-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathRNN_study_1.py
197 lines (151 loc) · 8.95 KB
/
RNN_study_1.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
# 参考文章：https://www.leiphone.com/news/201704/IlnwSvF6pGOZoHZq.html
# from __future__ import print_function, division
import numpy as np
import tensorflow as tf
import matplotlib.pyplot as plt
num_epochs = 100
total_series_length = 50000
truncated_backprop_length = 15
state_size = 4
num_classes = 2
echo_step = 3
batch_size = 5
num_batches = total_series_length//batch_size//truncated_backprop_length # 666 次，每次 15 个数
'''
return shape(batch_size, total_series_length / batch_size) = (5, 10000):
x = [[1 1 0 ... 1 1 1]
[0 0 1 ... 0 1 1]
[0 1 0 ... 0 0 0]
[1 0 0 ... 1 0 0]
[0 0 0 ... 1 0 0]]
y = [[0 0 0 ... 0 1 1]
[0 0 0 ... 1 1 1]
[0 0 0 ... 1 0 0]
[0 0 0 ... 0 0 1]
[0 0 0 ... 1 0 0]]
'''
def generateData():
x = np.array(np.random.choice(2, total_series_length, p=[0.5, 0.5]))
y = np.roll(x, echo_step) # 后 3 个放到前面。 例如：x = [1,2,3,4,5,6,7,8,9,10,11,12], y = [10,11,12,1,2,3,4,5,6,7,8,9]
# y[0:echo_step] = 0
x = x.reshape((batch_size, -1)) # 按 batch_size 数分段。例如：batch_size = 2, 那么 x = [[1,2,3,4,5,6], [7,8,9,10,11,12]]; y = [[10,11,12,1,2,3], [4,5,6,7,8,9]]
y = y.reshape((batch_size, -1))
y[:, :echo_step] = 0 # 前 3 个元素置为 0。 例如： y = [[0,0,0,1,2,3], [0,0,0,7,8,9]]
return (x, y)
batchX_placeholder = tf.placeholder(tf.float32, [batch_size, truncated_backprop_length]) # (5, 15)
batchY_placeholder = tf.placeholder(tf.int32, [batch_size, truncated_backprop_length])
'''
batchX = [[1 1 0 0 0 0 1 1 1 1 0 1 1 0 0]
[0 0 1 0 1 0 1 1 1 1 1 0 1 1 1]
[0 1 0 1 1 1 0 1 1 0 0 1 1 1 1]
[1 0 0 1 1 1 0 1 0 1 1 1 0 0 0]
[0 0 0 0 1 1 0 0 0 1 1 1 0 0 0]]
batchY = [[0 0 0 1 1 0 0 0 0 1 1 1 1 0 1]
[0 0 0 0 0 1 0 1 0 1 1 1 1 1 0]
[0 0 0 0 1 0 1 1 1 0 1 1 0 0 1]
[0 0 0 1 0 0 1 1 1 0 1 0 1 1 1]
[0 0 0 0 0 0 0 1 1 0 0 0 1 1 1]]
'''
init_state = tf.placeholder(tf.float32, [batch_size, state_size]) # (5, 4) 初值 = [[0. 0. 0. 0.] [0. 0. 0. 0.] [0. 0. 0. 0.] [0. 0. 0. 0.] [0. 0. 0. 0.]]
W = tf.Variable(np.random.rand(state_size + 1, state_size), dtype=tf.float32) # (5, 4)
b = tf.Variable(np.zeros((1, state_size)), dtype=tf.float32) # (1, 4)
W2 = tf.Variable(np.random.rand(state_size, num_classes),dtype=tf.float32) # (4, 2)
b2 = tf.Variable(np.zeros((1, num_classes)), dtype=tf.float32) # (1, 2)
# 把每个数拆成一个元素。Unpack columns
# inputs_series = tf.unpack(batchX_placeholder, axis=1) # 按列解包 (15, 5) 。tf.unpack 在 1.x 已被 tf.unstack 替代
# labels_series = tf.unpack(batchY_placeholder, axis=1)
# inputs_series = tf.transpose(batchX_placeholder, perm=[1, 0])
# labels_series = tf.transpose(batchY_placeholder, perm=[1, 0])
# inputs_series = np.transpose(batchX_placeholder)
# labels_series = np.transpose(batchY_placeholder)
inputs_series = tf.unstack(batchX_placeholder, axis=1) # [(5, 1), .... 共 15 个] 把每个数拆成一个元素
labels_series = tf.unstack(batchY_placeholder, axis=1)
'''
batchX = [[1, 1, 0, 0, 0, 0, 1, 1, 1, 1, 0, 1, 1, 0, 0],
[0, 0, 1, 0, 1, 0, 1, 1, 1, 1, 1, 0, 1, 1, 1],
[0, 1, 0, 1, 1, 1, 0, 1, 1, 0, 0, 1, 1, 1, 1],
[1, 0, 0, 1, 1, 1, 0, 1, 0, 1, 1, 1, 0, 0, 0],
[0, 0, 0, 0, 1, 1, 0, 0, 0, 1, 1, 1, 0, 0, 0]]
inputs_series = [[1, 0, 0, 1, 0], [1, 0, 1, 0, 0], [0, 1, 0, 0, 0], [0, 0, 1, 1, 0], [0, 1, 1, 1, 1], [0, 0, 1, 1, 1], [1, 1, 0, 0, 0], [1, 1, 1, 1, 0], [1, 1, 1, 0, 0], [1, 1, 0, 1, 1], [0, 1, 0, 1, 1], [1, 0, 1, 1, 1], [1, 1, 1, 0, 0], [0, 1, 1, 0, 0], [0, 1, 1, 0, 0]]
'''
print('inputs_series, info = ', inputs_series, '\n')
# inputs_series = tf.Print(inputs_series, [inputs_series, inputs_series.shape], 'inputs_series = ')
# Forward pass
current_state = init_state
states_series = []
index = 0
for current_input in inputs_series:
current_input = tf.Print(current_input, [current_input, current_input.shape, 'index = ' + str(index)], message='current_input0 = ', summarize=100)
# [1, 0, 0, 1, 0]
current_input = tf.reshape(current_input, [batch_size, 1]) # shape: (5, 1)。
current_input = tf.Print(current_input, [current_input, current_input.shape, 'index = ' + str(index)], message='current_input1 = ', summarize=100)
