Dropout 解决 overfitting
overfitting也被称为过度学习,过度拟合。他是机器学习中常见的问题。
图中的黑色曲线是正常模型,绿色曲线就是overfitting模型。尽管绿色曲线很精确的区分了所有的训练数据,但是并没有描述数据的整体特征,对新测试的数据适应性比较差。
举个Regression(回归)的例子.
第三条曲线存在overfitting问题,尽管它经过了所有的训练点,但是不能很好地反映数据的趋势,预测能力严重不足。tensorflow提供了强大的dropout方法来结局overfitting的问题。
建立Dropout层
本次内容需要安装适应sklearn数据库中的数据 没有安装sklearn的同学可以参考一下以下
import tensorflow as tf
from sklearn.datasets import load_digits
from sklearn.model_selection import train_test_split
from sklearn.preprocessing import LabelBinarizer
keep_prob = tf.placeholder(tf.float32)
...
...
Wx_plus_b = tf.nn.dropout(Wx_plus_b, keep_prob)
这里的keep_prob是保留概率,即我们要保留的结果所占的比例,他作为一个placeholder,在run时传入,当keep_prob=1的时候,相当于100%保留,就是drop没起作用。
下面准备数据:
digits = load_digits()
X = digits.data
y = digits.target
y = LabelBinarizer().fit_transform(y)
X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=.3)
其中X_train是训练数据,X_test是测试数据。然后添加隐藏层和输出层
# add output layer
l1 = add_layer(xs, 64, 50, 'l1', activation_function=tf.nn.tanh)
prediction = add_layer(l1, 50, 10, 'l2', activation_function=tf.nn.softmax)
添加层的激活函数选用tanh(其实我也不知道为什么,莫烦的课里讲不写tanh函数,其他的函数会报none错误,就不管啦)
训练
sess.run(train_step, feed_dict={xs: X_train, ys: y_train, keep_prob: 0.5})
#sess.run(train_step, feed_dict={xs: X_train, ys: y_train, keep_prob: 1})
保留概率是指保留下多少,我们尝试0.5&1.0 观察tensorboard生成的可视化结果
可视化结果
keep_prob=1时,会暴露overfitting的问题,keep_prob=0.5时,Dropout就发挥了作用。
当keep_prob=1时 模型对训练数据的适应性优于测试数据,存在overfitting。红线是train的误差,蓝线是test的误差。
当keep_prob=0.5时
可以明显看到loss的变化
完整代码以及部分注释如下
import tensorflow as tf
from sklearn.datasets import load_digits
from sklearn.model_selection import train_test_split
from sklearn.preprocessing import LabelBinarizer #载入数据
digits = load_digits()
X = digits.data
y = digits.target
#preprocessing.LabelBinarizer是一个很好用的工具。
# 比如可以把yes和no转化为0和1
# 或是把incident和normal转化为0和1
y = LabelBinarizer().fit_transform(y)
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 = 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
#here to drop
Wx_plus_b = tf.nn.dropout(Wx_plus_b, keep_prob)
if activation_function is None:
outputs = Wx_plus_b
else:
outputs = activation_function(Wx_plus_b,)
tf.summary.histogram(layer_name + 'outputs',outputs)
return outputs #define placeholder for inputs to network
keep_prob = tf.placeholder(tf.float32)#保留概率 即我们需要保留的结果所占比
xs = tf.placeholder(tf.float32,[None,64])#8*8
ys = tf.placeholder(tf.float32,[None,10]) #add output layer
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 loss between prediction and real data
#loss
cross_entropy = tf.reduce_mean(-tf.reduce_sum(ys*tf.log(prediction),reduction_indices=[1])) tf.summary.scalar('loss',cross_entropy)
train_step = tf.train.GradientDescentOptimizer(0.5).minimize(cross_entropy) sess = tf.Session()
merged = tf.summary.merge_all() #summary writer goes in here
train_writer = tf.summary.FileWriter("logs/train",sess.graph)
test_writer = tf.summary.FileWriter("logs/test",sess.graph) init = tf.global_variables_initializer() #变量初始化 sess.run(init) for i in range(500):
#here to determine the keeping probability
sess.run(train_step,feed_dict={xs:X_train,ys:y_train,keep_prob:0.5})
if i %50 ==0:
#record loss
train_result = sess.run(merged,feed_dict={xs:X_train,ys:y_train,keep_prob:1})
test_result = sess.run(merged,feed_dict={xs:X_test,ys:y_test,keep_prob:1})
train_writer.add_summary(test_result,i)
test_writer.add_summary(test_result,i)