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

Regularising neural networks using flexible multivariate activation function.

Mirko Solazzi1, Aurelio Uncini

  • 1Dipartimento di Elettronica e Automatica, University of Ancona Via Brecce Bianche, 60131 Ancona, Italy.

Neural Networks : the Official Journal of the International Neural Network Society
|March 24, 2004
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

Continual learning with invertible generative models.

Neural networks : the official journal of the International Neural Network Society·2023
Same author

A Meta-Learning Approach for Training Explainable Graph Neural Networks.

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

A Probabilistic Re-Intepretation of Confidence Scores in Multi-Exit Models.

Entropy (Basel, Switzerland)·2022
Same author

Structured Ensembles: An approach to reduce the memory footprint of ensemble methods.

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

An Information-Theoretic Perspective on Proper Quaternion Variational Autoencoders.

Entropy (Basel, Switzerland)·2021
Same author

Adaptive Propagation Graph Convolutional Network.

IEEE transactions on neural networks and learning systems·2020
Same journal

Exploiting audio-visual modalities in videos: Object detection via multi-stage bilateral coupling network.

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

Reliability-aware modality completion with cross-modal distillation for federated learning with missing modalities.

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

IGFD-Net: Illumination-guided frequency decoupling for polarization image fusion.

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

Multiple-Strategies dung beetle optimizer and its applications in engineering optimization and bankruptcy prediction.

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

Aggregating global-scale pixel-wise forgery cues within a graph.

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

Finite-Time intermittent control for secure synchronization of Neutral-Type stochastic delayed neural networks under aperiodic DoS attacks.

Neural networks : the official journal of the International Neural Network Society·2026
See all related articles

This study introduces a novel neural network architecture using adaptive spline activation functions. This design enhances generalization and learning speed for complex problems.

Area of Science:

  • Artificial Intelligence
  • Machine Learning
  • Computational Neuroscience

Background:

  • Traditional neural networks often struggle with high structural complexity relative to problem complexity.
  • Generalised regularisation networks offer a theoretical framework for improved network generalization.
  • Adaptive activation functions are key to enhancing learning efficiency and performance.

Purpose of the Study:

  • To introduce a novel general neural network architecture.
  • To leverage nonlinear flexible multivariate functions for improved network performance.
  • To develop a learning algorithm that enhances generalization and convergence speed.

Main Methods:

  • The proposed architecture utilizes multi-dimensional adaptive cubic spline basis activation functions.

Related Experiment Videos

  • Each activation function aggregates information from a subset of previous layer outputs.
  • A specialized learning algorithm adapts local parameters of the activation function.

    • Main Results:

    • The new architecture demonstrates significantly reduced structural complexity compared to problem complexity.
    • The adaptive spline activation functions improve network generalization capabilities.
    • The learning algorithm accelerates the convergence of the training process.

    Conclusions:

    • The proposed neural network architecture offers an effective approach for complex problems.
    • The use of adaptive spline activation functions leads to improved generalization and faster learning.
    • This framework provides a promising direction for the development of efficient neural networks.