tensorflow入门到没门

tensorflow 开发的基本步骤

入门

import tensorflow as tf
import numpy as np

# 创建数据
x_data = np.random.rand(100).astype(np.float32)
# 学习系数接近.1  bias 接近.3
y_data = x_data * .1 + .3

# 开始创建结构
# 一维的数列  范围-1到1
Weights = tf.Variable(tf.random_uniform([1], -1.0, 1.0))

biases = tf.Variable(tf.zeros([1]))
# 预测
y = Weights * x_data + biases

# 损失率
loss = tf.reduce_mean(tf.square(y - y_data))
# 优化器
optimizer = tf.train.GradientDescentOptimizer(.5)

# 减少误差 目的提高准确度
train = optimizer.minimize(loss)

# 初始值  初始变量
init = tf.global_variables_initializer()

# 结束创建结构

sess = tf.Session()

sess.run(init)  # 激活init

for step in range(201):
    sess.run(train)
    if step % 20 == 0:  # 每个20步打印一下结果
        print(step, sess.run(Weights), sess.run(biases))

Session会话

import tensorflow as tf
import numpy as np

matrix01 = tf.constant([[3, 3]])
print(matrix01)
matrix02 = tf.constant([
    [2],
    [2]
])
print(matrix02)

product=tf.matmul(matrix01,matrix02) # 矩阵乘法
print(product)

# 方法1
# sess=tf.Session()
# print(sess)
# result=sess.run(product)
# print(result)
# sess.close()

# 方法2

with tf.Session() as sess:
    result1=sess.run(product)
    print(result1)

变量Variable

import tensorflow as tf
import numpy as np

state = tf.Variable(0, name="counter")
print(state.name)
one = tf.constant(1)

new_value = tf.add(state , one)
print(new_value)
update = tf.assign(state, new_value)
print(update)

init=tf.global_variables_initializer() # 初始化所有变量

with tf.Session() as sess:
    sess.run(init)
    for _ in range(3):
        sess.run(update)
        print(sess.run(state))

传入值placeholder

import tensorflow as tf
import numpy as np

input01 = tf.placeholder(tf.float32)
input02 = tf.placeholder(tf.float32)

output = tf.multiply(input01, input02)

with tf.Session() as sess:
    print(sess.run(output, feed_dict={input01: [7.], input02: [8.]}))

激励函数

http://www.tensorfly.cn/tfdoc/api_docs/python/nn.html

添加结构

import tensorflow as tf
import numpy as np


# 添加层
def add_layer_func(inputs, input_size, output_size, activation_function=None):
    Weights = tf.Variable(tf.random_normal([input_size, output_size]))
    biases = tf.Variable(tf.zeros([1, output_size]) + 0.1)
    Wx_plus_biases = tf.matmul(inputs, Weights) + biases
    if activation_function is None:
        outputs = Wx_plus_biases
    else:
        outputs = activation_function(Wx_plus_biases)
    return outputs

建造神经网络

import tensorflow as tf
import numpy as np


# 添加神经层
def add_layer_func(inputs, input_size, output_size, activation_function=None):
    Weights = tf.Variable(tf.random_normal([input_size, output_size]))
    biases = tf.Variable(tf.zeros([1, output_size]) + 0.1)
    Wx_plus_biases = tf.matmul(inputs, Weights) + biases
    if activation_function is None:
        outputs = Wx_plus_biases
    else:
        outputs = activation_function(Wx_plus_biases)
    return outputs


# 输入层  创建数据
x_data = np.linspace(-1, 1, 300)[:, np.newaxis]
noise = np.random.normal(0, 0.05, x_data.shape)
y_data = np.square(x_data) - 0.5 + noise

# 隐藏层
xs = tf.placeholder(tf.float32, [None, 1])
ys = tf.placeholder(tf.float32, [None, 1])
hide = add_layer_func(xs, 1, 10, activation_function=tf.nn.relu)

# 输出层
prediction = add_layer_func(hide, 10, 1, activation_function=None)

# 预测
loss = tf.reduce_mean(tf.reduce_sum(tf.square(ys - prediction), reduction_indices=[1]))

train_step = tf.train.GradientDescentOptimizer(.1).minimize(loss)  # 学习效率  一般小于1

init = tf.global_variables_initializer()
sess = tf.Session()
sess.run(init)

for i in range(1000):
    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}))

可视化

优化器

https://www.tensorflow.org/api_docs/python/tf/train#Optimizers