深度学习Week17——优化器对比实验

文章目录
深度学习Week17——优化器对比实验
一、前言
二、我的环境
三、前期工作
1、配置环境
2、导入数据
2.1 加载数据
2.2 检查数据
2.3 配置数据集
2.4 数据可视化
四、构建模型
五、训练模型
1、将其嵌入model中
2、在Dataset数据集中进行数据增强
六、模型评估
1、Accuracy与Loss图
2、模型评估

一、前言

🍨 本文为🔗365天深度学习训练营中的学习记录博客
🍖 原作者：K同学啊 | 接辅导、项目定制

终于，基础篇打卡已经基本结束，深度学习的基础学习到了很多，相关函数已经可以比较熟练的使用，最后一篇主要是探究不同优化器、以及不同参数配置对模型的影响，在论文当中我们也可以进行优化器的比对，以增加论文工作量。

我记得在今年4月份，我和学长一同发表了一篇EI会议论文，里面关于优化器的对比就占了一定的篇幅，那么今天，通过学习，我一定也会自己清楚如何弄清楚优化器对比，本次学习由于临近期末周，仅做了基础学习，没有过多扩展，未来会抽一周补全。

二、我的环境

电脑系统：Windows 10
语言环境：Python 3.8.0
编译器：Pycharm2023.2.3
深度学习环境：TensorFlow
显卡及显存：RTX 3060 8G

三、前期工作

1、配置环境

import tensorflow as tf
gpus = tf.config.list_physical_devices("GPU")

if gpus:
    gpu0 = gpus[0] #如果有多个GPU，仅使用第0个GPU
    tf.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")

from tensorflow          import keras
import matplotlib.pyplot as plt
import pandas            as pd
import numpy             as np
import warnings,os,PIL,pathlib

warnings.filterwarnings("ignore")             #忽略警告信息
plt.rcParams['font.sans-serif']    = ['SimHei']  # 用来正常显示中文标签
plt.rcParams['axes.unicode_minus'] = False    # 用来正常显示负号

2、导入数据

导入所有猫狗图片数据，依次分别为训练集图片(train_images)、训练集标签(train_labels)、测试集图片(test_images)、测试集标签(test_labels)，数据集来源于K同学啊

2.1 加载数据

data_dir   = "/home/mw/input/dogcat3675/365-7-data"

data_dir    = pathlib.Path(data_dir)
image_count = len(list(data_dir.glob('*/*')))
print("图片总数为：",image_count)

图片总数为： 3400

img_height = 256
img_width  = 256
batch_size = 32

"""
关于image_dataset_from_directory()的详细介绍可以参考文章：https://mtyjkh.blog.csdn.net/article/details/117018789
"""
train_ds = tf.keras.preprocessing.image_dataset_from_directory(
    data_dir,
    validation_split=0.2,
    subset="training",
    seed=12,
    image_size=(img_height, img_width),
    batch_size=batch_size)

输出：

Found 3400 files belonging to 2 classes.
Using 2720 files for training.

"""
关于image_dataset_from_directory()的详细介绍可以参考文章：https://mtyjkh.blog.csdn.net/article/details/117018789
"""
val_ds = tf.keras.preprocessing.image_dataset_from_directory(
    data_dir,
    validation_split=0.2,
    subset="validation",
    seed=12,
    image_size=(img_height, img_width),
    batch_size=batch_size)

Found 3400 files belonging to 2 classes.
Using 680 files for validation.

我们可以通过class_names输出数据集的标签。标签将按字母顺序对应于目录名称。

class_names = train_ds.class_names
print(class_names)

[‘cat’, ‘dog’]

2.2 检查数据

for image_batch, labels_batch in train_ds:
    print(image_batch.shape)
    print(labels_batch.shape)
    break

(32, 256, 256, 3)
(32,)

2.3 配置数据集

AUTOTUNE = tf.data.AUTOTUNE

def train_preprocessing(image,label):
    return (image/255.0,label)

train_ds = (
    train_ds.cache()
    .shuffle(1000)
    .map(train_preprocessing)    # 这里可以设置预处理函数
#     .batch(batch_size)           # 在image_dataset_from_directory处已经设置了batch_size
    .prefetch(buffer_size=AUTOTUNE)
)

val_ds = (
    val_ds.cache()
    .shuffle(1000)
    .map(train_preprocessing)    # 这里可以设置预处理函数
#     .batch(batch_size)         # 在image_dataset_from_directory处已经设置了batch_size
    .prefetch(buffer_size=AUTOTUNE)
)

