-
大小: 11.06MB文件类型: .zip金币: 1下载: 0 次发布日期: 2023-07-06
- 语言: Python
- 标签: TensorFlow python
资源简介
python语言编写,利用TensorFlow建立两层卷积神经网络,数据集为手写体识别数据集MNIST,识别准确率99%
代码片段和文件信息
import tensorflow as tf
from tensorflow.examples.tutorials.mnist import input_data
mnist = input_data.read_data_sets(‘MNIST_data/‘ one_hot=True) #读取数据
#x为训练图像占位符,y为图片标签
x = tf.placeholder(tf.float32 [None 784])
y_label = tf.placeholder(tf.float32 [None 10])
#将单张图片从784维向量重新还原成28*28的矩阵图片-1表示形状第一维的大小是由x自动确定的
x_image = tf.reshape(x [-1 28 28 1])
#第一层卷积
def weight_variable(shape):
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)
def conv2d(x W):
return tf.nn.conv2d(x W strides=[1 1 1 1]padding=‘SAME‘)
def max_pool_2X2(x):
return tf.nn.max_pool(x ksize=[1 2 2 1] strides=[1 2 2 1]
padding=‘SAME‘)
W_conv1 = weight_variable([5 5 1 32])
b_conv1 = bias_variable([32])
h_conv1 = tf.nn.relu(conv2d(x_image W_conv1) + b_conv1)
h_pool1 = max_pool_2X2(h_conv1)
#第二层卷积
W_conv2 = weight_variable([5 5 32 64])
b_conv2 = bias_variable([64])
h_conv2 = tf.nn.relu(conv2d(h_pool1 W_conv2) + b_conv2)
h_pool2 = max_pool_2X2(h_conv2)
#全连接层
W_fc1 = weight_variable([7*7*64 1024])
b_fc1 = bias_variable([1024])
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)
#使用dropoutkeep_prop是一个占位符,训练时为0.5,测试时为1
keep_prop = tf.placeholder(tf.float32)
h_fc1_drop = tf.nn.dropout(h_fc1 keep_prop)
W_fc2 = weight_variable([1024 10])
b_fc2 = bias_variable([10])
y_conv = tf.matmul(h_fc1 W_fc2) + b_fc2
#不采用softmax计算交叉熵的方法
#采用
cross_entropy = tf.reduce_mean(
tf.nn.softmax_cross_entropy_with_logits(labels=y_label logits=y_conv))
#定义train_step
train_step = tf.train.AdamOptimizer(1e-4).minimize(cross_entropy)
#定义测试的准确率
correct_prediction = tf.equal(tf.argmax(y_conv 1) tf.argmax(y_label 1))
accuracy = tf.reduce_mean(tf.cast(correct_prediction tf.float32))
#创建session对变量初始化
sess = tf.InteractiveSession()
sess.run(tf.global_variables_initializer())
#训练2000步
for i in range(1500):
batch = mnist.train.next_batch(50)
#每100步报告一次准确率
if i%100 == 0:
train_accuracy = accuracy.eval(feed_dict={
x: batch[0] y_label: batch[1] keep_prop: 1.0})
print(‘step %d training accuracy %.2f‘%(i train_accuracy))
train_step.run(feed_dict={
x: batch[0] y_label: batch[1] keep_prop: 0.2})
#print(‘test accuracy %.3f‘%accuracy.eval(feed_dict={
# x: mnist.test.images y_label: mnist.test.labels keep_prop: 1.0}))
print(‘test accuracy %.2f‘%accuracy.eval(feed_dict={
x: mnist.test.images[0: 2000] y_label: mnist.test.labels[0: 2000] keep_prop: 1.0}))
属性 大小 日期 时间 名称
----------- --------- ---------- ----- ----
目录 0 2018-11-06 21:31 TF_MNIST_CONV\
目录 0 2019-01-24 21:11 TF_MNIST_CONV\.idea\
文件 576 2018-11-05 12:53 TF_MNIST_CONV\.idea\TF_MNIST_CONV.iml
目录 0 2018-10-31 19:57 TF_MNIST_CONV\.idea\libraries\
文件 128 2018-10-31 19:57 TF_MNIST_CONV\.idea\libraries\R_User_Library.xm
文件 185 2018-11-05 12:53 TF_MNIST_CONV\.idea\misc.xm
文件 278 2018-10-31 19:46 TF_MNIST_CONV\.idea\modules.xm
文件 22661 2019-01-24 21:11 TF_MNIST_CONV\.idea\workspace.xm
目录 0 2018-11-09 20:12 TF_MNIST_CONV\MNIST_data\
文件 1648877 2018-10-30 18:24 TF_MNIST_CONV\MNIST_data\t10k-images-idx3-ubyte.gz
文件 4542 2018-10-30 18:24 TF_MNIST_CONV\MNIST_data\t10k-labels-idx1-ubyte.gz
文件 165 2018-11-09 20:12 TF_MNIST_CONV\MNIST_data\tensor.py
文件 9912422 2018-10-30 18:24 TF_MNIST_CONV\MNIST_data\train-images-idx3-ubyte.gz
文件 28881 2018-10-30 18:24 TF_MNIST_CONV\MNIST_data\train-labels-idx1-ubyte.gz
文件 2998 2018-11-06 21:31 TF_MNIST_CONV\conv.py
评论
共有 条评论