-
Notifications
You must be signed in to change notification settings - Fork 13
/
ae.py
254 lines (171 loc) · 7.51 KB
/
ae.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
243
244
245
246
247
248
249
250
251
252
253
254
import tensorflow as tf
from data_reader import *
import numpy as np
from uncompress import *
import os
import tensorflow.contrib.layers as lays
import cPickle
def log(message,file_path=os.path.join('ae_logs_1','ae_log.txt')):
print message
f1=open(file_path, 'a+')
f1.write(message)
f1.close()
def autoencoder(inputs):
# encoder
# 50 x 86796 x 1 -> 50 x 10850 x 32
# 50 x 10850 x 32 -> 50 x 5825 x 16
# 50 x 5825 x 16 -> 50 x 1357 x 8
encoder_1 = lays.conv2d(inputs, 32, [5, 5], stride=(1,8), padding='SAME')
encoder_2 = lays.conv2d(encoder_1, 16, [5, 5], stride=(1,2), padding='SAME')
compressed = lays.conv2d(encoder_2, 8, [5, 5], stride=(1,4), padding='SAME')
# decoder
# 50 x 1357 x 8 -> 50 x 5825 x 16
# 50 x 5825 x 16 -> 50 x 10850 x 32
# 50 x 10850 x 32 -> 50 x 86796 x 1
decoder_1 = lays.conv2d_transpose(compressed, 16, [5, 5], stride=(1,4), padding='SAME')
decoder_2 = lays.conv2d_transpose(decoder_1, 32, [5, 5], stride=(1,2), padding='SAME')
decoder_3 = lays.conv2d_transpose(decoder_2, 1, [5, 5], stride=(1,8), padding='SAME', activation_fn=tf.nn.tanh)
return decoder_3[:,:,0:inputs.get_shape().as_list()[2],:],compressed,encoder_1,encoder_2,decoder_1,decoder_2
def compress(input,x=None,return_x=False):
input_size = 50
learning_rate = 0.00001
X = tf.placeholder(tf.float32, [None, input_size,86796,1])
reconstruction, compressed, encoder_1, encoder_2, decoder_1, decoder_2 = autoencoder(X)
# print encoder_1.get_shape,encoder_2.get_shape
# exit(0)
loss_op = tf.reduce_mean(tf.square(reconstruction - X))
optimizer = tf.train.AdamOptimizer(learning_rate=learning_rate)
train_op = optimizer.minimize(loss_op)
if(return_x):
saver = tf.train.Saver()
x = saver._var_list
print x
return x
#print type(x)
saver = tf.train.Saver(var_list=x)
# saver = tf.train.Saver(var_list=x)
with tf.Session() as sess1:
# Run the initializer
# sess1.run(init)
# print 'restoring ae session'
# saver.restore(sess1, "ae_logs_1/save.ckpt")
# print 'done loading'
# exit(0)
# init_new_vars_op = tf.variables_initializer([reconstruction, compressed])
init_new_op = tf.variables_initializer([v for v in tf.global_variables() if v.name.split(':')[0] in set(sess1.run(tf.report_uninitialized_variables()))])
sess1.run(init_new_op)
batch_x = uncompress(input,86796)
# Run optimization op (backprop)
[_,data_point] = sess1.run([reconstruction, compressed], feed_dict={X: batch_x})
# print np.stack(ae_output).shape
return data_point
def print_intermediate():
learning_rate = 0.00001
# np.set_printoptions(threshold=np.nan)
num_epoch = 1
batch_size = 1
display_step = 1
input_size = 50
X = tf.placeholder(tf.float32, [None, input_size,86796,1])
reconstruction, compressed, encoder_1, encoder_2, decoder_1, decoder_2 = autoencoder(X)
loss_op = tf.reduce_mean(tf.square(reconstruction - X))
tf.summary.scalar('loss',loss_op)
# optimizer = tf.train.AdamOptimizer(learning_rate=learning_rate)
# train_op = optimizer.minimize(loss_op)
init = tf.global_variables_initializer()
