def __init__(self, state_size, window_size, trend, skip):
self.state_size = state_size
self.window_size = window_size
self.half_window = window_size // 2
self.trend = trend
self.skip = skip
tf.reset_default_graph()
self.INITIAL_FEATURES = np.zeros((4, self.state_size))
self.X = tf.placeholder(tf.float32, (None, None, self.state_size))
self.Y = tf.placeholder(tf.float32, (None, self.OUTPUT_SIZE))
cell = tf.nn.rnn_cell.LSTMCell(self.LAYER_SIZE, state_is_tuple=False)
self.hidden_layer = tf.placeholder(tf.float32,
(None, 2 * self.LAYER_SIZE))
self.rnn, self.last_state = tf.nn.dynamic_rnn(
inputs=self.X,
cell=cell,
dtype=tf.float32,
initial_state=self.hidden_layer)
tensor_action, tensor_validation = tf.split(self.rnn[:, -1], 2, 1)
feed_action = tf.layers.dense(tensor_action, self.OUTPUT_SIZE)
feed_validation = tf.layers.dense(tensor_validation, 1)
self.logits = feed_validation + tf.subtract(
feed_action, tf.reduce_mean(feed_action, axis=1, keep_dims=True))
self.cost = tf.reduce_sum(tf.square(self.Y - self.logits))
self.optimizer = tf.train.AdamOptimizer(
learning_rate=self.LEARNING_RATE).minimize(self.cost)
self.sess = tf.InteractiveSession()
self.sess.run(tf.global_variables_initializer())
def __init__(self, state_size, window_size, trend, skip, batch_size):
self.state_size = state_size
self.window_size = window_size
self.half_window = window_size // 2
self.trend = trend
self.skip = skip
self.action_size = 3
self.batch_size = batch_size
self.memory = deque(maxlen=1000)
self.inventory = []
self.gamma = 0.95
self.epsilon = 0.5
self.epsilon_min = 0.01
self.epsilon_decay = 0.999
tf.reset_default_graph()
self.sess = tf.InteractiveSession()
self.X = tf.placeholder(tf.float32, [None, self.state_size])
self.Y = tf.placeholder(tf.float32, [None, self.action_size])
feed = tf.layers.dense(self.X, 512, activation=tf.nn.relu)
tensor_action, tensor_validation = tf.split(feed, 2, 1)
feed_action = tf.layers.dense(tensor_action, self.action_size)
feed_validation = tf.layers.dense(tensor_validation, 1)
self.logits = feed_validation + tf.subtract(
feed_action, tf.reduce_mean(feed_action, axis=1, keep_dims=True))
self.cost = tf.reduce_mean(tf.square(self.Y - self.logits))
self.optimizer = tf.train.GradientDescentOptimizer(1e-5).minimize(
self.cost)
self.sess.run(tf.global_variables_initializer())
def __init__(self, name, input_size, output_size, size_layer):
with tf.variable_scope(name):
self.X = tf.placeholder(tf.float32, (None, input_size))
feed_actor = tf.layers.dense(self.X, size_layer, activation = tf.nn.relu)
tensor_action, tensor_validation = tf.split(feed_actor,2,1)
feed_action = tf.layers.dense(tensor_action, output_size)
feed_validation = tf.layers.dense(tensor_validation, 1)
self.logits = feed_validation + tf.subtract(feed_action,
tf.reduce_mean(feed_action,axis=1,keep_dims=True))
def __init__(self, input_size, output_size, layer_size, learning_rate):
self.X = tf.placeholder(tf.float32, (None, input_size))
self.Y = tf.placeholder(tf.float32, (None, output_size))
feed = tf.layers.dense(self.X, layer_size, activation=tf.nn.relu)
tensor_action, tensor_validation = tf.split(feed, 2, 1)
feed_action = tf.layers.dense(tensor_action, output_size)
feed_validation = tf.layers.dense(tensor_validation, 1)
self.logits = feed_validation + tf.subtract(
feed_action, tf.reduce_mean(feed_action, axis=1, keep_dims=True))
self.cost = tf.reduce_sum(tf.square(self.Y - self.logits))
self.optimizer = tf.train.AdamOptimizer(
learning_rate=learning_rate).minimize(self.cost)
def __init__(self, name, input_size, output_size, size_layer):
with tf.variable_scope(name):
self.X = tf.placeholder(tf.float32, (None, None, input_size))
self.hidden_layer = tf.placeholder(tf.float32, (None, 2 * size_layer))
cell = tf.nn.rnn_cell.LSTMCell(size_layer, state_is_tuple = False)
self.rnn,self.last_state = tf.nn.dynamic_rnn(inputs=self.X, cell=cell,
dtype=tf.float32,
initial_state=self.hidden_layer)
tensor_action, tensor_validation = tf.split(self.rnn[:,-1],2,1)
feed_action = tf.layers.dense(tensor_action, output_size)
feed_validation = tf.layers.dense(tensor_validation, 1)
self.logits = feed_validation + tf.subtract(feed_action,
tf.reduce_mean(feed_action,axis=1,keep_dims=True))
def __init__(self, name, input_size, output_size, size_layer, learning_rate):
with tf.variable_scope(name):
self.X = tf.placeholder(tf.float32, (None, input_size))
self.Y = tf.placeholder(tf.float32, (None, output_size))
self.REWARD = tf.placeholder(tf.float32, (None, 1))
feed_critic = tf.layers.dense(self.X, size_layer, activation = tf.nn.relu)
