小白初学tensorflow入手教程——持续更新

81 阅读 0 评论 54 点赞

我是靠谱客的博主孝顺黑米，最近开发中收集的这篇文章主要介绍小白初学tensorflow入手教程——持续更新，觉得挺不错的，现在分享给大家，希望可以做个参考。

概述

看过很多视频和教程，适用与windows和tf2.0版本的教程很少。。决定从tensorflow中文社区重新开始学习。但是也会有很多报错，把诸多修改后的小代码如下所示。
变量：

# -*- coding: utf-8 -*-
"""
Created on Sat Nov  2 10:59:18 2019

@author: lsf81
"""

import tensorflow as tf
state=tf.Variable(0,name="counter")
one=tf.constant(1)
new_value=tf.add(state,one)
update=tf.compat.v1.assign(state,new_value)
init_op=tf.compat.v1.global_variables_initializer()
with tf.compat.v1.Session() as sess:
    sess.run(init_op)
    print(sess.run(state))
    for _ in range(3):
        sess.run(update)
        print(sess.run(state))

运行结果：在这里插入代码片
注意：原版update = tf.assign(state, new_value)都更新成了update=tf.compat.v1.assign(state,new_value)
Fetch:

import tensorflow as tf
input1 = tf.constant(3.0)
input2 = tf.constant(2.0)
input3 = tf.constant(5.0)
intermed = tf.compat.v1.add(input2, input3)
mul = tf.multiply(input1, intermed)

with tf.compat.v1.Session() as sess:
  result = sess.run([mul, intermed])
  print(result)

