TensorFlow實現神經網路

import tensorflow as tf
import numpy as np
import matplotlib.pyplot as plt

# 1.生成訓練樣本
dataset_size=128
X=np.random.RandomState(1).uniform(0,1,(dataset_size,2))
Y=[[int(x1+x2<1)] for (x1,x2) in X]

for i in range(len(X)):
	if Y[i][0]==1:
		plt.scatter(X[i][0],X[i][1],c='r')
	else:
		plt.scatter(X[i][0],X[i][1],c='k')
plt.show()

# 2.定義訓練資料batch大小
batchsize=8

# 3.定義神經網路引數
w1=tf.Variable(tf.random_normal([2,3],mean=0,stddev=1,seed=1))
w2=tf.Variable(tf.random_normal([3,1],mean=0,stddev=1,seed=1))

x=tf.placeholder(tf.float32,shape=(None,2),name='x-input')
y_=tf.placeholder(tf.float32,shape=(None,1),name='y-input')

# 前向傳播
a=tf.matmul(x,w1)
y=tf.matmul(a,w2)

# 損失函式
cross_entropy=-tf.reduce_mean(y_*tf.log(tf.clip_by_value(y,1e-10,1.0)))
train_step=tf.train.AdamOptimizer(0.001).minimize(cross_entropy)

# 建立會話執行tensorflow
with tf.Session() as sess:
	# 初始化變數
	init_op=tf.initialize_all_variables()
	sess.run(init_op)

	STEPS=5000
	print('Start training>............')
	for i in range(STEPS):
		start=(i*batchsize)%dataset_size
		end=min(start+batchsize,dataset_size)

		# 訓練
		sess.run(train_step,feed_dict={x:X[start:end],y_:Y[start:end]})
		# 計算交叉熵並輸出
		if i%1000==0:
			total_cross_entropy=sess.run(cross_entropy,feed_dict={x:X,y_:Y})
			print('第%d次訓練，總體交叉熵為：%f'%(i,total_cross_entropy))