© Springer-Verlag Berlin Heidelberg 2002. Learning classifier systems represent a technique by which various characteristics of a given problem space may be deduced and presented to the user in a readable format. In this paper we present results from the use of XCS on simple tasks with the general multi-variable features typically found in problems addressed by an Interactive Evolutionary Design process. That is, we examine the behaviour of XCS with versions of a well-known single-step task and consider the speed of learning and the ability to respond to changes. We introduce a simple form of supervised learning for XCS with the aim of improving its performance with respect to these two measures. Results show that improvements can be made under the new learning scheme and that other aspects of XCS can also play a significant role.
Bull, L., Wyatt, D., & Parmee, I. (2002). Initial modifications to XCS for use in interactive evolutionary design. In J. J. Merelo, P. Adamidis, & H. Beyer (Eds.), Parallel Problem Solving from Nature—PPSN VII. , (568-577). https://doi.org/10.1007/3-540-45712-7_55