必须要注意，在现在版本中，mul调用方式变成了multiply,and sub 变成了subtract。
Feed:


```python
import tensorflow as tf
input1 = tf.placeholder(tf.float32)
input2 = tf.placeholder(tf.float32)
output = tf.multiply(input1, input2)

with tf.compat.v1.Session() as sess:
  print(sess.run([output],feed_dict={input1:[7.],input2:[2.]}))

注意：官网上是input1 = tf.placeholder(tf.types.float32)，其中types是多余的。

mnist详细版教程
https://blog.csdn.net/cqrtxwd/article/details/79028264

input_data.py下载地址
https://blog.csdn.net/weixin_43159628/article/details/83241345

搞了一下午：最新更新如下：
出现Please use alternatives such as official/mnist/dataset.py from tensorflow/models.的错误。原因是新版本不再支持tensorflow/example/tutorials函数库了，全部更新到keras里面了，具体代码如下

import tensorflow as tf
#from tensorflow.examples.tutorials.mnist import input_data

#mnist=input_data.read_data_sets('./MNIST_data/',one_hot=True)
#sess=tf.InteractiveSession()
#print('Training data size: ', mnist.train.num_examples)
mnist=tf.keras.datasets.mnist
(X_train, y_train), (X_test, y_test) = mnist.load_data()
print(X_train.shape) # out: (60000, 28, 28)
print(y_train.shape) # out: (60000,)

参考博文：https://blog.csdn.net/u011106767/article/details/93879120
后续问题遇到了keras中mnist分类的问题。见以下博文：
https://blog.csdn.net/Yumi_huang/article/details/82351173

参考上面的博文：

在这里插入代# -*- coding: utf-8 -*-
"""
Created on Sat Nov  2 18:56:47 2019

@author: lsf81
"""
import numpy as np
import matplotlib.pyplot as plt

path = r"F:pythonAnacondaLibsite-packageskeras_applicationsexamplesmnist.npz"
f = np.load(path)
x_train, y_train = f['x_train'], f['y_train']
x_test, y_test = f['x_test'], f['y_test']
f.close()
print(x_train.shape)
print(x_test.shape)
for i in range(9):
    plt.subplot(3,3,i+1)
    plt.imshow(x_train[i], cmap='gray', interpolation='none')
    plt.title("Class {}".format(y_train[i]))
plt.show()


from keras.datasets import mnist
from keras.models import Sequential
from keras.layers.core import Dense, Activation, Dropout
from keras.utils import np_utils
import numpy as np
import matplotlib.pyplot as plt
from keras.optimizers import RMSprop

path = r"F:pythonAnacondaLibsite-packageskeras_applicationsexamplesmnist.npz"
f = np.load(path)

x_train, y_train = f['x_train'], f['y_train']
x_test, y_test = f['x_test'], f['y_test']
f.close()
# print(x_train.shape)
# print(x_test.shape)
# for i in range(9):
#     plt.subplot(3,3,i+1)
#     plt.imshow(x_train[i], cmap='gray', interpolation='none')
#     plt.title("Class {}".format(y_train[i]))
# plt.show()

#将二维数据变为一维
X_train = x_train.reshape(len(x_train), -1)
X_test = x_test.reshape(len(x_test), -1)
X_train = X_train.astype('float32')
X_test = X_test.astype('float32')
# Normalization.scaling it so that all values are in the [0, 1] interval. 
X_train = (X_train - 127) / 127
X_test = (X_test - 127) / 127

#one hot encoding
y_train = np_utils.to_categorical(y_train, num_classes=10)
y_test = np_utils.to_categorical(y_test, num_classes=10)
model = Sequential([
    Dense(512, input_dim=784),
    Activation('relu'),
    Dropout(0.2),
    # Dense(512),
    # Activation('relu'),
    # Dropout(0.2),
    Dense(10),
    Activation('softmax'),

])
rmsprop = RMSprop(lr=0.001, rho=0.9, epsilon=1e-08, decay=0.0)

# We add metrics to get more results you want to see
model.compile(optimizer=rmsprop,
              loss='categorical_crossentropy',
              metrics=['accuracy'])

print('Training ------------')
# Another way to train the model
model.fit(X_train, y_train, epochs=20, batch_size=100)

print('nTesting ------------')
# Evaluate the model with the metrics we defined earlier
loss, accuracy = model.evaluate(X_test, y_test)

print('test loss: ', loss)
print('test accuracy: ', accuracy)
码片

最后的结果为：

