A Gaussian Processes Reinforcement Learning Method in Large Discrete State Spaces

In order to solve the problem of "curse of dimensionality", which means that the state spaces will grow exponentially in the number of features, in large discrete state spaces in reinforcement learning, a reinforcement learning method based on Gaussian processes is proposed. The Gaussian processes model can represent the distributions of functions, and it can be used to get a distribution of the expectation instead of its value. The experiment result shows that the performance such as speed of convergence and final effect can be improved obviously. The "curse of dimensionality" in large discrete state spaces could be solved to ascertain extent with the GP regression model.