tensorflow学习：建立一个最简单的神经网络

程序功能：使用y= x^2 - 0.5加噪声生成1000个<x, y>样本点，然后搭建一个最简单的神经网络学习

import tensorflow as tf
import numpy as np


def add_layer(input_data, input_size, output_size, activation_function = None):
    '''
        input_data：输入数据
        input_size：输入的神经元个数
        output_size：输出的神经元个数
        activation_function：激励函数
    '''
    #产生input_size行output_size列的随机数
    Weights = tf.Variable(tf.random_normal([input_size, output_size]))
    #产生一行output_size列全为0.1的数
    biases = tf.Variable(tf.zeros([1, output_size]) + 0.1)
    #input_data * weights + biases
    Wx_plus_b = tf.matmul(input_data, Weights) + biases
    if activation_function is None:
        output = Wx_plus_b
    else:
        output = activation_function(Wx_plus_b)
    return output

#在-1到1之间均匀产生1000个数的list，并将其转成列向量
x_data = np.linspace(-1, 1, 1000)[:, np.newaxis]
noise = np.random.normal(0, 0.05, x_data.shape)
y_data = np.square(x_data) - 0.5 + noise

xs = tf.placeholder(tf.float32, [None, 1])
ys = tf.placeholder(tf.float32, [None, 1])

#隐藏层输出10个神经单元，现在还不知道该怎么设置这个神经单元数比较好
layer1 = add_layer(xs, 1, 10, tf.nn.relu)
prediction = add_layer(layer1, 10, 1, None)

#add_layer函数不需要自己实现，直接调下面函数也能达到相同的效果
# layer1_new = tf.layers.dense(xs, 10, tf.nn.relu)
# prediction_new = tf.layers.dense(layer1_new, 1)

#这两个loss到底有什么区别
loss = tf.reduce_mean(tf.square(ys - prediction))
# loss = tf.reduce_mean(tf.reduce_sum(tf.square(ys - prediction), reduction_indices=[1]))
train_step = tf.train.GradientDescentOptimizer(0.1).minimize(loss)

init = tf.global_variables_initializer()

with tf.Session() as sess:
    sess.run(init)
    
    for step in range(1001):
        #这个地方是用所有的数据进行梯度下降，这样做不合适
        sess.run(train_step, feed_dict = {xs: x_data, ys: y_data})
        if step % 50 == 0:
            print(step, sess.run(loss, feed_dict = {xs: x_data, ys:y_data}))