tf.keras学习笔记（一) mnist分类

最新推荐文章于 2025-05-14 18:43:46 发布
原创 最新推荐文章于 2025-05-14 18:43:46 发布 · 1.1k 阅读 · 2 ·
CC 4.0 BY-SA版权
版权声明：本文为博主原创文章，遵循 CC 4.0 BY-SA 版权协议，转载请附上原文出处链接和本声明。

这篇博客是tf.keras学习系列的第一部分，主要介绍了如何使用tf.keras进行MNIST手写数字数据集的分类。通过导入keras库并展示简单的代码实现，博主展示了模型训练和运行的基本流程。

摘要生成于 C知道 ，由 DeepSeek-R1 满血版支持， 前往体验 >

tf.keras学习笔记（一）

  • mnist分类

导入keras的包

from tensorflow.contrib import keras

另外一些常见的数据集keras.datasets提供了
在这里插入图片描述

  • 代码：
import tensorflow as tf
# import keras form tf
from tensorflow.contrib import keras

# load mnist
mnist = keras.datasets.mnist
(x_train, y_train), (x_test, y_test) = mnist.load_data()
x_train, x_test = x_train / 255.0, x_test / 255.0

# build model
model = keras.models.Sequential()
model.add(keras.layers.Flatten(input_shape=(28,28)))
model.add(keras.layers.Dense(512, activation=tf.nn.leaky_relu))
model.add(keras.layers.Dropout(0.2))
model.add(keras.layers.Dense(10,activation=tf.nn.softmax))

# define optimizer and loss
model.compile(optimizer='adam',
              loss='sparse_categorical_crossentropy',
              metrics=['accuracy'])

