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

Accuracy-based learning classifier systems: models, analysis and applications to classification tasks.

Ester Bernadó-Mansilla1, Josep M Garrell-Guiu

  • 1Computing Sciences, Bell Laboratories, Lucent Technologies, 600-700 Mountain Avenue, Murray Hill, NJ 07974-0636, USA. esterb@salleurl.edu

Evolutionary Computation
|October 16, 2003
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 journal

Computing Optimal Populations for Binary Problems using Logic Minimization.

Evolutionary computation·2026
Same journal

Enhancing Generalization and Scalability for Multi-Objective Optimization with Population Pre-Training.

Evolutionary computation·2026
Same journal

XCS for Sequential Perceptual Aliasing in Multi-Step Decision Making.

Evolutionary computation·2026
Same journal

A dynamic multi-objective evolutionary algorithm using dual-space prediction and surrogate-based sampling.

Evolutionary computation·2026
Same journal

Adapting MOEA/D to CMA-ES for Dealing with Ill-conditioned Multiobjective Problems.

Evolutionary computation·2026
Same journal

Editorial of the Special Issue: Parallel Problem Solving from Nature PPSN 2024 Extended Versions of Best Paper Candidates.

Evolutionary computation·2026
See all related articles

This study compares two accuracy-based Learning Classifier Systems (LCS), XCS and UCS, for classification tasks. UCS, using supervised learning for fitness, shows improved efficiency and accuracy, especially for complex datasets.

Area of Science:

  • Artificial Intelligence
  • Machine Learning
  • Data Mining

Background:

  • Learning Classifier Systems (LCS), particularly XCS, are effective for classification and data mining.
  • Existing methods often represent knowledge through complete action maps, which can be inefficient.

Purpose of the Study:

  • To investigate two accuracy-based LCS models: XCS and a novel system, UCS.
  • To compare their performance on diverse classification problems.
  • To analyze the influence of fitness pressure on classifier accuracy.

Main Methods:

  • Investigated XCS and proposed UCS, an alternative LCS.
  • UCS evolves a 'best action map' more efficiently than XCS's 'complete action map'.
  • Compared reinforcement learning (XCS) versus supervised learning (UCS) for fitness determination.

Related Experiment Videos

Main Results:

  • UCS demonstrated more efficient evolution of action maps.
  • Significant differences were observed in how fitness pressure drives accuracy.
  • UCS, with supervised fitness, is recommended for multi-class and imbalanced datasets.

Conclusions:

  • Accuracy-based LCS dynamics are better understood through this comparative analysis.
  • UCS offers a more efficient and potentially more accurate approach for specific classification challenges.
  • Further research into LCS complexity and improvements for classification tasks is suggested.