-
Notifications
You must be signed in to change notification settings - Fork 1
/
emnist_classifier.py
124 lines (112 loc) · 5.42 KB
/
emnist_classifier.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
#-----------------------------------------------------------------#
# Author: Nicholas Keen #
# #
# This is my Capstone Project for SUNY Potsdam, it is a program #
# that classifies MNIST and EMNIST data. It relies on tensorflow #
# to run. #
# #
# Further, you will need to alter the function extract_labels to #
# include the apropriate number of classes. #
# #
# This function can be found in the follwing pathway #
# tensorflow/contrib/learn/python/learn/datasets/mnist.py #
# #
# You will also need to either alter the names of the datasets #
# you want to classify to look like the standard MNIST names #
# #
# OR #
# #
# In that same mnist.py file alter TRAIN_IMAGES, TRAIN_LABELS, #
# TEST_IMAGES and TEST_LABELS in the read_data_sets function to #
# match the names of the datasets you want to classify #
# #
# This program was adapted from the tensorflow tutorials #
#-----------------------------------------------------------------#
#imports the input data function from tensorflow
from tensorflow.examples.tutorials.mnist import input_data
#loads mnist/emnist from a directory
emnist = input_data.read_data_sets('directory/containing/emnist/data', one_hot = True)
import tensorflow as tf
sess = tf.InteractiveSession()
#outputs determines the number of outputs, 10 for mnist, 47 for emnist,
#62 for byclass, etc. and text determines the name of the file
#that output will be written to
outputs = 10
text = "printResultsToThisFile.txt"
#weight variable
def weight_variable(shape):
initial = tf.truncated_normal(shape, stddev = 0.1)
return tf.Variable(initial)
#bias variable
def bias_variable(shape):
initial = tf.constant(0.1, shape = shape)
return tf.Variable(initial)
#convolution layer
def conv2d(x, W):
return tf.nn.conv2d(x, W, strides = [1, 1, 1, 1], padding = 'SAME')
#pooling layer
def max_pool_2x2(x):
return tf.nn.max_pool(x, ksize = [1, 2, 2, 1],
strides = [1, 2, 2, 1], padding = 'SAME')
#set initial x value
x = tf.placeholder(tf.float32, shape = [None, 784])
#set expected outputs
y_ = tf.placeholder(tf.float32, shape = [None, outputs])
#creates convolution layer from inputs
W_conv1 = weight_variable([5, 5, 1, 32])
b_conv1 = bias_variable([32])
x_image = tf.reshape(x, [-1, 28, 28, 1])
#hooks input to hidden layer 1 and applies a relu function
h_conv1 = tf.nn.relu(conv2d(x_image, W_conv1) + b_conv1)
h_pool1 = max_pool_2x2(h_conv1)
W_conv2 = weight_variable([5, 5, 32, 64])
b_conv2 = bias_variable([64])
#hooks hidden layer 1 to hidden layer 2, and applies a relu function
h_conv2 = tf.nn.relu(conv2d(h_pool1, W_conv2) + b_conv2)
h_pool2 = max_pool_2x2(h_conv2)
W_fc1 = weight_variable([7*7*64, 1024])
b_fc1 = bias_variable([1024])
h_pool2_flat = tf.reshape(h_pool2, [-1, 7*7*64])
#hooks hidden layer 2 to the final hidden layer and applies a relu function
h_fc1 = tf.nn.relu(tf.matmul(h_pool2_flat, W_fc1) + b_fc1)
keep_prob = tf.placeholder(tf.float32)
#applies dropout
h_fc1_drop = tf.nn.dropout(h_fc1, keep_prob)
W_fc2 = weight_variable([1024, outputs])
b_fc2 = bias_variable([outputs])
#output
y_conv = tf.matmul(h_fc1_drop, W_fc2) + b_fc2
#applies cross entropy loss on the expected and actual values
cross_entropy = tf.reduce_mean(
tf.nn.softmax_cross_entropy_with_logits(labels = y_, logits = y_conv))
#applies the Adam Optimizer
train_step = tf.train.AdamOptimizer(1e-4).minimize(cross_entropy)
correct_prediction = tf.equal(tf.argmax(y_conv, 1), tf.argmax(y_, 1))
accuracy = tf.reduce_mean(tf.cast(correct_prediction, tf.float32))
with tf.Session() as sess:
sess.run(tf.global_variables_initializer())
#batches data and trains
for i in range(20000):
batch = emnist.train.next_batch(50)
if i%100 == 0:
train_accuracy = accuracy.eval(feed_dict = {
x: batch[0], y_: batch[1], keep_prob: 1.0})
#evaluate feedback every 100 epochs
print('step %d, training accuracy %g' % (i, train_accuracy))
print('step %d, training accuracy %g' % (i, train_accuracy), file = open(text, "a"))
train_step.run(feed_dict = {x: batch[0], y_: batch[1], keep_prob: 0.5})
tot_accuracy = 0
#batches data and tests
for i in range(20000):
batch = emnist.test.next_batch(50)
if i%100 == 0:
test_accuracy = accuracy.eval(feed_dict = {
x: batch[0], y_: batch[1], keep_prob: 1.0})
#evaluate feedback every 100 epochs
print('step %d, test accuracy %g' % (i, test_accuracy))
print('step %d, test accuracy %g' % (i, test_accuracy), file = open(text, "a"))
tot_accuracy = tot_accuracy + test_accuracy
#find total accuracy and report it
final = tot_accuracy/200.0
print('Final test accuracy: %f' % final)
print('Final test accuracy: %f' % final, file = open(text, "a"))