机器学习实践——使用TensorFlow训练线性模型Y=2*X

登录后复制

import numpy as npimport tensorflow as tfimport matplotlib.pyplot as pltdef moving_average(a,w=10):    
if(len(a)<w):        return a[:]    return [val if idx <w else sum(a[idx-w:idx])/w for idx,
val in enumerate(a)]X_train=np.linspace(-1,1,100)Y_train=2*X_train+np.random.randn(*X_train.shape)*0.3
# 训练数据X=tf.placeholder('float')Y=tf.placeholder('float')# 占位符W=tf.Variable(tf.random_normal([1]),
name='weight')b=tf.Variable(tf.zeros([1]),name='bias')# 定义权重和偏置z=tf.multiply(X,W)+b# 定义前向结构
# 反向模型的搭建即反向优化loss=tf.reduce_mean(tf.square(Y-z))
# 定义学习率:代表调整参数的速度，这个值一般小于一，这个值越大，表明调整幅度的速度越大，但不精确
# 这个值越小，调整幅度越小，但是速度慢learn_rate=0.01
# 定义优化器：GridientDescentOptimizer梯度下降算法optimizer=tf.train.GradientDescentOptimizer(learn_rate)
.minimize(loss)# 迭代训练模型,初始化所有变量init=tf.global_variables_initializer()
# 定义训练次数training_epochs=20# 定义显示信息display_step=2with tf.Session() as sess:    
sess.run(init)    plot_data={'batch_size':[],'loss_value':[]}    
for epoch in range(training_epochs):        for x,y in zip(X_train,Y_train):            
sess.run(optimizer,feed_dict={X:x,Y:y})        if epoch % display_step == 0:            
loss_value = sess.run(loss, feed_dict={X:x, Y:y})            print('Epoch:', epoch + 1, '
Loss=', loss_value, 'w=',sess.run(W),'b=',sess.run(b))            if not (loss == 'NA'):                
plot_data['batch_size'].append(epoch)                plot_data['loss_value'].append(loss_value)    
print('Finished!')    # 可视化模型    plt.plot(X_train, Y_train, 'ro', label='Origin data')    
plt.plot(X_train, sess.run(W) * X_train + sess.run(b),label='FittedLine')    plt.legend()    
plt.show()    plot_data['avgloss']=moving_average(plot_data['loss_value'])    plt.figure(1)    
plt.subplot(211)    plt.plot(plot_data['batch_size'],plot_data['avgloss'],'b',linewidth=1.5)    
plt.xlabel('Minbatch number')    plt.ylabel('Loss')    plt.title('Minibatch run vs Training loss')    
plt.show()1.2.3.4.5.6.7.8.9.10.11.12.13.14.15.16.17.18.19.20.21.22.23.24.25.26.27.28.29.30.31.32.33
.34.35.36.37.38.39.40.41.42.43.44.45.46.47.48.49.50.51.52.53.54.55.56.57.58.59.60.61.62.63.64.65.66.67.68.69.

免责声明：本文系网络转载或改编，未找到原创作者，版权归原作者所有。如涉及版权，请联系删