# fit the training data
model.fit(x_train, y_train, epochs=5)

# test the test data
model.evaluate(x_test, y_test)
  • 运行结果：
/usr/bin/python3.6 /media/todd/38714CA0C89E958E/147/yl_tmp/readingbook/pipeline/gesture_reco/gesture_reco.py
Downloading data from https://storage.googleapis.com/tensorflow/tf-keras-datasets/mnist.npz
    8192/11490434 [..............................] - ETA: 0s
   57344/11490434 [..............................] - ETA: 28s
  163840/11490434 [..............................] - ETA: 13s
  221184/11490434 [..............................] - ETA: 16s
  442368/11490434 [>.............................] - ETA: 9s 
  573440/11490434 [>.............................] - ETA: 8s
  655360/11490434 [>.............................] - ETA: 8s
 1212416/11490434 [==>...........................] - ETA: 4s
 1572864/11490434 [===>..........................] - ETA: 3s
 1785856/11490434 [===>..........................] - ETA: 4s
 2490368/11490434 [=====>........................] - ETA: 2s
 3571712/11490434 [========>.....................] - ETA: 2s
 3588096/11490434 [========>.....................] - ETA: 2s
 3604480/11490434 [========>.....................] - ETA: 2s
 4145152/11490434 [=========>....................] - ETA: 2s
 4194304/11490434 [=========>....................] - ETA: 2s
 6619136/11490434 [================>.............] - ETA: 0s
 6635520/11490434 [================>.............] - ETA: 1s
 6684672/11490434 [================>.............] - ETA: 1s
 6832128/11490434 [================>.............] - ETA: 1s
 8028160/11490434 [===================>..........] - ETA: 0s
 9404416/11490434 [=======================>......] - ETA: 0s
 9977856/11490434 [=========================>....] - ETA: 0s
10092544/11490434 [=========================>....] - ETA: 0s
10797056/11490434 [===========================>..] - ETA: 0s
11493376/11490434 [==============================] - 2s 0us/step
Epoch 1/5
   32/60000 [..............................] - ETA: 2:50:32 - loss: 2.3859 - acc: 0.0938
  256/60000 [..............................] - ETA: 21:26 - loss: 1.9522 - acc: 0.3477  
  576/60000 [..............................] - ETA: 9:34 - loss: 1.4732 - acc: 0.5573 
  864/60000 [..............................] - ETA: 6:24 - loss: 1.2103 - acc: 0.6366
 1120/60000 [..............................] - ETA: 4:58 - loss: 1.0855 - acc: 0.6687
 1408/60000 [..............................] - ETA: 3:58 - loss: 0.9779 - acc: 0.6960
 1728/60000 [..............................] - ETA: 3:14 - loss: 0.8756 - acc: 0.7251
 2016/60000 [>.............................] - ETA: 2:47 - loss: 0.8299 - acc: 0.7391
 2272/60000 [>.............................] - ETA: 2:29 - loss: 0.7826 - acc: 0.7544
 2560/60000 [>.............................] - ETA: 2:13 - loss: 0.7444 - acc: 0.7680
 2880/60000 [>.............................] - ETA: 1:58 - loss: 0.7013 - acc: 0.7812
 3232/60000 [>.............................] - ETA: 1:46 - loss: 0.6679 - acc: 0.7915
 3552/60000 [>.............................] - ETA: 1:36 - loss: 0.6362 - acc: 0.8029
 3872/60000 [>.............................] - ETA: 1:29 - loss: 0.6127 - acc: 0.8102
 4160/60000 [=>............................] - ETA: 1:23 - loss: 0.5971 - acc: 0.8161
 4512/60000 [=>............................] - ETA: 1:16 - loss: 0.5778 - acc: 0.8225
 4800/60000 [=>............................] - ETA: 1:12 - loss: 0.5588 - acc: 0.8290
