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 new data mining scheme using artificial neural networks.

S M Kamruzzaman1, A M Jehad Sarkar

  • 1Department of Electronics Engineering, Hankuk University of Foreign Studies, 89 Wangsan-ri, Mohyeon-myon, Yongin-si, Kyonggi-do, 449-791, Korea. smzaman@hufs.ac.kr

Sensors (Basel, Switzerland)
|December 14, 2011
PubMed
Summary

This study introduces a new algorithm to extract understandable symbolic rules from artificial neural networks (ANNs). This method enhances the explainability of ANNs for data mining classification tasks.

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

GABA content and an antioxidant profile positively correlated with the anticonvulsive activity of Microcos paniculata in acute seizure mice.

Heliyon·2023
Same author

A framework for supervising lifestyle diseases using long-term activity monitoring.

Sensors (Basel, Switzerland)·2012
Same author

Ginsenoside-Rp1 inhibits platelet activation and thrombus formation via impaired glycoprotein VI signalling pathway, tyrosine phosphorylation and MAPK activation.

British journal of pharmacology·2012
Same author

An intelligent tool for activity data collection.

Sensors (Basel, Switzerland)·2011
Same author

Antiplatelet activity of Phellinus baummii methanol extract is mediated by cyclic AMP elevation and inhibition of collagen-activated integrin-α(IIb) β₃ and MAP kinase.

Phytotherapy research : PTR·2011
Same author

Inhibitory effects of Bulnesia sarmienti aqueous extract on agonist-induced platelet activation and thrombus formation involves mitogen-activated protein kinases.

Journal of ethnopharmacology·2010

Area of Science:

  • Computer Science
  • Data Mining
  • Machine Learning

Background:

  • Artificial Neural Networks (ANNs) are powerful machine learning tools but often function as "black boxes", limiting their interpretability in data mining.
  • Current ANN applications lack explicit symbolic rules, hindering verification and interpretation by human experts.

Purpose of the Study:

  • To develop a novel algorithm for extracting concise and accurate symbolic rules from trained ANNs.
  • To improve the explainability and utility of ANNs in data mining classification problems.

Main Methods:

  • A new algorithm was developed to extract symbolic rules from ANNs.
  • The approach focuses on generating easily interpretable rules without sacrificing accuracy.

Main Results:

Keywords:
clusteringconstructive algorithmdata miningneural networkspruningrule extractionsymbolic rulesweight freezing

Related Experiment Videos

  • The proposed algorithm successfully extracts concise symbolic rules from ANNs.
  • Extracted rules demonstrate high accuracy and are comparable to existing methods in terms of rule count and complexity.
  • The method enhances the explainability of ANN predictions for classification tasks.

Conclusions:

  • The novel algorithm effectively addresses the "black box" nature of ANNs in data mining.
  • The extracted symbolic rules provide a clear and verifiable explanation for ANN classifications.
  • This approach significantly improves the practical application of ANNs in data mining classification.