tensor_action, tensor_validation = tf.split(feed_critic,2,1)
feed_action = tf.layers.dense(tensor_action, output_size)
feed_validation = tf.layers.dense(tensor_validation, 1)
feed_critic = feed_validation + tf.subtract(feed_action,tf.reduce_mean(feed_action,axis=1,keep_dims=True))
feed_critic = tf.nn.relu(feed_critic) + self.Y
feed_critic = tf.layers.dense(feed_critic, size_layer//2, activation = tf.nn.relu)
self.logits = tf.layers.dense(feed_critic, 1)
self.cost = tf.reduce_mean(tf.square(self.REWARD - self.logits))
self.optimizer = tf.train.AdamOptimizer(learning_rate).minimize(self.cost)
def __init__(self, name, input_size, output_size, size_layer, learning_rate):
with tf.variable_scope(name):
self.X = tf.placeholder(tf.float32, (None, None, input_size))
self.Y = tf.placeholder(tf.float32, (None, output_size))
self.hidden_layer = tf.placeholder(tf.float32, (None, 2 * size_layer))
self.REWARD = tf.placeholder(tf.float32, (None, 1))
feed_critic = tf.layers.dense(self.X, size_layer, activation = tf.nn.relu)
cell = tf.nn.rnn_cell.LSTMCell(size_layer, state_is_tuple = False)
self.rnn,self.last_state = tf.nn.dynamic_rnn(inputs=self.X, cell=cell,
dtype=tf.float32,
initial_state=self.hidden_layer)
tensor_action, tensor_validation = tf.split(self.rnn[:,-1],2,1)
feed_action = tf.layers.dense(tensor_action, output_size)
feed_validation = tf.layers.dense(tensor_validation, 1)
feed_critic = feed_validation + tf.subtract(feed_action,tf.reduce_mean(feed_action,axis=1,keep_dims=True))
feed_critic = tf.nn.relu(feed_critic) + self.Y
feed_critic = tf.layers.dense(feed_critic, size_layer//2, activation = tf.nn.relu)
self.logits = tf.layers.dense(feed_critic, 1)
self.cost = tf.reduce_mean(tf.square(self.REWARD - self.logits))
self.optimizer = tf.train.AdamOptimizer(learning_rate).minimize(self.cost)
def __init__(self, state_size, window_size, trend, skip):
self.state_size = state_size
self.window_size = window_size
self.half_window = window_size // 2
self.trend = trend
self.skip = skip
tf.reset_default_graph()
self.X = tf.placeholder(tf.float32, (None, self.state_size))
self.Y = tf.placeholder(tf.float32, (None, self.state_size))
self.ACTION = tf.placeholder(tf.float32, (None))
self.REWARD = tf.placeholder(tf.float32, (None))
self.batch_size = tf.shape(self.ACTION)[0]
with tf.variable_scope('curiosity_model'):
action = tf.reshape(self.ACTION, (-1, 1))
state_action = tf.concat([self.X, action], axis=1)
save_state = tf.identity(self.Y)
feed = tf.layers.dense(state_action, 32, activation=tf.nn.relu)
self.curiosity_logits = tf.layers.dense(feed, self.state_size)
self.curiosity_cost = tf.reduce_sum(
tf.square(save_state - self.curiosity_logits), axis=1)
self.curiosity_optimizer = tf.train.RMSPropOptimizer(
self.LEARNING_RATE).minimize(
tf.reduce_mean(self.curiosity_cost))
total_reward = tf.add(self.curiosity_cost, self.REWARD)
with tf.variable_scope("q_model"):
with tf.variable_scope("eval_net"):
x_action = tf.layers.dense(self.X, 128, tf.nn.relu)
tensor_action, tensor_validation = tf.split(x_action, 2, 1)
feed_action = tf.layers.dense(tensor_action, self.OUTPUT_SIZE)
feed_validation = tf.layers.dense(tensor_validation, 1)
self.logits = feed_validation + tf.subtract(
feed_action,
tf.reduce_mean(feed_action, axis=1, keep_dims=True))
with tf.variable_scope("target_net"):
y_action = tf.layers.dense(self.Y, 128, tf.nn.relu)
tensor_action, tensor_validation = tf.split(y_action, 2, 1)
feed_action = tf.layers.dense(tensor_action, self.OUTPUT_SIZE)
feed_validation = tf.layers.dense(tensor_validation, 1)
y_q = feed_validation + tf.subtract(
feed_action,
tf.reduce_mean(feed_action, axis=1, keep_dims=True))
q_target = total_reward + self.GAMMA * tf.reduce_max(y_q, axis=1)
action = tf.cast(self.ACTION, tf.int32)
action_indices = tf.stack(
[tf.range(self.batch_size, dtype=tf.int32), action], axis=1)
q = tf.gather_nd(params=self.logits, indices=action_indices)
self.cost = tf.losses.mean_squared_error(labels=q_target,
predictions=q)
self.optimizer = tf.train.RMSPropOptimizer(
self.LEARNING_RATE).minimize(
self.cost,
var_list=tf.get_collection(
tf.GraphKeys.TRAINABLE_VARIABLES, "q_model/eval_net"))
t_params = tf.get_collection(tf.GraphKeys.GLOBAL_VARIABLES,
scope='q_model/target_net')
e_params = tf.get_collection(tf.GraphKeys.GLOBAL_VARIABLES,
scope='q_model/eval_net')
self.target_replace_op = [
tf.assign(t, e) for t, e in zip(t_params, e_params)
]
self.sess = tf.InteractiveSession()
self.sess.run(tf.global_variables_initializer())