tensorflow从入门到精通——Alexnet网络实现分类

导入包和库

import tensorflow as tf
from tensorflow.examples.tutorials.mnist import input_data

数据定义

# 输入数据
minst = input_data.read_data_sets("./data/",one_hot=True)

网络超参数定义

# 定义网络的超参数

learning_rate = 0.001
training_iters = 200000
batch_size =128
display_step = 10

网络参数

# 定义网路的参数
n_input = 784
n_classes = 10
dropout = 0.75

数据预定义占位符

# 输入占位符

x = tf.placeholder(tf.float32,[None,n_input])
y = tf.placeholder(tf.float32,[None,n_classes])
# 留着dropout
keep_prob = tf.placeholder(tf.float32) # dropout

定义卷积操作

def conv2d(name,x,w,b,strides=1):

    x = tf.nn.conv2d(x,w,strides=[1,strides,strides,1],padding="SAME")
    x = tf.nn.bias_add(x,b)
    # relu激活函数
    result = tf.nn.relu(x, name=name)
    return result

# 定义池化层
def maxpool2d(name,x,k=2):

    result = tf.nn.max_pool(x,ksize=[1,k,k,1],strides=[1,k,k,1],padding='SAME',name=name)
    return result

# 规范化操作,lrn
def norm(name,l_input,lsize=4):

    result = tf.nn.lrn(l_input,lsize,bias=1.0,alpha=0.001/9.0,beta=0.75,name=name)

    return result

定义网络参数

# 定义所有的网络参数

weights = {
     
    "wc1":tf.Variable(tf.random_normal([11,11,1,96])),
    "wc2": tf.Variable(tf.random_normal([5,5, 96, 256])),
    "wc3": tf.Variable(tf.random_normal([3, 3, 256, 384])),
    "wc4": tf.Variable(tf.random_normal([3, 3, 384, 384])),
    "wc5": tf.Variable(tf.random_normal([3, 3, 384, 256])),
    "wd1": tf.Variable(tf.random_normal([2*2*256,4096])),
    "wd2": tf.Variable(tf.random_normal([4096,4096])),
    "out": tf.Variable(tf.random_normal([4096,10])),
}
biases = {
     
    "bc1":tf.Variable(tf.random_normal([96])),
    "bc2":tf.Variable(tf.random_normal([256])),
    "bc3":tf.Variable(tf.random_normal([384])),
    "bc4":tf.Variable(tf.random_normal([384])),
    "bc5":tf.Variable(tf.random_normal([256])),
    "bd1":tf.Variable(tf.random_normal([4096])),
    "bd2":tf.Variable(tf.random_normal([4096])),
    "out":tf.Variable(tf.random_normal([n_classes])),
}

定义AlexNet

# 定义整个网络
def alex_net(x,weights,biases,dropout):
    # reshape
    x = tf.reshape(x,shape=[-1,28,28,1])

    # 第一层卷积
    conv1 = conv2d('conv1',x,weights['wc1'],biases['bc1'])
    # 下采样
    pool1 = maxpool2d('pool1',conv1,k=2)
    # 规范化
    norm1 = norm("morm1",pool1,lsize=4)

    # 第二层卷积
    conv2 = conv2d('conv2', norm1, weights['wc2'], biases['bc2'])
    # 下采样
    pool2 = maxpool2d('pool2', conv2, k=2)
    # 规范化
    norm2 = norm("morm2", pool2, lsize=4)

    # 第三层卷积
    conv3 = conv2d('conv3', norm2, weights['wc3'], biases['bc3'])
    # 下采样
    pool3 = maxpool2d('pool3', conv3, k=2)
    # 规范化
    norm3 = norm("morm3", pool3, lsize=4)

    # 第四层卷积
    conv4 = conv2d('conv4', norm3, weights['wc4'], biases['bc4'])
    conv5 = conv2d('conv5',conv4, weights['wc5'], biases['bc5'])

    # 下采样
    pool5 = maxpool2d('pool5', conv5, k=2)
    # 规范化
    norm5 = norm("morm5", pool5, lsize=4)


    fc1 = tf.reshape(norm5,[-1,weights["wd1"].get_shape().as_list()[0]])

    fc1 = tf.add(tf.matmul(fc1,weights['wd1']),biases['bd1'])
    fc1 = tf.nn.relu(fc1)
    # dropout
    fc1 = tf.nn.dropout(fc1,dropout)

