机器学习---Deep Q-Network (DQN)

1. RL_brain

import numpy as np
import pandas as pd
import tensorflow as tf

np.random.seed(1)
tf.set_random_seed(1)

# Deep Q Network off-policy
class DeepQNetwork:
    def __init__(
            self,
            n_actions,
            n_features,
            learning_rate=0.01,
            reward_decay=0.9,
            e_greedy=0.9,
            replace_target_iter=300,
            memory_size=500,
            batch_size=32,
            e_greedy_increment=None,
            output_graph=False,
    ):
        self.n_actions = n_actions
        self.n_features = n_features
        self.lr = learning_rate
        self.gamma = reward_decay
        self.epsilon_max = e_greedy
        self.replace_target_iter = replace_target_iter
        self.memory_size = memory_size
        self.batch_size = batch_size
        self.epsilon_increment = e_greedy_increment
        self.epsilon = 0 if e_greedy_increment is not None else self.epsilon_max

        # total learning step
        self.learn_step_counter = 0

        # initialize zero memory [s, a, r, s_]
        self.memory = np.zeros((self.memory_size, n_features * 2 + 2))

        # consist of [target_net, evaluate_net]
        self._build_net()
        t_params = tf.get_collection('target_net_params')
        e_params = tf.get_collection('eval_net_params')
        self.replace_target_op = [tf.assign(t, e) for t, e in zip(t_params, e_params)]

        self.sess = tf.Session()

        if output_graph:
            # $ tensorboard --logdir=logs
            # tf.train.SummaryWriter soon be deprecated, use following
            tf.summary.FileWriter("logs/", self.sess.graph)

        self.sess.run(tf.global_variables_initializer())
        self.cost_his = []

    def _build_net(self):
        # ------------------ build evaluate_net ------------------
        self.s = tf.placeholder(tf.float32, [None, self.n_features], name='s')  # input
        self.q_target = tf.placeholder(tf.float32, [None, self.n_actions], name='Q_target')  # for calculating loss
        with tf.variable_scope('eval_net'):
            # c_names(collections_names) are the collections to store variables
            c_names, n_l1, w_initializer, b_initializer = \
                ['eval_net_params', tf.GraphKeys.GLOBAL_VARIABLES], 10, \
                tf.random_normal_initializer(0., 0.3), tf.constant_initializer(0.1)  # config of layers

            # first layer. collections is used later when assign to target net
            with tf.variable_scope('l1'):
                w1 = tf.get_variable('w1', [self.n_features, n_l1], initializer=w_initializer, collections=c_names)
                b1 = tf.get_variable('b1', [1, n_l1], initializer=b_initializer, collections=c_names)
                l1 = tf.nn.relu(tf.matmul(self.s, w1) + b1)

            # second layer. collections is used later when assign to target net
            with tf.variable_scope('l2'):
                w2 = tf.get_variable('w2', [n_l1, self.n_actions], initializer=w_initializer, collections=c_names)
                b2 = tf.get_variable('b2', [1, self.n_actions], initializer=b_initializer, collections=c_names)
                self.q_eval = tf.matmul(l1, w2) + b2

        with tf.variable_scope('loss'):
            self.loss = tf.reduce_mean(tf.squared_difference(self.q_target, self.q_eval))
        with tf.variable_scope('train'):
            self._train_op = tf.train.RMSPropOptimizer(self.lr).minimize(self.loss)

        # ------------------ build target_net ------------------
        self.s_ = tf.placeholder(tf.float32, [None, self.n_features], name='s_')    # input
        with tf.variable_scope('target_net'):
            # c_names(collections_names) are the collections to store variables
            c_names = ['target_net_params', tf.GraphKeys.GLOBAL_VARIABLES]

            # first layer. collections is used later when assign to target net
            with tf.variable_scope('l1'):
                w1 = tf.get_variable('w1', [self.n_features, n_l1], initializer=w_initializer, collections=c_names)
                b1 = tf.get_variable('b1', [1, n_l1], initializer=b_initializer, collections=c_names)
                l1 = tf.nn.relu(tf.matmul(self.s_, w1) + b1)

