tensorflow学习笔记

fashion_mnist训练集

切分训练集和测试集

(train_image,train_label),(test_image,test_label) = tf.keras.datasets.fashion_mnist.load_data()

归一化数据集

train_image = train_image/255
test_image=test_image/255

建立网络模型

model = tf.keras.Sequential()
model.add(tf.keras.layers.Flatten(input_shape = (28,28)))
model.add(tf.keras.layers.Dense(128,activation='relu'))
model.add(tf.keras.layers.Dense(10,activation='softmax'))
model.compile(optimizer='adam',loss='sparse_categorical_crossentropy',metrics=['acc'])
##也可以自定义优化器的学习速率
##model.compile(optimizer=tf.keras.optimizers.Adam(learning_rate=0.001),loss='sparse_categorical_crossentropy',metrics=['acc'])

训练网络，加入验证集

history = model.fit(train_image,train_label,epochs=10,validation_data=(test_image,test_label))
##history用来记录历史数据

评估测试集

model.evaluate(test_image,test_label)

预测单张图片

import numpy as np
img = test_image[0]
img = np.expand_dims(img,0)##升维至（1，28，28）
predict = model.predict(img)
##predict为array([[3.9570288e-08, 1.9165432e-07, 2.3744233e-09, 5.4760602e-11,
##       2.4855660e-09, 9.2416611e-03, 8.0690410e-07, 1.2171428e-02,
##       2.7652545e-09, 9.7858584e-01]], dtype=float32)
np.argmax(predict)
##获得概率最大的一位，结果为9

画图比较loss和val_loss,accuracy和val_accuracy

import matplotlib.pyplot as plt
plt.plot(history.epoch,history.history.get('loss'),label = 'loss')
plt.plot(history.epoch,history.history.get('val_loss'),label='val_loss')
plt.legend()##增加标签图例