Convolutional Neural Network (ver. python)
출처¶
http://jorditorres.org/first-contact-with-tensorflow/#cap5 (First Contact with tensorflow)
The MNIST data-set¶
import tensorflow as tf
from tensorflow.examples.tutorials.mnist import input_data
mnist = input_data.read_data_sets("~/MNIST_data/", one_hot=True)
모형에 대한 자세한 설명은 생략하도록 하겠습니다.¶
CNN(convolutional neural network), convolutions, max-pooling, ReLU, softmax, cross entropy, Adam
input -> conv1 -> pool1 -> conv2 -> pool2 -> [inner product -> relu] -> dropout -> [inner product -> softmax] -> output
Loss : cross entropy , Optimizer : Adam
x = tf.placeholder("float", shape = [None, 784])
y_ = tf.placeholder("float", shape = [None, 10])
x_image = tf.reshape(x, [-1, 28, 28, 1])
print("x_image=", x_image)
함수 정의 : weight_variable - truncated normal distribution에서 난수 발생해서 원하는 모양으로 weight tensor를 만드는 함수.
def weight_variable(shape):
initial = tf.truncated_normal(shape)
return tf.Variable(initial)
함수 정의 : bias_variable - 원하는 모양으로 bias tensor를 만드는 함수.
def bias_variable(shape):
initial = tf.constant(1.0, shape = shape)
return tf.Variable(initial)
함수 정의 : conv2d - 2-D convolution 계산하는 함수
def conv2d(x, W):
return tf.nn.conv2d(x, W, strides = [1, 1, 1, 1], padding = 'SAME')
함수 정의 : max_pool_2x2 - max-pooling 하는 함수
def max_pool_2x2(x):
return tf.nn.max_pool(x, ksize = [1, 2, 2, 1], strides = [1, 2, 2, 1], padding = 'SAME')
모형 설정¶
convolution 1
W_conv1 = weight_variable([5, 5, 1, 32])
b_conv1 = bias_variable([32])
h_conv1 = tf.nn.relu(conv2d(x_image, W_conv1) + b_conv1)
max-pooling 1
h_pool1 = max_pool_2x2(h_conv1)
convolution 2
W_conv2 = weight_variable([5, 5, 32, 64])
b_conv2 = bias_variable([64])
h_conv2 = tf.nn.relu(conv2d(h_pool1, W_conv2) + b_conv2)
max-pooling 2
h_pool2 = max_pool_2x2(h_conv2)
reshaping and [inner product - ReLU (activate function)] 1
W_fc1 = weight_variable([7*7*64, 1024])
b_fc1 = bias_variable([1024])
h_pool2_flat = tf.reshape(h_pool2, [-1, 7*7*64])
h_fc1 = tf.nn.relu(tf.matmul(h_pool2_flat, W_fc1) + b_fc1)
dropping out
keep_prob = tf.placeholder("float")
h_fc1_drop = tf.nn.dropout(h_fc1, keep_prob)
[inner product - softmax (activate function)] 2
W_fc2 = weight_variable([1024, 10])
b_fc2 = bias_variable([10])
y_conv = tf.nn.softmax(tf.matmul(h_fc1_drop, W_fc2) + b_fc2)
Loss 와 Optimizer 설정¶
cross_entropy = -tf.reduce_sum(y_*tf.log(y_conv))
train_step = tf.train.AdamOptimizer(1e-4).minimize(cross_entropy)
defining accuracy
correct_prediction = tf.equal(tf.argmax(y_conv, 1), tf.argmax(y_, 1))
accuracy = tf.reduce_mean(tf.cast(correct_prediction, "float"))
Session 반복 실행¶
sess = tf.Session()
sess.run(tf.global_variables_initializer())
for i in range(201):
batch = mnist.train.next_batch(50)
if i%10 == 0:
train_accuracy = sess.run(accuracy, feed_dict = {x: batch[0], y_: batch[1], keep_prob: 1.0})
print("step ", i, "training accuracy ", train_accuracy)
sess.run(train_step, feed_dict={x: batch[0], y_: batch[1], keep_prob: 0.5})
if i%50 == 0:
print("test accuracy ", sess.run(accuracy, feed_dict = {x: mnist.test.images, y_: mnist.test.labels, keep_prob: 1.0}), "step", i)
같이보기¶
Convolutional Neural Network (ver. R)