 5056/60000 [=>............................] - ETA: 1:09 - loss: 0.5496 - acc: 0.8319
 5376/60000 [=>............................] - ETA: 1:05 - loss: 0.5360 - acc: 0.8358
 5760/60000 [=>............................] - ETA: 1:00 - loss: 0.5275 - acc: 0.8398
 6080/60000 [==>...........................] - ETA: 57s - loss: 0.5181 - acc: 0.8426 
 6400/60000 [==>...........................] - ETA: 55s - loss: 0.5122 - acc: 0.8442
 6752/60000 [==>...........................] - ETA: 52s - loss: 0.5036 - acc: 0.8479
 7072/60000 [==>...........................] - ETA: 50s - loss: 0.4966 - acc: 0.8498
 7360/60000 [==>...........................] - ETA: 48s - loss: 0.4901 - acc: 0.8519
 7584/60000 [==>...........................] - ETA: 47s - loss: 0.4856 - acc: 0.8536
 7808/60000 [==>...........................] - ETA: 45s - loss: 0.4809 - acc: 0.8554
 8128/60000 [===>..........................] - ETA: 44s - loss: 0.4762 - acc: 0.8575
 8416/60000 [===>..........................] - ETA: 42s - loss: 0.4718 - acc: 0.8591
 8640/60000 [===>..........................] - ETA: 41s - loss: 0.4657 - acc: 0.8604
 8832/60000 [===>..........................] - ETA: 40s - loss: 0.4617 - acc: 0.8611
 9024/60000 [===>..........................] - ETA: 40s - loss: 0.4580 - acc: 0.8623
 9280/60000 [===>..........................] - ETA: 39s - loss: 0.4538 - acc: 0.8636
 9504/60000 [===>..........................] - ETA: 38s - loss: 0.4521 - acc: 0.8641
 9792/60000 [===>..........................] - ETA: 37s - loss: 0.4466 - acc: 0.8658
10016/60000 [====>.........................] - ETA: 36s - loss: 0.4443 - acc: 0.8663
10304/60000 [====>.........................] - ETA: 35s - loss: 0.4387 - acc: 0.8679
10560/60000 [====>.........................] - ETA: 34s - loss: 0.4344 - acc: 0.8697
10848/60000 [====>.........................] - ETA: 33s - loss: 0.4306 - acc: 0.8708
11168/60000 [====>.........................] - ETA: 32s - loss: 0.4278 - acc: 0.8718
11424/60000 [====>.........................] - ETA: 32s - loss: 0.4243 - acc: 0.8730
11680/60000 [====>.........................] - ETA: 31s - loss: 0.4214 - acc: 0.8736
11904/60000 [====>.........................] - ETA: 31s - loss: 0.4178 - acc: 0.8747
12128/60000 [=====>........................] - ETA: 30s - loss: 0.4165 - acc: 0.8756
12416/60000 [=====>........................] - ETA: 29s - loss: 0.4118 - acc: 0.8769
12640/60000 [=====>........................] - ETA: 29s - loss: 0.4084 - acc: 0.8777
12896/60000 [=====>........................] - ETA: 28s - loss: 0.4040 - acc: 0.8790
13216/60000 [=====>........................] - ETA: 28s - loss: 0.4011 - acc: 0.8801
13504/60000 [=====>........................] - ETA: 27s - loss: 0.3982 - acc: 0.8809
13728/60000 [=====>........................] - ETA: 27s - loss: 0.3977 - acc: 0.8813
13984/60000 [=====>........................] - ETA: 26s - loss: 0.3959 - acc: 0.8821
14240/60000 [======>.......................] - ETA: 26s - loss: 0.3956 - acc: 0.8822
14464/60000 [======>.......................] - ETA: 25s - loss: 0.3940 - acc: 0.8822
14720/60000 [======>.......................] - ETA: 25s - loss: 0.3923 - acc: 0.8827
14944/60000 [======>.......................] - ETA: 25s - loss: 0.3910 - acc: 0.8830
15200/60000 [======>.......................] - ETA: 24s - loss: 0.3881 - acc: 0.8838
15456/60000 [======>.......................] - ETA: 24s - loss: 0.3872 - acc: 0.8837
