forked from thu-ml/zhusuan
-
Notifications
You must be signed in to change notification settings - Fork 0
/
sbn_vimco.py
167 lines (148 loc) · 6.85 KB
/
sbn_vimco.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
#!/usr/bin/env python
# -*- coding: utf-8 -*-
from __future__ import absolute_import
from __future__ import print_function
from __future__ import division
import os
import time
import tensorflow as tf
from tensorflow.contrib import layers
from six.moves import range
import numpy as np
import zhusuan as zs
from examples import conf
from examples.utils import dataset
@zs.reuse('model')
def sbn(observed, n, n_x, n_h, n_particles, is_training):
with zs.BayesianNet(observed=observed) as model:
h3_logits = tf.zeros([n, n_h])
h3 = zs.Bernoulli('h3', h3_logits, n_samples=n_particles,
group_event_ndims=1)
h2_logits = layers.fully_connected(
tf.to_float(h3), n_h, activation_fn=None)
h2 = zs.Bernoulli('h2', h2_logits, group_event_ndims=1)
h1_logits = layers.fully_connected(
tf.to_float(h2), n_h, activation_fn=None)
h1 = zs.Bernoulli('h1', h1_logits, group_event_ndims=1)
x_logits = layers.fully_connected(
tf.to_float(h1), n_x, activation_fn=None)
x = zs.Bernoulli('x', x_logits, group_event_ndims=1)
return model
def q_net(x, n_h, n_particles, is_training):
with zs.BayesianNet() as variational:
h1_logits = layers.fully_connected(
tf.to_float(x), n_h, activation_fn=None)
h1 = zs.Bernoulli('h1', h1_logits, n_samples=n_particles,
group_event_ndims=1)
h2_logits = layers.fully_connected(
tf.to_float(h1), n_h, activation_fn=None)
h2 = zs.Bernoulli('h2', h2_logits, group_event_ndims=1)
h3_logits = layers.fully_connected(
tf.to_float(h2), n_h, activation_fn=None)
h3 = zs.Bernoulli('h3', h3_logits, group_event_ndims=1)
return variational
if __name__ == "__main__":
tf.set_random_seed(1237)
# Load MNIST
data_path = os.path.join(conf.data_dir, 'mnist.pkl.gz')
x_train, t_train, x_valid, t_valid, x_test, t_test = \
dataset.load_mnist_realval(data_path)
x_train = np.vstack([x_train, x_valid]).astype('float32')
np.random.seed(1234)
x_test = np.random.binomial(1, x_test, size=x_test.shape).astype('float32')
n_x = x_train.shape[1]
# Define model parameters
n_h = 200
# Define training/evaluation parameters
lb_samples = 10
ll_samples = 1000
epoches = 3000
batch_size = 24
test_batch_size = 100
iters = x_train.shape[0] // batch_size
test_iters = x_test.shape[0] // test_batch_size
test_freq = 10
learning_rate = 0.001
anneal_lr_freq = 200
anneal_lr_rate = 0.75
# Build the computation graph
is_training = tf.placeholder(tf.bool, shape=[], name='is_training')
n_particles = tf.placeholder(tf.int32, shape=[], name='n_particles')
x_orig = tf.placeholder(tf.float32, shape=[None, n_x], name='x')
x_bin = tf.cast(tf.less(tf.random_uniform(tf.shape(x_orig), 0, 1), x_orig),
tf.int32)
x = tf.placeholder(tf.int32, shape=[None, n_x], name='x')
x_obs = tf.tile(tf.expand_dims(x, 0), [n_particles, 1, 1])
n = tf.shape(x)[0]
def log_joint(observed):
model = sbn(observed, n, n_x, n_h, n_particles, is_training)
log_ph3, log_ph2_h3, log_ph1_h2, log_px_h1 = model.local_log_prob(
['h3', 'h2', 'h1', 'x'])
return log_ph3 + log_ph2_h3 + log_ph1_h2 + log_px_h1
variational = q_net(x, n_h, n_particles, is_training)
qh3_samples, log_qh3 = variational.query('h3', outputs=True,
local_log_prob=True)
qh2_samples, log_qh2 = variational.query('h2', outputs=True,
local_log_prob=True)
qh1_samples, log_qh1 = variational.query('h1', outputs=True,
local_log_prob=True)
cost, lower_bound = zs.vimco(log_joint, {'x': x_obs},
{'h3': [qh3_samples, log_qh3],
'h2': [qh2_samples, log_qh2],
'h1': [qh1_samples, log_qh1]}, axis=0)
lower_bound = tf.reduce_mean(lower_bound)
cost = tf.reduce_mean(cost)
log_likelihood = tf.reduce_mean(
zs.is_loglikelihood(log_joint, {'x': x_obs},
{'h3': [qh3_samples, log_qh3],
'h2': [qh2_samples, log_qh2],
'h1': [qh1_samples, log_qh1]}, axis=0))
learning_rate_ph = tf.placeholder(tf.float32, shape=[], name='lr')
optimizer = tf.train.AdamOptimizer(learning_rate_ph, epsilon=1e-4)
grads = optimizer.compute_gradients(cost)
infer = optimizer.apply_gradients(grads)
params = tf.trainable_variables()
for i in params:
print(i.name, i.get_shape())
# Run the inference
with tf.Session() as sess:
sess.run(tf.global_variables_initializer())
for epoch in range(1, epoches + 1):
time_epoch = -time.time()
if epoch % anneal_lr_freq == 0:
learning_rate *= anneal_lr_rate
np.random.shuffle(x_train)
lbs = []
for t in range(iters):
x_batch = x_train[t * batch_size:(t + 1) * batch_size]
x_batch_bin = sess.run(x_bin, feed_dict={x_orig: x_batch})
_, lb = sess.run([infer, lower_bound],
feed_dict={x: x_batch_bin,
learning_rate_ph: learning_rate,
n_particles: lb_samples,
is_training: True})
lbs.append(lb)
time_epoch += time.time()
print('Epoch {} ({:.1f}s): Lower bound = {}'.format(
epoch, time_epoch, np.mean(lbs)))
if epoch % test_freq == 0:
time_test = -time.time()
test_lbs = []
test_lls = []
for t in range(test_iters):
test_x_batch = x_test[
t * test_batch_size: (t + 1) * test_batch_size]
test_lb = sess.run(lower_bound,
feed_dict={x: test_x_batch,
n_particles: lb_samples,
is_training: False})
test_ll = sess.run(log_likelihood,
feed_dict={x: test_x_batch,
n_particles: ll_samples,
is_training: False})
test_lbs.append(test_lb)
test_lls.append(test_ll)
time_test += time.time()
print('>>> TEST ({:.1f}s)'.format(time_test))
print('>> Test lower bound = {}'.format(np.mean(test_lbs)))
print('>> Test log likelihood = {}'.format(np.mean(test_lls)))