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

Extracting rules from neural networks by pruning and hidden-unit splitting

R Setiono1

  • 1Department of Information Systems and Computer Science, National University of Singapore, Kent Ridge, Republic of Singapore.

Neural Computation
|January 1, 1997
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

Recursive neural network rule extraction for data with mixed attributes.

IEEE transactions on neural networks·2008
Same author

A connectionist approach to generating oblique decision trees.

IEEE transactions on systems, man, and cybernetics. Part B, Cybernetics : a publication of the IEEE Systems, Man, and Cybernetics Society·2008
Same author

Extracting M-of-N rules from trained neural networks.

IEEE transactions on neural networks·2008
Same author

Extraction of rules from artificial neural networks for nonlinear regression.

IEEE transactions on neural networks·2008
Same author

Feedforward neural network construction using cross validation.

Neural computation·2001
Same author

A comparison between two neural network rule extraction techniques for the diagnosis of hepatobiliary disorders.

Artificial intelligence in medicine·2000

This study introduces a novel algorithm for extracting understandable rules from neural networks. The method prunes networks to identify key inputs, enabling accurate rule generation for complex data.

Area of Science:

  • Artificial Intelligence
  • Machine Learning
  • Computational Biology

Background:

  • Standard three-layer feedforward neural networks can be complex and opaque.
  • Extracting human-readable rules from trained neural networks is challenging.
  • Understanding the decision-making process of neural networks is crucial for applications in science and engineering.

Purpose of the Study:

  • To propose a novel algorithm for extracting symbolic rules from trained feedforward neural networks.
  • To enhance the interpretability of neural network models.
  • To demonstrate the algorithm's effectiveness on real-world datasets.

Main Methods:

  • The algorithm prunes trained neural networks to remove redundant connections and identify relevant inputs.

Related Experiment Videos

  • It generates rules by analyzing activation values at hidden units.
  • If a hidden unit has too many inputs, it is split, and a new subnetwork is trained and pruned iteratively.

    • Main Results:

    • The algorithm successfully extracts compact rule sets from complex datasets.
    • The extracted rules demonstrate high predictive accuracy.
    • The method was validated using real-world data from molecular biology and signal processing.

    Conclusions:

    • The proposed algorithm provides an effective method for rule extraction from neural networks.
    • This approach enhances the interpretability of complex machine learning models.
    • The technique shows promise for applications requiring both predictive accuracy and transparency.