Learning classifier systems

J. Holland introduced the concept of Learning Classifier Systems (LCS) more than 30 years ago as an extension to evolutionary computing [11:10].

Learning classifier systems are a kind of rule-based system with general mechanisms for processing rules in parallel, for the adaptive generation of new rules, and for testing the effectiveness of new rules.

However, the concept started to get the attention of computer scientists only a few years ago, with the introduction of several variants of the original concept, including Extended Learning Classifiers (XCS). Learning classifier systems are interesting because they combine rules, reinforcement learning, and genetic algorithms.

Note

Disclaimer

The implementation of the extended learning classifier is presented for informational purposes only. Validating XCS against a known and labeled population of rules is a very significant endeavor. The source code snippet is presented only to illustrate the different components of...