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

Localized generalization error model and its application to architecture selection for radial basis function neural

Daniel S Yeung1, Wing W Y Ng, Defeng Wang

  • 1Media and Life Science, Department of Computer Science and Technology, Shenzhen Graduate School, Harbin Institute of Technology, Shenzhen 518055, China. csdaniel@comp.polyu.edu.hk

IEEE Transactions on Neural Networks
|January 29, 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

AI framework for multidisease detection via retinal imaging.

Nature medicine·2026
Same author

Effects of dietary allicin supplementation on nutrient digestion and gastrointestinal health of Guizhou black goats.

Animal nutrition (Zhongguo xu mu shou yi xue hui)·2026
Same author

Sustained release of quercetin through homogeneous poly (lactic-co-glycolic acid) microspheres to enhance MSCs biofunctions for regenerative therapy.

Regenerative therapy·2026
Same author

Correction: Effects of <i>Flammulina velutipes</i> mushroom residues on growth performance, apparent digestibility, serum biochemical indicators, rumen fermentation and microbial of Guizhou black goat.

Frontiers in microbiology·2026
Same author

Comparing three induction regimens for nasopharyngeal carcinoma: a propensity score-matched analysis.

Therapeutic advances in medical oncology·2026
Same author

Risk Factors for Osteoporotic Vertebral Compression Fracture and Evaluation of Clinical Outcomes of Minimally Invasive Vertebral Augmentation.

Global spine journal·2026
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

This study introduces a localized generalization error model for improved classifier performance. The new method enhances accuracy and reduces training time by focusing on unseen samples near training data.

Area of Science:

  • Machine Learning
  • Computational Intelligence
  • Pattern Recognition

Background:

  • Current generalization error models provide loose bounds for classifiers, particularly for local learning machines like Support Vector Machines (SVM), Radial Basis Function Neural Networks (RBFNN), and Multilayer Perceptron Neural Networks (MLPNN).
  • These models often consider the entire input space, neglecting the localized nature of these algorithms where unseen samples near training data are more critical.

Purpose of the Study:

  • To propose a localized generalization error model that provides tighter error bounds within a neighborhood of training samples.
  • To develop an architecture selection technique for classifiers based on this localized model, optimizing for maximal coverage of unseen samples within a specified generalization error threshold.

Main Methods:

  • Developed a localized generalization error model utilizing stochastic sensitivity measure to bound generalization error near training samples.

Related Experiment Videos

  • Integrated this model into an architecture selection technique for classifiers.
  • Evaluated the technique on 17 University of California at Irvine (UCI) datasets.

    • Main Results:

    • The proposed localized error model and architecture selection technique consistently outperformed cross-validation (CV), sequential learning, and other methods in testing classification accuracy.
    • Achieved superior results with fewer hidden neurons and reduced training time compared to existing methods.
    • Demonstrated the effectiveness of focusing on localized error bounds for classifier optimization.

    Conclusions:

    • The localized generalization error model offers a more effective approach for classifier architecture selection compared to traditional methods.
    • This technique leads to improved classification accuracy and computational efficiency, especially for local learning machines.
    • The findings suggest a paradigm shift towards localized error analysis in machine learning for better real-world performance.