tensorflow简单的回归
#回归
import tensorflow as tf
import numpy as np
import matplotlib.pyplot as plt
#使用numpy生成200个随机点
x_data = np.linspace(-0.5, 0.5, 200)[:,np.newaxis] #200*1的二维矩阵
'''
np.linspace(start, stop, num)生成[start,stop]内等距离的num个点
np.newaxis 在使用和功能上等价于 None,其实就是 None 的一个别名;
当np.newaxis在[,]中,前面时,变为列扩展的二维数组
当np.newaxis在[,]中,后面时,变为行扩展的二维数组
'''
noise = np.random.normal(0, 0.02, x_data.shape) #生成形状与x一样的随机干扰
y_data = np.square(x_data) + noise #上述为定义的样本
#定义两个placeholder(根据样本来定义的)
x = tf.placeholder(tf.float32, [None, 1]) #x形状:行不确定,列为1
y = tf.placeholder(tf.float32, [None, 1])
'''构建模型,使得输入一个x,经过神经网络。得到y,与真实的样本y_data接近'''
'''
Weights = tf.Variable(tf.random_normal([in_size,out_size])) #Weight中都是随机变量
biases = tf.Variable(tf.zeros([1,out_size])+0.1) #biases推荐初始值不为0
'''
#构建神经网络中间层
Weights_L1 = tf.Variable(tf.random_normal([1,10])) #输入是一维的输入层一个神经元,中间层设置10个神经元
biases_L1 = tf.Variable(tf.zeros([1, 10]))
Wx_plus_b_L1 = tf.matmul(x, Weights_L1) + biases_L1 #信号的总和
L1 = tf.nn.tanh(Wx_plus_b_L1) #用双曲正切函数作为中间层的激活函数
#定义神经网络输出层
Weights_L2 = tf.Variable(tf.random_normal([10, 1]))
biases_L2 = tf.Variable(tf.zeros([1, 1]))
Wx_plus_b_L2 = tf.matmul(L1, Weights_L2) + biases_L2
prediction = tf.nn.tanh(Wx_plus_b_L2)
#二次代价函数
loss = tf.reduce_mean(tf.square(prediction - y))
#使用梯度下降法
train_step = tf.train.GradientDescentOptimizer(0.1).minimize(loss)
with tf.Session() as sess:
#初始化变量
sess.run(tf.global_variables_initializer())
for _ in range(2000):
sess.run(train_step, feed_dict={x:x_data, y:y_data})
#获得预测值
prediction_value = sess.run(prediction, feed_dict={x:x_data})
#画图
plt.figure()
plt.scatter(x_data, y_data)
plt.plot(x_data, prediction_value, 'r-', lw=5)
plt.show()
import tensorflow as tf
import numpy as np
import matplotlib.pyplot as plt
def add_layer(inputs,in_size,out_size,activation_function=None): #activation_function=None线性函数
Weights = tf.Variable(tf.random_normal([in_size,out_size])) #Weight中都是随机变量
biases = tf.Variable(tf.zeros([1,out_size])+0.1) #biases推荐初始值不为0
Wx_plus_b = tf.matmul(inputs,Weights)+biases #inputs*Weight+biases
if activation_function is None:
outputs = Wx_plus_b
else:
outputs = activation_function(Wx_plus_b)
return outputs
#创建数据x_data,y_data
x_data = np.linspace(-1,1,300)[:,np.newaxis] #[-1,1]区间,300个单位,np.newaxis增加维度
noise = np.random.normal(0,0.05,x_data.shape) #噪点
y_data = np.square(x_data)-0.5+noise
xs = tf.placeholder(tf.float32,[None,1])
ys = tf.placeholder(tf.float32,[None,1])
#三层神经,输入层(1个神经元),隐藏层(10神经元),输出层(1个神经元)
l1 = add_layer(xs,1,10,activation_function=tf.nn.tanh) #中间层
prediction = add_layer(l1,10,1,activation_function=None) #输出层
#predition值与y_data差别
loss = tf.reduce_mean(tf.reduce_sum(tf.square(ys-prediction),reduction_indices=[1])) #square()平方,sum()求和,mean()平均值
train_step = tf.train.GradientDescentOptimizer(0.1).minimize(loss) #0.1学习效率,minimize(loss)减小loss误差
init = tf.initialize_all_variables()
sess = tf.Session()
sess.run(init) #先执行init
#可视化
fig = plt.figure()
ax = fig.add_subplot(1,1,1)
ax.scatter(x_data,y_data)
plt.ion() #不让show() block
plt.show()
#训练1k次
for i in range(1000):
sess.run(train_step,feed_dict={xs:x_data,ys:y_data})
if i%50==0:
try:
ax.lines.remove(lines[0]) #lines建一个抹除一个
except Exception:
pass
#print(sess.run(loss,feed_dict={xs:x_data,ys:y_data})) #输出loss值
#可视化
prediction_value = sess.run(prediction,feed_dict={xs:x_data,ys:y_data})
lines = ax.plot(x_data,prediction_value,'r-',lw=5) #x_data X轴,prediction_value Y轴,'r-'红线,lw=5线宽5
plt.pause(0.5) #暂停0.5秒