Tensorflow框架學習(1)--CNN的實現

阿新 • • 發佈：2020-12-23

 1 # CNN神經網路的程式碼實現
 2 import tensorflow as tf
 3 # 匯入資料集
 4 from tensorflow.examples.tutorials.mnist import input_data
 5 # number 1 to 10 data:讀取資料
 6 mnist = input_data.read_data_sets("MNIST_data", one_hot=True)
 7 
 8 sess = tf.Session()
 9 
10 #計算準確率
11 def compute_accuracy(v_xs, v_ys):
12     global 
 prediction
13     y_pre = sess.run(prediction, feed_dict={xs: v_xs, ys: v_ys, keep_prob: 0.5})   # keep_prob的設定是為了防止發生過擬合的現象
14     correct_prediction = tf.equal(tf.argmax(y_pre, 1), tf.argmax(v_ys, 1))
15     accuracy = tf.reduce_mean(tf.cast(correct_prediction, tf.float32))
16     result = sess.run(accuracy, feed_dict={xs: v_xs, ys: v_ys, keep_prob: 0.5})
 
17     return result
18 
19 # 定義並初始化weight_variable
20 def weight_variable(shape):
21     initial = tf.truncated_normal(shape, stddev=0.1)
22     return tf.Variable(initial)
23 
24 # 定義並初始化bias_variable
25 def bias_variable(shape):
26     initial = tf.constant(0.1, shape=shape)
27     return tf.Variable(initial)
 
28 
29 #定義卷積層
30 def conv2d(x,W):
31     # strides=[1,x_movement,y_movement,1]
32     # strides[0] = strides[3] = 0
33     return tf.nn.conv2d(x, W, strides=[1, 1, 1, 1], padding="SAME")
34 
35 #定義池化層
36 def max_pool_2x2(x):
37     # strides=[1,x_movement,y_movement,1]
38     # strides[0] = strides[3] = 0
39     # ksize=[1,x,x,1]:其中表示池化的大小
40     # 在池化得我過程中，不能有重疊，否則會出現報錯 --栽溝裡了！！！
41     return tf.nn.max_pool(x, ksize=[1, 2, 2, 1], strides=[1, 2, 2, 1], padding="SAME")
42 
43 # 定義placeholder for inputs to network,None表示輸入的樣本數待定
44 xs = tf.placeholder(tf.float32, [None, 784])     # None表示輸入的樣本數待定（待定）
45 ys = tf.placeholder(tf.float32, [None, 10])
46 keep_prob = tf.placeholder(tf.float32)
47 
48 # 將輸入的資料變換成矩陣的形式
49 # 此時x_image表示的是[n_samples,28,28,1]，其中1表示channel
50 x_image = tf.reshape(xs, [-1, 28, 28, 1])        # -1表示輸入的樣本數待定（待定）
51 
52 # 定義conv1 layer--卷積層1
53 # [5,5,1,32]
54 # 5*5:patch size  1:channel  32:filter的數量
55 W_conv1 = weight_variable([5, 5, 1, 32])
56 b_conv1 = bias_variable([32])
57 h_conv1 = tf.nn.relu(conv2d(x_image, W_conv1) + b_conv1)   #output_size:28*28*32
58 h_pool1 = max_pool_2x2(h_conv1)                           #output_size:14*14*32
59 
60 # 定義conv2 layer
61 # 32：第二層卷積中卷積核的深度   64：卷積核的個數
62 W_conv2 = weight_variable([5, 5, 32, 64])
63 b_conv2 = bias_variable([64])
64 h_conv2 = tf.nn.relu(conv2d(h_pool1, W_conv2) + b_conv2)   #output_size:14*14*64
65 h_pool2 = max_pool_2x2(h_conv2)                           #output_size:7*7*64
66 
67 
68 # 定義func1 layer:全卷積層1
69 W_fc1 = weight_variable([7*7*64, 1024])    #全連線層的in_size:7*7*64  out_size:1024
70 b_fc1 = bias_variable([1024])
71 # 將con2 layer的輸出拉平:[n_sample,7,7,64] ->> [n_sample,7*7*64]
72 h_pool2_flat = tf.reshape(h_pool2, [-1, 7*7*64])        # -1表示輸入的樣本數（待定）
73 h_fc1 = tf.nn.relu(tf.matmul(h_pool2_flat, W_fc1) + b_fc1)
74 # 為了防止出現過擬合，使用tf.nn.dropout()
75 h_fc1_drop = tf.nn.dropout(h_fc1, keep_prob)
76 
77 # 定義func2 layer
78 W_fc2 = weight_variable([1024, 10])    #全連線層的in_size:1024  out_size:10(分類對應10個數據)
79 b_fc2 = bias_variable([10])
80 # 將con2 layer的輸出拉平:[n_sample,7,7,64] ->> [n_sample,7*7*64]
81 prediction = tf.nn.softmax(tf.matmul(h_fc1_drop, W_fc2)+b_fc2)
82 
83 # the error between prediction and real data
84 cross_entropy = tf.reduce_mean(-tf.reduce_sum(ys * tf.log(prediction), reduction_indices=[1]))
85 # 優化器的選擇
86 train_step = tf.train.AdamOptimizer(1e-4).minimize(cross_entropy)
87 
88 # 定義變數需要初始化所有的引數  -- very important
89 sess.run(tf.initialize_all_variables())
90 
91 # 資料的訓練
92 for i in range(1000):
93     batch_xs, batch_ys = mnist.train.next_batch(100)
94     sess.run(train_step, feed_dict={xs: batch_xs, ys: batch_ys, keep_prob: 0.5})
95     if i % 50 == 0:
96         #    每訓練50次，計算一下準確略：函式需要自定義
97         print(compute_accuracy(
98             mnist.test.images, mnist.test.labels))