15680/60000 [======>.......................] - ETA: 24s - loss: 0.3862 - acc: 0.8842
15872/60000 [======>.......................] - ETA: 23s - loss: 0.3853 - acc: 0.8846
16128/60000 [=======>......................] - ETA: 23s - loss: 0.3836 - acc: 0.8852
16416/60000 [=======>......................] - ETA: 23s - loss: 0.3806 - acc: 0.8862
16608/60000 [=======>......................] - ETA: 22s - loss: 0.3793 - acc: 0.8866
16864/60000 [=======>......................] - ETA: 22s - loss: 0.3775 - acc: 0.8871
17120/60000 [=======>......................] - ETA: 22s - loss: 0.3757 - acc: 0.8876
17344/60000 [=======>......................] - ETA: 21s - loss: 0.3748 - acc: 0.8879
17600/60000 [=======>......................] - ETA: 21s - loss: 0.3732 - acc: 0.8884
17824/60000 [=======>......................] - ETA: 21s - loss: 0.3717 - acc: 0.8887
18112/60000 [========>.....................] - ETA: 20s - loss: 0.3704 - acc: 0.8891
18400/60000 [========>.....................] - ETA: 20s - loss: 0.3684 - acc: 0.8898
18656/60000 [========>.....................] - ETA: 20s - loss: 0.3676 - acc: 0.8899
18848/60000 [========>.....................] - ETA: 20s - loss: 0.3658 - acc: 0.8905
19072/60000 [========>.....................] - ETA: 19s - loss: 0.3642 - acc: 0.8911
19264/60000 [========>.....................] - ETA: 19s - loss: 0.3634 - acc: 0.8916
19488/60000 [========>.....................] - ETA: 19s - loss: 0.3619 - acc: 0.8919
19744/60000 [========>.....................] - ETA: 19s - loss: 0.3604 - acc: 0.8924
19936/60000 [========>.....................] - ETA: 19s - loss: 0.3592 - acc: 0.8928
20192/60000 [=========>....................] - ETA: 18s - loss: 0.3578 - acc: 0.8937
20416/60000 [=========>....................] - ETA: 18s - loss: 0.3556 - acc: 0.8942
20640/60000 [=========>....................] - ETA: 18s - loss: 0.3546 - acc: 0.8944
20928/60000 [=========>....................] - ETA: 18s - loss: 0.3536 - acc: 0.8947
21216/60000 [=========>....................] - ETA: 17s - loss: 0.3518 - acc: 0.8951
21472/60000 [=========>....................] - ETA: 17s - loss: 0.3498 - acc: 0.8957
21728/60000 [=========>....................] - ETA: 17s - loss: 0.3479 - acc: 0.8961
21984/60000 [=========>....................] - ETA: 17s - loss: 0.3465 - acc: 0.8964
22272/60000 [==========>...................] - ETA: 16s - loss: 0.3455 - acc: 0.8969
22528/60000 [==========>...................] - ETA: 16s - loss: 0.3443 - acc: 0.8974
22816/60000 [==========>...................] - ETA: 16s - loss: 0.3432 - acc: 0.8977
23072/60000 [==========>...................] - ETA: 16s - loss: 0.3420 - acc: 0.8980
23328/60000 [==========>...................] - ETA: 15s - loss: 0.3411 - acc: 0.8982
23584/60000 [==========>...................] - ETA: 15s - loss: 0.3390 - acc: 0.8989
23840/60000 [==========>...................] - ETA: 15s - loss: 0.3376 - acc: 0.8993
24096/60000 [===========>..................] - ETA: 15s - loss: 0.3367 - acc: 0.8997
24384/60000 [===========>..................] - ETA: 15s - loss: 0.3355 - acc: 0.9001
24640/60000 [===========>..................] - ETA: 14s - loss: 0.3342 - acc: 0.9004
24928/60000 [===========>..................] - ETA: 14s - loss: 0.3325 - acc: 0.9011
25184/60000 [===========>..................] - ETA: 14s - loss: 0.3315 - acc: 0.9013
25472/60000 [===========>..................] - ETA: 14s - loss: 0.3296 - acc: 0.9016
25728/60000 [===========>..................] - ETA: 14s - loss: 0.3288 - acc: 0.9018
