二、构建CNN网络模型三、编译模型四、训练模型五、预测六、模型评估

前期工作

tf.config.list_physical_devices(GPU)if

gpus:gpu0

#如果有多个GPU仅使用第0个GPUtf.config.experimental.set_memory_growth(gpu0,

True)

#设置GPU显存用量按需使用tf.config.set_visible_devices([gpu0],GPU)2.

导入数据

datasets.cifar10.load_data()3.归一化

train_images,

255.0train_images.shape,test_images.shape,train_labels.shape,test_labels.shape4.可视化

class_names

truck]plt.figure(figsize(20,10))

for

range(20):plt.subplot(5,10,i1)plt.xticks([])plt.yticks([])plt.grid(False)plt.imshow(train_images[i],

cmapplt.cm.binary)plt.xlabel(class_names[train_labels[i][0]])

model

models.Sequential([layers.Conv2D(32,

(3,

#卷积层1卷积核3*3layers.MaxPooling2D((2,

2)),

#卷积层2卷积核3*3layers.MaxPooling2D((2,

2)),

#Flatten层连接卷积层与全连接层layers.Dense(64,

activationrelu),

_________________________________________________________________

Layer

_________________________________________________________________

max_pooling2d

_________________________________________________________________

conv2d_1

_________________________________________________________________

max_pooling2d_1

_________________________________________________________________

conv2d_2

_________________________________________________________________

flatten

_________________________________________________________________

dense

_________________________________________________________________

dense_1

_________________________________________________________________三、编译模型

model.compile(optimizeradam,losstf.keras.losses.SparseCategoricalCrossentropy(from_logitsTrue),metrics[accuracy])四、训练模型

history

通过模型进行预测得到的是每一个类别的概率数字越大该图片为该类别的可能性越大

plt.imshow(test_images[10])输出测试集中第一张图片的预测结果

import

print(class_names[np.argmax(pre[10])])313/313

3ms/step

pltplt.plot(history.history[accuracy],

labelaccuracy)

plt.plot(history.history[val_accuracy],

label

verbose2)print(test_acc)0.7166000008583069