Learning Classifier System

A Learning Classifier System (LCS, German learning classifier systems ) is a machine learning process in which evolutionary algorithms are combined with classic learning algorithms in order to generate adaptive systems.

These systems are based on rules that traditionally take the form of the conditional (if-then) statement . These rules are (the best behavior because of a particular input English input run). To do this, they are adapted using an evolutionary algorithm. In principle, the Pittsburgh-Type-LCS , in which separate populations of rules are adapted, are to be distinguished from the Michigan-Style-LCS . In the latter, individual rules are evolved and not complete sets of rules.

literature

• Larry Bull, Tim Kovacs: Foundations of Learning Classifier Systems , Springer-Verlag , Berlin Heidelberg New York 2005, ISBN 3-540-25073-5