手写数字识别——Softmax Regression

    今天对照Tensorflow的书，实现了一个简单的手写数字识别的算法Softmax Regression。

    其中的mnist数据集，下载自http://yann.lecun.com/exdb/mnist/

   

from tensorflow.examples.tutorials.mnist import input_data
mnist = input_data.read_data_sets("MNIST_data/", one_hot=True)

print(mnist.train.images.shape, mnist.train.labels.shape)
print(mnist.test.images.shape, mnist.test.labels.shape)
print(mnist.validation.images.shape, mnist.validation.labels.shape)

import tensorflow as tf

#设置session
sess = tf.InteractiveSession()
x = tf.placeholder(tf.float32, [None, 784])

#创建Softmax Regression模型中的weights和biases
W = tf.Variable(tf.zeros([784, 10]))
b = tf.Variable(tf.zeros([10]))

#Softmax Regression的公式
y = tf.nn.softmax(tf.matmul(x, W) + b)

#损失函数采用cross-entropy
y_ = tf.placeholder(tf.float32, [None, 10])
cross_entropy = tf.reduce_mean(-tf.reduce_sum(y_ * tf.log(y),
                                              reduction_indices=[1]))
#随机梯度下降的步长
train_step = tf.train.GradientDescentOptimizer(0.5).minimize(cross_entropy)

tf.global_variables_initializer().run()

for i in range(1000):
    batch_xs, batch_ys = mnist.train.next_batch(100)
    train_step.run({x: batch_xs, y_: batch_ys})

#判断是否准确
correct_prediction = tf.equal(tf.argmax(y, 1), tf.argmax(y_, 1))
accuracy = tf.reduce_mean(tf.cast(correct_prediction, tf.float32))

#输入测试数据并进行评估
print(accuracy.eval({x: mnist.test.images, y_: mnist.test.labels}))

准确率达到91.7%