RL = DuelingDQN(
env.n_actions,
env.n_features, #observation/state 的属性，如长宽高
learning_rate=0.01,
reward_decay=0.9,
e_greedy=0.9,
dueling=True,
replace_target_iter=200, # 每 200 步替换一次 target_net 的参数
memory_size=2000, # 记忆上限
# output_graph=True # 是否输出 tensorboard 文件
)
env.after(100, run_maze) #进行强化学习训练
env.mainloop()
# 观看训练时间曲线
his_dueling = np.vstack((episodes, steps))
file = open('his_dueling', 'wb')
pickle.dump(his_dueling, file)
file.close()
plt.plot(his_dueling[0, :],
his_dueling[1, :] - his_dueling[1, 0],
c='b',
label='Dueling DQN')
plt.legend(loc='best') # legend图例，其中’loc’参数有多种，’best’表示自动分配最佳位置
plt.ylabel('total training time')
plt.xlabel('episode')
plt.grid() # 显示网格线 1=True=默认显示；0=False=不显示
plt.show()
RL.plot_cost() # 观看神经网络的误差曲线
# RL.plot_q() # 观看q_max值变化曲线
RL.learn()
# swap observation
observation = observation_
# break while loop when end of this episode
if done:
break
step += 1
print('game over')
env.destroy()
if __name__ == '__main__':
env = Maze()
RL = DuelingDQN(
env.n_actions,
env.n_features,
learning_rate=0.01,
reward_decay=0.9,
e_greedy=0.9,
replace_target_iter=200,
memory_size=2000,
# output_graph=True
)
env.after(100, run_maze)
env.mainloop()
RL.plot_cost()