/
laa.py
133 lines (121 loc) · 5.01 KB
/
laa.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
'''
Created on Dec 4, 2016
@author: jianhua
'''
import random
import math
import tensorflow as tf
import numpy as np
import matplotlib.pyplot as plt
from numpy.linalg import cholesky
#definition=================================================================================================
source_num = 5
source_label_per_object = 2
object_num = 4
simulate_lambda = 0.2
cate_num = 2;
#true_label==================================================================================================
true_label = [0 for x in range(object_num)]
for i in range(object_num):
if i < object_num / 2:
true_label[i] = 1
else:
true_label[i] = 2
#print(true_label)
#simulate data==============================================================================================
def initial(source_ability_mu,source_ability_sigma):
label = [([0] * source_num * 2) for i in range(object_num)]
simulate_source_ability = np.random.normal(source_ability_mu, source_ability_sigma, source_num)
simulate_source_rate = 1 / (1 + np.exp(-simulate_lambda * simulate_source_ability));
#instance_difficulty = np.random.normal(instance_difficulty_mu,instance_difficulty_sigma, object_num)
'''ddad'''
tmp_sort = []
for i in range(0, source_num):
tmp_sort.append(i)
for i in range(object_num):
random.shuffle(tmp_sort)
tmp_source_pool = tmp_sort[0:source_label_per_object]
tmp_rate = simulate_source_rate[tmp_source_pool]
tmp_rand = np.random.rand(source_label_per_object);
#print(tmp_rate)
#print(tmp_rand)
for j in range(source_label_per_object):
if tmp_rate[j] < tmp_rand[j]:
label[i][2 * tmp_source_pool[j] + 2 - true_label[i]] = 1;
else:
label[i][2 * tmp_source_pool[j] + true_label[i] - 1] = 1;
#print(simulate_source_rate)
#print (label)
return label
#=============================================================================================================
label = initial(1.5,16)
#print label[:][1]
#=============================================================================================================
tmp_encoder_h2 = [([0.0] * cate_num) for i in range(source_num * 2)]
for i in range (cate_num):
for j in range(source_num * 2):
if (i + j) % 2 == 0:
tmp_encoder_h2[j][i] = 1.0
#print (tmp_encoder_h2)
tmp_decoder = [([0.0] * source_num * 2) for i in range(source_num * 2)]
for i in range (source_num * 2):
tmp_decoder[i][i] = 1.0
if i % 2 == 0:
tmp_decoder[i][i + 1] = 1.0
tmp_decoder[i + 1][i] = 1.0
print (tmp_decoder)
#=============================================================================================================
display_step = 1
learning_rate = 0.01
training_epochs = 20
n_input = source_num * 2
n_hidden_left = n_input
n_hidden_right = 2
#print(n_input)
X = tf.placeholder("float", [None, n_input])
theta = tf.constant(0.5)
weights = {
'encoder_h1': tf.Variable(tf.random_normal([n_input, n_hidden_left],mean = 0.0, stddev = 0.01)),
'encoder_h2': tf.Variable(tmp_encoder_h2),
'decoder_h1': tf.Variable(tf.random_normal([n_hidden_left, n_input],mean = 0.0, stddev = 0.01)),
'decoder_h2': tf.Variable(tf.random_normal([n_hidden_right, n_input],mean = 0.0, stddev = 0.01)),
}
biases = {
'encoder_b1': tf.constant(0.0 ,shape=[n_hidden_left]),
'encoder_b2': tf.constant(0.0, shape=[n_hidden_right]),
'decoder_b1': tf.constant(0.0, shape=[n_input]),
'decoder_b2': tf.constant(0.0, shape=[n_input]),
}
tmp_div = tf.Variable(tmp_decoder)
print (weights['encoder_h1'])
def encoder(x):
print (x)
# Encoder Hidden layer with sigmoid activation #1
layer_left = tf.nn.sigmoid(tf.add(tf.matmul(x, weights['encoder_h1']),
biases['encoder_b1']))
# Decoder Hidden layer with sigmoid activation #2
layer_right = tf.nn.softmax(tf.add(tf.matmul(x, weights['encoder_h2']),
biases['encoder_b2']))
return layer_left,layer_right
def decoder(x,y):
tmp = tf.exp(tf.add(tf.matmul(x, weights['decoder_h1']), biases['decoder_b1']) + tf.add(tf.matmul(y, weights['decoder_h2']),
biases['decoder_b2']))
layer_final = tf.div(tmp,tmp * tmp_div)
return layer_final
encoder_left, encoder_right = encoder(X)
decoder_op = decoder(encoder_left, encoder_right)
y_pred = decoder_op
y_true = X
#cost = tf.reduce_mean(tf.pow(y_true - y_pred, 2))
cost = tf.nn.sigmoid_cross_entropy_with_logits(y_pred, y_true) + tf.mul(theta, tf.nn.l2_loss(weights['encoder_h1'] - tmp_div))
optimizer = tf.train.RMSPropOptimizer(learning_rate).minimize(cost)
init = tf.initialize_all_variables()
with tf.Session() as sess:
sess.run(init)
# Training cycle
for epoch in range(training_epochs):
for i in range(object_num):
_, c = sess.run([optimizer, cost], feed_dict={X: [label[i][:]]})
if epoch % display_step == 0:
print("Epoch:", '%04d' % (epoch+1),
"cost=", "{0}".format(c))