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

A quantitative study of pruning by optimal brain damage

J Gorodkin1, L K Hansen, A Krogh

  • 1CONNECT, Niels Bohr Institute, København, Denmark.

International Journal of Neural Systems
|June 1, 1993
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

No association between peripheral serotonin-gene-related DNA methylation and brain serotonin neurotransmission in the healthy and depressed state.

Clinical epigenetics·2024
Same author

Oral contraceptives and the serotonin 4 receptor: a molecular brain imaging study in healthy women.

Acta psychiatrica Scandinavica·2020
Same author

Chitin-calcium alginate composite fibers for wound care dressings spun from ionic liquid solution.

Journal of materials chemistry. B·2020
Same author

Molecular imaging of neuroinflammation in patients after mild traumatic brain injury: a longitudinal <sup>123</sup> I-CLINDE single photon emission computed tomography study.

European journal of neurology·2019
Same author

Two subgroups of antipsychotic-naive, first-episode schizophrenia patients identified with a Gaussian mixture model on cognition and electrophysiology.

Translational psychiatry·2017
Same author

BDNF val66met association with serotonin transporter binding in healthy humans.

Translational psychiatry·2017

Optimal Brain Damage (OBD) effectively prunes feedforward networks, improving learning curves and identifying smaller, accurate architectures. This method aligns with Occam's Razor, optimizing network design for better performance.

Area of Science:

  • Artificial Intelligence
  • Machine Learning
  • Neural Network Architectures

Background:

  • Feedforward networks are prone to overfitting.
  • Network pruning is essential for efficient model design.
  • Optimal Brain Damage (OBD) is a known pruning technique.

Purpose of the Study:

  • Implement and apply the Optimal Brain Damage (OBD) scheme.
  • Evaluate OBD's impact on the contiguity classification problem.
  • Investigate the relationship between network size and test error.

Main Methods:

  • Applied Optimal Brain Damage (OBD) pruning to feedforward networks.
  • Utilized the contiguity classification problem for evaluation.
  • Analyzed pruned network architectures and their corresponding test errors.

Related Experiment Videos

Main Results:

  • OBD implementation improved the learning curve (reduced test error).
  • Smaller networks, pruned via OBD, exhibited the lowest test error, supporting Occam's Razor.
  • A novel heuristic for selecting the smallest successful architecture was proposed and validated.

Conclusions:

  • OBD is an effective method for pruning feedforward networks.
  • The study validates the principle of Occam's Razor in neural network design.
  • The proposed heuristic efficiently optimizes network architecture selection.