取自于TensorFlow实战
'''
本文代码中的AlexNet不是完整的，书中也只给了卷积网络的构建，全连接层是博主自己加的，后面会分享完整的AlexNet
因为设备运算速度原因，只计算前向和反向的计算时间。
AlexNet使用了五个卷积层，各卷积层使用了relu、LRN，maxpool等trick
每层的卷积核大小和卷积步长需要经验和实践尝试
LRN层现在很少使用了，因为效果不是很明显，可以不加
maxpool层有ksize=[1,3,3,1],strides=[1,2,2,1],有重叠，可以增加特征的丰富性
'''
from datetime import datetime
import tensorflow as tf
import math
import time
#设置TensorFlow的CPU运算优先级
import os
os.environ['TF_CPP_MIN_LOG_LEVEL']='2'
#设置训练参数
batch_size=128
num_batches=100
#节点信息打印
def print_activation(t):
print(t.op.name,' ',t.get_shape().as_list())
#网络构建
def inference(images):
#参数信息列表
parameters=[]
#第一卷积层
with tf.name_scope('conv1') as scope:
#卷积核[11,11,3,64]，卷积步长[1,4,4,1]
kernel=tf.Variable(tf.truncated_normal([11,11,3,64],dtype=tf.float32,stddev=1e-1),name='weights')
conv=tf.nn.conv2d(images,kernel,[1,4,4,1],padding='SAME')
#加偏置
biases=tf.Variable(tf.constant(0,shape=[64],dtype=tf.float32),trainable=True,name='biases')
bias=tf.nn.bias_add(conv,biases)
#整流线性激活
conv1=tf.nn.relu(bias,name=scope)
#参数存入参数列表
parameters+=[kernel,biases]
#LRN标准化，池化，pool1节点信息显示
lrn1=tf.nn.lrn(conv1,4,bias=1,alpha=0.001/9,beta=0.75,name='lrn1')
pool1=tf.nn.max_pool(lrn1,ksize=[1,3,3,1],strides=[1,2,2,1],padding='VALID',name='pool1')
print_activation(pool1)
#第二卷积层，同上，卷积核[5, 5, 64, 192]，卷积步长[1,1,1,1,1]
with tf.name_scope('conv2') as scope:
kernel = tf.Variable(tf.truncated_normal([5, 5, 64, 192], dtype=tf.float32, stddev=1e-1), name='weights')
conv = tf.nn.conv2d(pool1, kernel, [1, 1, 1, 1], padding='SAME')
biases = tf.Variable(tf.constant(0, shape=[192], dtype=tf.float32), trainable=True, name='biases')
bias = tf.nn.bias_add(conv, biases)
conv2 = tf.nn.relu(bias, name=scope)
parameters += [kernel, biases]
print_activation(conv2)
lrn2 = tf.nn.lrn(conv2, 4, bias=1, alpha=0.001 / 9, beta=0.75, name='lrn2')
pool2 = tf.nn.max_pool(lrn2, ksize=[1, 3, 3, 1], strides=[1, 2, 2, 1], padding='VALID', name='pool2')
print_activation(pool2)
#第三卷积层卷积核[3, 3, 192, 384]，卷积步长[1, 1, 1, 1]，不做LRN和池化
with tf.name_scope('conv3') as scope:
kernel = tf.Variable(tf.truncated_normal([3, 3, 192, 384], dtype=tf.float32, stddev=1e-1), name='weights')
conv = tf.nn.conv2d(pool2, kernel, [1, 1, 1, 1], padding='SAME')
biases = tf.Variable(tf.constant(0, shape=[384], dtype=tf.float32), trainable=True, name='biases')
bias = tf.nn.bias_add(conv, biases)
conv3 = tf.nn.relu(bias, name=scope)
parameters += [kernel, biases]
print_activation(conv3)
# 第四卷积层，卷积核[3, 3, 384, 256]，卷积步长[1, 1, 1, 1]，不做LRN和池化
with tf.name_scope('conv4') as scope:
kernel = tf.Variable(tf.truncated_normal([3, 3, 384, 256], dtype=tf.float32, stddev=1e-1), name='weights')
conv = tf.nn.conv2d(conv3, kernel, [1, 1, 1, 1], padding='SAME')
biases = tf.Variable(tf.constant(0, shape=[256], dtype=tf.float32), trainable=True, name='biases')
bias = tf.nn.bias_add(conv, biases)
conv4 = tf.nn.relu(bias, name=scope)
parameters += [kernel, biases]
print_activation(conv4)