    # 全连接2
    fc2 =  tf.reshape(fc1, [-1, weights["wd2"].get_shape().as_list()[0]])
    fc2 = tf.add(tf.matmul(fc2, weights['wd2']), biases['bd2'])
    fc2 = tf.nn.relu(fc2)
    # dropout
    fc2 = tf.nn.dropout(fc2, dropout)

    # 输出层
    out = tf.add(tf.matmul(fc2,weights["out"]),biases['out'])
    return out

模型图定义以及损失函数

• tf.reduce_mean(tf.nn.softmax_cross_entropy_with_logits(logits=pred,labels=y))，版本问题需要现在指定参数传入。

# 构建模型定义损失函数
pred = alex_net(x,weights,biases,keep_prob)
# 定义损失函数
cost = tf.reduce_mean(tf.nn.softmax_cross_entropy_with_logits(logits=pred,labels=y))
optimizer = tf.train.AdamOptimizer(learning_rate=learning_rate).minimize(cost)

# 评估函数
correct_pred = tf.equal(tf.argmax(pred,1),tf.argmax(y,1))
# 分类得分
accuracy = tf.reduce_mean(tf.cast(correct_pred,tf.float32))

模型训练以及测试

# 初始化变量
if __name__ == '__main__':

    init = tf.global_variables_initializer()

    with tf.Session() as sess:
        with tf.device("gpu:0"):
            sess.run(init)
            # 开始训练
            step = 1
            while step*batch_size<training_iters:
                batch_x,batch_y = minst.train.next_batch(batch_size)
                sess.run(optimizer,feed_dict={
     x:batch_x,y:batch_y,keep_prob:dropout})
                if step%display_step==0:
                    loss,acc = sess.run([cost,accuracy],feed_dict={
     x:batch_x,
                                                                   y:batch_y,
                                                                   keep_prob:1.})
                    print("Iter"+str(step*batch_size)+
                          ",Minibatch Loss="+"{:.6f}".format(loss)+
                          ",Training Accuracy="+
                        "{:.5f}".format(acc))
                step+=1
            print("Optimization Finished!")
            print("Testing Accuracy:",
                  sess.run(accuracy,feed_dict={
     
                      x:minst.test.images[:256],
                      y:minst.test.labels[:256],
                      keep_prob:1.
                  }))

完整代码

# -*- coding: utf-8 -*-
"""
@Time ： 2021/8/12 上午9:10
@Auth ： 陈伟峰
@File ：read_data.py
@phone: 15882085601
@IDE ：PyCharm
@Motto：ABC(Always Be Coding)

"""
import tensorflow as tf
from tensorflow.examples.tutorials.mnist import input_data



# 输入数据
minst = input_data.read_data_sets("./data/",one_hot=True)

# 定义网络的超参数

learning_rate = 0.001
training_iters = 200000
batch_size =128
display_step = 10


# 定义网路的参数

n_input = 784
n_classes = 10
dropout = 0.75

# 输入占位符

x = tf.placeholder(tf.float32,[None,n_input])
y = tf.placeholder(tf.float32,[None,n_classes])
keep_prob = tf.placeholder(tf.float32) # dropout



def conv2d(name,x,w,b,strides=1):

    x = tf.nn.conv2d(x,w,strides=[1,strides,strides,1],padding="SAME")
    x = tf.nn.bias_add(x,b)
    # relu激活函数
    result = tf.nn.relu(x, name=name)
    return result

# 定义池化层
def maxpool2d(name,x,k=2):

    result = tf.nn.max_pool(x,ksize=[1,k,k,1],strides=[1,k,k,1],padding='SAME',name=name)
    return result

# 规范化操作
def norm(name,l_input,lsize=4):

    result = tf.nn.lrn(l_input,lsize,bias=1.0,alpha=0.001/9.0,beta=0.75,name=name)

    return result


# 定义所有的网络参数

weights = {
     
    "wc1":tf.Variable(tf.random_normal([11,11,1,96])),
    "wc2": tf.Variable(tf.random_normal([5,5, 96, 256])),
    "wc3": tf.Variable(tf.random_normal([3, 3, 256, 384])),
    "wc4": tf.Variable(tf.random_normal([3, 3, 384, 384])),
    "wc5": tf.Variable(tf.random_normal([3, 3, 384, 256])),
    "wd1": tf.Variable(tf.random_normal([2*2*256,4096])),
    "wd2": tf.Variable(tf.random_normal([4096,4096])),
    "out": tf.Variable(tf.random_normal([4096,10])),
}
biases = {
     
