State generalization method with support vector machines in reinforcement learning

The conventional reinforcement learning assumes discrete state space. Therefore, it is necessary to make states discrete in order to handle continuous state environments. However, if a simple discretization is applied, the number of states increases exponentially with the dimension of the state space, and the learning time increases. In this paper, we propose a state generalization that is able to quickly adapt to environments by using Support Vector Machines. We conducted an experiment on the simulation task that navigates a robot to a goal. As a result of the experiment, the proposed method adapted to environment by a few trials.