Bucket brigade performance. II. Default hierarchies  [1987]

Riolo, R.L.

Access the full text:

Learning systems that operate in environments with huge numbers of states must be able to categorize the states into equivalence classes that can be treated alike. Holland-type classifier systems can learn to categorize states by building default hierarchies of classifiers (rules). However, for default hierarchies to work properly classifiers that implement exception rules must be able to control the system when they are applicable, thus preventing the default rules from making mistakes. This paper presents results that show the standard bucket brigade algorithm does not lead to correct exception rules always winning the competition with the default rules they protect. A simple modification to the bucket brigade algorithm is suggested, and results are presented that show this modification works as desired: default hierarchies can be made to achieve payoff rates as near to optimal as desired.

Other subjects

  • bucket brigade algorithm
  • algorithms
  • computer analysis
  • genetic models

From the journal

Genetic algorithms and their applications : proceedings of the second International Conference on Genetic Algorithms : July 28-31, 1987 at the Massachusetts Institute of Technology, Cambridge, MA