TensorFlow实现单隐层神经网络

这里使用MNIST数据集，MNIST数据集的下载地址http://yann.lecun.com/exdb/mnist/

from tensorflow.examples.tutorials.mnist import input_data
import tensorflow as tf
mnist = input_data.read_data_sets(r"D:\PycharmProjects\tensorflow\MNIST_data", one_hot=True)

# 创建默认的InteractiveSession，这样后面执行的各项操作就无需指定Session了
sess = tf.InteractiveSession()

in_units = 784 # 输入节点数
h1_units = 300 # 隐含单元数
W1 = tf.Variable(tf.truncated_normal([in_units, h1_units], stddev=0.1))
b1 = tf.Variable(tf.zeros([h1_units]))
W2 = tf.Variable(tf.zeros([h1_units, 10])) # 因为是识别数字，输出单元数为10
b2 = tf.Variable(tf.zeros([10]))

x = tf.placeholder(tf.float32, [None, in_units])
keep_prob = tf.placeholder(tf.float32)

hidden1 = tf.nn.relu(tf.matmul(x, W1) + b1) # 使用relu激活函数可以解决梯度弥散
hidden1_drop = tf.nn.dropout(hidden1, keep_prob) # 使用Dropout方法，keep_prob是保留节点的概率
y = tf.nn.softmax(tf.matmul(hidden1_drop, W2) + b2) # 预测的标签

y_ = tf.placeholder(tf.float32, [None, 10]) # 正确的标签
cross_entropy = tf.reduce_mean(-tf.reduce_sum(y_ * tf.log(y), reduction_indices=[1])) # 损失函数
train_step = tf.train.AdagradOptimizer(0.3).minimize(cross_entropy) # 学习率是0.3

# 训练
tf.global_variables_initializer().run()
# 一共采用3000个batch，每个batch100个样本，一共300000个样本
# 一个数据集55000个样本，相当于对全数据集进行5轮（epoch）迭代
for i in range(3000):
  batch_xs, batch_ys = mnist.train.next_batch(100)
  train_step.run({x: batch_xs, y_: batch_ys, keep_prob: 0.75})

# 测试
correct_prediction = tf.equal(tf.argmax(y, 1), tf.argmax(y_, 1))
accuracy = tf.reduce_mean(tf.cast(correct_prediction, tf.float32))
# 预测时keep_prob应该等于1，即使用全部特征来预测样本的类别
print(accuracy.eval({x: mnist.test.images, y_: mnist.test.labels, keep_prob: 1.0}))