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

Figure 12

Experimental results. (a) Hit rate of DQN. (b) Episode average reward of DQN. (c) Hit rate of double DQN. (d) Episode average reward of double DQN. (e) Hit rate of dueling DQN. (f) Episode average reward of dueling DQN.

(a)

(b)

(c)

(d)

(e)

(f)