上一講筆者和大家一起學習了如何使用 Tensorflow 構建一個卷積神經(jīng)網(wǎng)絡模型。本節(jié)我們將繼續(xù)利用 Tensorflow 的便捷性完成 mnist 手寫數(shù)字數(shù)據(jù)集的識別實戰(zhàn)。mnist 數(shù)據(jù)集是 Yann Lecun 大佬基于美國國家標準技術研究所構建的一個研究深度學習的手寫數(shù)字的數(shù)據(jù)集。mnist 由 70000 張不同人手寫的 0-9 10個數(shù)字的灰度圖組成。本節(jié)筆者就和大家一起研究如何利用 Tensorflow 搭建一個 CNN 模型來識別這些手寫的數(shù)字。

數(shù)據(jù)導入

mnist 作為標準深度學習數(shù)據(jù)集，在各大深度學習開源框架中都默認有進行封裝。所以我們直接從 Tensorflow 中導入相關的模塊即可：

import tensorflow as tf
from tensorflow.examples.tutorials.mnist 
import input_data
# load mnist data
mnist = input_data.read_data_sets('MNIST_data', one_hot=True)

快速搭建起一個簡易神經(jīng)網(wǎng)絡模型

數(shù)據(jù)導入之后即可按照 Tensorflow 的范式創(chuàng)建相應的 Tensor 變量然后創(chuàng)建會話：

# create the session
sess = tf.InteractiveSession()
# create variables and run the session
x = tf.placeholder('float', shape=[None, 784])
y_ = tf.placeholder('float', shape=[None, 10])
W = tf.Variable(tf.zeros([784, 10]))
b = tf.Variable(tf.zeros([10]))
sess.run(tf.global_variables_initializer())

定義前向傳播過程和損失函數(shù)：

#definethenetandlossfunctiony=tf.nn.softmax(tf.matmul(x,W)+b)
cross_entropy=-tf.reduce_sum(y_*tf.log(y))

進行模型訓練：

# train the model
train_step = tf.train.GradientDescentOptimizer(0.01).minimize(cross_entropy)
for i in range(1000):
 batch = mnist.train.next_batch(50)
 train_step.run(feed_dict={x: batch[0], y_: batch[1]})

使用訓練好的模型對測試集進行預測：

# evaluate the model
correct_prediction = tf.equal(tf.argmax(y,1), tf.argmax(y_,1))
accuracy = tf.reduce_mean(tf.cast(correct_prediction, "float"))
print(accuracy.eval(feed_dict={x: mnist.test.images, y_: mnist.test.labels}))

預測準確率為 0.9，雖然說也是一個很高的準確率了，但對于 mnist 這種標準數(shù)據(jù)集來說，這樣的結果還有很大的提升空間。所以我們繼續(xù)優(yōu)化模型結構，為模型添加卷積結構。

搭建卷積神經(jīng)網(wǎng)絡模型

定義初始化模型權重函數(shù)：

# initilize the weight
def weight_variable(shape):
  initial = tf.truncated_normal(shape, stddev=0.1) 
  return tf.Variable(initial)
  
def bias_variable(shape):
  initial = tf.constant(0.1, shape=shape) 
  return tf.Variable(initial)

定義卷積和池化函數(shù)：

# convolutional and pooling
def conv2d(x, W): 
  return tf.nn.conv2d(x, W, strides=[1, 1, 1, 1], padding='SAME')

def max_pool_2x2(x): 
  return tf.nn.max_pool(x, ksize=[1, 2, 2, 1],
            strides=[1, 2, 2, 1], padding='SAME')

搭建第一層卷積：

# the first convolution layer
W_conv1 = weight_variable([5, 5, 1, 32])
b_conv1 = bias_variable([32])
x_image = tf.reshape(x, [-1,28,28,1])
h_conv1 = tf.nn.relu(conv2d(x_image, W_conv1) + b_conv1)
h_pool1 = max_pool_2x2(h_conv1)

搭建第二層卷積：

# the second convolution layer
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)
h_pool2 = max_pool_2x2(h_conv2)

搭建全連接層：

# dense layer/full_connected layer
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)

設置 dropout 防止過擬合：

# dropout to prevent overfitting
keep_prob = tf.placeholder("float")
h_fc1_drop = tf.nn.dropout(h_fc1, keep_prob)

對輸出層定義 softmax ：

# model output
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)

訓練模型并進行預測：

# model trainning and evaluating
cross_entropy = -tf.reduce_sum(y_*tf.log(y_conv))
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, "float"))
sess.run(tf.initialize_all_variables())
for i in range(20000):
  batch = mnist.train.next_batch(50) 
    if i%100 == 0:
      train_accuracy = accuracy.eval(feed_dict={
        x:batch[0], y_: batch[1], keep_prob: 1.0})
  print("step %d, training accuracy %g"%(i, train_accuracy))
  train_step.run(feed_dict={x: batch[0], y_: batch[1], keep_prob: 0.5})

print("test accuracy %g"%accuracy.eval(feed_dict={
  x: mnist.test.images, y_: mnist.test.labels, keep_prob: 1.0}))

部分迭代過程和預測結果如下：

經(jīng)過添加兩層卷積之后我們的模型預測準確率達到了 0.9931，模型訓練的算是比較好了