学习TensorFlow CNN手写数字识别

数据集：MNIST，0~9的28×28像素的灰度手写数字图片（训练集和测试集，图片和标签共4个压缩文件）

参见：https://blog.csdn.net/Lee_01/article/details/90294292

代码：

import tensorflow as tf
import tensorflow.examples.tutorials.mnist.input_data as input_data

# 使用tensorflow下载MNIST数据集
mnist = input_data.read_data_sets('MNIST_data', one_hot=True)

# 获取训练数据
train_data = mnist.train
# 获取验证数据
validation_data = mnist.validation
# 获取测试数据
test_data = mnist.test

x = tf.placeholder("float", shape=[None, 784])
y_ = tf.placeholder("float", shape=[None, 10])


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)


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')


# 卷积核5x5，32个卷积（提取原图的32个特征）
W_conv1 = weight_variable([5, 5, 1, 32])
b_conv1 = bias_variable([32])

x_image = tf.reshape(x, [-1, 28, 28, 1])

# 使用relu激活函数，保证结果输出的非线性化
# 修正线性单元会计算输入的线性函数，如果值是正数，则输出其值，如果是负数，则返回0
h_conv1 = tf.nn.relu(conv2d(x_image, W_conv1) + b_conv1)
# 池化卷积核2x2，取最大值
h_pool1 = max_pool_2x2(h_conv1)

# 卷积核5x5，64个卷积
W_conv2 = weight_variable([5, 5, 32, 64])
b_conv2 = bias_variable([64])

# 使用relu激活函数
h_conv2 = tf.nn.relu(conv2d(h_pool1, W_conv2) + b_conv2)
# 池化卷积核2x2，取最大值
h_pool2 = max_pool_2x2(h_conv2)

# 将7x7x64的输入通过全连接层，将输出转换成1024维的向量
W_fc1 = weight_variable([7 * 7 * 64, 1024])
b_fc1 = bias_variable([1024])

# 将1024维的输入向量转换成一个10维的输出，每个维度值代表属于某一个数字的概率
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)

keep_prob = tf.placeholder("float")
h_fc1_drop = tf.nn.dropout(h_fc1, keep_prob)

W_fc2 = weight_variable([1024, 10])
b_fc2 = bias_variable([10])

# softmax层
y_conv = tf.nn.softmax(tf.matmul(h_fc1_drop, W_fc2) + b_fc2)

cross_entropy = -tf.reduce_sum(y_ * tf.log(y_conv))
# 使用adam优化器来以0.0001的学习率来进行微调
train_step = tf.train.AdamOptimizer(1e-4).minimize(cross_entropy)
# 判断预测与实际的标签是否一致
correct_prediction = tf.equal(tf.argmax(y_conv, 1), tf.argmax(y_, 1))
accuracy = tf.reduce_mean(tf.cast(correct_prediction, "float"))

# saver = tf.train.Saver() # 定义saver

with tf.Session() as sess:
    sess.run(tf.global_variables_initializer())
    # 循环训练20000次
    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})

    # saver.save(sess, './model.ckpt')  # 模型储存位置
    print("test accuracy %g" % accuracy.eval(
        feed_dict={x: mnist.test.images, y_: mnist.test.labels, keep_prob: 1.0}))

运行结果：准确度能够达到99%

参考文章：

https://blog.csdn.net/red_stone1/article/details/78651063

https://blog.csdn.net/greedystar/article/details/80444519

https://blog.csdn.net/wills798/article/details/80890990