2.4 数据可视化

plt.figure(figsize=(10, 8))  # 图形的宽为10高为5
plt.suptitle("数据展示")

for images, labels in train_ds.take(1):
    for i in range(15):
        plt.subplot(4, 5, i + 1)
        plt.xticks([])
        plt.yticks([])
        plt.grid(False)

        # 显示图片
        plt.imshow(images[i])
        # 显示标签
        plt.xlabel(class_names[labels[i]])

plt.show()

在这里插入图片描述

四、构建模型

from tensorflow.keras.layers import Dropout,Dense,BatchNormalization
from tensorflow.keras.models import Model

def create_model(optimizer='adam'):
    # 加载预训练模型
    vgg16_base_model = tf.keras.applications.vgg16.VGG16(weights='imagenet',
                                                                include_top=False,
                                                                input_shape=(img_width, img_height, 3),
                                                                pooling='avg')
    for layer in vgg16_base_model.layers:
        layer.trainable = False

    X = vgg16_base_model.output
    
    X = Dense(170, activation='relu')(X)
    X = BatchNormalization()(X)
    X = Dropout(0.5)(X)

    output = Dense(len(class_names), activation='softmax')(X)
    vgg16_model = Model(inputs=vgg16_base_model.input, outputs=output)

    vgg16_model.compile(optimizer=optimizer,
                        loss='sparse_categorical_crossentropy',
                        metrics=['accuracy'])
    return vgg16_model

model1 = create_model(optimizer=tf.keras.optimizers.Adam())
model2 = create_model(optimizer=tf.keras.optimizers.SGD())
model2.summary()

Downloading data from https://storage.googleapis.com/tensorflow/keras-applications/vgg16/vgg16_weights_tf_dim_ordering_tf_kernels_notop.h5
58892288/58889256 [==============================] - 3s 0us/step
Model: "model_1"
_________________________________________________________________
Layer (type)                 Output Shape              Param #   
=================================================================
input_2 (InputLayer)         [(None, 256, 256, 3)]     0         
_________________________________________________________________
block1_conv1 (Conv2D)        (None, 256, 256, 64)      1792      
_________________________________________________________________
block1_conv2 (Conv2D)        (None, 256, 256, 64)      36928     
_________________________________________________________________
block1_pool (MaxPooling2D)   (None, 128, 128, 64)      0         
_________________________________________________________________
block2_conv1 (Conv2D)        (None, 128, 128, 128)     73856     
_________________________________________________________________
block2_conv2 (Conv2D)        (None, 128, 128, 128)     147584    
_________________________________________________________________
block2_pool (MaxPooling2D)   (None, 64, 64, 128)       0         
_________________________________________________________________
block3_conv1 (Conv2D)        (None, 64, 64, 256)       295168    
_________________________________________________________________
block3_conv2 (Conv2D)        (None, 64, 64, 256)       590080    
_________________________________________________________________
block3_conv3 (Conv2D)        (None, 64, 64, 256)       590080    
_________________________________________________________________
block3_pool (MaxPooling2D)   (None, 32, 32, 256)       0         
_________________________________________________________________
block4_conv1 (Conv2D)        (None, 32, 32, 512)       1180160   
_________________________________________________________________
block4_conv2 (Conv2D)        (None, 32, 32, 512)       2359808   
_________________________________________________________________
block4_conv3 (Conv2D)        (None, 32, 32, 512)       2359808   
_________________________________________________________________
block4_pool (MaxPooling2D)   (None, 16, 16, 512)       0         
_________________________________________________________________
block5_conv1 (Conv2D)        (None, 16, 16, 512)       2359808   
_________________________________________________________________
block5_conv2 (Conv2D)        (None, 16, 16, 512)       2359808   
_________________________________________________________________
block5_conv3 (Conv2D)        (None, 16, 16, 512)       2359808   
_________________________________________________________________
block5_pool (MaxPooling2D)   (None, 8, 8, 512)         0         
_________________________________________________________________
global_average_pooling2d_1 ( (None, 512)               0         
_________________________________________________________________
dense_2 (Dense)              (None, 170)               87210     
_________________________________________________________________
batch_normalization_1 (Batch (None, 170)               680       
_________________________________________________________________
dropout_1 (Dropout)          (None, 170)               0         
_________________________________________________________________
dense_3 (Dense)              (None, 2)                 342       
=================================================================
Total params: 14,802,920
Trainable params: 87,892
Non-trainable params: 14,715,028
_________________________________________________________________

