Resolving the Conflict between Competitive and Cooperative Behavior in Michigan-Type Fuzzy Classifier Systems

A vital problem for fuzzy classifier systems of the Michigan type is the conflict of competition and cooperation of rules. Whereas the classical approach of a classifier system circumvents this complicacy by the total lack of collaboration of classifiers, the fuzzification approach has to deal with it. This paper proposes a solution to this dilemma by introducing a special encoding of the classifiers and by performing a selection algorithm on a sub-group of firing rules.