Tensorflow 搭建自己的神经网络(二)

MNIST简单分类

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

mnist = input_data.read_data_sets('MNIST_data', one_hot=True)
#print mnist.train.num_examples
#print mnist.test.num_examples
#print mnist.validation.num_examples
#print mnist.train.images[0]
#print mnist.train.labels[0]

def add_layer(inputs, in_size, out_size, activation_function=None):
    Weights = tf.Variable(tf.random_normal([in_size, out_size]))
    biases = tf.Variable(tf.zeros([1, out_size]) + 0.1)
    Wx_plus_b = tf.matmul(inputs, Weights) + biases
    if activation_function is None:
        outputs = Wx_plus_b
    else:
        outputs = activation_function(Wx_plus_b)
    return outputs

def compute_accuracy(v_xs, v_ys):
    # no need ys
    y_pre = sess.run(prediction, feed_dict={xs: v_xs})
    # tf.argmax(x, 1): find max value in each row here_return 100*1
    correct_prediction = tf.equal(tf.argmax(y_pre, 1), tf.argmax(v_ys, 1))
    # tf.cast(x, dtype): transfer format of x to dtype
    accuracy = tf.reduce_mean(tf.cast(correct_prediction,tf.float32))
    result = sess.run(accuracy, feed_dict = {xs: v_xs, ys: v_ys})
    return result
    

# define placeholder for inputs to network
xs = tf.placeholder(tf.float32, shape = (None, 784))#28*28
ys = tf.placeholder(tf.float32, [None, 10])

# add output layer
prediction = add_layer(xs, 784, 10, activation_function = tf.nn.softmax)

# the error between prediction and real data
cross_entropy = tf.reduce_mean(-tf.reduce_sum(ys * tf.log(
        tf.clip_by_value(prediction, 1e-10, 1)),reduction_indices=[1]))
train_step = tf.train.GradientDescentOptimizer(0.5).minimize(cross_entropy)

with tf.Session() as sess:
    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})
        if i % 100 == 0:
            print(compute_accuracy(
                    mnist.test.images, mnist.test.labels))

 

Tensorflow解决overfitting——dropout    转自：莫烦Tensorflow教程（1~14）

Summary：所有需要在TensorBoard上展示的统计结果。
tf.name_scope()：为Graph中的Tensor添加层级，TensorBoard会按照代码指定的层级进行展示，初始状态下只绘制最高层级的效果，点击后可展开层级看到下一层的细节。
tf.summary.scalar()：添加标量统计结果。
tf.summary.histogram()：添加任意shape的Tensor，统计这个Tensor的取值分布。
tf.summary.merge_all()：添加一个操作，代表执行所有summary操作，这样可以避免人工执行每一个summary op。
tf.summary.FileWrite：用于将Summary写入磁盘，需要制定存储路径logdir，如果传递了Graph对象，则在Graph Visualization会显示Tensor Shape Information。执行summary op后，将返回结果传递给add_summary()方法即可。

未使用dropout：

import tensorflow as tf
from sklearn.datasets import load_digits
from sklearn.model_selection import train_test_split
from sklearn.preprocessing import LabelBinarizer

#load data
digits=load_digits()
#0~9的图像
X=digits.data
#y是binary的，表示数字1，就在第二个位置放上1，其余都为0
y=digits.target
y=LabelBinarizer().fit_transform(y)
#切分
X_train,X_test,y_train,y_test=train_test_split(X,y,test_size=3)

def add_layer(inputs,in_size,out_size,layer_name,activation_function=None):
    #Weights是一个矩阵，[行，列]为[in_size,out_size]
    Weights=tf.Variable(tf.random_normal([in_size,out_size]))#正态分布
    #初始值推荐不为0，所以加上0.1，一行，out_size列
    biases=tf.Variable(tf.zeros([1,out_size])+0.1)
    #Weights*x+b的初始化的值，也就是未激活的值
    Wx_plus_b=tf.matmul(inputs,Weights)+biases

    #激活

    if activation_function is None:
        #激活函数为None，也就是线性函数
        outputs=Wx_plus_b
    else:
        outputs=activation_function(Wx_plus_b)
    # 下面的表示outputs的值
    tf.summary.histogram(layer_name+'/outputs',outputs)

    return outputs

#define placeholder for inputs to network
xs=tf.placeholder(tf.float32,[None,64])
ys=tf.placeholder(tf.float32,[None,10])

#add output layer
# l1为隐藏层，为了更加看出overfitting，所以输出给了100
l1=add_layer(xs,64,100,'l1',activation_function=tf.nn.tanh)

prediction=add_layer(l1,100,10,'l2',activation_function=tf.nn.softmax)

