MNIST全连接实现（keras）

keras全连接网络实现(Sequential)

一 加载keras中的MNIST数据集

from keras.datasets import mnist

Using TensorFlow backend.

(train_images,train_labels),(test_images,test_labels)=mnist.load_data()

A local file was found, but it seems to be incomplete or outdated because the auto file hash does not match the original value of 8a61469f7ea1b51cbae51d4f78837e45 so we will re-download the data.
Downloading data from https://s3.amazonaws.com/img-datasets/mnist.npz
11493376/11490434 [==============================] - 700s 61us/step

train_images.shape

(60000, 28, 28)

len(train_labels)

60000

train_labels

array([5, 0, 4, ..., 5, 6, 8], dtype=uint8)

test_images.shape

(10000, 28, 28)

len(test_labels)

10000

test_labels

array([7, 2, 1, ..., 4, 5, 6], dtype=uint8)

二 网络架构

from keras import models #模型
from keras import layers #层

network = models.Sequential()
network.add(layers.Dense(512,activation='relu',input_shape=(28*28,))) #添加全连接层
network.add(layers.Dense(10,activation='softmax'))

三 编译步骤

network.compile(optimizer='rmsprop', #优化器-随机梯度下降的某个变体
               loss='categorical_crossentropy',#损失函数-分类交叉熵损失函数
               metrics=['accuracy']) #监控的指标

四 准备图像数据

train_images = train_images.reshape((60000,28*28))
train_images = train_images.astype('float32')/255 #归一化

test_images = test_images.reshape((10000,28*28))
test_images = test_images.astype('float32')/255 #归一化（便于计算）

准备标签

from keras.utils import to_categorical

train_labels = to_categorical(train_labels)
test_labels = to_categorical(test_labels)

开始训练（fit）

network.fit(train_images,train_labels,epochs=5,batch_size=128)

Epoch 1/5
60000/60000 [==============================] - 4s 66us/step - loss: 0.2578 - acc: 0.9251
Epoch 2/5
60000/60000 [==============================] - 4s 66us/step - loss: 0.1039 - acc: 0.9691
Epoch 3/5
60000/60000 [==============================] - 4s 67us/step - loss: 0.0683 - acc: 0.9795
Epoch 4/5
60000/60000 [==============================] - 4s 67us/step - loss: 0.0498 - acc: 0.9849
Epoch 5/5
60000/60000 [==============================] - 4s 67us/step - loss: 0.0374 - acc: 0.9884





<keras.callbacks.History at 0x7f596a807dd8>

开始测试

test_loss,test_acc = network.evaluate(test_images,test_labels)
print('test_acc:',test_acc)

10000/10000 [==============================] - 0s 41us/step
test_acc: 0.9798