[TensorFlow] Tutorial 1. MNIST For ML Beginners

텐서플로우 튜토리얼 따라하기 1편.

코드 복사해서 돌려보고 분석하기.

#-*- coding: utf-8 -*-

#MNIST For ML Beginners
import tensorflow as tf

from tensorflow.examples.tutorials.mnist import input_data
mnist = input_data.read_data_sets("MNIST_data/", one_hot=True)

# train data y = xW+b
# placeholder와 variable 차이는 session 생성 전에 초기화 유무.
x = tf.placeholder(tf.float32, [None, 784]) # 28*28 = 784 dim
W = tf.Variable(tf.zeros([784, 10]))        # 784 * 10 class
b = tf.Variable(tf.zeros([10]))
y = tf.nn.softmax(tf.matmul(x, W) + b)

# label
y_ = tf.placeholder(tf.float32, [None, 10]) # none은 행의 수가 한정되지 않음을 의미.

# define loss
cross_entropy = tf.reduce_mean(-tf.reduce_sum(y_ * tf.log(y), reduction_indices=[1]))

# define weight update method
train_step = tf.train.GradientDescentOptimizer(0.5).minimize(cross_entropy)

# define init variable 변수는 초기화 필수
init = tf.initialize_all_variables()

# create session
sess = tf.Session()

# initialization
sess.run(init)

# training / batch size : 100 / iteration 100
for i in range(1000):
  batch_xs, batch_ys = mnist.train.next_batch(100)
  # python 문법. feed_dict 파라미터에 직접 dictionary 넣어주기.
  sess.run(train_step, feed_dict={x: batch_xs, y_: batch_ys})

# define correctness
correct_prediction = tf.equal(tf.argmax(y,1), tf.argmax(y_,1))

# bool을 float으로 캐스팅하고 평균을 구함. 즉 맞은 확률 계산.
accuracy = tf.reduce_mean(tf.cast(correct_prediction, tf.float32))

# run & print
print(sess.run(accuracy, feed_dict={x: mnist.test.images, y_: mnist.test.labels}))