# 第五卷积层同上，卷积核[3, 3, 256, 256]，卷积步长[1, 1, 1, 1]，池化
with tf.name_scope('conv5') as scope:
kernel = tf.Variable(tf.truncated_normal([3, 3, 256, 256], dtype=tf.float32, stddev=1e-1), name='weights')
conv = tf.nn.conv2d(conv4, kernel, [1, 1, 1, 1], padding='SAME')
biases = tf.Variable(tf.constant(0, shape=[256], dtype=tf.float32), trainable=True, name='biases')
bias = tf.nn.bias_add(conv, biases)
conv5 = tf.nn.relu(bias, name=scope)
parameters += [kernel, biases]
print_activation(conv5)
pool5 = tf.nn.max_pool(conv5, ksize=[1, 3, 3, 1], strides=[1, 2, 2, 1], padding='VALID', name='pool5')
print_activation(pool5)
# 第一全连接层，一维化、线性、激活
with tf.name_scope('fc1') as scope:
reshape = tf.reshape(pool5, [batch_size, -1],name='reshape')
dim = reshape.get_shape()[1].value
w1 = tf.Variable(tf.truncated_normal([dim,4096], dtype=tf.float32, stddev=1e-1), name='weights')
b1 = tf.Variable(tf.constant(0.1, shape=[4096],dtype=tf.float32), trainable=True, name='biases')
fc1 = tf.nn.relu(tf.matmul(reshape, w1) + b1,name=scope)
parameters += [w1, b1]
print_activation(fc1)
# 第二全连接层，线性、激活
with tf.name_scope('fc2') as scope:
w2 = tf.Variable(tf.truncated_normal([4096,4096], dtype=tf.float32, stddev=1e-1), name='weights')
b2 = tf.Variable(tf.constant(0.1, shape=[4096],dtype=tf.float32), trainable=True, name='biases')
fc2 = tf.nn.relu(tf.matmul(fc1, w2) + b2,name=scope)
parameters += [w2, b2]
print_activation(fc2)
# 第三全连接层、线性，激活函数为softmax
with tf.name_scope('fc3') as scope:
w3 = tf.Variable(tf.truncated_normal([4096,1000], dtype=tf.float32, stddev=1e-1), name='weights')
b3 = tf.Variable(tf.constant(0.1, shape=[1000],dtype=tf.float32), trainable=True, name='biases')
fc3 = tf.nn.softmax(tf.matmul(fc2, w3) + b3,name=scope)
parameters += [w3, b3]
print_activation(fc3)
#返回结果和参数列表
return fc3,parameters
#预计算节点计算耗时函数
def time_tensorflow_run(session,target,info_string):
num_steps_burn_in=10
total_duration=0
total_duration_squared=0
for i in range(num_batches+num_steps_burn_in):
start_time=time.time()
_=session.run(target)
duration=time.time()-start_time
if i>num_steps_burn_in:
if not i%10:
print('{:s}:step{:d},duration={:.3f}'.format(datetime.now(),i-num_steps_burn_in,duration))
total_duration+=duration
total_duration_squared+=duration*duration
mn=total_duration/num_batches
vr=total_duration_squared/num_batches-mn*mn
sd=math.sqrt(vr)
print('{:s}:{:s} across {:d} steps,{:.3f}+/-{:.3f} sec/batch'.format(datetime.now(),info_string,num_batches,mn,sd))
#计算耗时
def run_benckmark():
with tf.Graph().as_default():
#加载训练数据和网络
image_size=224
images=tf.Variable(tf.random_normal([batch_size,image_size,image_size,3],dtype=tf.float32,stddev=0.1))
fc3,parameters=inference(images)
#初始化
init=tf.global_variables_initializer()
sess=tf.Session()
sess.run(init)
#计算前向耗时
time_tensorflow_run(sess,fc3,'Forward')
#定义损失函数，计算反向耗时
objective=tf.nn.l2_loss(fc3)
grad=tf.gradients(objective,parameters)
time_tensorflow_run(sess,grad,'Forward-backward')
run_benckmark()