五、训练模型

NO_EPOCHS = 50

history_model1  = model1.fit(train_ds, epochs=NO_EPOCHS, verbose=1, validation_data=val_ds)
history_model2  = model2.fit(train_ds, epochs=NO_EPOCHS, verbose=1, validation_data=val_ds)

Epoch 1/50
85/85 [==============================] - 40s 230ms/step - loss: 0.2183 - accuracy: 0.8948 - val_loss: 0.1916 - val_accuracy: 0.9912
Epoch 2/50
85/85 [==============================] - 15s 182ms/step - loss: 0.0372 - accuracy: 0.9901 - val_loss: 0.0799 - val_accuracy: 0.9956
Epoch 3/50
85/85 [==============================] - 16s 184ms/step - loss: 0.0213 - accuracy: 0.9941 - val_loss: 0.0456 - val_accuracy: 0.9897
Epoch 4/50
85/85 [==============================] - 16s 185ms/step - loss: 0.0191 - accuracy: 0.9916 - val_loss: 0.0220 - val_accuracy: 0.9941
Epoch 5/50
85/85 [==============================] - 16s 187ms/step - loss: 0.0177 - accuracy: 0.9938 - val_loss: 0.0319 - val_accuracy: 0.9868
Epoch 6/50
85/85 [==============================] - 16s 188ms/step - loss: 0.0135 - accuracy: 0.9952 - val_loss: 0.0139 - val_accuracy: 0.9941
Epoch 7/50
85/85 [==============================] - 16s 189ms/step - loss: 0.0108 - accuracy: 0.9965 - val_loss: 0.0343 - val_accuracy: 0.9868
Epoch 8/50
85/85 [==============================] - 16s 190ms/step - loss: 0.0088 - accuracy: 0.9979 - val_loss: 0.0919 - val_accuracy: 0.9676
Epoch 9/50
85/85 [==============================] - 16s 191ms/step - loss: 0.0141 - accuracy: 0.9958 - val_loss: 0.0476 - val_accuracy: 0.9824
Epoch 10/50
85/85 [==============================] - 16s 192ms/step - loss: 0.0093 - accuracy: 0.9985 - val_loss: 0.0067 - val_accuracy: 0.9985
Epoch 11/50
85/85 [==============================] - 16s 193ms/step - loss: 0.0055 - accuracy: 0.9995 - val_loss: 0.0160 - val_accuracy: 0.9941
Epoch 12/50
85/85 [==============================] - 16s 194ms/step - loss: 0.0085 - accuracy: 0.9986 - val_loss: 0.0076 - val_accuracy: 0.9956
Epoch 13/50
85/85 [==============================] - 17s 195ms/step - loss: 0.0073 - accuracy: 0.9982 - val_loss: 0.0131 - val_accuracy: 0.9971
Epoch 14/50
85/85 [==============================] - 17s 195ms/step - loss: 0.0060 - accuracy: 0.9999 - val_loss: 0.0126 - val_accuracy: 0.9941
Epoch 15/50
85/85 [==============================] - 17s 196ms/step - loss: 0.0031 - accuracy: 0.9993 - val_loss: 0.0053 - val_accuracy: 0.9985
Epoch 16/50
85/85 [==============================] - 17s 196ms/step - loss: 0.0031 - accuracy: 0.9999 - val_loss: 0.0108 - val_accuracy: 0.9941
Epoch 17/50
85/85 [==============================] - 17s 196ms/step - loss: 0.0038 - accuracy: 0.9989 - val_loss: 0.0054 - val_accuracy: 0.9985
Epoch 18/50
85/85 [==============================] - 17s 195ms/step - loss: 0.0014 - accuracy: 1.0000 - val_loss: 0.0048 - val_accuracy: 0.9985
Epoch 19/50
85/85 [==============================] - 17s 195ms/step - loss: 0.0017 - accuracy: 1.0000 - val_loss: 0.0060 - val_accuracy: 0.9985
Epoch 20/50
85/85 [==============================] - 17s 195ms/step - loss: 0.0023 - accuracy: 0.9988 - val_loss: 0.0045 - val_accuracy: 0.9985
Epoch 21/50
85/85 [==============================] - 16s 194ms/step - loss: 0.0066 - accuracy: 0.9980 - val_loss: 0.0048 - val_accuracy: 1.0000
Epoch 22/50
