神经网络(输入:1个神经元)(中间层:10个神经元)(输出:1个神经元)
一、创造二次函数并加入噪声
import tensorflow as tf
import matplotlib.pyplot as plt
import numpy as np
#创造二次函数并加入噪声(神经网络输入及真实输出)
x_data = np.linspace(-0.5,0.5,200)[:,np.newaxis]
np.shape(x_data)
noise = np.random.normal(0,0.02,x_data.shape)
y_data = np.square(x_data) + noise
二、#定义两个占位符(神经网络输入及输出)
x = tf.placeholder(tf.float32,[None,1])
y = tf.placeholder(tf.float32,[None,1])
三、#定义神经网络中间层
#定义神经网络中间层
Weights_L1 = tf.Variable(tf.random_normal([1,10])) #神经元参数
biases_L1 = tf.Variable(tf.zeros([1,10])) #偏置项
print(Weights_L1,biases_L1)
Wx_plus_b_L1 = tf.matmul(x,Weights_L1) + biases_L1
L1 = tf.nn.tanh(Wx_plus_b_L1) #激活函数
四、定义网络输出层
#定义网络输出层
Weights_L2 = tf.Variable(tf.random_normal([10,1])) #神经元参数
biases_L2 = tf.Variable(tf.zeros([1,1])) #偏置项
Wx_plus_b_L2 = tf.matmul(L1,Weights_L2) + biases_L2 #激活项
prediction = tf.nn.tanh(Wx_plus_b_L2)
五、代价函数并进行梯度下降
#二次代价函数
loss = tf.reduce_mean(tf.square(y-prediction))
train_step = tf.train.GradientDescentOptimizer(0.1).minimize(loss) #梯度下降函数
六、运行神经网络并绘图
with tf.Session() as sess:
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()