ABSTRACT
The learning classifier system (LCS) (Goldberg 1989, Holland et al 1986) is often referred to as the primary machine learning technique that employs genetic algorithms (GAs). It is also often described as a production system framework with a genetic algorithm as the primary rule discovery method. However, the details of LCS operation vary widely from one implementation to another. In fact, no standard version of the LCS exists. In many ways, the LCS is more of a concept than an algorithm. To explain details of the LCS concept, this article will begin by introducing the type of machine learning problem most often associated with the LCS. This discussion will be followed by a overview of the LCS, in its most common form. Final sections will introduce the more complex issues involved in LCSs.