    "bc1":tf.Variable(tf.random_normal([96])),
    "bc2":tf.Variable(tf.random_normal([256])),
    "bc3":tf.Variable(tf.random_normal([384])),
    "bc4":tf.Variable(tf.random_normal([384])),
    "bc5":tf.Variable(tf.random_normal([256])),
    "bd1":tf.Variable(tf.random_normal([4096])),
    "bd2":tf.Variable(tf.random_normal([4096])),
    "out":tf.Variable(tf.random_normal([n_classes])),
}

# 定义整个网络
def alex_net(x,weights,biases,dropout):
    # reshape
    x = tf.reshape(x,shape=[-1,28,28,1])

    # 第一层卷积
    conv1 = conv2d('conv1',x,weights['wc1'],biases['bc1'])
    # 下采样
    pool1 = maxpool2d('pool1',conv1,k=2)
    # 规范化
    norm1 = norm("morm1",pool1,lsize=4)

    # 第二层卷积
    conv2 = conv2d('conv2', norm1, weights['wc2'], biases['bc2'])
    # 下采样
    pool2 = maxpool2d('pool2', conv2, k=2)
    # 规范化
    norm2 = norm("morm2", pool2, lsize=4)

    # 第三层卷积
    conv3 = conv2d('conv3', norm2, weights['wc3'], biases['bc3'])
    # 下采样
    pool3 = maxpool2d('pool3', conv3, k=2)
    # 规范化
    norm3 = norm("morm3", pool3, lsize=4)

    # 第四层卷积
    conv4 = conv2d('conv4', norm3, weights['wc4'], biases['bc4'])
    conv5 = conv2d('conv5',conv4, weights['wc5'], biases['bc5'])

    # 下采样
    pool5 = maxpool2d('pool5', conv5, k=2)
    # 规范化
    norm5 = norm("morm5", pool5, lsize=4)


    fc1 = tf.reshape(norm5,[-1,weights["wd1"].get_shape().as_list()[0]])

    fc1 = tf.add(tf.matmul(fc1,weights['wd1']),biases['bd1'])
    fc1 = tf.nn.relu(fc1)
    # dropout
    fc1 = tf.nn.dropout(fc1,dropout)

    # 全连接2
    fc2 =  tf.reshape(fc1, [-1, weights["wd2"].get_shape().as_list()[0]])
    fc2 = tf.add(tf.matmul(fc2, weights['wd2']), biases['bd2'])
    fc2 = tf.nn.relu(fc2)
    # dropout
    fc2 = tf.nn.dropout(fc2, dropout)

    # 输出层
    out = tf.add(tf.matmul(fc2,weights["out"]),biases['out'])
    return out


# 构建模型定义损失函数
pred = alex_net(x,weights,biases,keep_prob)
# 定义损失函数
cost = tf.reduce_mean(tf.nn.softmax_cross_entropy_with_logits(logits=pred,labels=y))
optimizer = tf.train.AdamOptimizer(learning_rate=learning_rate).minimize(cost)

# 评估函数
correct_pred = tf.equal(tf.argmax(pred,1),tf.argmax(y,1))
accuracy = tf.reduce_mean(tf.cast(correct_pred,tf.float32))


# 初始化变量
if __name__ == '__main__':

    init = tf.global_variables_initializer()

    with tf.Session() as sess:
        with tf.device("gpu:0"):
            sess.run(init)
            # 开始训练
            step = 1
            while step*batch_size<training_iters:
                batch_x,batch_y = minst.train.next_batch(batch_size)
                sess.run(optimizer,feed_dict={
     x:batch_x,y:batch_y,keep_prob:dropout})
                if step%display_step==0:
                    loss,acc = sess.run([cost,accuracy],feed_dict={
     x:batch_x,
                                                                   y:batch_y,
                                                                   keep_prob:1.})
                    print("Iter"+str(step*batch_size)+
                          ",Minibatch Loss="+"{:.6f}".format(loss)+
                          ",Training Accuracy="+
                        "{:.5f}".format(acc))
                step+=1
            print("Optimization Finished!")
            print("Testing Accuracy:",
                  sess.run(accuracy,feed_dict={
     
                      x:minst.test.images[:256],
                      y:minst.test.labels[:256],
                      keep_prob:1.
                  }))

• 虽然tensorflow对于模型的构建步骤相比于pytorch那些框架来说比较复杂，但是tensorflow的模型构建方便自己知道模型中的参数，就连大小计算量都能够更加清晰明了。用起来也不错！
•