|
(March 25) Proving Release SP4 has been released. In Sp4, we've fixed the remaining problems with keepaway (we hope!). As well, we have added the capability for running proving runs with keepaway! Check out the downloads page for more details!
 Put this button on your home page with the following code: <a href=http://rl-competition.org><img border=0 src="http://rl-competition.org/images/stories/participantbutton.png"></a>
Also, please download and print the RL-Competition poster and put it up in your department! http://rl-competition.org/images/poster.pdf Welcome to the official website for the Second Annual Reinforcement Learning Competition. Building on last year's competition and the benchmarking events that preceded it, this event will be a forum for reinforcement learning researchers to rigorously compare the performance of their methods on a suite of challenging domains. These problems will be significantly more complex and difficult than those used in previous years. They include: the game of Tetris; robot soccer keepaway, based on the RoboCup simulator; a real-time strategy (RTS) game; and a helicopter control problem, based on the work of Andrew Ng and collaborators. See the Domains Page for more details on competition events. In addition, this year's competition will utilize new evaluation paradigms designed to encourage algorithms that generalize well to previously unseen tasks. In particular, each domain will be parameterized and test parameters will differ from those used for training. As a result, only learning algorithms that are robust across a range of parameters can expect to perform well. Algorithms will be compared based on cumulative reward achieved while interacting with the competition domains (online learning). See the Rules Page for more details on these evaluation paradigms. The competition is underway and will conclude in July of 2008 with an event at the 2008 International Conference on Machine Learning in Helsinki, Finland, at which the winners will be announced. Competitors will be invited to attend and present their methods. The event will also feature invited speakers and discussions about the best way to perform empirical comparisons in reinforcement learning and the future of the competition. Please peruse this web site to learn more about the format of the competition and domains it will include. Please direct all questions to the Discussion Page. Previous, related workshops |
