![](https://img-blog.csdnimg.cn/20201014180756913.png?x-oss-process=image/resize,m_fixed,h_64,w_64)
TensorFlow
粉皮皮
这个作者很懒,什么都没留下…
展开
-
优化器Optimier
SGD为最基础的,Momentum和AdaGrad对其进行了优化,RMSProp结合了前两者的特征,而Adam是完全结合了Momentum和AdaGrad本质是learning rate 的改变原创 2021-07-02 17:43:30 · 160 阅读 · 0 评论 -
结果可视化plot_result
import matplotlib.pyplot as pltimport tensorflow.compat.v1 as tfcimport tensorflow as tfimport numpy as np# tensorflow 2.5.0# 添加层函数def add_layer(inputs, in_size, out_size, activation_function=None): Weights = tf.Variable(tf.random.normal([in_siz原创 2021-07-02 17:17:00 · 1079 阅读 · 0 评论 -
搭建神经网络
import tensorflow.compat.v1 as tfcimport tensorflow as tfimport numpy as np# tensorflow 2.5.0def add_layer(inputs, in_size, out_size, activation_function=None): Weights = tf.Variable(tf.random.normal([in_size, out_size])) biases = tf.Variable(原创 2021-07-02 11:53:09 · 66 阅读 · 0 评论 -
定义一个添加层函数
import tensorflow.compat.v1 as tfcimport tensorflow as tf# 定义一个添加神经层的函数# None说明激励函数是线性的def add_layer(inputs, in_size, out_size, activation_function=None): # 随机变量的一个矩阵 Weights = tf.Variable(tf.random.normal([in_size, out_size])) # 加0.1是因为在神原创 2021-07-02 11:11:32 · 63 阅读 · 0 评论 -
激励函数active function
原创 2021-07-02 11:00:13 · 180 阅读 · 0 评论 -
placeholder传入值
placeholder是Tensorflow中的占位符,暂时储存变量如果想要从外部传入data,就需要tf.placeholder(),然后用sess.run(paraA, feed_dict={})的形式传递数据import tensorflow.compat.v1 as tfimport tensorflowinput1 = tf.placeholder(tf.float32) # typeinput2 = tf.placeholder(tf.float32)output = tens原创 2021-07-02 10:50:54 · 317 阅读 · 0 评论 -
Variable变量
在 Tensorflow 中,定义了某字符串是变量,它才是变量,这一点是与 Python 所不同的。定义语法:state = tf.Variable(0, name=‘counter’),0是初始值在sess中激活变量sess起到类似指针的作用import tensorflow as tf# tensorflow 2.5.0state = tf.Variable(0, name='counter') # 变量# print(state.name)one = tf.constant(1)原创 2021-07-02 10:40:18 · 87 阅读 · 0 评论 -
Session会话控制
import tensorflow as tf# tensorflow 2.5.0# two methods for opening Sessionmatrix1 = tf.constant([[3, 3]])matrix2 = tf.constant([[2], [2]])product = tf.matmul(matrix1, matrix2) # matrix multiply np.dot(m1, m2)# method 1# sess原创 2021-07-02 10:28:13 · 55 阅读 · 0 评论 -
拟合例子1
import tensorflow as tfimport numpy as np# tensorflow 2.5.0# creat datax_data = np.random.rand(100).astype(np.float32)y_data = x_data*0.1 + 0.3### create tensorflow structure start ###Weights = tf.Variable(tf.random.uniform([1], -1.0, 1.0))biases原创 2021-07-02 09:59:45 · 68 阅读 · 0 评论 -
tensorflow基本概念与函数
请戳原创 2021-07-01 20:37:40 · 43 阅读 · 0 评论