MNIST手写数据集简单版本

最新推荐文章于 2024-08-02 12:58:05 发布

沐风的程序人生

最新推荐文章于 2024-08-02 12:58:05 发布

阅读量788

点赞数 1

分类专栏：深度学习TensorFlow 文章标签： MNIST手写数字分类

本文链接：https://blog.csdn.net/any1234567890/article/details/92658590

版权

深度学习TensorFlow 专栏收录该内容

4 篇文章 0 订阅

订阅专栏

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

# 载入数据集
mnist = input_data.read_data_sets("MNIST_data",one_hot=True)  #图片的路径需要与运行文件的路径在同一级目录，或者也可以使用绝对路径

# 定义一个批次
batch_size = 100
n_batch = mnist.train.num_examples//batch_size  # 这里//表示整除

# 定义两个占位符placeholder
x = tf.placeholder(tf.float32, [None, 784])  # 784=28*28 表示手写数据图片的像素个数
y = tf.placeholder(tf.float32, [None, 10])  # 10表示0~9,10分类

# 创建一个简单的神经网络
w = tf.Variable(tf.zeros([784, 10]))  # 10个神经元
b = tf.Variable(tf.zeros([10]))  # 偏置
prediction = tf.nn.softmax(tf.matmul(x, w)+b) 

# 二次代价函数
loss = tf.reduce_mean(tf.square(y - prediction))
# 使用梯度下降法
train_step = tf.train.GradientDescentOptimizer(0.2).minimize(loss)

# 初始化变量
init = tf.global_variables_initializer()

# 结果存放在一个布尔类型列表中
correct_prediction = tf.equal(tf.argmax(y, 1), tf.argmax(prediction, 1))  # argmax()返回一维张量中最大值所在的位置
# 求准确率
accuracy = tf.reduce_mean(tf.cast(correct_prediction, tf.float32))  # cast()函数：类型转换

with tf.Session() as sess:
    sess.run(init)
    for epoch in range(21):
        for batch in range(n_batch):
            batch_xs, batch_ys = mnist.train.next_batch(batch_size)
            sess.run(train_step, feed_dict={x:batch_xs, y:batch_ys})
            
        acc = sess.run(accuracy, feed_dict={x:mnist.test.images, y:mnist.test.labels})
        print("Inter:"+str(epoch)+", Testing accuracy:"+str(acc))

运行结果如下：

Extracting MNIST_data\train-images-idx3-ubyte.gz
Extracting MNIST_data\train-labels-idx1-ubyte.gz
Extracting MNIST_data\t10k-images-idx3-ubyte.gz
Extracting MNIST_data\t10k-labels-idx1-ubyte.gz
Inter:0, Testing accuracy:0.8309
Inter:1, Testing accuracy:0.8712
Inter:2, Testing accuracy:0.8815
Inter:3, Testing accuracy:0.8883
Inter:4, Testing accuracy:0.8949
Inter:5, Testing accuracy:0.8974
Inter:6, Testing accuracy:0.9002
Inter:7, Testing accuracy:0.9009
Inter:8, Testing accuracy:0.9032
Inter:9, Testing accuracy:0.9046
Inter:10, Testing accuracy:0.9062
Inter:11, Testing accuracy:0.9071
Inter:12, Testing accuracy:0.9075
Inter:13, Testing accuracy:0.9092
Inter:14, Testing accuracy:0.9101
Inter:15, Testing accuracy:0.9114
Inter:16, Testing accuracy:0.9113
Inter:17, Testing accuracy:0.9121
Inter:18, Testing accuracy:0.9136
Inter:19, Testing accuracy:0.9133
Inter:20, Testing accuracy:0.914