85/85 [==============================] - 16s 194ms/step - loss: 0.0019 - accuracy: 0.9999 - val_loss: 0.0212 - val_accuracy: 0.9941
Epoch 23/50
85/85 [==============================] - 16s 193ms/step - loss: 0.0038 - accuracy: 0.9990 - val_loss: 0.0049 - val_accuracy: 0.9985
Epoch 24/50
85/85 [==============================] - 16s 193ms/step - loss: 0.0037 - accuracy: 0.9987 - val_loss: 0.0054 - val_accuracy: 0.9985
Epoch 25/50
85/85 [==============================] - 16s 193ms/step - loss: 0.0017 - accuracy: 1.0000 - val_loss: 0.0839 - val_accuracy: 0.9721
Epoch 26/50
85/85 [==============================] - 16s 193ms/step - loss: 0.0037 - accuracy: 0.9983 - val_loss: 0.0084 - val_accuracy: 0.9985
Epoch 27/50
85/85 [==============================] - 16s 193ms/step - loss: 0.0022 - accuracy: 1.0000 - val_loss: 0.0078 - val_accuracy: 0.9985
Epoch 28/50
85/85 [==============================] - 16s 193ms/step - loss: 0.0030 - accuracy: 0.9992 - val_loss: 0.0044 - val_accuracy: 0.9985
Epoch 29/50
85/85 [==============================] - 16s 193ms/step - loss: 0.0011 - accuracy: 0.9999 - val_loss: 0.0047 - val_accuracy: 0.9971
Epoch 30/50
85/85 [==============================] - 16s 193ms/step - loss: 0.0023 - accuracy: 1.0000 - val_loss: 0.0082 - val_accuracy: 0.9985
Epoch 31/50
85/85 [==============================] - 16s 193ms/step - loss: 0.0015 - accuracy: 0.9998 - val_loss: 0.0099 - val_accuracy: 0.9956
Epoch 32/50
85/85 [==============================] - 16s 193ms/step - loss: 0.0014 - accuracy: 1.0000 - val_loss: 0.0058 - val_accuracy: 0.9985
Epoch 33/50
85/85 [==============================] - 16s 193ms/step - loss: 0.0012 - accuracy: 1.0000 - val_loss: 0.0067 - val_accuracy: 0.9985
Epoch 34/50
85/85 [==============================] - 16s 193ms/step - loss: 9.7591e-04 - accuracy: 1.0000 - val_loss: 0.0058 - val_accuracy: 0.9985
Epoch 35/50
85/85 [==============================] - 16s 193ms/step - loss: 6.5250e-04 - accuracy: 1.0000 - val_loss: 0.0106 - val_accuracy: 0.9971
Epoch 36/50
85/85 [==============================] - 16s 193ms/step - loss: 0.0016 - accuracy: 0.9997 - val_loss: 0.0140 - val_accuracy: 0.9926
Epoch 37/50
85/85 [==============================] - 16s 192ms/step - loss: 0.0070 - accuracy: 0.9960 - val_loss: 0.0359 - val_accuracy: 0.9868
Epoch 38/50
85/85 [==============================] - 16s 192ms/step - loss: 0.0023 - accuracy: 0.9996 - val_loss: 0.0309 - val_accuracy: 0.9897
Epoch 39/50
85/85 [==============================] - 16s 193ms/step - loss: 0.0023 - accuracy: 0.9991 - val_loss: 0.0062 - val_accuracy: 0.9985
Epoch 40/50
85/85 [==============================] - 16s 193ms/step - loss: 0.0014 - accuracy: 0.9999 - val_loss: 0.0133 - val_accuracy: 0.9956
Epoch 41/50
85/85 [==============================] - 16s 193ms/step - loss: 0.0028 - accuracy: 0.9986 - val_loss: 0.1411 - val_accuracy: 0.9603
Epoch 42/50
85/85 [==============================] - 16s 193ms/step - loss: 0.0036 - accuracy: 0.9991 - val_loss: 0.0233 - val_accuracy: 0.9941
Epoch 43/50
85/85 [==============================] - 16s 192ms/step - loss: 0.0025 - accuracy: 0.9995 - val_loss: 0.0018 - val_accuracy: 1.0000
Epoch 44/50
85/85 [==============================] - 16s 193ms/step - loss: 0.0015 - accuracy: 1.0000 - val_loss: 0.0043 - val_accuracy: 0.9985
