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

Dynamics of the evolution of learning algorithms by selection.

Juan Pablo Neirotti1, Nestor Caticha

  • 1Departamento de Física Geral, Instituto de Física, Universidade de São Paulo, Rua do Matão Travessa R 187, Brazil.

Physical Review. E, Statistical, Nonlinear, and Soft Matter Physics
|June 6, 2003
PubMed
Summary

This study explores artificial learning system evolution using genetic programming. Key findings reveal a strict temporal order in the emergence of learning structures, with surprise detection preceding performance assessment for improved generalization.

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

Legislative rebellions and impeachments in a neural network society.

Physical review. E·2024
Same author

Entropic Dynamics in Neural Networks, the Renormalization Group and the Hamilton-Jacobi-Bellman Equation.

Entropy (Basel, Switzerland)·2020
Same author

Exploration and recency as the main proximate causes of probability matching: a reinforcement learning analysis.

Scientific reports·2017
Same author

Backward renormalization-group inference of cortical dipole sources and neural connectivity efficacy.

Physical review. E·2017
Same author

t-Test at the Probe Level: An Alternative Method to Identify Statistically Significant Genes for Microarray Data.

Microarrays (Basel, Switzerland)·2016
Same author

Supervised Variational Relevance Learning, An Analytic Geometric Feature Selection with Applications to Omic Datasets.

IEEE/ACM transactions on computational biology and bioinformatics·2015

Area of Science:

  • Artificial Intelligence
  • Machine Learning
  • Evolutionary Computation

Background:

  • Studying the evolution of artificial learning systems provides insights into optimizing algorithm design.
  • Supervised learning scenarios are crucial for developing effective neural network classifiers.

Purpose of the Study:

  • To investigate the evolutionary dynamics of learning algorithms using genetic programming.
  • To identify functional structures that drive improvements in artificial learning processes.

Main Methods:

  • Utilizing genetic programming to evolve populations of programs for neural network classifiers.
  • Monitoring evolutionary dynamics through phenotypic and genotypic entropies.
  • Analyzing the temporal order of emergent functional structures.

Related Experiment Videos

Main Results:

  • Identified specific combinations of variables and operators that enhance rule extraction and learning schedules.
  • Discovered structures signaling surprise (difference between predicted and correct classification) and performance assessment.
  • Observed a strict temporal order: surprise-measuring structures emerge before performance-measuring structures.

Conclusions:

  • The evolution of artificial learning systems exhibits distinct dynamical transitions.
  • A specific temporal order in the discovery of functional structures is critical for effective learning.
  • Asymptotic generalization ability approaches Bayesian optimal results.