runfile('E:/Users/lsf81/Desktop/cnnmnist2.py', wdir='E:/Users/lsf81/Desktop')
(60000, 28, 28)
(10000, 28, 28)


Using TensorFlow backend.
WARNING:tensorflow:From F:pythonAnacondaenvstensorflowlibsite-packageskerasbackendtensorflow_backend.py:74: The name tf.get_default_graph is deprecated. Please use tf.compat.v1.get_default_graph instead.

WARNING:tensorflow:From F:pythonAnacondaenvstensorflowlibsite-packageskerasbackendtensorflow_backend.py:517: The name tf.placeholder is deprecated. Please use tf.compat.v1.placeholder instead.

WARNING:tensorflow:From F:pythonAnacondaenvstensorflowlibsite-packageskerasbackendtensorflow_backend.py:4138: The name tf.random_uniform is deprecated. Please use tf.random.uniform instead.

WARNING:tensorflow:From F:pythonAnacondaenvstensorflowlibsite-packageskerasbackendtensorflow_backend.py:133: The name tf.placeholder_with_default is deprecated. Please use tf.compat.v1.placeholder_with_default instead.

WARNING:tensorflow:From F:pythonAnacondaenvstensorflowlibsite-packageskerasbackendtensorflow_backend.py:3445: calling dropout (from tensorflow.python.ops.nn_ops) with keep_prob is deprecated and will be removed in a future version.
Instructions for updating:
Please use `rate` instead of `keep_prob`. Rate should be set to `rate = 1 - keep_prob`.
WARNING:tensorflow:From F:pythonAnacondaenvstensorflowlibsite-packageskerasoptimizers.py:790: The name tf.train.Optimizer is deprecated. Please use tf.compat.v1.train.Optimizer instead.

WARNING:tensorflow:From F:pythonAnacondaenvstensorflowlibsite-packageskerasbackendtensorflow_backend.py:3295: The name tf.log is deprecated. Please use tf.math.log instead.

Training ------------
WARNING:tensorflow:From F:pythonAnacondaenvstensorflowlibsite-packagestensorflowpythonopsmath_grad.py:1250: add_dispatch_support.<locals>.wrapper (from tensorflow.python.ops.array_ops) is deprecated and will be removed in a future version.
Instructions for updating:
Use tf.where in 2.0, which has the same broadcast rule as np.where
Epoch 1/20
60000/60000 [==============================] - 5s 79us/step - loss: 0.4220 - acc: 0.8750
Epoch 2/20
60000/60000 [==============================] - 3s 58us/step - loss: 0.1797 - acc: 0.9443
Epoch 3/20
60000/60000 [==============================] - 5s 80us/step - loss: 0.1379 - acc: 0.9586
Epoch 4/20
60000/60000 [==============================] - 9s 146us/step - loss: 0.1224 - acc: 0.9636
Epoch 5/20
60000/60000 [==============================] - 9s 155us/step - loss: 0.1066 - acc: 0.9683
Epoch 6/20
60000/60000 [==============================] - 9s 148us/step - loss: 0.1005 - acc: 0.9701
Epoch 7/20
60000/60000 [==============================] - 9s 154us/step - loss: 0.0929 - acc: 0.9728
Epoch 8/20
60000/60000 [==============================] - 9s 155us/step - loss: 0.0866 - acc: 0.9754
Epoch 9/20
60000/60000 [==============================] - 9s 145us/step - loss: 0.0829 - acc: 0.9761
Epoch 10/20
60000/60000 [==============================] - 9s 154us/step - loss: 0.0766 - acc: 0.9773
Epoch 11/20
60000/60000 [==============================] - 9s 147us/step - loss: 0.0740 - acc: 0.9789
Epoch 12/20
60000/60000 [==============================] - 9s 153us/step - loss: 0.0683 - acc: 0.9814
Epoch 13/20
60000/60000 [==============================] - 7s 118us/step - loss: 0.0698 - acc: 0.9808
Epoch 14/20
60000/60000 [==============================] - 10s 158us/step - loss: 0.0652 - acc: 0.9823
Epoch 15/20
60000/60000 [==============================] - 9s 153us/step - loss: 0.0637 - acc: 0.9830
Epoch 16/20
60000/60000 [==============================] - 9s 144us/step - loss: 0.0634 - acc: 0.9831
Epoch 17/20
60000/60000 [==============================] - 9s 153us/step - loss: 0.0605 - acc: 0.9836
