A Novel Reinforcement Learning Architecture for Continuous State and Action Spaces

Figure 1

The actor-critic architecture.