SparkSnail

调用LSTM进行MNIST手写数字识别。import tensorflow as tffrom tensorflow.examples.tutorials.mnist import input_datafrom tensorflow.contrib import rnnmnist = input_data.read_data_sets('MNIST_data', one_hot=Tr

import tensorflow as tfv1 = tf.Variable(0, dtype = tf.float32)#定义变量用来计算滑动平均，初始化为0，shadow_variable = 0step = tf.Variable(0, trainable = False)#定义滑动平均的step，初始化为0#定义滑动平均的类，初始化时给定衰减率0.99,step为0ema =

模拟非线性回归，给定一些二维点，y = x^2 + noise，用梯度下降进行训练，实线前向传播神经网络。import tensorflow as tfimport numpy as npimport matplotlib.pyplot as plt#生成随机变量x_data = np.linspace(-1, 1, 200)#生成200个随机点，范围为-1 --> 1x

1.reduce_mean(input_tensor, reduction_indices=None, keep_dims=False, name=None)功能：求平均值参数：input_tensor：输入reduction_indices:指定维度2.clip_by_value()功能：将一个张量中的数值限制在一个范围内例：v = tf.constant([1.

实现一个简单的MNIST手写数字识别，不带隐藏层的前向传播神经网络，使用梯度下降进行训练。准确率0.92import tensorflow as tffrom tensorflow.examples.tutorials.mnist import input_datamnist = input_data.read_data_sets('MNIST_data', one_hot=Tru

1.加入了两个隐层2.学习率衰减3.加入反向传播4.dropout防止过拟合准确率0.98import tensorflow as tffrom tensorflow.examples.tutorials.mnist import input_datamnist = input_data.read_data_sets('MNIST_data', one_hot=True

使用两个卷积层，两个全连接层。import tensorflow as tffrom tensorflow.examples.tutorials.mnist import input_datamnist = input_data.read_data_sets('MNIST_data', one_hot=True)batch_size = 100 #每个批次大小n_batch = m