            # second layer. collections is used later when assign to target net
            with tf.variable_scope('l2'):
                w2 = tf.get_variable('w2', [n_l1, self.n_actions], initializer=w_initializer, collections=c_names)
                b2 = tf.get_variable('b2', [1, self.n_actions], initializer=b_initializer, collections=c_names)
                self.q_next = tf.matmul(l1, w2) + b2

    def store_transition(self, s, a, r, s_):
        if not hasattr(self, 'memory_counter'):
            self.memory_counter = 0

        transition = np.hstack((s, [a, r], s_))

        # replace the old memory with new memory
        index = self.memory_counter % self.memory_size
        self.memory[index, :] = transition

        self.memory_counter += 1

    def choose_action(self, observation):
        # to have batch dimension when feed into tf placeholder
        observation = observation[np.newaxis, :]

        if np.random.uniform() < self.epsilon:
            # forward feed the observation and get q value for every actions
            actions_value = self.sess.run(self.q_eval, feed_dict={self.s: observation})
            action = np.argmax(actions_value)
        else:
            action = np.random.randint(0, self.n_actions)
        return action

    def learn(self):
        # check to replace target parameters
        if self.learn_step_counter % self.replace_target_iter == 0:
            self.sess.run(self.replace_target_op)
            print('\ntarget_params_replaced\n')

        # sample batch memory from all memory
        if self.memory_counter > self.memory_size:
            sample_index = np.random.choice(self.memory_size, size=self.batch_size)
        else:
            sample_index = np.random.choice(self.memory_counter, size=self.batch_size)
        batch_memory = self.memory[sample_index, :]

        q_next, q_eval = self.sess.run(
            [self.q_next, self.q_eval],
            feed_dict={
                self.s_: batch_memory[:, -self.n_features:],  # fixed params
                self.s: batch_memory[:, :self.n_features],  # newest params
            })

        # change q_target w.r.t q_eval's action
        q_target = q_eval.copy()

        batch_index = np.arange(self.batch_size, dtype=np.int32)
        eval_act_index = batch_memory[:, self.n_features].astype(int)
        reward = batch_memory[:, self.n_features + 1]

        q_target[batch_index, eval_act_index] = reward + self.gamma * np.max(q_next, axis=1)

        """
        For example in this batch I have 2 samples and 3 actions:
        q_eval =
        [[1, 2, 3],
         [4, 5, 6]]

        q_target = q_eval =
        [[1, 2, 3],
         [4, 5, 6]]

        Then change q_target with the real q_target value w.r.t the q_eval's action.
        For example in:
            sample 0, I took action 0, and the max q_target value is -1;
            sample 1, I took action 2, and the max q_target value is -2:
        q_target =
        [[-1, 2, 3],
         [4, 5, -2]]

        So the (q_target - q_eval) becomes:
        [[(-1)-(1), 0, 0],
         [0, 0, (-2)-(6)]]

        We then backpropagate this error w.r.t the corresponding action to network,
        leave other action as error=0 cause we didn't choose it.
        """

        # train eval network
        _, self.cost = self.sess.run([self._train_op, self.loss],
                                     feed_dict={self.s: batch_memory[:, :self.n_features],
                                                self.q_target: q_target})
        self.cost_his.append(self.cost)

        # increasing epsilon
        self.epsilon = self.epsilon + self.epsilon_increment if self.epsilon < self.epsilon_max else self.epsilon_max
        self.learn_step_counter += 1

    def plot_cost(self):
        import matplotlib.pyplot as plt
        plt.plot(np.arange(len(self.cost_his)), self.cost_his)
        plt.ylabel('Cost')
        plt.xlabel('training steps')
        plt.show()

