Report on the 2008 Reinforcement Learning Competition

Abstract

This article reports on the 2008 Reinforcement Learning Competition,
  which began in November 2007 and ended with a workshop at the
  International Conference on Machine Learning (ICML) in July 2008 in
  Helsinki, Finland.  Researchers from around the world developed
  reinforcement learning agents to compete in six problems of various
  complexity and difficulty.  The competition employed fundamentally
  redesigned evaluation frameworks that, unlike those in previous
  competitions, aimed to systematically encourage the submission of
  robust learning methods. We describe the unique challenges of
  empirical evaluation in reinforcement learning and briefly review
  the history of the previous competitions and the evaluation
  frameworks they employed.  We also describe the novel frameworks
  developed for the 2008 competition as well as the software
  infrastructure on which they rely.  Furthermore, we describe the six
  competition domains and present a summary of selected competition
  results.  Finally, we discuss the implications of these results and
  outline ideas for the future of the competition.