-
Notifications
You must be signed in to change notification settings - Fork 3
/
model.py
242 lines (186 loc) · 10.9 KB
/
model.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
"""
Implemetación del algoritmo DDGP (Deep Neural Network Policy Gradient)
para el caso de un sistema de trading sobre un solo activo
"""
import tensorflow as tf
import numpy as np
import tflearn
from buffer import ReplayBuffer
class ActorNetwork(object):
"""
La entrada a red es el estado
La salida es la acción bajo una política determinística mu(s)
"""
def __init__(self, sess, state_dim, action_dim, learning_rate, tau, batch_size):
self.sess = sess
self.s_dim = state_dim
self.a_dim = action_dim
self.learning_rate = learning_rate
self.tau = tau
self.batch_size = batch_size
self.initializer = tflearn.initializations.variance_scaling(factor=1.0,
mode='FAN_IN',
uniform=True,
seed=None,
dtype=tf.float32)
# Red Actuadora
self.inputs, self.softmax = self.create_actor_network()
self.network_params = tf.trainable_variables()
# Red objetivo
self.target_inputs, self.target_softmax = self.create_actor_network()
self.target_network_params = tf.trainable_variables()[len(self.network_params):]
# Operación para actualizar periodicamente la red objetivo con los pesos de la red en línea
self.update_target_network_params = [
self.target_network_params[i].assign(tf.multiply(self.network_params[i], self.tau) +
tf.multiply(self.target_network_params[i], 1.0 - self.tau))
for i in range(len(self.target_network_params))]
# Este gradiente es provisto por lar red crítica
self.action_gradient = tf.placeholder(tf.float32, [None, self.a_dim])
# Combinar los gradientes
self.unnormalized_actor_gradients = tf.gradients(self.softmax, self.network_params, -self.action_gradient)
grads = []
for var, grad in zip(self.network_params, self.unnormalized_actor_gradients):
if grad is None:
grads.append(tf.zeros_like(var))
else:
grads.append(grad)
self.unnormalized_actor_gradients = grads
self.actor_gradients = list(map(lambda x: tf.div(x, self.batch_size),
self.unnormalized_actor_gradients))
# Operación de optimización
self.optimize = tf.train.AdamOptimizer(self.learning_rate).apply_gradients(zip(self.actor_gradients,
self.network_params))
self.num_trainable_vars = len(self.network_params) + len(self.target_network_params)
def create_actor_network(self):
inputs = tf.placeholder(tf.float32, [None, self.s_dim[0], self.s_dim[1], self.s_dim[2]])
net = tflearn.layers.conv.conv_2d(incoming=inputs,
nb_filter=2,
filter_size=[1, 3],
strides=1,
padding="valid",
activation="relu",
weights_init= self.initializer,
weight_decay=0.0)
width = net.get_shape()[2]
net = tflearn.layers.conv.conv_2d(incoming=net,
nb_filter=10,
filter_size=[1, width],
strides=1,
padding="valid",
activation="relu",
weights_init=self.initializer,
regularizer="L2",
weight_decay=5e-09)
net = tflearn.layers.conv.conv_2d(incoming=net,
nb_filter=self.a_dim,
filter_size=1,
padding="valid",
weights_init=self.initializer,
regularizer="L2",
weight_decay=5e-08)
net = tflearn.fully_connected(net, self.a_dim, weights_init=self.initializer)
softmax = tf.nn.softmax(net)
return inputs, softmax
def train(self, inputs, a_gradient):
self.sess.run(self.optimize, feed_dict={self.inputs: inputs,
self.action_gradient: a_gradient})
def predict(self, inputs):
return self.sess.run(self.softmax, feed_dict={self.inputs: inputs})
def predict_target(self, inputs):
return self.sess.run(self.target_softmax, feed_dict={self.target_inputs: inputs})
def update_target_network(self):
self.sess.run(self.update_target_network_params)
def get_num_trainable_vars(self):
return self.num_trainable_vars
class CriticNetwork(object):
"""
La entrada a la red viene dada por el estado y la acción.
