深度強化學習篇1：神經網絡迴歸實例---複雜函數擬合

擅長一件事，喜歡一件事，工作一件事。如果三件事重合，就不要躊躇了。

一、TensorFlow2

TensorFlow的張量與數據流圖：一維張量如向量，二維張量如矩陣，以此類推。TensorFlow的數據流圖如下。

二、神經網絡基礎編程實例

1、TensorFlow例程1——TensorFlow結構，優化器逼近函數參數

import tensorflow as tf
import numpy as np

# 數據
x_data = np.random.rand(100).astype(np.float32)
y_data = x_data*0.1 + 0.3
# 權重和偏置
Weights = tf.Variable(tf.random_uniform([1], -1.0, 1.0))
biases = tf.Variable(tf.zeros([1]))
# 估計y
y = Weights*x_data+biases
loss = tf.reduce_mean(tf.square(y-y_data))
# 設置優化器
optimizer = tf.train.GradientDescentOptimizer(0.5)
train = optimizer.minimize(loss)
# 初始化全部變量
init = tf.initialize_all_variables()
# 設置對話
sess = tf.Session()
sess.run(init)
for step in range(201):
    sess.run(train)
    if step % 20 == 0:
        print(step,sess.run(Weights),sess.run(biases))

2、TensorFlow例程2——構建三層神經網絡預測函數輸出(300一維數據輸入，輸出預測值，損失函數爲誤差平和方均值)

一維數據輸入，全連接層10個神經元。

import tensorflow.compat.v1 as tf
tf.disable_v2_behavior()
import numpy as np
import matplotlib.pyplot as plt
plt.ion()
plt.show()

def add_layer(inputs, in_size, out_size, activation_function=None):
    Weights = tf.Variable(tf.random_normal([in_size, out_size]))
    biases = tf.Variable(tf.zeros([1, out_size]) + 0.1)
    # 神經網絡預測值
    Wx_plus_b = tf.matmul(inputs, Weights) + biases
    # 激勵函數
    if activation_function is None:
        outputs = Wx_plus_b
    else:
        outputs = activation_function(Wx_plus_b)
    return outputs

# 構建數據
x_data = np.linspace(-1,1,300, dtype=np.float32)[:, np.newaxis]
noise = np.random.normal(0, 0.05, x_data.shape).astype(np.float32)
y_data = np.square(x_data) - 0.5 + noise

# 定義輸入
xs = tf.placeholder(tf.float32, [None, 1])
ys = tf.placeholder(tf.float32, [None, 1])

# 定義隱藏層
l1 = add_layer(xs, 1, 10, activation_function=tf.nn.relu)

# 定義輸出層
prediction = add_layer(l1, 10, 1, activation_function=None)

# 損失函數：誤差平方和求均值
loss = tf.reduce_mean(tf.reduce_sum(tf.square(ys - prediction), reduction_indices=[1]))

# 優化器最小化損失函數
train_step = tf.train.GradientDescentOptimizer(0.1).minimize(loss)

# 初始化變量
init = tf.global_variables_initializer()

# 開啓會話
sess = tf.Session()
sess.run(init)

# 顯示圖片框
fig = plt.figure()

# 連續畫圖ax,數據點
ax = fig.add_subplot(1,1,1)
ax.scatter(x_data, y_data)

# 訓練
for i in range(5000):
    sess.run(train_step, feed_dict={xs: x_data, ys: y_data})
    if i % 50 == 0:
        # print(sess.run(loss, feed_dict={xs: x_data, ys: y_data}))
        # 除了第一次,抹除之前預測線
        try:
            ax.lines.remove(lines[0])
        except Exception:
            pass
        prediction_value = sess.run(prediction, feed_dict={xs: x_data})
        # 把預測用一條線表示
        lines = ax.plot(x_data, prediction_value, 'r-', lw=5)
        plt.pause(0.1)
        print(sess.run(loss, feed_dict={xs: x_data, ys: y_data}))