-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathautoencoder.py
157 lines (121 loc) · 5.49 KB
/
autoencoder.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
""" Deep Auto-Encoder implementation
An auto-encoder works as follows:
Data of dimension k is reduced to a lower dimension j using a matrix multiplication:
softmax(W*x + b) = x'
where W is matrix from R^k --> R^j
A reconstruction matrix W' maps back from R^j --> R^k
so our reconstruction function is softmax'(W' * x' + b')
Now the point of the auto-encoder is to create a reduction matrix (values for W, b)
that is "good" at reconstructing the original data.
Thus we want to minimize ||softmax'(W' * (softmax(W *x+ b)) + b') - x||
A deep auto-encoder is nothing more than stacking successive layers of these reductions.
"""
#code refrence https://gist.github.com/saliksyed/593c950ba1a3b9dd08d5
import tensorflow as tf
import numpy as np
import math
import random
import pandas as pd
import matplotlib.pyplot as plt
import sys
from sklearn.cross_validation import train_test_split
def create(x, layer_sizes):
# Build the encoding layers
next_layer_input = x
encoding_matrices = []
for dim in layer_sizes:
input_dim = int(next_layer_input.get_shape()[1])
# Initialize W using random values in interval [-1/sqrt(n) , 1/sqrt(n)]
W = tf.Variable(tf.random_uniform([input_dim, dim], -1.0 / math.sqrt(input_dim), 1.0 / math.sqrt(input_dim)))
#W = tf.Variable(tf.zeros([input_dim,dim]))
# Initialize b to zero
b = tf.Variable(tf.zeros([dim]))
# We are going to use tied-weights so store the W matrix for later reference.
encoding_matrices.append(W)
#output = tf.nn.tanh(tf.matmul(next_layer_input,W) + b)
output = tf.matmul(next_layer_input,W) + b
# the input into the next layer is the output of this layer
next_layer_input = output
# The fully encoded x value is now stored in the next_layer_input
encoded_x = next_layer_input
# build the reconstruction layers by reversing the reductions
layer_sizes.reverse()
encoding_matrices.reverse()
for i, dim in enumerate(layer_sizes[1:] + [ int(x.get_shape()[1])]) :
# we are using tied weights, so just lookup the encoding matrix for this step and transpose it
W = tf.transpose(encoding_matrices[i])
b = tf.Variable(tf.zeros([dim]))
output = tf.matmul(next_layer_input,W) + b
next_layer_input = output
# the fully encoded and reconstructed value of x is here:
reconstructed_x = next_layer_input
return {
'encoded': encoded_x,
'decoded': reconstructed_x,
'cost' : tf.sqrt(tf.reduce_mean(tf.square(x-reconstructed_x)))
}
def error_thresh(x):
return x>=29
#return x>=0.45
def log_var(x):
if x[7] == 0:
#rep = 0.00000001
rep=x[7]
else:
rep = x[7]
return (x[0], x[1], x[2], x[3], x[4], x[5], x[6], rep, x[8], x[9])
def deep_test(data):
sess = tf.Session()
random.seed(1)
train, test = train_test_split(data,test_size=0.1)
print "test",test.shape
print "train",train.shape
data_filtered = train[train['TARGET']==0].ix[:,:9]
test_pos = test[test['TARGET']==0]
test_neg = test[test['TARGET']==1]
print "no of filtered rows",data_filtered.shape
total_pos = test_pos.shape[0]
total_neg = test_neg.shape[0]
print "no of positive rows",total_pos
print "no of negative rows",total_neg
start_dim = data_filtered.shape[1]
x = tf.placeholder("float", [None, start_dim])
autoencoder = create(x, [9, 7, 5, 3, 2])
# lyr = range(10,350)
# lyr.reverse()
# autoencoder = create(x,lyr)
init = tf.initialize_all_variables()
sess.run(init)
train_step = tf.train.GradientDescentOptimizer(0.00000001).minimize(autoencoder['cost'])
for i in range(20000):
data_filtered.reindex(np.random.permutation(data_filtered.index))
batch = data_filtered.iloc[:,:].values[:100]
sess.run(train_step, feed_dict={x: batch})
#print i, " cost", sess.run(autoencoder['cost'], feed_dict={x: batch})
#print i, " test cost_pos", sess.run(autoencoder['cost'], feed_dict={x:test_pos.iloc[:,:9].values})
#print i, " test cost_neg", sess.run(autoencoder['cost'], feed_dict={x:test_neg.iloc[:,:9].values})
test_pos_r = [sess.run(autoencoder['cost'], feed_dict={x: np.reshape(np.array(test_pos.ix[i,0:9]),(1,9))}) for i in test_pos.index]
#test_pos_r = [sess.run(autoencoder['cost'], feed_dict={x: np.reshape(np.array(test_pos.ix[i,:]),(1,371))}) for i in test_pos.index]
test_neg_r = [sess.run(autoencoder['cost'], feed_dict={x: np.reshape(np.array(test_neg.ix[i,0:9]),(1,9))}) for i in test_neg.index]
pos_above_thresh= sum(1 for x in test_pos_r if error_thresh(x))
neg_above_thresh= sum(1 for x in test_neg_r if error_thresh(x))
print "% positives above threshold", (float(pos_above_thresh)/total_pos)*100 , "%"
print "% negatives above threshold", (float(neg_above_thresh)/total_neg)*100 , "%"
plt.plot(test_pos_r,'ro',color='b')
plt.plot(test_neg_r,'ro',color='r')
plt.show()
# print "testing begins"
# for i in test.index:
# print test.loc[i,'TARGET'],sess.run(autoencoder['cost'], feed_dict={x: np.reshape(np.array(test.ix[i,0:9]),(1,9))})
if __name__ == '__main__':
train = pd.read_csv("./santander/train.csv")
features = ['var15', 'ind_var5', 'ind_var8_0', 'ind_var30', 'num_var5', 'num_var30', 'num_var42', 'var36', 'num_meses_var5_ult3']
data = train[features+['TARGET']]
#cols_to_norm = ['var15', 'num_var5', 'num_var30', 'num_var42', 'var36', 'num_meses_var5_ult3']
#data[cols_to_norm] = data[cols_to_norm].apply(lambda x: (x - x.mean()) / (x.max() - x.min()))
#data_trans = data.apply(log_var, axis=1)
# data['var38mc'] = np.isclose(data.var38, 117310.979016)
# data['logvar38'] = data.loc[~data['var38mc'], 'var38'].map(np.log)
# data.loc[data['var38mc'], 'logvar38'] = 0
# data.drop('var38', axis=1, inplace=True)
deep_test(data)