Evaluation of a family of reinforcement learning cross-domain optimization heuristics

In our participation to the Cross-Domain Heuristic Search Challenge (CHeSC 2011) [1] we developed an approach based on Reinforcement Learning for the automatic, on-line selection of low-level heuristics across different problem domains. We tested different memory models and learning techniques to improve the results of the algorithm. In this paper we report our design choices and a comparison of the different algorithms we developed.