Epoch 45/50
85/85 [==============================] - 16s 192ms/step - loss: 0.0040 - accuracy: 0.9988 - val_loss: 0.0049 - val_accuracy: 0.9985
Epoch 46/50
85/85 [==============================] - 16s 193ms/step - loss: 0.0011 - accuracy: 1.0000 - val_loss: 0.0034 - val_accuracy: 0.9985
Epoch 47/50
85/85 [==============================] - 16s 193ms/step - loss: 0.0012 - accuracy: 1.0000 - val_loss: 0.0055 - val_accuracy: 0.9971
Epoch 48/50
85/85 [==============================] - 16s 193ms/step - loss: 0.0023 - accuracy: 0.9989 - val_loss: 0.0639 - val_accuracy: 0.9824
Epoch 49/50
85/85 [==============================] - 16s 193ms/step - loss: 8.3466e-04 - accuracy: 1.0000 - val_loss: 0.0049 - val_accuracy: 0.9985
Epoch 50/50
85/85 [==============================] - 16s 193ms/step - loss: 0.0021 - accuracy: 0.9993 - val_loss: 0.0163 - val_accuracy: 0.9926
Epoch 1/50
85/85 [==============================] - 17s 194ms/step - loss: 0.3235 - accuracy: 0.8649 - val_loss: 0.4082 - val_accuracy: 0.8088
Epoch 2/50
85/85 [==============================] - 16s 192ms/step - loss: 0.0971 - accuracy: 0.9641 - val_loss: 0.2085 - val_accuracy: 0.9882
Epoch 3/50
85/85 [==============================] - 16s 193ms/step - loss: 0.0705 - accuracy: 0.9773 - val_loss: 0.1053 - val_accuracy: 0.9926
Epoch 4/50
85/85 [==============================] - 16s 193ms/step - loss: 0.0675 - accuracy: 0.9780 - val_loss: 0.0565 - val_accuracy: 0.9926
Epoch 5/50
85/85 [==============================] - 16s 193ms/step - loss: 0.0510 - accuracy: 0.9841 - val_loss: 0.0317 - val_accuracy: 0.9941
Epoch 6/50
85/85 [==============================] - 16s 192ms/step - loss: 0.0466 - accuracy: 0.9802 - val_loss: 0.0229 - val_accuracy: 0.9956
Epoch 7/50
85/85 [==============================] - 16s 192ms/step - loss: 0.0424 - accuracy: 0.9869 - val_loss: 0.0160 - val_accuracy: 0.9985
Epoch 8/50
85/85 [==============================] - 16s 192ms/step - loss: 0.0488 - accuracy: 0.9843 - val_loss: 0.0152 - val_accuracy: 0.9956
Epoch 9/50
85/85 [==============================] - 16s 192ms/step - loss: 0.0405 - accuracy: 0.9892 - val_loss: 0.0134 - val_accuracy: 0.9971
Epoch 10/50
85/85 [==============================] - 16s 193ms/step - loss: 0.0398 - accuracy: 0.9875 - val_loss: 0.0128 - val_accuracy: 0.9956
Epoch 11/50
85/85 [==============================] - 16s 193ms/step - loss: 0.0387 - accuracy: 0.9856 - val_loss: 0.0139 - val_accuracy: 0.9956
Epoch 12/50
85/85 [==============================] - 16s 192ms/step - loss: 0.0334 - accuracy: 0.9900 - val_loss: 0.0155 - val_accuracy: 0.9956
Epoch 13/50
85/85 [==============================] - 16s 192ms/step - loss: 0.0321 - accuracy: 0.9915 - val_loss: 0.0119 - val_accuracy: 0.9956
Epoch 14/50
85/85 [==============================] - 16s 192ms/step - loss: 0.0358 - accuracy: 0.9878 - val_loss: 0.0116 - val_accuracy: 0.9971
Epoch 15/50
85/85 [==============================] - 16s 193ms/step - loss: 0.0270 - accuracy: 0.9898 - val_loss: 0.0098 - val_accuracy: 0.9985
Epoch 16/50
85/85 [==============================] - 16s 192ms/step - loss: 0.0233 - accuracy: 0.9936 - val_loss: 0.0102 - val_accuracy: 0.9956
Epoch 17/50
