Inthesummerof2008, reinforcementlearningresearchersfromaroundtheworld gathered in the north of France for a week of talks and discussions on reinfor- ment learning, on how it could be made more e?cient, applied to a broader range of applications, and utilized at more abstract and symbolic levels. As a participant in this 8th European Workshop on Reinforcement Learning, I was struck by both the quality and quantity of the presentations. There were four full days of short talks, over 50 in all, far more than there have been at any p- vious meeting on reinforcement learning in Europe, or indeed, anywhere else in the world. There was an air of excitement as substantial progress was reported in many areas including Computer Go, robotics, and ?tted methods. Overall, the work reported seemed to me to be an excellent, broad, and representative sample of cutting-edge reinforcement learning research. Some of the best of it is collected and published in this volume. The workshopandthe paperscollectedhere provideevidence thatthe ?eldof reinforcement learning remains vigorous and varied. It is appropriate to re?ect on some of the reasons for this. One is that the ?eld remains focused on a pr- lem - sequential decision making - without prejudice as to solution methods. Another is the existence of a common terminology and body of theory

Reinforcement learning is the problem faced by an agent that must learn behavior through trial-and-error interactions with a dynamic environment. Usually, the problem to be solved contains subtasks that repeat at different regions of the state space. Without any guidance an agent has to learn the solutions of all subtask instances independently, which in turn degrades the performance of the learning process. In this work, we propose two novel approaches for building the connections between different regions of the search space. The first approach efficiently discovers abstractions in the form of conditionally terminating sequences and represents these abstractions compactly as a single tree structure; this structure is then used to determine the actions to be executed by the agent. In the second approach, a similarity function between states is defined based on the number of common action sequences; by using this similarity function, updates on the action-value function of a state are re�ected to all similar states that allows experience acquired during learning be applied to a broader context. The effectiveness of both approaches is demonstrated empirically over various domains.