Tensorflow2.0学习笔记-mnist手写数据集训练

mnist手写数据集训练

mnist数据集：
　　提供了6万张28 * 28像素点的0~9的手写数字图片，用于测试
　　提供了1万张28 * 28像素点的0~9的手写数字图片，用于测试
导入mnist数据集：
　　mnist = tf.keras.datasets.mnist
　　(x_train, t_train), (x_test, y_test) = mnisti.load_data()
作为输入特征，输入神经网络时，将数据拉直成了一维数组：
　　tf.keras.layers.Flatten()
训练代码如下：

1. 使用顺序搭建

import tensorflow as tf

#  使用pycharm调用mnisi数据集时，会自动下载
mnist = tf.keras.datasets.mnist
(x_train, y_train), (x_test, y_test) = mnist.load_data()
#对输入特征进行归一化，使原本为0~255的灰度值，变成0~1之间的数字
x_train, x_test = x_train / 255.0, x_test / 255.0

model = tf.keras.models.Sequential([
    tf.keras.layers.Flatten(), #  拉直为一维数组
    tf.keras.layers.Dense(126, activation='relu'), #  定义第一层网络有126个神经元，使用激活函数为relu
    tf.keras.layers.Dense(10, activation='softmax')  #  定义第二层网络有10个神经元，使输出为符合概率分布。
])
#  定义的神经元个数会对最后输出的正确率有影响
model.compile(optimizer='adam',  #  使用adam优化器
              loss=tf.keras.losses.SparseCategoricalCrossentropy(from_logits=False),  #  输出不是直接输出所以为False
              metrics=['sparse_categorical_accuracy'])  #  数据集中的标签是数值，输出为概率分布。

model.fit(x_train, y_train, batch_size=32, epochs=5, validation_data=(x_test, y_test), validation_freq=1)
model.summary()

2. 使用类搭建

import tensorflow as tf
from tensorflow.keras.layers import Dense, Flatten
from tensorflow.keras import Model

mnist = tf.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


class MnistModel(Model):
    def __init__(self):
        super(MnistModel, self).__init__()
        self.flatten = Flatten()
        self.d1 = Dense(126, activation='relu')
        self.d2 = Dense(10, activation='softmax')

    def call(self, x):
        x = self.flatten(x)
        x = self.d1(x)
        y = self.d2(x)
        return y


model = MnistModel()

model.compile(optimizer='adam',
              loss=tf.keras.losses.SparseCategoricalCrossentropy(from_logits=False),
              metrics=['sparse_categorical_accuracy'])

model.fit(x_train, y_train, batch_size=32, epochs=5, validation_data=(x_test, y_test), validation_freq=1)
model.summary()

3. 输出结果：

.......
60000/60000 [==============================] - 9s 143us/sample - loss: 0.0454 - sparse_categorical_accuracy: 0.9860 - val_loss: 0.0738 - val_sparse_categorical_accuracy: 0.9776
Model: "sequential"
_________________________________________________________________
Layer (type)                 Output Shape              Param #   
=================================================================
flatten (Flatten)            multiple                  0         
_________________________________________________________________
dense (Dense)                multiple                  98910     
_________________________________________________________________
dense_1 (Dense)              multiple                  1270      
=================================================================
Total params: 100,180
Trainable params: 100,180
Non-trainable params: 0
_________________________________________________________________

Process finished with exit code 0