针对 Deep Q Learning 可能无法收敛的问题，这里提出了一种 fix target 的方法，就是冻结现实神经网络，延时更新参数。
这个方法的初衷是这样的：
1. 之前我们每个（批）记忆都会更新参数，这是一种实时更新神经网络参数的方法，这个方法有个问题，就是每次都更新，由于样本都是随机的，可能存在各种不正常现象，比如你考试得了90分，妈妈奖励了你，但是也有可能是考了90分，被臭骂一顿，因为别人都考了95分以上，当然这只是个例子，正是各种异常现象，可能导致损失忽小忽大，参数来回震荡，无法收敛。
2. fix target 方法搭了2个神经网络，一个是估计，一个是现实，估计神经网络实时更新，而现实神经网络暂时冻结，q估计用估计神经网络，q现实用现实神经网络，冻结现实神经网络，就是我先不动，然后派很多人去做尝试，回来给我汇报，我根据汇报总结经验，然后再行动，这是我们正常的处事逻辑，这样显得神经网络更稳重，容易收敛。
核心代码如下
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'): # n_l1 第一隐层的神经元个数 w b 初始化 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_): # 存储记忆,固定大小的记忆库 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): # q learning 选择动作 e贪心 observation = observation[np.newaxis, :] if np.random.uniform() < self.epsilon: 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): # 每隔一定步数更新，代表积累了一定的经验才进行总结，这样显得不那么武断，对应到神经网络，就是更新效率高，容易收敛 if self.learn_step_counter % self.replace_target_iter == 0: # 更新神经网络参数 self.sess.run(self.replace_target_op) print(' target_params_replaced ') ##### 取样本 ## memory 是 初始化一个 memory_size 的数据表，当记忆大于这个表时，表已被填满，随机从表中选择记忆即可， ## 当记忆小于这个表时，表未被填满，只能从记忆里随机选样本 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, :] ##### 训练神经网络 # 利用神经网络 计算 q_next 下个状态的q值，q_eval 当前状态的q值 q_next, q_eval = self.sess.run([self.q_next, self.q_eval], feed_dict={ self.s_: batch_memory[:, -self.n_features:], # 下一个状态 self.s: batch_memory[:, :self.n_features], # 当前状态 }) q_target = q_eval.copy() # q_eval 是q 估计 ### 更新 q learning 的 q表 ## 每个真实实验有个动作和奖励 batch_index = np.arange(self.batch_size, dtype=np.int32) eval_act_index = batch_memory[:, self.n_features].astype(int) # 动作 reward = batch_memory[:, self.n_features + 1] # 奖励 ## 更新q表 q_target[batch_index, eval_act_index] = reward + self.gamma * np.max(q_next, axis=1) # q 现实 # 注意这里的动作当前状态选择的动作，而不是下个状态基于贪心的动作 # 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
图示如下