25984/60000 [===========>..................] - ETA: 14s - loss: 0.3273 - acc: 0.9021
26240/60000 [============>.................] - ETA: 13s - loss: 0.3261 - acc: 0.9026
26496/60000 [============>.................] - ETA: 13s - loss: 0.3248 - acc: 0.9030
26688/60000 [============>.................] - ETA: 13s - loss: 0.3244 - acc: 0.9031
26912/60000 [============>.................] - ETA: 13s - loss: 0.3233 - acc: 0.9036
27168/60000 [============>.................] - ETA: 13s - loss: 0.3224 - acc: 0.9039
27488/60000 [============>.................] - ETA: 13s - loss: 0.3219 - acc: 0.9040
27712/60000 [============>.................] - ETA: 12s - loss: 0.3213 - acc: 0.9040
27968/60000 [============>.................] - ETA: 12s - loss: 0.3203 - acc: 0.9043
28224/60000 [=============>................] - ETA: 12s - loss: 0.3192 - acc: 0.9047
28512/60000 [=============>................] - ETA: 12s - loss: 0.3179 - acc: 0.9051
28768/60000 [=============>................] - ETA: 12s - loss: 0.3175 - acc: 0.9051
28960/60000 [=============>................] - ETA: 12s - loss: 0.3167 - acc: 0.9054
29216/60000 [=============>................] - ETA: 12s - loss: 0.3160 - acc: 0.9056
29472/60000 [=============>................] - ETA: 11s - loss: 0.3156 - acc: 0.9057
29664/60000 [=============>................] - ETA: 11s - loss: 0.3149 - acc: 0.9059
29920/60000 [=============>................] - ETA: 11s - loss: 0.3138 - acc: 0.9064
30176/60000 [==============>...............] - ETA: 11s - loss: 0.3125 - acc: 0.9067
30496/60000 [==============>...............] - ETA: 11s - loss: 0.3114 - acc: 0.9071
30720/60000 [==============>...............] - ETA: 11s - loss: 0.3113 - acc: 0.9072
30976/60000 [==============>...............] - ETA: 11s - loss: 0.3106 - acc: 0.9073
31232/60000 [==============>...............] - ETA: 10s - loss: 0.3103 - acc: 0.9075
31488/60000 [==============>...............] - ETA: 10s - loss: 0.3093 - acc: 0.9078
31712/60000 [==============>...............] - ETA: 10s - loss: 0.3083 - acc: 0.9081
31968/60000 [==============>...............] - ETA: 10s - loss: 0.3075 - acc: 0.9083
32224/60000 [===============>..............] - ETA: 10s - loss: 0.3067 - acc: 0.9085
32448/60000 [===============>..............] - ETA: 10s - loss: 0.3059 - acc: 0.9086
32672/60000 [===============>..............] - ETA: 10s - loss: 0.3053 - acc: 0.9087
32896/60000 [===============>..............] - ETA: 10s - loss: 0.3041 - acc: 0.9091
33152/60000 [===============>..............] - ETA: 9s - loss: 0.3039 - acc: 0.9091 
33408/60000 [===============>..............] - ETA: 9s - loss: 0.3029 - acc: 0.9095
33600/60000 [===============>..............] - ETA: 9s - loss: 0.3028 - acc: 0.9096
33792/60000 [===============>..............] - ETA: 9s - loss: 0.3021 - acc: 0.9098
34080/60000 [================>.............] - ETA: 9s - loss: 0.3011 - acc: 0.9099
34304/60000 [================>.............] - ETA: 9s - loss: 0.3003 - acc: 0.9101
34592/60000 [================>.............] - ETA: 9s - loss: 0.2996 - acc: 0.9104
34816/60000 [================>.............] - ETA: 9s - loss: 0.2990 - acc: 0.9106
35104/60000 [================>.............] - ETA: 8s - loss: 0.2987 - acc: 0.9107
35424/60000 [================>.............] - ETA: 8s - loss: 0.2982 - acc: 0.9108
35648/60000 [================>.............] - ETA: 8s - loss: 0
最低0.47元/天 解锁文章

