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

Estimations of error bounds for neural-network function approximators.

N W Townsend1, L Tarassenko

  • 1Neural Networks Research Group, Department of Engineering Science, Oxford University, Oxford, OX1 3PJ, UK.

IEEE Transactions on Neural Networks
|February 7, 2008
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

Self-monitoring of blood pressure following a stroke or transient ischaemic attack (TASMIN5S): a randomised controlled trial.

BMC cardiovascular disorders·2024
Same author

Contactless skin perfusion monitoring with video cameras: tracking pharmacological vasoconstriction and vasodilation using photoplethysmographic changes.

Physiological measurement·2022
Same author

Correction to: Digital messaging to support control for type 2 diabetes (StAR2D): a multicentre randomised controlled trial.

BMC public health·2022
Same author

Digital messaging to support control for type 2 diabetes (StAR2D): a multicentre randomised controlled trial.

BMC public health·2021
Same author

Automated method for detecting and reading seven-segment digits from images of blood glucose metres and blood pressure monitors.

Journal of medical engineering & technology·2019
Same author

Digital blood glucose monitoring could provide new objective assessments of blood glucose control in women with gestational diabetes.

Diabetic medicine : a journal of the British Diabetic Association·2015
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

Neural networks lack confidence measures for function approximation. This study presents a perturbation analysis method to determine output bounds considering input errors and training imperfections, enhancing neural network reliability.

Area of Science:

  • Artificial Intelligence
  • Machine Learning
  • Numerical Analysis

Background:

  • Neural networks are widely applied for function approximation tasks.
  • A significant limitation is the absence of reliable confidence measures for their output estimates.
  • Existing methods often rely on network characteristics like the Hessian matrix to predict output bounds.

Purpose of the Study:

  • To investigate the impact of input vector errors and noise on neural network function approximation.
  • To develop a novel method for determining output bounds that accounts for both input perturbations and training-related weight imperfections.
  • To enhance the trustworthiness and interpretability of neural network predictions.

Main Methods:

  • Utilized perturbation analysis to systematically examine the effect of input errors.

Related Experiment Videos

  • Developed a technique to quantify output bound uncertainties based on input vector noise.
  • Incorporated the influence of imperfections in trained neural network weights into the bound estimation.
  • Demonstrated the proposed method's efficacy through practical application.

    • Main Results:

    • The study successfully established a method to quantify output bounds influenced by input errors.
    • The proposed technique effectively integrates the impact of both input noise and weight imperfections.
    • Demonstrated that output bounds can be reliably estimated by considering these error sources.
    • Provided a more comprehensive understanding of neural network output uncertainty.

    Conclusions:

    • The developed perturbation analysis method offers a robust way to estimate confidence bounds for neural network outputs.
    • Accounting for input vector errors and weight imperfections is crucial for accurate uncertainty quantification.
    • This approach significantly improves the reliability and interpretability of neural network predictions in function approximation.