使用TensorFlow搭建一个全连接神经网络教程

说明

本例子利用TensorFlow搭建一个全连接神经网络，实现对MNIST手写数字的识别。

先上代码

from tensorflow.examples.tutorials.mnist import input_dataimport tensorflow as tf# prepare datamnist = input_data.read_data_sets('MNIST_data', one_hot=True)xs = tf.placeholder(tf.float32, [None, 784])ys = tf.placeholder(tf.float32, [None, 10])# the model of the fully-connected networkweights = tf.Variable(tf.random_normal([784, 10]))biases = tf.Variable(tf.zeros([1, 10]) + 0.1)outputs = tf.matmul(xs, weights) + biasespredictions = tf.nn.softmax(outputs)cross_entropy = tf.reduce_mean(-tf.reduce_sum(ys * tf.log(predictions),            reduction_indices=[1]))train_step = tf.train.GradientDescentOptimizer(0.5).minimize(cross_entropy)# compute the accuracycorrect_predictions = tf.equal(tf.argmax(predictions, 1), tf.argmax(ys, 1))accuracy = tf.reduce_mean(tf.cast(correct_predictions, tf.float32))with tf.Session() as sess: init = tf.global_variables_initializer() sess.run(init) for i in range(1000):  batch_xs, batch_ys = mnist.train.next_batch(100)  sess.run(train_step, feed_dict={   xs: batch_xs,   ys: batch_ys  })  if i % 50 == 0:   print(sess.run(accuracy, feed_dict={    xs: mnist.test.images,    ys: mnist.test.labels   }))

代码解析

1. 读取MNIST数据

mnist = input_data.read_data_sets('MNIST_data', one_hot=True)

2. 建立占位符

xs = tf.placeholder(tf.float32, [None, 784])ys = tf.placeholder(tf.float32, [None, 10])

xs 代表图片像素数据, 每张图片(28×28)被展开成(1×784), 有多少图片还未定, 所以shape为None×784.

ys 代表图片标签数据, 0-9十个数字被表示成One-hot形式, 即只有对应bit为1, 其余为0.

3. 建立模型

weights = tf.Variable(tf.random_normal([784, 10]))biases = tf.Variable(tf.zeros([1, 10]) + 0.1)outputs = tf.matmul(xs, weights) + biasespredictions = tf.nn.softmax(outputs)cross_entropy = tf.reduce_mean(-tf.reduce_sum(ys * tf.log(predictions),            reduction_indices=[1]))train_step = tf.train.GradientDescentOptimizer(0.5).minimize(cross_entropy)

使用Softmax函数作为激活函数：

4. 计算正确率

correct_predictions = tf.equal(tf.argmax(predictions, 1), tf.argmax(ys, 1))accuracy = tf.reduce_mean(tf.cast(correct_predictions, tf.float32))

5. 使用模型

with tf.Session() as sess: init = tf.global_variables_initializer() sess.run(init) for i in range(1000):  batch_xs, batch_ys = mnist.train.next_batch(100)  sess.run(train_step, feed_dict={   xs: batch_xs,   ys: batch_ys  })  if i % 50 == 0:   print(sess.run(accuracy, feed_dict={    xs: mnist.test.images,    ys: mnist.test.labels   }))

运行结果

训练1000个循环, 准确率在87%左右.