Learning classifier system, learning classifier generation method and program

摘要

PROBLEM TO BE SOLVED: To optimize a classifier when series data is classified by the classifier.SOLUTION: A learning classifier system obtains a collation set selected from a classifier that a classifier storage section has stored based on input series data, obtains an action set based on the value of an action section of the classifier in the collation set, updates the fitness value of the classifier in the action set by learning to subsume it when a plurality of classifiers in the action set can be subsumed into a single classifier, and selects or modifies the subsumed classifier based on a certain condition. In addition, the system reduces a condition section that the classifier owns to make the reduced classifier update the classifier stored by the storage section, and further, when an appropriate classifier is not available, creates a new classifier and store it in the storage section.SELECTED DRAWING: Figure 1