# [[1][0][0][1][0]]
print('current, info = ', current_input, '\n')
current_state = tf.Print(current_state, [current_state, current_state.shape, 'index = ' + str(index)], message='current_state = ', summarize=100)
# [[0.788977802 -0.758756518 0.878002822 0.829297543][0.485389352 -0.58057 0.708371818 -0.412607044][0.551199 0.24081631 0.932751775 0.228438258][-0.816147685 -0.0264752097 -0.915767789 -0.879304767][-0.751862705 -0.0714274198 0.888504 0.431245238]][5 4]
# 列后追加
# input_and_state_concatenated = tf.concat(1, [current_input, current_state]) # Increasing number of columns
input_and_state_concatenated = tf.concat([current_input, current_state], 1) # shape (5, 5)
input_and_state_concatenated = tf.Print(input_and_state_concatenated, [input_and_state_concatenated, input_and_state_concatenated.shape, 'index = ' + str(index)], message='input_and_state_concatenated = ', summarize=100)
# [[1 -0.732565463 -0.0681669414 0.908388376 0.454913557][0 0.644927859 0.211452574 -0.751372695 -0.90508461][0 -0.737988293 0.234725475 -0.907563686 -0.87096405][1 0.661614358 -0.143374264 -0.873949289 -0.301904887][0 -0.71041739 -0.154791594 -0.74115169 -0.793806553]][5 5]
next_state = tf.tanh(tf.matmul(input_and_state_concatenated, W) + b) # Broadcasted addition
next_state = tf.Print(next_state, [next_state, next_state.shape, 'index = ' + str(index)], message='next_state = ', summarize=100)
# [[0.831956625 0.16452603 -0.891076565 -0.345829368][-0.718315482 -0.751014352 0.862468779 0.130369157][-0.562301695 0.132066295 -0.847182155 -0.863753378][-0.752997875 -0.124342352 0.488403261 0.874053717][-0.733458877 -0.0517550744 -0.960962713 -0.207119182]][5 4]
states_series.append(next_state)
current_state = next_state
index += 1
# loss
logits_series = [tf.matmul(state, W2) + b2 for state in states_series] #Broadcasted addition
print('logits_series = ', logits_series)
predictions_series = [tf.nn.softmax(logits) for logits in logits_series]
print('predictions_series = ', predictions_series)
losses = [tf.nn.sparse_softmax_cross_entropy_with_logits(logits=logits, labels=labels) for logits, labels in zip(logits_series,labels_series)]
print('losses = ', losses)
total_loss = tf.reduce_mean(losses)
train_step = tf.train.AdagradOptimizer(0.3).minimize(total_loss)
def plot(loss_list, predictions_series, batchX, batchY):
plt.subplot(2, 3, 1)
plt.cla()
plt.plot(loss_list)
for batch_series_idx in range(5):
one_hot_output_series = np.array(predictions_series)[:, batch_series_idx, :]
single_output_series = np.array([(1 if out[0] < 0.5 else 0) for out in one_hot_output_series])
plt.subplot(2, 3, batch_series_idx + 2)
plt.cla()
plt.axis([0, truncated_backprop_length, 0, 2])
left_offset = range(truncated_backprop_length)
plt.bar(left_offset, batchX[batch_series_idx, :], width=1, color="blue")
plt.bar(left_offset, batchY[batch_series_idx, :] * 0.5, width=1, color="red")
plt.bar(left_offset, single_output_series * 0.3, width=1, color="green")
plt.draw()
plt.pause(0.0001)
with tf.Session() as sess:
sess.run(tf.global_variables_initializer())
plt.ion()
plt.figure()
plt.show()
loss_list = []
for epoch_idx in range(1): # range(num_epochs) 做 100 轮，每次都新生成 50000 个数据
x, y = generateData()
_current_state = np.zeros((batch_size, state_size)) # (5, 4)
print(" ++++++++ New data, epoch", epoch_idx)
print("x = ", x, "\ny = ", y, "\n_current_state = ", _current_state)
for batch_idx in range(2): # range(num_batches)
start_idx = batch_idx * truncated_backprop_length
end_idx = start_idx + truncated_backprop_length
print("start_idx = ", start_idx, "\nend_idx = ", end_idx)
batchX = x[:,start_idx:end_idx]
batchY = y[:,start_idx:end_idx]
print("batchX = ", batchX, "\nbatchY = ", batchY)
_total_loss, _train_step, _current_state, _predictions_series = sess.run([total_loss, train_step, current_state, predictions_series],
feed_dict = {
batchX_placeholder : batchX,
batchY_placeholder : batchY,
init_state : _current_state
})
loss_list.append(_total_loss)
if batch_idx % 100 == 0:
print("Step",batch_idx, "Loss", _total_loss)
plot(loss_list, _predictions_series, batchX, batchY)
plt.ioff()
plt.show()