/
one_fc.py
153 lines (121 loc) · 6.13 KB
/
one_fc.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
import tensorflow as tf
import numpy as np
from scipy.io import loadmat as load
import utils
class Networks():
def __init__(self, num_hidden_node, num_class, batch_size, epoch, learning_rate):
'''
Assign values to model hyperparameters
'''
self.num_hidden_node = num_hidden_node
self.num_class = num_class
self.batch_size = batch_size
self.epoch = epoch
self.learning_rate = learning_rate
self.graph = tf.Graph()
self.global_step = tf.get_variable(name='global_step', initializer=tf.constant(0),trainable=False)
def import_dataset(self):
train = load('train_32x32.mat')
test = load('test_32x32.mat')
train_data = train['X']
train_labels = train['y']
test_data = test['X']
test_labels = test['y']
train_data = np.transpose(train_data, [3,0,1,2])
train_data = utils.rgb2gray(train_data)
train_data = utils.normalize(train_data,-1,1)
train_shape = (train_data.shape[0], train_data.shape[1]*train_data.shape[2])
train_data = np.reshape(train_data, train_shape)
train_labels = utils.one_hot_coding(train_labels)
test_data = np.transpose(test_data,[3,0,1,2])
test_data = utils.rgb2gray(test_data)
test_data = utils.normalize(test_data,-1,1)
test_shape = (test_data.shape[0], test_data.shape[1]*test_data.shape[2])
test_data = np.reshape(test_data, test_shape)
test_labels = utils.one_hot_coding(test_labels)
self.im_size = train_data.shape[1]
#Create datasets from the above tensors
self.train_dataset = tf.data.Dataset.from_tensor_slices((train_data, train_labels))
self.test_dataset = tf.data.Dataset.from_tensor_slices((test_data, test_labels))
def build_graph(self):
'''
import data
'''
self.iterator = self.train_dataset.make_initializable_iterator()
inputs, labels = self.iterator.get_next()
'''
hidden layer
hidden layer weigths shape: input_length * num_hidden_node
hidden layer bias shape: 1 * num_hidden_node
'''
self.hidden_layer_weights = tf.get_variable(name='hidden_layer_weights',
shape=[inputs.shape[0],self.num_hidden_node],
initializer=tf.truncated_normal_initializer(
mean=0.0,
stddev=0.01))
self.hidden_layer_bias = tf.get_variable(name='hidden_layer_bias',
shape=[1,self.num_hidden_node],
initializer=tf.constant_initializer(0))
hidden_layer_output = tf.tensordot(self.hidden_layer_weights, inputs, axes=(0,0))
hidden_layer_output = tf.transpose(hidden_layer_output) + self.hidden_layer_bias
hidden_layer_output = tf.nn.relu(hidden_layer_output)
'''
output layer
output layer weights shape: hidden_layer_output.shape[1] * num_class
output layer bias shape: 1 * num_class
'''
self.output_layer_weights = tf.get_variable(name='output_layer_weights',
shape=[hidden_layer_output.shape[1], self.num_class],
initializer=tf.truncated_normal_initializer(
mean=0.0,
stddev=0.01))
self.output_layer_bias = tf.get_variable(name='output_layer_bias',
shape=[1, self.num_class],
initializer=tf.constant_initializer(0))
output_layer_output = tf.matmul(hidden_layer_output, self.output_layer_weights) \
+ self.output_layer_bias
correct_preds = tf.equal(tf.argmax(output_layer_output, 1), tf.argmax(labels,1))
self.accuracy = tf.reduce_sum(tf.cast(correct_preds, tf.float32))
'''
loss
'''
self.loss = tf.losses.softmax_cross_entropy(labels,output_layer_output)
'''optimizer'''
#self.optimizer = tf.train.GradientDescentOptimizer(
#learning_rate=self.learning_rate).minimize(self.loss, global_step=self.global_step)
self.optimizer = tf.train.AdamOptimizer().minimize(self.loss)
def train(self):
#Create a session and start to train and optimize weights and bias
with tf.Session() as sess:
sess.run(self.iterator.initializer)
sess.run(tf.global_variables_initializer())
for i in range(self.epoch):
total_loss = 0
step = 0
accurate = 0
try:
while(True):
one_accurate,oneloss,_= sess.run([self.accuracy, self.loss, self.optimizer])
step += 1
total_loss += oneloss
accurate += one_accurate
if(step % 500 == 0):
print('Average loss at step {}, epoch {} : {:5.1f} / {:5.1f}'.
format(step, i, total_loss/step, accurate/step))
except tf.errors.OutOfRangeError:
sess.run(self.iterator.initializer)
def test(self):
#Create a session and start to test
pass
def accuracy(self):
pass
if __name__=='__main__':
num_hidden_node = 100
num_class = 10
batch_size = 128
epoch = 10
learning_rate = 0.05
nt = Networks(num_hidden_node,num_class, batch_size,epoch, learning_rate)
nt.import_dataset()
nt.build_graph()
nt.train()