Kernel-based Bayesian methods for Reinforcement Learning (RL) such as Gaussian Process Temporal Difference (GPTD) are particularly promising because they rigorously treat uncertainty in the value func