Epoch 18/20
60000/60000 [==============================] - 8s 140us/step - loss: 0.0559 - acc: 0.9845
Epoch 19/20
60000/60000 [==============================] - 9s 149us/step - loss: 0.0569 - acc: 0.9848
Epoch 20/20
60000/60000 [==============================] - 7s 118us/step - loss: 0.0552 - acc: 0.9859

Testing ------------
10000/10000 [==============================] - 1s 63us/step
test loss:  0.11286064563859545
test accuracy:  0.9791

刚刚试着修改一下那些警告，升完级后发现不能运行了。又把版本改回来了，暂时先这样。自己的第一个程序。
上面mnist博客中的程序，已经跑完，代码如下

import input_data
import tensorflow as tf
#读取数据
mnist = input_data.read_data_sets('MNIST_data', one_hot=True)
sess=tf.compat.v1.InteractiveSession()
#构建cnn网络结构
#自定义卷积函数（后面卷积时就不用写太多）
def conv2d(x,w):
 return tf.nn.conv2d(x,w,strides=[1,1,1,1],padding='SAME')
#自定义池化函数
def max_pool_2x2(x):
 return tf.nn.max_pool2d(x,ksize=[1,2,2,1],strides=[1,2,2,1],padding='SAME')
#设置占位符，尺寸为样本输入和输出的尺寸
x=tf.compat.v1.placeholder(tf.float32,[None,784])
y_=tf.placeholder(tf.float32,[None,10])
x_img=tf.reshape(x,[-1,28,28,1])

#设置第一个卷积层和池化层
w_conv1=tf.Variable(tf.random.truncated_normal([3,3,1,32],stddev=0.1))
b_conv1=tf.Variable(tf.constant(0.1,shape=[32]))
h_conv1=tf.nn.relu(conv2d(x_img,w_conv1)+b_conv1)
h_pool1=max_pool_2x2(h_conv1)

#设置第二个卷积层和池化层
w_conv2=tf.Variable(tf.truncated_normal([3,3,32,50],stddev=0.1))
b_conv2=tf.Variable(tf.constant(0.1,shape=[50]))
h_conv2=tf.nn.relu(conv2d(h_pool1,w_conv2)+b_conv2)
h_pool2=max_pool_2x2(h_conv2)

#设置第一个全连接层
w_fc1=tf.Variable(tf.truncated_normal([7*7*50,1024],stddev=0.1))
b_fc1=tf.Variable(tf.constant(0.1,shape=[1024]))
h_pool2_flat=tf.reshape(h_pool2,[-1,7*7*50])
h_fc1=tf.nn.relu(tf.matmul(h_pool2_flat,w_fc1)+b_fc1)

#dropout（随机权重失活）
keep_prob=tf.placeholder(tf.float32)
h_fc1_drop=tf.nn.dropout(h_fc1,keep_prob)

#设置第二个全连接层
w_fc2=tf.Variable(tf.truncated_normal([1024,10],stddev=0.1))
b_fc2=tf.Variable(tf.constant(0.1,shape=[10]))
y_out=tf.nn.softmax(tf.matmul(h_fc1_drop,w_fc2)+b_fc2)

#建立loss function，为交叉熵
loss=tf.reduce_mean(-tf.reduce_sum(y_*tf.math.log(y_out),reduction_indices=[1]))
#配置Adam优化器，学习速率为1e-4
train_step=tf.compat.v1.train.AdamOptimizer(1e-4).minimize(loss)

#建立正确率计算表达式
correct_prediction=tf.equal(tf.argmax(y_out,1),tf.argmax(y_,1))
accuracy=tf.reduce_mean(tf.cast(correct_prediction,tf.float32))

#开始喂数据，训练
tf.compat.v1.global_variables_initializer().run()
for i in range(20000):
    batch=mnist.train.next_batch(50)
    if i%100==0:
        train_accuracy=accuracy.eval(feed_dict={x:batch[0],y_:batch[1],keep_prob:1})
        print("step %d,train_accuracy= %g"%(i,train_accuracy))
    train_step.run(feed_dict={x:batch[0],y_:batch[1],keep_prob:0.5})

#训练之后，使用测试集进行测试，输出最终结果
print("test_accuracy= %g"%accuracy.eval(feed_dict={x:mnist.test.images,y_:mnist.test.labels,keep_prob:1}))

结果为

F:pythonpycharmvenvScriptspython.exe F:/python/code/cnntest1.py
Successfully downloaded train-images-idx3-ubyte.gz 9912422 bytes.
Extracting MNIST_datatrain-images-idx3-ubyte.gz
Successfully downloaded train-labels-idx1-ubyte.gz 28881 bytes.
Extracting MNIST_datatrain-labels-idx1-ubyte.gz
Successfully downloaded t10k-images-idx3-ubyte.gz 1648877 bytes.
Extracting MNIST_datat10k-images-idx3-ubyte.gz
Successfully downloaded t10k-labels-idx1-ubyte.gz 4542 bytes.
Extracting MNIST_datat10k-labels-idx1-ubyte.gz
WARNING:tensorflow:From F:/python/code/cnntest1.py:5: The name tf.InteractiveSession is deprecated. Please use tf.compat.v1.InteractiveSession instead.

2019-11-05 09:27:27.623410: I tensorflow/core/platform/cpu_feature_guard.cc:142] Your CPU supports instructions that this TensorFlow binary was not compiled to use: AVX2
WARNING:tensorflow:From F:/python/code/cnntest1.py:14: The name tf.placeholder is deprecated. Please use tf.compat.v1.placeholder instead.

