tensorflow weight_variable going

 1 # coding: utf-8
 2 import tensorflow as tf
 3 from tensorflow.examples.tutorials.mnist import input_data
 4 
 5 
 6 def weight_variable(shape):
 7     initial = tf.truncated_normal(shape, stddev=0.1)
 8     return tf.Variable(initial)
 9 
10 
11 def bias_variable(shape):
12     initial = tf.constant(0.1, shape=shape)
13     return tf.Variable(initial)
14 
15 
16 def conv2d(x, W):
17     return tf.nn.conv2d(x, W, strides=[1, 1, 1, 1], padding='SAME')
18 
19 
20 def max_pool_2x2(x):
21     return tf.nn.max_pool(x, ksize=[1, 2, 2, 1],
22                           strides=[1, 2, 2, 1], padding='SAME')
23 
24 
25 if __name__ == '__main__':
26     # 读入数据
27     mnist = input_data.read_data_sets("MNIST_data/", one_hot=True)
28     # x为训练图像的占位符、y_为训练图像标签的占位符
29     x = tf.placeholder(tf.float32, [None, 784])
30     y_ = tf.placeholder(tf.float32, [None, 10])
31 
32     # 将单张图片从784维向量重新还原为28x28的矩阵图片
33     x_image = tf.reshape(x, [-1, 28, 28, 1])
34 
35     # 第一层卷积层
36     W_conv1 = weight_variable([5, 5, 1, 32])
37     b_conv1 = bias_variable([32])
38     h_conv1 = tf.nn.relu(conv2d(x_image, W_conv1) + b_conv1)
39     h_pool1 = max_pool_2x2(h_conv1)
40 
41     # 第二层卷积层
42     W_conv2 = weight_variable([5, 5, 32, 64])
43     b_conv2 = bias_variable([64])
44     h_conv2 = tf.nn.relu(conv2d(h_pool1, W_conv2) + b_conv2)
45     h_pool2 = max_pool_2x2(h_conv2)
46 
47     # 全连接层，输出为1024维的向量
48     W_fc1 = weight_variable([7 * 7 * 64, 1024])
49     b_fc1 = bias_variable([1024])
50     h_pool2_flat = tf.reshape(h_pool2, [-1, 7 * 7 * 64])
51     h_fc1 = tf.nn.relu(tf.matmul(h_pool2_flat, W_fc1) + b_fc1)
52     # 使用Dropout，keep_prob是一个占位符，训练时为0.5，测试时为1
53     keep_prob = tf.placeholder(tf.float32)
54     h_fc1_drop = tf.nn.dropout(h_fc1, keep_prob)
55 
56     # 把1024维的向量转换成10维，对应10个类别
57     W_fc2 = weight_variable([1024, 10])
58     b_fc2 = bias_variable([10])
59     y_conv = tf.matmul(h_fc1_drop, W_fc2) + b_fc2
60 
61     # 我们不采用先Softmax再计算交叉熵的方法，而是直接用tf.nn.softmax_cross_entropy_with_logits直接计算
62     cross_entropy = tf.reduce_mean(
63         tf.nn.softmax_cross_entropy_with_logits(labels=y_, logits=y_conv))
64     # 同样定义train_step
65     train_step = tf.train.AdamOptimizer(1e-4).minimize(cross_entropy)
66 
67     # 定义测试的准确率
68     correct_prediction = tf.equal(tf.argmax(y_conv, 1), tf.argmax(y_, 1))
69     accuracy = tf.reduce_mean(tf.cast(correct_prediction, tf.float32))
70 
71     sess = tf.InteractiveSession()
72     sess.run(tf.global_variables_initializer())
73 
74     for i in range(20000):
75         batch = mnist.train.next_batch(50)
76         # 每100步报告一次在验证集上的准确度
77         if i % 100 == 0:
78             train_accuracy = accuracy.eval(feed_dict={
79                 x: batch[0], y_: batch[1], keep_prob: 1.0})
80             print("step %d, training accuracy %g" % (i, train_accuracy))
81         train_step.run(feed_dict={x: batch[0], y_: batch[1], keep_prob: 0.5})
82 
83     # 训练结束后报告在测试集上的准确度
84     print("test accuracy %g" % accuracy.eval(feed_dict={
85         x: mnist.test.images, y_: mnist.test.labels, keep_prob: 1.0}))

#coding: utf-8
from tensorflow.examples.tutorials.mnist import input_data
import scipy.misc
import os

# 读取MNIST数据集。如果不存在会事先下载。
mnist = input_data.read_data_sets("MNIST_data/", one_hot=True)

# 我们把原始图片保存在MNIST_data/raw/文件夹下
# 如果没有这个文件夹会自动创建
save_dir = 'MNIST_data/raw/'
if os.path.exists(save_dir) is False:
os.makedirs(save_dir)

# 保存前20张图片
for i in range(20):
# 请注意，mnist.train.images[i, :]就表示第i张图片（序号从0开始）
image_array = mnist.train.images[i, :]
# TensorFlow中的MNIST图片是一个784维的向量，我们重新把它还原为28x28维的图像。
image_array = image_array.reshape(28, 28)
# 保存文件的格式为 mnist_train_0.jpg, mnist_train_1.jpg, ... ,mnist_train_19.jpg
filename = save_dir + 'mnist_train_%d.jpg' % i
# 将image_array保存为图片
# 先用scipy.misc.toimage转换为图像，再调用save直接保存。
scipy.misc.toimage(image_array, cmin=0.0, cmax=1.0).save(filename)

print('Please check: %s ' % save_dir)