TensotFlow 应用实例：11-使用CNN预测手写数字MNIST
本文是我在学习TensotFlow 的时候所记录的笔记，共享出来希望能够帮助一些需要的人。
什么是卷积神经网络 CNN (深度学习)?
What is Convolutional Neural Networks (deep learning)?
卷积神经网络 最常应用于 图片识别
卷积是说神经网络不在对每一个点的数据进行处理，而是对一个区域进行处理
import tensorflow as tf
from tensorflow.examples.tutorials.mnist import input_data
# 什么是卷积神经网络 CNN (深度学习)?
# What is Convolutional Neural Networks (deep learning)?
# 卷积神经网络 最常应用于 图片识别
# 卷积 神经网络
# 卷积是说神经网络不在对每一个点的数据进行处理，
# 而是对一个区域进行处理
# Google 自己的 CNN 教程
# https://classroom.udacity.com/courses/ud730/lessons/6377263405/concepts/63796332430923
# number 1 to 10 image data
# 如果本地没有相应的数据包，会先下载，然后解压数据包
# MNIST_data 是下载数据要保存的位置
mnist = input_data.read_data_sets('MNIST_data', one_hot=True)
# 添加神经层
def add_layer(inputs, in_size, out_size, activation_function=None):
# Weights define
# 权重，尽量要是一个随机变量
# 随机变量在生成初始变量的时候比全部为零效果要好的很多
Weights = tf.Variable(tf.random_normal([in_size, out_size]))
# biases define
# 偏值项，是一个列表，不是矩阵，默认设置为0 + 0.1
biases = tf.Variable(tf.zeros([1, out_size]) + 0.1)
# W * x + b
Wx_plus_b = tf.matmul(inputs, Weights) + biases
# 如果activation_function是空的时候就表示是一个线性关系直接放回即可
if activation_function is None:
outputs = Wx_plus_b
else:
outputs = activation_function(Wx_plus_b)
return outputs
# 计算精确度
# compute_accuracy 要使用
def compute_accuracy(v_xs, v_ys):
global prediction
y_pre = sess.run(prediction, feed_dict={xs: v_xs, keep_prob: 1})
correct_prediction = tf.equal(tf.argmax(y_pre, 1), tf.argmax(v_ys, 1))
accuracy = tf.reduce_mean(tf.cast(correct_prediction, tf.float32))
# result 是一个百分比，百分比越高证明越准确
result = sess.run(accuracy, feed_dict={xs: v_xs, ys: v_ys, keep_prob: 1})
return result
def weight_variable(shape):
# normal 产生随机变量
# stddev: A 0-D Tensor or Python value of type `dtype`. The standard deviation
# of the truncated normal distribution.
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)
# x input W is weight
def conv2d(x, W):
# strides [1, x_movement, y_movement, 1]
# 前后都要为1
# VALID SAME padding方式
# VALID 较小， SAME 和原图一样
return tf.nn.conv2d(x, W, strides=[1, 1, 1, 1], padding='SAME')
def max_pool_2x2(x):
result = tf.nn.max_pool(x,
ksize=[1, 2, 2, 1],
strides=[1, 2, 2, 1],
padding='SAME')
return result
# 定义 placeholder
xs = tf.placeholder(tf.float32, [None, 784])/255.
ys = tf.placeholder(tf.float32, [None, 10])
keep_prob = tf.placeholder(tf.float32)
# 将输入的xs转换为图片的形式
# -1 不管维度
# 28*28 像素点
# 1 channel 是黑白
x_image = tf.reshape(xs, [-1, 28, 28, 1])
# print(x_image.shape) # [n_samples, 28, 28, 1]
# conv1 layer
# 5 * 5 patch ，长*宽
# in size is 1， image的厚度，输入的厚度
# out is 32， 输出的深度，厚度
W_conv1 = weight_variable([5, 5, 1, 32])
# 32个输出，所有b为32
b_conv1 = bias_variable([32])
# conv2d output size 28x28X32
h_conv1 = tf.nn.relu(conv2d(x_image, W_conv1) + b_conv1)
# max_pool_2x2 output 14x14x32
h_pool1 = max_pool_2x2(h_conv1)
# conv2 layer
# out is 64， 输出的深度，厚度
W_conv2 = weight_variable([5, 5, 32, 64])
# 32个输出，所有b为64
b_conv2 = bias_variable([64])
# conv2d output size 14x14x64
h_conv2 = tf.nn.relu(conv2d(h_pool1, W_conv2) + b_conv2)
# max_pool_2x2 output 7x7x64
h_pool2 = max_pool_2x2(h_conv2)
# func1 layer
W_fc1 = weight_variable([7*7*64, 1024])
b_fc1 = bias_variable([1024])
# [n_samples, 7, 7, 64] >> [n_samples, 7*7*64]
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)
# drop out
h_fc1_drop = tf.nn.dropout(h_fc1, keep_prob)
# func2 layer
W_fc2 = weight_variable([1024, 10])
b_fc2 = bias_variable([10])
prediction = tf.nn.softmax(tf.matmul(h_fc1_drop, W_fc2) + b_fc2)
# prediction = add_layer(xs, 784, 10, activation_function=tf.nn.softmax)
# cross_entropy 分类的时候经常使用softmax + cross_entropy来计算的
cross_entropy = tf.reduce_mean(-tf.reduce_sum(ys * tf.log(prediction),
reduction_indices=[1]))
# train_step = tf.train.GradientDescentOptimizer(0.5).minimize(cross_entropy)
# AdamOptimizer 需要的学习速率应该更小
train_step = tf.train.AdamOptimizer(1e-4).minimize(cross_entropy)
sess = tf.Session()
# important step
# tf.initialize_all_variables() no long valid from
# "2017-03-02", "Use `tf.global_variables_initializer` instead."
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, keep_prob: 0.5})
if i % 50 == 0:
print(compute_accuracy(mnist.test.images, mnist.test.labels))
本文代码GitHub地址 tensorflow_learning_notes