WARNING:tensorflow:From F:/python/code/cnntest1.py:19: The name tf.truncated_normal is deprecated. Please use tf.random.truncated_normal instead.

WARNING:tensorflow:From F:/python/code/cnntest1.py:12: The name tf.nn.max_pool is deprecated. Please use tf.nn.max_pool2d instead.

WARNING:tensorflow:From F:/python/code/cnntest1.py:38: calling dropout (from tensorflow.python.ops.nn_ops) with keep_prob is deprecated and will be removed in a future version.
Instructions for updating:
Please use `rate` instead of `keep_prob`. Rate should be set to `rate = 1 - keep_prob`.
WARNING:tensorflow:From F:/python/code/cnntest1.py:46: The name tf.log is deprecated. Please use tf.math.log instead.

WARNING:tensorflow:From F:/python/code/cnntest1.py:48: The name tf.train.AdamOptimizer is deprecated. Please use tf.compat.v1.train.AdamOptimizer instead.

WARNING:tensorflow:From F:/python/code/cnntest1.py:55: The name tf.global_variables_initializer is deprecated. Please use tf.compat.v1.global_variables_initializer instead.

step 0,train_accuracy= 0.08
step 100,train_accuracy= 0.84
step 200,train_accuracy= 0.92
step 300,train_accuracy= 0.84
step 400,train_accuracy= 0.94
step 500,train_accuracy= 0.84
step 600,train_accuracy= 1
step 700,train_accuracy= 0.94
step 800,train_accuracy= 0.9
step 900,train_accuracy= 1
step 1000,train_accuracy= 0.96
step 1100,train_accuracy= 0.96
step 1200,train_accuracy= 0.96
step 1300,train_accuracy= 0.94
step 1400,train_accuracy= 0.94
step 1500,train_accuracy= 0.96
step 1600,train_accuracy= 0.96
step 1700,train_accuracy= 0.96
step 1800,train_accuracy= 0.98
step 1900,train_accuracy= 0.96
step 2000,train_accuracy= 0.98
step 2100,train_accuracy= 1
step 2200,train_accuracy= 0.96
step 2300,train_accuracy= 1
step 2400,train_accuracy= 1
step 2500,train_accuracy= 0.96
step 2600,train_accuracy= 0.98
step 2700,train_accuracy= 0.98
step 2800,train_accuracy= 0.94
step 2900,train_accuracy= 0.96
step 3000,train_accuracy= 0.98
step 3100,train_accuracy= 0.98
step 3200,train_accuracy= 0.94
step 3300,train_accuracy= 1
step 3400,train_accuracy= 1
step 3500,train_accuracy= 1
step 3600,train_accuracy= 0.96
step 3700,train_accuracy= 1
step 3800,train_accuracy= 0.98
step 3900,train_accuracy= 0.96
step 4000,train_accuracy= 0.98
step 4100,train_accuracy= 0.96
step 4200,train_accuracy= 0.96
step 4300,train_accuracy= 0.98
step 4400,train_accuracy= 1
step 4500,train_accuracy= 0.98
step 4600,train_accuracy= 0.96
step 4700,train_accuracy= 0.96
step 4800,train_accuracy= 0.92
step 4900,train_accuracy= 0.96
step 5000,train_accuracy= 1
step 5100,train_accuracy= 1
step 5200,train_accuracy= 1
step 5300,train_accuracy= 1
step 5400,train_accuracy= 1
step 5500,train_accuracy= 0.98
step 5600,train_accuracy= 1
step 5700,train_accuracy= 0.98
step 5800,train_accuracy= 0.98
step 5900,train_accuracy= 1
step 6000,train_accuracy= 1
step 6100,train_accuracy= 1
step 6200,train_accuracy= 1
step 6300,train_accuracy= 1
step 6400,train_accuracy= 0.96
step 6500,train_accuracy= 0.98
step 6600,train_accuracy= 0.94
step 6700,train_accuracy= 1
step 6800,train_accuracy= 1
step 6900,train_accuracy= 1
step 7000,train_accuracy= 1
step 7100,train_accuracy= 0.98
step 7200,train_accuracy= 1
step 7300,train_accuracy= 1
step 7400,train_accuracy= 1
step 7500,train_accuracy= 1
step 7600,train_accuracy= 1
step 7700,train_accuracy= 1
step 7800,train_accuracy= 1
step 7900,train_accuracy= 1
step 8000,train_accuracy= 0.98
step 8100,train_accuracy= 1
step 8200,train_accuracy= 0.96
step 8300,train_accuracy= 0.98
step 8400,train_accuracy= 1
step 8500,train_accuracy= 1
step 8600,train_accuracy= 0.98
step 8700,train_accuracy= 1
step 8800,train_accuracy= 1
