Relational instance based regression for relational reinforcement learning
Kurt Driessens, Jan Ramon · Lirias · 2003
Relational reinforcement learning (RRL) is a Q-learning technique which uses first or-der regression techniques to generalize the Q-function. Both the relational setting and the Q-learning context introduce a number of dif-ficulties which must be dealt with. In this paper we investigate a few different meth-ods that do incremental relational instance based regression and can be used for RRL. This leads us to different approaches which limit both memory consumption and pro-cessing times. We implemented a number of these approaches and experimentally evalu-ated and compared them to each other and an existing RRL algorithm. These experiments show relational instance based regression to work well and to add robustness to RRL. 1.