A Novel Reinforcement Learning Architecture for Continuous State and Action Spaces

Figure 7

Learning curves of the SARSA A²C algorithm using different numbers of radial basis functions.