A Dynamic Adjusting Reward Function Method for Deep Reinforcement Learning with Adjustable Parameters

Figure 8

Accumulated reward per episode of DQN and DQN with MSR.