卷积神经网络CNN用于MNIST数据分类
import tensorflow as tf
from tensorflow.examples.tutorials.mnist import input_data #手写数字相关的数据包
# 载入数据集
mnist = input_data.read_data_sets("MNIST_data",one_hot=True) #载入数据,{数据集包路径,把标签转化为只有0和1的形式}
#定义变量,即每个批次的大小
batch_size = 100 #一次放100章图片进去
n_batch = mnist.train.num_examples // batch_size #计算一共有多少个批次;训练集数量(整除)一个批次大小
#初始化权值
def weight_variable(shape):
initial = tf.truncated_normal(shape,stddev=0.1) #生成一个截断的正态分布
return tf.Variable(initial)
#初始化偏置
def bias_variable(shape):
initial = tf.constant(0.1,shape=shape)
return tf.Variable(initial)
#卷积层
def conv2d(x,W):
#x input tensor of shape '[batch, in_height, in_width, in_channels]'
#W filter / kernel tensor of shape [filter_height, filter_width, in_channels, out_channels]
#'strides[0] = strides[3] = 1', strides[1]代表x方向的步长,strides[2]代表y方向的步长
#padding:A 'string' frome: '"SAME"(补0), "VALID"(不补0)'
return tf.nn.conv2d(x,W,strides=[1,1,1,1],padding='SAME')
#池化层
def max_pool_2x2(x):
#ksize [1,x,y,1](窗口大小)
return tf.nn.max_pool(x,ksize=[1,2,2,1],strides=[1,2,2,1],padding='SAME')
#定义两个placeholder
x = tf.placeholder(tf.float32,[None,784]) #[行不确定,列为784]:28*28
y = tf.placeholder(tf.float32,[None,10]) #数字为0-9,则为10
#改变x的格式转为4D的向量[batch, in_height, in_width, in_channels]
x_image = tf.reshape(x,[-1,28,28,1])
#初始化第一个卷积层的权值和偏置
W_conv1 = weight_variable([5,5,1,32]) #5*5的采样窗口,32个卷积核从1个平面抽取特征
b_conv1 = bias_variable([32]) #每个卷积核一个偏置
#把x_image和权值向量进行卷积,再加上偏置值,然后应用于relu激活函数
h_conv1 = tf.nn.relu(conv2d(x_image,W_conv1) + b_conv1)
h_pool1 = max_pool_2x2(h_conv1) #进行max-pooling
#初始化第二个卷积层的权值和偏置
W_conv2 = weight_variable([5,5,32,64]) #5*5的采样窗口,32个卷积核从1个平面抽取特征
b_conv2 = bias_variable([64]) #每个卷积核一个偏置
#把h_pool1和权值向量进行卷积,再加上偏置值,然后应用于relu激活函数
h_conv2 = tf.nn.relu(conv2d(h_pool1,W_conv2) + b_conv2)
h_pool2 = max_pool_2x2(h_conv2) #进行max-pooling
#28*28的图片第一次卷积后还是28*28,第一次池化后变为14*14
#第二次卷积后为14*14,第二次池化后变为7*7
#经过上面的操作后得到64张7*7的平面
#初始化第一个全连接层的权值
W_fc1 = weight_variable([7*7*64,1024]) #上一层有7*7*64个神经元,全连接层有1024个神经元
b_fc1 = bias_variable([1024]) #1024个节点
#把池化层2的输出扁平化为1维
h_pool2_flat = tf.reshape(h_pool2,[-1,7*7*64])
#求第一个全连接层的输出
h_fc1 = tf.nn.relu(tf.matmul(h_pool2_flat,W_fc1) + b_fc1)
#keep_prob用来表示神经元的输出概率
keep_prob = tf.placeholder(tf.float32)
h_fc1_drop = tf.nn.dropout(h_fc1,keep_prob)
#初始化第二个全连接层
W_fc2 = weight_variable([1024,10])
b_fc2 = bias_variable([10])
#计算输出
prediction = tf.nn.softmax(tf.matmul(h_fc1_drop,W_fc2) + b_fc2)
#定义交叉熵代价函数
cross_entropy = tf.reduce_mean(tf.nn.softmax_cross_entropy_with_logits(labels=y,logits=prediction))
#使用AdamOptimizer进行优化
train_step = tf.train.AdamOptimizer(1e-4).minimize(cross_entropy)
#准确数,结果存放在一个布尔型列表中
correct_prediction = tf.equal(tf.argmax(prediction,1),tf.argmax(y,1)) #比较两个参数大小是否相同,同则返回为true,不同则返回为false;argmax():返回张量中最大的值所在的位置
#求准确率
accuracy = tf.reduce_mean(tf.cast(correct_prediction,tf.float32)) #cast():将布尔型转换为32位的浮点型;(比方说9个T和1个F,则为9个1,1个0,即准确率为90%)
with tf.Session() as sess:
sess.run(tf.global_variables_initializer())
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,keep_prob:0.7})
acc = sess.run(accuracy,feed_dict={x:mnist.test.images,y:mnist.test.labels,keep_prob:1.0})
print("Iter" + str(epoch) + ",Testing Accuracy" + str(acc))
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
Iter0,Testing Accuracy0.863
Iter1,Testing Accuracy0.8757
Iter2,Testing Accuracy0.881
Iter3,Testing Accuracy0.8833
点赞,关注,收藏👍,➕,👀点赞,关注,收藏👍,➕,👀点赞,关注,收藏👍,➕,👀
😘😘😘😘😘😘😘😘😘😘😘😘😘😘😘😘😘😘😘😘😘😘😘😘😘😘😘😘😘😘😘😘
💪💪💪💪💪💪💪💪💪💪💪💪💪💪💪💪💪💪💪💪💪💪💪💪💪💪💪💪💪💪💪💪
欢迎大噶关注我的微信公众号