深度强化学习-策略梯度算法(Reinforce)代码

引言

本文主要采用Pytorch来实现策略梯度算法，算法的原理可以参考我的这篇博文：深度强化学习-策略梯度算法推导，里面对该算法进行了详细推导。如果想深入理解策略梯度算法公式，可以参考我的另一篇博文：深度强化学习-策略梯度算法深入理解，里面将其与手写数字识别问题进行了类比，深入剖析了策略梯度算法公式。代码已经上传到我的Github上，喜欢的话可以点个小星星噢。

代码：/indigoLovee/Reinforce_pytorch

1 Reinforce算法

强化学习的目标在于最大化累积奖励。采用含参函数近似最优策略，沿着策略梯度的方向，更新策略参数，可以实现累积奖励最大化。策略梯度定理如下：

策略梯度定理：

Reinforce算法的伪代码如下：

2 Reinforce算法实现

Reinforce算法代码如下(Reinforce_discrete.py脚本)：

import torch as Timport torch.nn as nnimport torch.optim as optimfrom torch.distributions import Categoricaldevice = T.device("cuda:0" if T.cuda.is_available() else "cpu")class PolicyNetwork(nn.Module):def __init__(self, alpha, state_dim, action_dim, fc1_dim, fc2_dim):super(PolicyNetwork, self).__init__()self.fc1 = nn.Linear(state_dim, fc1_dim)self.fc2 = nn.Linear(fc1_dim, fc2_dim)self.prob = nn.Linear(fc2_dim, action_dim)self.optimizer = optim.Adam(self.parameters(), lr=alpha)self.to(device)def forward(self, state):x = T.relu(self.fc1(state))x = T.relu(self.fc2(x))prob = T.softmax(self.prob(x), dim=-1)return probdef save_checkpoint(self, checkpoint_file):T.save(self.state_dict(), checkpoint_file, _use_new_zipfile_serialization=False)def load_checkpoint(self, checkpoint_file):self.load_state_dict(T.load(checkpoint_file))class Reinforce:def __init__(self, alpha, state_dim, action_dim, fc1_dim, fc2_dim, ckpt_dir, gamma=0.99):self.gamma = gammaself.checkpoint_dir = ckpt_dirself.reward_memory = []self.log_prob_memory = []self.policy = PolicyNetwork(alpha=alpha, state_dim=state_dim, action_dim=action_dim,fc1_dim=fc1_dim, fc2_dim=fc2_dim)def choose_action(self, observation):state = T.tensor([observation], dtype=T.float).to(device)probabilities = self.policy.forward(state)dist = Categorical(probabilities)action = dist.sample()log_prob = dist.log_prob(action)self.log_prob_memory.append(log_prob)return action.item()def store_reward(self, reward):self.reward_memory.append(reward)def learn(self):G_list = []G_t = 0for item in self.reward_memory[::-1]:G_t = self.gamma * G_t + itemG_list.append(G_t)G_list.reverse()G_tensor = T.tensor(G_list, dtype=T.float).to(device)loss = 0for g, log_prob in zip(G_tensor, self.log_prob_memory):loss += -g * log_probself.policy.optimizer.zero_grad()loss.backward()self.policy.optimizer.step()self.reward_memory.clear()self.log_prob_memory.clear()def save_models(self, episode):self.policy.save_checkpoint(self.checkpoint_dir + 'Reinforce_policy_{}.pth'.format(episode))print('Saved the policy network successfully!')def load_models(self, episode):self.policy.load_checkpoint(self.checkpoint_dir + 'Reinforce_policy_{}.pth'.format(episode))print('Loaded the policy network successfully!')

算法仿真环境为gym库中的LunarLander-v2，因此需要先配置好gym库。进入Anaconda3中对应的Python环境中，执行下面的指令

pip install gym

但是，这样安装的gym库只包括少量的内置环境，如算法环境、简单文字游戏和经典控制环境，无法使用LunarLander-v2。因此还需要安装一些其他依赖项，具体可以参考我的这篇博文：AttributeError: module ‘gym.envs.box2d‘ has no attribute ‘LunarLander‘ 解决办法。

让智能体在环境中训练3000轮，训练代码如下(train.py脚本)：

import gymimport numpy as npimport argparsefrom utils import plot_learning_curvefrom Reinforce_discrete import Reinforceparser = argparse.ArgumentParser()parser.add_argument('--max_episodes', type=int, default=3000)parser.add_argument('--reward_path', type=str, default='./output_images/reward.png')parser.add_argument('--ckpt_dir', type=str, default='./checkpoints/Reinforce_discrete/')args = parser.parse_args()def main():env = gym.make('LunarLander-v2')agent = Reinforce(alpha=0.0005, state_dim=env.observation_space.shape[0],action_dim=env.action_space.n, fc1_dim=128, fc2_dim=128,ckpt_dir=args.ckpt_dir, gamma=0.99)total_rewards, avg_rewards = [], []for episode in range(args.max_episodes):total_reward = 0done = Falseobservation = env.reset()while not done:action = agent.choose_action(observation)observation_, reward, done, info = env.step(action)agent.store_reward(reward)total_reward += rewardobservation = observation_agent.learn()total_rewards.append(total_reward)avg_reward = np.mean(total_rewards[-100:])avg_rewards.append(avg_reward)print('EP:{} reward:{} avg_reward:{}'.format(episode + 1, total_reward, avg_reward))if (episode + 1) % 300 == 0:agent.save_models(episode + 1)episodes = [i for i in range(args.max_episodes)]plot_learning_curve(episodes, avg_rewards, 'Reward', 'reward', args.reward_path)if __name__ == '__main__':main()

训练时还会用到画图函数和创建文件夹函数，它们均放置在utils.py脚本中，具体代码如下：

import osimport matplotlib.pyplot as pltimport numpy as npdef plot_learning_curve(episodes, records, title, ylabel, figure_file):plt.figure()plt.plot(episodes, records, linestyle='-', color='r')plt.title(title)plt.xlabel('episode')plt.ylabel(ylabel)plt.show()plt.savefig(figure_file)def create_directory(path: str, sub_dirs: list):for sub_dir in sub_dirs:if os.path.exists(path + sub_dir):print(path + sub_dir + ' is already exist!')else:os.makedirs(path + sub_dir, exist_ok=True)print(path + sub_dir + ' create successfully!')def scale_action(action, high, low):action = np.clip(action, -1, 1)weight = (high - low) / 2bias = (high + low) / 2action_ = action * weight + biasreturn action_

3 仿真结果

LunarLander-v2环境中动作空间为离散形式，仿真结果如下图所示。

可以看出累积奖励在不断上升，说明通过策略梯度算法，可以不断改善智能体的策略。

其实，策略梯度算法主要针对的是连续问题。因此，我们在连续动作空间的环境LunarLanderContinuous-v2中对Reinforce算法进行了测试，但是测试效果不太理想，这部分的代码也已经放在我的Github里面，这里就不贴在博文中了。后面我们会介绍策略梯度算法的改善版本，敬请期待把！