La salida es valor de función valor Q(s,a)
La acción debe ser obtenida desde la salida de la red Actuadora
"""
def __init__(self, sess, state_dim, action_dim, learning_rate, tau, gamma, num_actor_vars):
self.sess = sess
self.s_dim = state_dim
self.a_dim = action_dim
self.learning_rate = learning_rate
self.tau = tau
self.gamma = gamma
self.initializer = tflearn.initializations.variance_scaling(factor=1.0,
mode='FAN_IN',
uniform=True,
seed=None,
dtype=tf.float32)
# Crear la red crítica
self.inputs, self.action, self.out = self.create_critic_network()
self.network_params = tf.trainable_variables()[num_actor_vars:]
# Red Objetivo
self.target_inputs, self.target_action, self.target_out = self.create_critic_network()
self.target_network_params = tf.trainable_variables()[(len(self.network_params) + num_actor_vars):]
# Operación para actualizar periódicamente las redes objetivo los pesos de la red en línea
self.update_target_network_params = [
self.target_network_params[i].assign(tf.multiply(self.target_network_params[i], self.tau) +
tf.multiply(self.target_network_params[i], 1.0 - self.tau))
for i in range(len(self.target_network_params))]
# Objetivo de la red (y_i)
self.predicted_q_value = tf.placeholder(tf.float32, [None, 1])
# Operación de optmización con pérdida
self.loss = tflearn.mean_square(self.predicted_q_value, self.out)
self.optimize = tf.train.AdamOptimizer(self.learning_rate).minimize(self.loss)
# Optener el gradiente de la red con respecto a la acción. Para cada acción en el minibatch esto sumará
# los gradientes de cada salida de la red crítrica en el minibatch con respecto dicha acción. Cada salida
# es independiente de todas las acciones excepto por una
self.action_grads = tf.gradients(self.out, self.action)
def create_critic_network(self):
inputs = tf.placeholder(tf.float32, [None, self.s_dim[0], self.s_dim[1], self.s_dim[2]])
action = tf.placeholder(tf.float32, [None, self.a_dim])
net = tflearn.layers.conv.conv_2d(incoming=inputs,
nb_filter=2,
filter_size=[1, 3],
strides=1,
padding="valid",
activation="relu",
weights_init=self.initializer,
weight_decay=0.0)
width = net.get_shape()[2]
net = tflearn.layers.conv.conv_2d(incoming=net,
nb_filter=10,
filter_size=[1, width],
strides=1,
padding="valid",
activation="relu",
weights_init=self.initializer,
regularizer="L2",
weight_decay=5e-09)
net = tflearn.layers.conv.conv_2d(incoming=net,
nb_filter=1,
filter_size=1,
padding="valid",
weights_init=self.initializer,
regularizer="L2",
weight_decay=5e-08)
net = tflearn.fully_connected(net, 20, weights_init=self.initializer,)
net = tflearn.layers.normalization.batch_normalization(net)
net = tflearn.activations.relu(net)
# Agregar el tensor de acción
# Se utilizan dos capas temporales para obtener los pasos y sesgos correspondientes
t1 = tflearn.fully_connected(net, 10, weights_init=self.initializer,)
t2 = tflearn.fully_connected(action, 10, weights_init=self.initializer,)
net = tflearn.activation(tf.matmul(net, t1.W) + tf.matmul(action, t2.W) + t2.b, activation="relu")
out = tflearn.fully_connected(net, 1, weights_init=self.initializer,)
return inputs, action, out
def train(self, inputs, action, predicted_q_value):
return self.sess.run([self.out, self.loss, self.optimize],
feed_dict={self.inputs: inputs,
self.action: action,
self.predicted_q_value: predicted_q_value})
def predict(self, inputs, action):
return self.sess.run(self.out, feed_dict={self.inputs: inputs,
self.action: action})
def predict_target(self, inputs, action):
return self.sess.run(self.target_out, feed_dict={self.target_inputs: inputs,
self.target_action: action})
def action_gradients(self, inputs, actions):
return self.sess.run(self.action_grads, feed_dict={self.inputs: inputs,
self.action: actions})
def update_target_network(self):
self.sess.run(self.update_target_network_params)