step 8900,train_accuracy= 1
step 9000,train_accuracy= 1
step 9100,train_accuracy= 1
step 9200,train_accuracy= 1
step 9300,train_accuracy= 1
step 9400,train_accuracy= 1
step 9500,train_accuracy= 1
step 9600,train_accuracy= 1
step 9700,train_accuracy= 0.98
step 9800,train_accuracy= 1
step 9900,train_accuracy= 0.98
step 10000,train_accuracy= 0.98
step 10100,train_accuracy= 1
step 10200,train_accuracy= 0.96
step 10300,train_accuracy= 1
step 10400,train_accuracy= 1
step 10500,train_accuracy= 1
step 10600,train_accuracy= 0.98
step 10700,train_accuracy= 1
step 10800,train_accuracy= 1
step 10900,train_accuracy= 0.98
step 11000,train_accuracy= 0.98
step 11100,train_accuracy= 1
step 11200,train_accuracy= 1
step 11300,train_accuracy= 0.98
step 11400,train_accuracy= 1
step 11500,train_accuracy= 0.98
step 11600,train_accuracy= 1
step 11700,train_accuracy= 0.98
step 11800,train_accuracy= 1
step 11900,train_accuracy= 0.98
step 12000,train_accuracy= 1
step 12100,train_accuracy= 1
step 12200,train_accuracy= 0.96
step 12300,train_accuracy= 0.98
step 12400,train_accuracy= 1
step 12500,train_accuracy= 1
step 12600,train_accuracy= 1
step 12700,train_accuracy= 1
step 12800,train_accuracy= 1
step 12900,train_accuracy= 1
step 13000,train_accuracy= 1
step 13100,train_accuracy= 1
step 13200,train_accuracy= 1
step 13300,train_accuracy= 1
step 13400,train_accuracy= 1
step 13500,train_accuracy= 1
step 13600,train_accuracy= 1
step 13700,train_accuracy= 1
step 13800,train_accuracy= 1
step 13900,train_accuracy= 1
step 14000,train_accuracy= 0.98
step 14100,train_accuracy= 1
step 14200,train_accuracy= 1
step 14300,train_accuracy= 0.98
step 14400,train_accuracy= 1
step 14500,train_accuracy= 1
step 14600,train_accuracy= 1
step 14700,train_accuracy= 1
step 14800,train_accuracy= 0.98
step 14900,train_accuracy= 1
step 15000,train_accuracy= 1
step 15100,train_accuracy= 1
step 15200,train_accuracy= 1
step 15300,train_accuracy= 1
step 15400,train_accuracy= 1
step 15500,train_accuracy= 1
step 15600,train_accuracy= 1
step 15700,train_accuracy= 1
step 15800,train_accuracy= 1
step 15900,train_accuracy= 1
step 16000,train_accuracy= 1
step 16100,train_accuracy= 1
step 16200,train_accuracy= 1
step 16300,train_accuracy= 1
step 16400,train_accuracy= 1
step 16500,train_accuracy= 1
step 16600,train_accuracy= 1
step 16700,train_accuracy= 1
step 16800,train_accuracy= 1
step 16900,train_accuracy= 1
step 17000,train_accuracy= 1
step 17100,train_accuracy= 1
step 17200,train_accuracy= 1
step 17300,train_accuracy= 1
step 17400,train_accuracy= 1
step 17500,train_accuracy= 1
step 17600,train_accuracy= 1
step 17700,train_accuracy= 1
step 17800,train_accuracy= 0.98
step 17900,train_accuracy= 1
step 18000,train_accuracy= 1
step 18100,train_accuracy= 1
step 18200,train_accuracy= 1
step 18300,train_accuracy= 1
step 18400,train_accuracy= 1
step 18500,train_accuracy= 1
step 18600,train_accuracy= 1
step 18700,train_accuracy= 1
step 18800,train_accuracy= 1
step 18900,train_accuracy= 1
step 19000,train_accuracy= 1
step 19100,train_accuracy= 1
step 19200,train_accuracy= 1
step 19300,train_accuracy= 1
step 19400,train_accuracy= 1
step 19500,train_accuracy= 1
step 19600,train_accuracy= 1
step 19700,train_accuracy= 1
step 19800,train_accuracy= 1
step 19900,train_accuracy= 1
2019-11-05 09:41:01.897069: W tensorflow/core/framework/allocator.cc:107] Allocation of 1003520000 exceeds 10% of system memory.
2019-11-05 09:41:02.535947: W tensorflow/core/framework/allocator.cc:107] Allocation of 250880000 exceeds 10% of system memory.
2019-11-05 09:41:02.763879: W tensorflow/core/framework/allocator.cc:107] Allocation of 392000000 exceeds 10% of system memory.
test_accuracy= 0.9908

Process finished with exit code 0