TensorFlow 实现 mnist 手写字识别

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
# 加载TF 并加载数据
import tensorflow as tf
import numpy as np
from tensorflow.examples.tutorials.mnist import input_data

#设置输入参数
batch_size = 128
test_size = 256


# 初始化权值与定义网络结构,建构一个3个卷积层和3个池化层,一个全连接层和一个输出层的卷积神经网络
# 首先定义初始化权重函数
def init_weights(shape):
return tf.Variable(tf.random_normal(shape, stddev=0.01))


# 第一组卷积层以及池化层,最后 droupout是为了防止过拟合,在模型训练的时候丢掉一些神经元
# padding表示对边界的处理,SAME表示卷积的输入和输出保持同样尺寸
def model(X, w, w2, w3, w4, w_o, p_keep_conv, p_keep_hidden):
l1a = tf.nn.relu(tf.nn.conv2d(X, w,
strides=[1, 1, 1, 1], padding='SAME'))
l1 = tf.nn.max_pool(l1a, ksize=[1, 2, 2, 1], # l1 shape=(?, 14, 14, 32)
strides=[1, 2, 2, 1], padding='SAME')
l1 = tf.nn.dropout(l1, p_keep_conv)

# 第二组卷积层及池化层,最后dropout一些神经元
l2a = tf.nn.relu(tf.nn.conv2d(l1, w2, # l2a shape=(?, 14, 14, 64)
strides=[1, 1, 1, 1], padding='SAME'))
l2 = tf.nn.max_pool(l2a, ksize=[1, 2, 2, 1], # l2 shape=(?, 7, 7, 64)
strides=[1, 2, 2, 1], padding='SAME')
l2 = tf.nn.dropout(l2, p_keep_conv)

# 第三组卷积神经网络及池化层,同样,最后dropout一些神经元
l3a = tf.nn.relu(tf.nn.conv2d(l2, w3, # l3a shape=(?, 7, 7, 128)
strides=[1, 1, 1, 1], padding='SAME'))
l3 = tf.nn.max_pool(l3a, ksize=[1, 2, 2, 1], # l3 shape=(?, 4, 4, 128)
strides=[1, 2, 2, 1], padding='SAME')
l3 = tf.reshape(l3, [-1, w4.get_shape().as_list()[0]]) # reshape to (?, 2048)
l3 = tf.nn.dropout(l3, p_keep_conv)
# 全连接层
l4 = tf.nn.relu(tf.matmul(l3, w4))
l4 = tf.nn.dropout(l4, p_keep_hidden)
# 输出层
pyx = tf.matmul(l4, w_o)
return pyx


# 导入数据
mnist = input_data.read_data_sets("MNIST_data/", one_hot=True)
# 定义四个变量,分别为输入训练图像矩阵及其标签,输入测试图像矩阵及其标签
trX, trY, teX, teY = mnist.train.images, mnist.train.labels, mnist.test.images, mnist.test.labels
# -1表示布考虑输入图片的数量,28*28为图片的像素数,1是通道(channel)的数量,
# 因MNIST图片为黑白,彩色图片通道是3
trX = trX.reshape(-1, 28, 28, 1) # 28x28x1
teX = teX.reshape(-1, 28, 28, 1) # 28x28x1

X = tf.placeholder("float", [None, 28, 28, 1])
Y = tf.placeholder("float", [None, 10]) # 10为识别图片的类别从0到9,共10个取值

# 定义模型函数
# 神经网络模型的构建函数,传入以下参数
# X:输入数据
# w: 每一层权重
w = init_weights([3, 3, 1, 32]) # 大小为3*3,输入的维度为1 ,输出维度为32
w2 = init_weights([3, 3, 32, 64]) # 大小为3*3,输入维度为32,输出维度为64
w3 = init_weights([3, 3, 64, 128]) # 大小为3*3,输入维度为64,输出维度为128
w4 = init_weights([128 * 4 * 4, 625]) # 全连接层,输入维度为128*4*4,也就是上一层的输出,输出维度为625
w_o = init_weights([625, 10]) # 输出层,输入的维度为625, 输出110维,代表10类(labels)

# p_keep_conv,p_keep_hidden:dropout 保留神经元比例
# 定义dropout的占位符keep_conv,表示一层中有多少比例的神经元被保留,生成网络模型,得到预测数据
# 在训练的时候把设定比例的节点改为0,避免过拟合
p_keep_conv = tf.placeholder("float")
p_keep_hidden = tf.placeholder("float")
py_x = model(X, w, w2, w3, w4, w_o, p_keep_conv, p_keep_hidden)

# 定义损失函数,采用tf.nn.softmax_cross_entropy_with_logists,作为比较预测值和真实值的差距
# 定义训练操作(train_op) 采用RMSProp算法作为优化器,
cost = tf.reduce_mean(tf.nn.softmax_cross_entropy_with_logits(logits=py_x, labels=Y))
train_op = tf.train.RMSPropOptimizer(0.001, 0.9).minimize(cost)
predict_op = tf.argmax(py_x, 1)


#9.4.3训练模型和评估模型

#在会话中定义图,开始训练和评估
# Launch the graph in a session
with tf.Session() as sess:
# you need to initialize all variabels
tf.global_variables_initializer().run()

for i in range(100):
training_batch=zip(range(0,len(trX),batch_size),
range(batch_size,len(trX)+1,batch_size))
for start, end in training_batch:
sess.run(train_op, feed_dict={X: trX[start:end], Y: trY[start:end],
p_keep_conv: 0.8, p_keep_hidden: 0.5})

test_indices = np.arange(len(teX)) # Get A Test Batch
np.random.shuffle(test_indices)
test_indices = test_indices[0:test_size]

print(i, np.mean(np.argmax(teY[test_indices], axis=1) ==
sess.run(predict_op, feed_dict={X: teX[test_indices],
p_keep_conv: 1.0,
p_keep_hidden: 1.0})))
#预测的时候设置为1 即对全部样本进行迭代训练
1
2
3
# 贡献者:{沙舟}
# 数据集下载地址:https://github.com/nlintz/TensorFlow-Tutorials/blob/master/05_convolutional_net.py
# 代码来自于《TensorFlow技术解析与实战》