初始化函数__init__：在类的初始化函数中，定义了一些参数，如学习率、折扣因子、探索率等。

同时，初始化了一个记忆库（memory），用于存储经验（状态、动作、奖励、下一个状态）。此

外，还创建了两个神经网络（评估网络和目标网络），并设置了训练过程。

_build_net方法：这个方法用于构建评估网络和目标网络。评估网络用于计算当前状态下每个动作

的Q值，而目标网络用于计算下一个状态下每个动作的Q值。这两个网络的结构相同，都是由两个

全连接层组成。store_transition方法：这个方法用于将经验（状态、动作、奖励、下一个状态）存

储到记忆库中。当记忆库满了之后，新的经验会替换掉最早的经验。

choose_action方法：这个方法用于根据当前状态选择动作。首先，通过评估网络计算当前状态下

每个动作的Q值。然后，根据当前的探索率（epsilon）来决定是选择具有最大Q值的动作（贪婪策

略），还是随机选择一个动作（探索策略）。

首先，检查是否需要替换目标参数。如果学习步数（self.learn_step_counter）是替换目标迭代次

数（self.replace_target_iter）的倍数，则执行替换操作，并打印出提示信息。接下来，从记忆中

随机抽取一批样本。根据记忆库的大小和当前的记忆计数器（self.memory_counter），选择相应

的索引进行抽样。然后，根据这些索引从记忆中获取对应的批次数据（batch_memory）。使用当

前的评估网络（self.q_eval）和下一个状态（self.s_）计算每个动作的Q值（q_next），同时使用

最新的状态（self.s）和评估网络计算每个动作的Q值（q_eval）。

根据评估网络的动作选择结果，更新目标Q值（q_target）。具体来说，将评估网络的Q值复制到

目标Q值中，并根据实际奖励和折扣因子（self.gamma）以及下一个状态的最大Q值来更新对应动

作的目标Q值。使用梯度下降法训练评估网络。通过计算目标Q值和评估网络的Q值之间的误差，

并将其反向传播到网络中，以更新网络的权重。在每次学习步骤之后，增加探索率

（self.epsilon）。探索率用于控制智能体在探索和利用之间平衡的程度。随着学习的进行，探索率

逐渐增加，以便智能体更多地利用已知的知识而不是继续探索。最后，将学习步数加一，以便在下

一次调用learn方法时进行适当的替换操作。

2. maze_env

import numpy as np
import time
import sys
if sys.version_info.major == 2:
    import Tkinter as tk
else:
    import tkinter as tk

UNIT = 40   # pixels
MAZE_H = 4  # grid height
MAZE_W = 4  # grid width

class Maze(tk.Tk, object):
    def __init__(self):
        super(Maze, self).__init__()
        self.action_space = ['u', 'd', 'l', 'r']
        self.n_actions = len(self.action_space)
        self.n_features = 2
        self.title('maze')
        self.geometry('{0}x{1}'.format(MAZE_H * UNIT, MAZE_H * UNIT))
        self._build_maze()

    def _build_maze(self):
        self.canvas = tk.Canvas(self, bg='white',
                           height=MAZE_H * UNIT,
                           width=MAZE_W * UNIT)

        # create grids
        for c in range(0, MAZE_W * UNIT, UNIT):
            x0, y0, x1, y1 = c, 0, c, MAZE_H * UNIT
            self.canvas.create_line(x0, y0, x1, y1)
        for r in range(0, MAZE_H * UNIT, UNIT):
            x0, y0, x1, y1 = 0, r, MAZE_W * UNIT, r
            self.canvas.create_line(x0, y0, x1, y1)

        # create origin
        origin = np.array([20, 20])

        # hell
        hell1_center = origin + np.array([UNIT * 2, UNIT])
        self.hell1 = self.canvas.create_rectangle(
            hell1_center[0] - 15, hell1_center[1] - 15,
            hell1_center[0] + 15, hell1_center[1] + 15,
            fill='black')
        # hell
        # hell2_center = origin + np.array([UNIT, UNIT * 2])
        # self.hell2 = self.canvas.create_rectangle(
        #     hell2_center[0] - 15, hell2_center[1] - 15,
        #     hell2_center[0] + 15, hell2_center[1] + 15,
        #     fill='black')

