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

Using additive noise in back-propagation training.

L Holmstrom1, P Koistinen

  • 1Rolf Nevanlinna Inst., Helsinki Univ.

IEEE Transactions on Neural Networks
|January 1, 1992
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

Primordial Symptoms and ECG Among Sudden Cardiac Death Victims Due to Primary Myocardial Fibrosis.

Clinical cardiology·2025
Same author

Age-related trends of ischemic sudden cardiac death in women.

International journal of cardiology·2024
Same author

Characteristics of women with ischemic sudden cardiac death.

Annals of medicine·2023
Same author

Determinants of survival in sudden cardiac arrest manifesting with pulseless electrical activity.

Resuscitation·2023
Same author

ReCOV: recovery and rehabilitation during and after COVID-19 - a study protocol of a longitudinal observational study on patients, next of kin and health care staff.

BMC sports science, medicine & rehabilitation·2021
Same author

United Nations Secretary-General's Mechanism.

Revue scientifique et technique (International Office of Epizootics)·2018
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

Adding noise to training data can enhance neural network generalization. This method, viewed as kernel estimation, offers rules for noise selection in pattern classification and error-prone data mapping.

Area of Science:

  • Artificial Intelligence
  • Machine Learning
  • Neural Networks

Background:

  • Neural networks often struggle with generalization, leading to poor performance on unseen data.
  • Improving generalization is crucial for reliable application of neural networks in various domains.

Purpose of the Study:

  • To investigate the impact of additive noise on neural network generalization.
  • To develop mathematically justified guidelines for noise introduction during training.

Main Methods:

  • Analyzing feedforward layered neural networks trained with the back-propagation algorithm.
  • Interpreting additive noise as a method for kernel estimation of training data distribution.
  • Considering applications in pattern classification and data corrupted by measurement errors.

Related Experiment Videos

Main Results:

  • Additive noise can improve neural network generalization, though not universally guaranteed.
  • Mathematically derived rules for selecting noise characteristics are proposed.
  • Asymptotic consistency results are established using mathematical statistics.

Conclusions:

  • The proposed method of introducing additive noise offers a principled approach to enhance neural network generalization.
  • Numerical simulations validate the practical applicability of this training strategy.