data_X,_ = load_data()
indices = np.random.permutation(np.arange(data_X.shape[0]))
data_X = data_X[indices,:,:]
tf.contrib.layers.summarize_collection(tf.GraphKeys.TRAINABLE_VARIABLES)
merged = tf.summary.merge_all()
saver = tf.train.Saver()
with tf.Session() as sess:
train_writer = tf.summary.FileWriter("ae_logs_1/",
sess.graph)
# Run the initializer
sess.run(init)
i = 0
print 'started training'
for epoch in range(num_epoch):
for step in range(data_X.shape[0]/batch_size):
batch_x = data_X[step*batch_size:(step+1)*batch_size]
i+=1
batch_x = uncompress(batch_x,86796)
print 'restoring ae session'
saver.restore(sess, "ae_logs_1/save.ckpt")
print 'done loading'
init_new_op = tf.variables_initializer([v for v in tf.global_variables() if v.name.split(':')[0] in set(sess.run(tf.report_uninitialized_variables()))])
sess.run(init_new_op)
print 'input - '
print 'input shape - ', batch_x.shape
print batch_x
stage1,stage2,stage3 = sess.run([encoder_1, encoder_2,compressed],feed_dict={X: batch_x})
print '******stage 1******'
print '******stage 1 shape', stage1.shape
print stage1
print '******stage 2******'
print '******stage 2 shape', stage2.shape
print stage2
print '******stage 3******'
print '******stage 3 shape', stage3.shape
print stage3
exit(0)
print 'done!'
if __name__=='__main__':
os.environ['CUDA_VISIBLE_DEVICES'] = ''
#compress()
# x = compress(None,None,True)
# print x
# batch_x = np.identity((2,50,50))
# batch_x = compress(batch_x,x)
# exit(0)
print_intermediate()
exit(0)
learning_rate = 0.00001
num_epoch = 1
batch_size = 1
display_step = 1
input_size = 50
X = tf.placeholder(tf.float32, [None, input_size,86796,1])
reconstruction, compressed, encoder_1, encoder_2, decoder_1, decoder_2 = autoencoder(X)
loss_op = tf.reduce_mean(tf.square(reconstruction - X))
tf.summary.scalar('loss',loss_op)
optimizer = tf.train.AdamOptimizer(learning_rate=learning_rate)
train_op = optimizer.minimize(loss_op)
init = tf.global_variables_initializer()
data_X,_ = load_data()
indices = np.random.permutation(np.arange(data_X.shape[0]))
data_X = data_X[indices,:,:]
tf.contrib.layers.summarize_collection(tf.GraphKeys.TRAINABLE_VARIABLES)
merged = tf.summary.merge_all()
saver = tf.train.Saver()
with tf.Session() as sess:
train_writer = tf.summary.FileWriter("ae_logs_1/",
sess.graph)
# Run the initializer
sess.run(init)
'''
print 'restoring session'
saver.restore(sess, "logs3/epoch0i180.ckpt")
print 'done loading'
# exit(0) '''
i = 0
print 'started training'
for epoch in range(num_epoch):
for step in range(data_X.shape[0]/batch_size):
batch_x = data_X[step*batch_size:(step+1)*batch_size]
i+=1
batch_x = uncompress(batch_x,86796)
# Run optimization op (backprop)
_,summary = sess.run([train_op,merged], feed_dict={X: batch_x})
train_writer.add_summary(summary, i)
if step % display_step == 0:
# Calculate batch loss and accuracy
loss = sess.run(loss_op, feed_dict={X: batch_x})
log("LR : "+str(learning_rate)+" Epoch : " + str(epoch) + " Step " + str(step) + ", Loss= " + \
"{:.4f}".format(loss))
if i%20 == 0:
print 'saving checkpoint'
save_path = saver.save(sess, os.path.join('ae_logs_1','save.ckpt'))
print("Model saved in path: %s" % save_path)
print 'done!'