#the error between prediction and real data
cross_entropy=tf.reduce_mean(-tf.reduce_sum(ys*tf.log(prediction),reduction_indices=[1]))
#添加标量统计结果
tf.summary.scalar('loss',cross_entropy)
train_step=tf.train.GradientDescentOptimizer(0.6).minimize(cross_entropy)

init=tf.global_variables_initializer()

with tf.Session() as sess:
    sess.run(init)
    #添加一个操作，代表执行所有summary操作，这样可以避免人工执行每一个summary op
    merged=tf.summary.merge_all()
    #summary writer goes in here
    train_writer=tf.summary.FileWriter("logs/train",sess.graph)#train为log的子文件夹
    test_writer=tf.summary.FileWriter("logs/test",sess.graph)
    for i in range(500):
        sess.run(train_step,feed_dict={xs:X_train,ys:y_train})
        if i%50==0:
            #record loss
            train_result=sess.run(merged,feed_dict={xs:X_train,ys:y_train})
            test_result = sess.run(merged, feed_dict={xs: X_test, ys: y_test})
            train_writer.add_summary(train_result,i)
            test_writer.add_summary(test_result,i)

红色：testdata

绿色：trainingdata

使用dropout：

train_result=sess.run(merged,feed_dict{xs:X_train,ys:y_train,keep_prob:1})

也就是训练数据未使用dropout时，仍然有过拟合

 

train_result=sess.run(merged,feed_dict{xs:X_train,ys:y_train,keep_prob:0.5})

此时两者拟合的很好

import tensorflow as tf
from sklearn.datasets import load_digits
from sklearn.model_selection import train_test_split
from sklearn.preprocessing import LabelBinarizer

#load data
digits=load_digits()
#0~9的图像
X=digits.data
#y是binary的，表示数字1，就在第二个位置放上1，其余都为0
y=digits.target
y=LabelBinarizer().fit_transform(y)
#切分
X_train,X_test,y_train,y_test=train_test_split(X,y,test_size=3)

def add_layer(inputs,in_size,out_size,layer_name,activation_function=None):
    #Weights是一个矩阵，[行，列]为[in_size,out_size]
    Weights=tf.Variable(tf.random_normal([in_size,out_size]))#正态分布
    #初始值推荐不为0，所以加上0.1，一行，out_size列
    biases=tf.Variable(tf.zeros([1,out_size])+0.1)
    #Weights*x+b的初始化的值，也就是未激活的值
    Wx_plus_b=tf.matmul(inputs,Weights)+biases
    #dropout 主功能，drop掉50%的结果，输出更新后的结果
    Wx_plus_b=tf.nn.dropout(Wx_plus_b,keep_prob)
    #激活

    if activation_function is None:
        #激活函数为None，也就是线性函数
        outputs=Wx_plus_b
    else:
        outputs=activation_function(Wx_plus_b)
    # 下面的表示outputs的值
    tf.summary.histogram(layer_name+'/outputs',outputs)

    return outputs

#define placeholder for inputs to network
"""dropout"""
# 确定保留多少结果不被舍弃掉
keep_prob=tf.placeholder(tf.float32)

xs=tf.placeholder(tf.float32,[None,64])
ys=tf.placeholder(tf.float32,[None,10])

#add output layer
# l1为隐藏层，为了更加看出overfitting，所以输出给了100
l1=add_layer(xs,64,50,'l1',activation_function=tf.nn.tanh)

prediction=add_layer(l1,50,10,'l2',activation_function=tf.nn.softmax)

#the error between prediction and real data
cross_entropy=tf.reduce_mean(-tf.reduce_sum(ys*tf.log(prediction),reduction_indices=[1]))
#添加标量统计结果
tf.summary.scalar('loss',cross_entropy)
train_step=tf.train.GradientDescentOptimizer(0.6).minimize(cross_entropy)

init=tf.global_variables_initializer()

with tf.Session() as sess:
    sess.run(init)
    #添加一个操作，代表执行所有summary操作，这样可以避免人工执行每一个summary op
    merged=tf.summary.merge_all()
    #summary writer goes in here
    train_writer=tf.summary.FileWriter("logs/train",sess.graph)#train为log的子文件夹
    test_writer=tf.summary.FileWriter("logs/test",sess.graph)
    for i in range(500):
        # drop掉60%，保持40%不被drop掉
        sess.run(train_step,feed_dict={xs:X_train,ys:y_train,keep_prob:0.4})
        if i%50==0:
            #record loss（不要drop掉任何东西，所以为1）
            train_result=sess.run(merged,feed_dict={xs:X_train,ys:y_train,keep_prob:0.5})
            test_result = sess.run(merged, feed_dict={xs: X_test, ys: y_test,keep_prob:1})
            train_writer.add_summary(train_result,i)
            test_writer.add_summary(test_result,i)