Recent trends in learning classifier systems research

  • Authors:
  • Pier Luca Lanzi;Rick L. Riolo

  • Affiliations:
  • Artificial Intelligence & Robotics Laboratory, Dipartimento di Elettronica e Informazione, Politecnico di Milano, Milan, Italy;Center for Study of Complex Systems, University of Michigan, Ann Arbor, MI

  • Venue:
  • Advances in evolutionary computing
  • Year:
  • 2003

Quantified Score

Hi-index 0.00

Visualization

Abstract

In this chapter we review recent advances and trends in learning classifier systems (LCS) research. These advances fall in three main areas: (i) improved allocation and use of credit assigned to rules, which stems in part from utilizing connections with well-established reinforcement learning algorithms, and from using rule predictive accuracy as the "fitness" value guiding the genetic algorithm's search for better rules; (ii) research on alternative LCS architectures, including alternative rule syntax and semantics, as well as work on both simpler and more complex LCS; and (iii) increases in both the number and the range of LCS applications. We feel these advances have led to the resurgence of LCS research in the past five years, and in a final section we list some of the most immediate challenges facing LCS researchers at this time.