200万优质内容无限畅学
MNIST数据集分类
qq_41005604的博客
03-31 523
import numpy as np from keras.datasets import mnist from keras.utils import np_utils from keras.models import Sequential from keras.layers import Dense from keras.optimizers import SGD # 载入数据 ''' (x_train【训练集的数据】,y_train【训练集的标签】),(x_test,y_test) = mnist.l
tensorflow学习笔记（使用 tf.data 加载 NumPy 数据）
DNDAW的博客
09-03 863
本文章是对tensorflow的理解 tensorflow学习笔记（使用 tf.data 加载 NumPy 数据 本教程提供了一个将数据从 NumPy 数组加载到 tf.data.Dataset 中的示例。 此示例从 .npz（ NumPy Zipped Data） 文件加载 MNIST 数据集。 导入模块 import numpy as np import tensorflow as tf import tensorflow_datasets as tfds 从 .npz 文件中加载 从网页DATA.
使用tf.keras搭建mnist手写数字识别网络
AI吃大瓜的博客
03-04 5302
使用tf.keras搭建mnist手写数字识别网络 目录 使用tf.keras搭建mnist手写数字识别网络 1.使用tf.keras.Sequential搭建序列模型 1.1tf.keras.Sequential模型 1.2搭建mnist手写数字识别网络（序列模型） 1.3完整的训练代码 2.构建高级模型 2.1 函数式 API 2.2搭建mnist手写数字识别网络（...
【机器学习实战】图像识别：使用mnist数据集训练一个图像分类模型
最新发布
qq_36453147的博客
05-14 1993
MNIST数据集是机器学习领域中用于手写数字图像分类的经典基准数据集，广泛应用于数字识别系统、机器学习算法研究、图像预处理技术研究、模型评估与比较基准以及作为计算机视觉系统的基础模块。
MNIST 数据集分类
qq_51544319的博客
10-14 1986
MNIST 数据集分类：构建简单的CNN对 mnist 数据集进行分类
(三)--MNIST数据集分类(简单版本)
qq_40147092的博客
10-15 337
[toc] MNIST数据集分类(简单版本)import tensorflow as tffrom tensorflow.examples.tutorials.mnist import input_data#载入数据集mnist=input_data.read_data_sets("MNIST_data",one_hot=Tru...
【Tensorflow2.x学习笔记】- Keras高层接口
qq_42433311的博客
01-03 2755
目录 1.Keras简介 2.常见功能模块 2.1 常见网络层类 2.2 网路容器 3.模型装配、训练、测试 3.1 模型装配 3.2 模型训练 3.3模型测试 4.模型的保存与加载 4.1 张量方式 4.2网络方式 4.3 SaveModel方式 5.自定义网络 5.1自定义网络层 5.2自定义网络 6.测量工具 7.可视化 7.1模型端 7.2浏览器端 1.Keras简介 Keras是一个主要由 Python语言开发的开源神经网络计...
使用tf.keras从零开始实现Softmax回归
大雄没有叮当猫的博客
08-28 767
公众号：数据挖掘与机器学习笔记 3.6 使用tf.keras从零开始实现Softmax回归   这一节我们来动手实现softmax回归。首先导入本节实现所需的包或者模块 %matplotlib inline import tensorflow as tf import matplotlib.pyplot as plt from IPython import display from tensorflow import keras import tensorflow.data as tfdata tf.e
学习笔记||tensorflow-过拟合及解决
Amber_15的博客
08-02 1126
model.compile(optimizer=tf.keras.optimizers.Adam(learning_rate=0.001), loss='categorical_crossentropy', metrics=['acc']) history = model.fit(train_image, train_label_onehot, epochs=10, val.
3.MNIST数据集分类
booze_的博客
07-05 5388
大多数示例使用手写数字的MNIST数据集。该数据集包含60,000个用于训练的示例和10,000个用于测试的示例。 每一张图片包含28*28个像素，在MNIST训练数据集中是一个形状为[60000,28,28]的张量，我们首先需要把数据集转成[60000,784],然后才能放到网络中训练。第一个维度数字用来索引图片，第二个维度数字用来索引每张图片中的像素点。一般我们还需要把图片中的数据归一化0~1之间。 MNIST数据集的标签是介于0-9的数字，我们要把标签转化为"one-hotvectors"。一个o
SVM实现MNIST数据集分类
01-19
本代码为matlab实现的支持向量机模式识别算法，对MNIST数据集进行三种样本分类，如果电脑内存够大，可以实现更多的类别分类，另外exclass是对随机的两个样本分类，也可以参考，exmuticlassall为对MNIST数据集分类，注释掉的代码也可以实现随即三样本分类并作图。
SVM算法对MNIST数据集分类
12-24
代码使用matlab编写，压缩包中包含MNIST数据集及其读取函数、SVM算法实现和ReadMe.txt。 SVM算法中使用了PCA降维处理数据减少运行时间，但是由于使用的SMO算法较为原始，运行速度仍然较慢。由于实现未经过任何优化，仅适合初学者使用，使用高斯核正确率在75%以上，有部分注释。请结合ReadMe文件使用。
使用knn对MNIST分类
01-13
有不少同学看到我的《Python代码实现简单的MNIST手写数字识别（适合初学者看）》博客，跟我要源代码和数据，还得邮箱一个一个回复，我直接放在资源里吧。另外还有根据knn原理写的代码，没有使用sklearn库，也上传在我的资源里了。
tensorflow.keras.datasets.mnist加载的mnist数据集
04-14
通过tensorflow.keras.datasets.mnist.load_data()导入数据集时，由于需要google的服务器，国内无法访问资源。
MNIST数据集分类&Dropout
qq_38149535的博客
12-14 486
import numpy as np from keras.datasets import mnist #MNIST数据集 from keras.utils import np_utils from keras.models import Sequential from keras.layers import Dense from keras.optimizers import SGD #载入...
TensorFlow（五）keras mnist分类
Abecedarian_CLF的博客
05-30 393
from keras.datasets import mnist from keras.utils import np_utils from keras.models import Sequential from keras.layers import Dense, Activation from keras.optimizers import RMSprop # download the mn...
第三章 mnist分类
张晓峰的博客
07-04 1016
分类实例
CNN简单应用-MNIST数字分类
qq_64334741的博客
11-28 636
MNIST数字分类项目旨在使用机器学习技术来构建一个模型，能够自动识别手写数字的图像。这个项目是一个经典的图像分类任务，常用于入门级机器学习和深度学习示例。我们会使用MNIST数据集，这个数据集包含了一系列28x28像素的手写数字图像，从0到9。项目的目标是训练一个模型，能够准确地将这些手写数字图像分类到正确的数字标签。
TensorFlow 2.1实战：MNIST图像分类入门
- 使用`tensorflow.keras.datasets.mnist.load_data()`可以从TensorFlow库中直接加载MNIST数据集。这将返回两个元组，分别包含训练集和测试集的图像和对应的标签。 - 训练集图像`train_images`是一个形状为`(60000...
评论
添加红包

请填写红包祝福语或标题

个

红包个数最小为10个

元

红包金额最低5元

当前余额3.43元 前往充值 >
需支付：10.00元
成就一亿技术人!
领取后你会自动成为博主和红包主的粉丝 规则
hope_wisdom
发出的红包
实付元
使用余额支付
点击重新获取
扫码支付
钱包余额 0

抵扣说明：

1.余额是钱包充值的虚拟货币，按照1:1的比例进行支付金额的抵扣。
2.余额无法直接购买下载，可以购买VIP、付费专栏及课程。

余额充值