        # create oval
        oval_center = origin + UNIT * 2
        self.oval = self.canvas.create_oval(
            oval_center[0] - 15, oval_center[1] - 15,
            oval_center[0] + 15, oval_center[1] + 15,
            fill='yellow')

        # create red rect
        self.rect = self.canvas.create_rectangle(
            origin[0] - 15, origin[1] - 15,
            origin[0] + 15, origin[1] + 15,
            fill='red')

        # pack all
        self.canvas.pack()

    def reset(self):
        self.update()
        time.sleep(0.1)
        self.canvas.delete(self.rect)
        origin = np.array([20, 20])
        self.rect = self.canvas.create_rectangle(
            origin[0] - 15, origin[1] - 15,
            origin[0] + 15, origin[1] + 15,
            fill='red')
        # return observation
        return (np.array(self.canvas.coords(self.rect)[:2]) - np.array(self.canvas.coords(self.oval)[:2]))/(MAZE_H*UNIT)

    def step(self, action):
        s = self.canvas.coords(self.rect)
        base_action = np.array([0, 0])
        if action == 0:   # up
            if s[1] > UNIT:
                base_action[1] -= UNIT
        elif action == 1:   # down
            if s[1] < (MAZE_H - 1) * UNIT:
                base_action[1] += UNIT
        elif action == 2:   # right
            if s[0] < (MAZE_W - 1) * UNIT:
                base_action[0] += UNIT
        elif action == 3:   # left
            if s[0] > UNIT:
                base_action[0] -= UNIT

        self.canvas.move(self.rect, base_action[0], base_action[1])  # move agent

        next_coords = self.canvas.coords(self.rect)  # next state

        # reward function
        if next_coords == self.canvas.coords(self.oval):
            reward = 1
            done = True
        elif next_coords in [self.canvas.coords(self.hell1)]:
            reward = -1
            done = True
        else:
            reward = 0
            done = False
        s_ = (np.array(next_coords[:2]) - np.array(self.canvas.coords(self.oval)[:2]))/(MAZE_H*UNIT)
        return s_, reward, done

    def render(self):
        # time.sleep(0.01)
        self.update()

使用tkinter库创建的迷宫环境类。在这个环境中，智能体需要从红色矩形到达黄色圆形，同时避免

黑色矩形区域。环境提供了以下方法：

init(self): 初始化迷宫环境，包括设置窗口大小、标题等。

_build_maze(self): 构建迷宫，包括绘制网格线、创建障碍物和目标点。

reset(self): 重置迷宫环境，将红色矩形放回初始位置，并返回当前状态。

step(self, action): 根据给定动作，更新红色矩形的位置，并返回下一个状态、奖励值和是否结束。

render(self): 更新迷宫环境的显示。

3. DQN_modified

import numpy as np
import tensorflow as tf

np.random.seed(1)
tf.set_random_seed(1)

# Deep Q Network off-policy
class DeepQNetwork:
    def __init__(
            self,
            n_actions,
            n_features,
            learning_rate=0.01,
            reward_decay=0.9,
            e_greedy=0.9,
            replace_target_iter=300,
            memory_size=500,
            batch_size=32,
            e_greedy_increment=None,
            output_graph=False,
    ):
        self.n_actions = n_actions
        self.n_features = n_features
        self.lr = learning_rate
        self.gamma = reward_decay
        self.epsilon_max = e_greedy
        self.replace_target_iter = replace_target_iter
        self.memory_size = memory_size
        self.batch_size = batch_size
        self.epsilon_increment = e_greedy_increment
        self.epsilon = 0 if e_greedy_increment is not None else self.epsilon_max

