# -*- coding:utf-8 -*-#功能： 使用tensorflow實現一個簡單的邏輯回歸import tensorflow as tfimport numpy as npimport matplotlib.pyplot as plt #創建占位符X=tf.placeholder(tf.float32)Y=tf.placeholder(tf.float32) #創建變量#tf.random_normal([1])返回一個符合正太分布的隨機數w=tf.Variable(tf.random_normal([1],name='weight'))b=tf.Variable(tf.random_normal([1],name='bias')) y_predict=tf.sigmoid(tf.add(tf.mul(X,w),b))num_samples=400cost=tf.reduce_sum(tf.pow(y_predict-Y,2.0))/num_samples #學習率lr=0.01optimizer=tf.train.AdamOptimizer().minimize(cost) #創建session 并初始化所有變量num_epoch=500cost_accum=[]cost_prev=0#np.linspace（）創建agiel等差數組，元素個素為num_samplesxs=np.linspace(-5,5,num_samples)ys=np.sin(xs)+np.random.normal(0,0.01,num_samples) with tf.Session() as sess:  #初始化所有變量  sess.run(tf.initialize_all_variables())  #開始訓練  for epoch in range(num_epoch):    for x,y in zip(xs,ys):      sess.run(optimizer,feed_dict={X:x,Y:y})    train_cost=sess.run(cost,feed_dict={X:x,Y:y})    cost_accum.append(train_cost)    print "train_cost is:",str(train_cost)     #當誤差小于10-6時 終止訓練    if np.abs(cost_prev-train_cost)<1e-6:      break    #保存最終的誤差    cost_prev=train_cost#畫圖 畫出每一輪訓練所有樣本之后的誤差plt.plot(range(len(cost_accum)),cost_accum,'r')plt.title('Logic Regression Cost Curve')plt.xlabel('epoch')plt.ylabel('cost')plt.show()