, Germany

Neural Networks : the Official Journal of the International Neural Network Society
|March 1, 1997
PubMed
Summary

Smaller feedforward neural networks can outperform larger ones. Conditions on the transfer function f ensure optimal smaller nets perform better on finite test sets, regardless of larger net complexity.

Related Concept Videos

You might also read

Related Articles

Articles linked to this work by shared authors, journal, and citation graph.

Sort by
Same journal

Injecting reasoning into vision-language models via weight-decomposed merging.

Neural networks : the official journal of the International Neural Network Society·2026
Same journal

A Non-equilibrium thermodynamic framework for neural networks: A principled correspondence and parameter dynamics.

Neural networks : the official journal of the International Neural Network Society·2026
Same journal

Dual-pathway mask ranking guided selective fine-tuning for backdoor purification.

Neural networks : the official journal of the International Neural Network Society·2026
Same journal

Discriminative transfer feature learning for unsupervised domain adaptation.

Neural networks : the official journal of the International Neural Network Society·2026
Same journal

ARetinex-Net: Low-light image enhancement with adaptive retinex model and global illumination representation.

Neural networks : the official journal of the International Neural Network Society·2026
Same journal

A Multi-Task Learning Framework with Physics Embedded for Signal Reconstruction and State Prediction in Underground Multi-Robot Systems.

Neural networks : the official journal of the International Neural Network Society·2026

Area of Science:

  • Artificial Intelligence
  • Machine Learning
  • Computational Neuroscience

Background:

  • Feedforward multilayered neural networks are fundamental to machine learning.
  • Understanding the relationship between network structure and mapping capabilities is crucial.
  • The impact of network depth on performance is a key research question.

Purpose of the Study:

  • To establish conditions for unique definition of feedforward neural nets by their mappings.
  • To identify sufficient conditions for smaller neural networks to outperform larger ones.
  • To demonstrate the existence of test sets where smaller nets achieve lower error.

Main Methods:

  • Analysis of transfer function properties in feedforward neural networks.
  • Theoretical investigation of network structure and performance.
  • Development of conditions for comparative network performance evaluation.

Main Results:

  • Conditions are provided for feedforward neural nets to be uniquely defined by their mappings.
  • Sufficient conditions are established on the transfer function f for smaller nets to excel.
  • Demonstration that for certain transfer functions, smaller nets can achieve superior performance on finite test sets.

Conclusions:

  • Network architecture and transfer function choice significantly impact performance.
  • Smaller, optimally configured neural networks can be more effective than larger, arbitrarily configured ones.
  • The study provides theoretical underpinnings for designing more efficient neural network architectures.

Related Experiment Videos