        # total learning step
        self.learn_step_counter = 0

        # initialize zero memory [s, a, r, s_]
        self.memory = np.zeros((self.memory_size, n_features * 2 + 2))

        # consist of [target_net, evaluate_net]
        self._build_net()

        t_params = tf.get_collection(tf.GraphKeys.GLOBAL_VARIABLES, scope='target_net')
        e_params = tf.get_collection(tf.GraphKeys.GLOBAL_VARIABLES, scope='eval_net')

        with tf.variable_scope('hard_replacement'):
            self.target_replace_op = [tf.assign(t, e) for t, e in zip(t_params, e_params)]

        self.sess = tf.Session()

        if output_graph:
            # $ tensorboard --logdir=logs
            tf.summary.FileWriter("logs/", self.sess.graph)

        self.sess.run(tf.global_variables_initializer())
        self.cost_his = []

    def _build_net(self):
        # ------------------ all inputs ------------------------
        self.s = tf.placeholder(tf.float32, [None, self.n_features], name='s')  # input State
        self.s_ = tf.placeholder(tf.float32, [None, self.n_features], name='s_')  # input Next State
        self.r = tf.placeholder(tf.float32, [None, ], name='r')  # input Reward
        self.a = tf.placeholder(tf.int32, [None, ], name='a')  # input Action

        w_initializer, b_initializer = tf.random_normal_initializer(0., 0.3), tf.constant_initializer(0.1)

        # ------------------ build evaluate_net ------------------
        with tf.variable_scope('eval_net'):
            e1 = tf.layers.dense(self.s, 20, tf.nn.relu, kernel_initializer=w_initializer,
                                 bias_initializer=b_initializer, name='e1')
            self.q_eval = tf.layers.dense(e1, self.n_actions, kernel_initializer=w_initializer,
                                          bias_initializer=b_initializer, name='q')

        # ------------------ build target_net ------------------
        with tf.variable_scope('target_net'):
            t1 = tf.layers.dense(self.s_, 20, tf.nn.relu, kernel_initializer=w_initializer,
                                 bias_initializer=b_initializer, name='t1')
            self.q_next = tf.layers.dense(t1, self.n_actions, kernel_initializer=w_initializer,
                                          bias_initializer=b_initializer, name='t2')

        with tf.variable_scope('q_target'):
            q_target = self.r + self.gamma * tf.reduce_max(self.q_next, axis=1, name='Qmax_s_')    # shape=(None, )
            self.q_target = tf.stop_gradient(q_target)
        with tf.variable_scope('q_eval'):
            a_indices = tf.stack([tf.range(tf.shape(self.a)[0], dtype=tf.int32), self.a], axis=1)
            self.q_eval_wrt_a = tf.gather_nd(params=self.q_eval, indices=a_indices)    # shape=(None, )
        with tf.variable_scope('loss'):
            self.loss = tf.reduce_mean(tf.squared_difference(self.q_target, self.q_eval_wrt_a, name='TD_error'))
        with tf.variable_scope('train'):
            self._train_op = tf.train.RMSPropOptimizer(self.lr).minimize(self.loss)

    def store_transition(self, s, a, r, s_):
        if not hasattr(self, 'memory_counter'):
            self.memory_counter = 0
        transition = np.hstack((s, [a, r], s_))
        # replace the old memory with new memory
        index = self.memory_counter % self.memory_size
        self.memory[index, :] = transition
        self.memory_counter += 1

    def choose_action(self, observation):
        # to have batch dimension when feed into tf placeholder
        observation = observation[np.newaxis, :]

        if np.random.uniform() < self.epsilon:
            # forward feed the observation and get q value for every actions
            actions_value = self.sess.run(self.q_eval, feed_dict={self.s: observation})
            action = np.argmax(actions_value)
        else:
            action = np.random.randint(0, self.n_actions)
        return action

    def learn(self):
        # check to replace target parameters
        if self.learn_step_counter % self.replace_target_iter == 0:
            self.sess.run(self.target_replace_op)
            print('\ntarget_params_replaced\n')

        # sample batch memory from all memory
        if self.memory_counter > self.memory_size:
            sample_index = np.random.choice(self.memory_size, size=self.batch_size)
        else:
            sample_index = np.random.choice(self.memory_counter, size=self.batch_size)
        batch_memory = self.memory[sample_index, :]

        _, cost = self.sess.run(
            [self._train_op, self.loss],
            feed_dict={
                self.s: batch_memory[:, :self.n_features],
                self.a: batch_memory[:, self.n_features],
                self.r: batch_memory[:, self.n_features + 1],
                self.s_: batch_memory[:, -self.n_features:],
            })

        self.cost_his.append(cost)

        # increasing epsilon
        self.epsilon = self.epsilon + self.epsilon_increment if self.epsilon < self.epsilon_max else self.epsilon_max
        self.learn_step_counter += 1

    def plot_cost(self):
        import matplotlib.pyplot as plt
        plt.plot(np.arange(len(self.cost_his)), self.cost_his)
        plt.ylabel('Cost')
        plt.xlabel('training steps')
        plt.show()

if __name__ == '__main__':
    DQN = DeepQNetwork(3,4, output_graph=True)

初始化函数__init__：定义了神经网络的参数，如学习率、折扣因子、探索率等。同时初始化了经

验回放存储器，用于存储过去的经验。

_build_net方法：构建了两个神经网络，一个是评估网络（eval_net），另一个是目标网络

（target_net）。这两个网络的结构相同，但参数不同。评估网络用于计算当前状态下的动作值，

目标网络用于计算下一状态下的动作值。

store_transition方法：将经验（状态、动作、奖励、下一状态）存储到经验回放存储器中。

choose_action方法：根据当前状态选择动作。首先计算所有动作的值，然后根据贪婪策略选择一

个动作。如果随机数小于当前的探索率，则随机选择一个动作；否则选择具有最大动作值的动作。

learn方法：从经验回放存储器中随机抽取一批经验，然后使用这些经验更新评估网络的参数。同

时，每隔一定步数，将评估网络的参数复制到目标网络中。

plot_cost方法：绘制损失函数随训练步数的变化曲线，用于观察模型的学习过程。

4. run

from maze_env import Maze
from RL_brain import DeepQNetwork

def run_maze():
    step = 0
    for episode in range(300):
        # initial observation
        observation = env.reset()

        while True:
            # fresh env
            env.render()

            # RL choose action based on observation
            action = RL.choose_action(observation)

            # RL take action and get next observation and reward
            observation_, reward, done = env.step(action)

            RL.store_transition(observation, action, reward, observation_)

            if (step > 200) and (step % 5 == 0):
                RL.learn()

            # swap observation
            observation = observation_

            # break while loop when end of this episode
            if done:
                break
            step += 1

    # end of game
    print('game over')
    env.destroy()

if __name__ == "__main__":
    # maze game
    env = Maze()
    RL = DeepQNetwork(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()

这段代码使用了两个模块：maze_env和RL_brain。maze_env模块定义了迷宫环境，而RL_brain

模块定义了深度Q网络。在run_maze()函数中，首先初始化了一个迷宫环境对象env和一个深度Q

网络对象RL。然后进行了300个回合的训练。每个回合开始时，通过调用env.reset()方法获取初始

观察值observation。接下来进入一个循环，直到当前回合结束。

在循环中，首先调用env.render()方法来刷新环境显示。然后，根据当前的观察值observation，使

用深度Q网络选择动作action。接着，执行该动作并获取下一个观察值observation_、奖励值

reward以及是否结束的标志done。将这个经验（状态、动作、奖励、下一个状态）存储到经验回

放存储器中。

如果步数大于200且是5的倍数，则进行学习操作，即调用RL.learn()方法更新深度Q网络的参数。

然后将下一个观察值赋值给当前的观察值，以便进行下一步的动作选择。如果当前回合结束，则跳

出循环。最后，打印"game over"并销毁迷宫环境对象。在主程序部分，首先创建迷宫环境对象env

和深度Q网络对象RL。然后设置迷宫游戏的环境参数，如学习率、奖励衰减、贪婪策略等。接着，

使用env.after(100, run_maze)方法在100毫秒后启动迷宫游戏的主循环。最后，调用

env.mainloop()方法进入主循环，并在训练结束后绘制损失函数曲线。