85/85 [==============================] - 16s 192ms/step - loss: 0.0274 - accuracy: 0.9915 - val_loss: 0.0106 - val_accuracy: 0.9971
Epoch 18/50
85/85 [==============================] - 16s 192ms/step - loss: 0.0233 - accuracy: 0.9942 - val_loss: 0.0090 - val_accuracy: 0.9971
Epoch 19/50
85/85 [==============================] - 16s 193ms/step - loss: 0.0284 - accuracy: 0.9894 - val_loss: 0.0111 - val_accuracy: 0.9971
Epoch 20/50
85/85 [==============================] - 16s 193ms/step - loss: 0.0232 - accuracy: 0.9934 - val_loss: 0.0098 - val_accuracy: 0.9956
Epoch 21/50
85/85 [==============================] - 16s 192ms/step - loss: 0.0262 - accuracy: 0.9928 - val_loss: 0.0108 - val_accuracy: 0.9941
Epoch 22/50
85/85 [==============================] - 16s 193ms/step - loss: 0.0239 - accuracy: 0.9940 - val_loss: 0.0112 - val_accuracy: 0.9941
Epoch 23/50
85/85 [==============================] - 16s 193ms/step - loss: 0.0251 - accuracy: 0.9927 - val_loss: 0.0089 - val_accuracy: 0.9956
Epoch 24/50
85/85 [==============================] - 16s 193ms/step - loss: 0.0251 - accuracy: 0.9910 - val_loss: 0.0086 - val_accuracy: 0.9956
Epoch 25/50
85/85 [==============================] - 16s 192ms/step - loss: 0.0254 - accuracy: 0.9904 - val_loss: 0.0102 - val_accuracy: 0.9956
Epoch 26/50
85/85 [==============================] - 16s 192ms/step - loss: 0.0245 - accuracy: 0.9916 - val_loss: 0.0084 - val_accuracy: 0.9985
Epoch 27/50
85/85 [==============================] - 16s 193ms/step - loss: 0.0226 - accuracy: 0.9932 - val_loss: 0.0086 - val_accuracy: 0.9985
Epoch 28/50
85/85 [==============================] - 16s 192ms/step - loss: 0.0289 - accuracy: 0.9905 - val_loss: 0.0088 - val_accuracy: 0.9971
Epoch 29/50
85/85 [==============================] - 16s 192ms/step - loss: 0.0219 - accuracy: 0.9928 - val_loss: 0.0091 - val_accuracy: 0.9956
Epoch 30/50
85/85 [==============================] - 16s 193ms/step - loss: 0.0215 - accuracy: 0.9943 - val_loss: 0.0077 - val_accuracy: 0.9985
Epoch 31/50
85/85 [==============================] - 16s 192ms/step - loss: 0.0178 - accuracy: 0.9952 - val_loss: 0.0074 - val_accuracy: 0.9971
Epoch 32/50
85/85 [==============================] - 16s 193ms/step - loss: 0.0231 - accuracy: 0.9929 - val_loss: 0.0117 - val_accuracy: 0.9956
Epoch 33/50
85/85 [==============================] - 16s 192ms/step - loss: 0.0183 - accuracy: 0.9962 - val_loss: 0.0078 - val_accuracy: 0.9971
Epoch 34/50
85/85 [==============================] - 16s 192ms/step - loss: 0.0186 - accuracy: 0.9950 - val_loss: 0.0076 - val_accuracy: 0.9971
Epoch 35/50
85/85 [==============================] - 16s 192ms/step - loss: 0.0185 - accuracy: 0.9949 - val_loss: 0.0089 - val_accuracy: 0.9956
Epoch 36/50
85/85 [==============================] - 16s 192ms/step - loss: 0.0199 - accuracy: 0.9924 - val_loss: 0.0080 - val_accuracy: 0.9971
Epoch 37/50
85/85 [==============================] - 16s 192ms/step - loss: 0.0165 - accuracy: 0.9960 - val_loss: 0.0076 - val_accuracy: 0.9971
Epoch 38/50
85/85 [==============================] - 16s 193ms/step - loss: 0.0180 - accuracy: 0.9945 - val_loss: 0.0083 - val_accuracy: 0.9985
Epoch 39/50
85/85 [==============================] - 16s 192ms/step - loss: 0.0219 - accuracy: 0.9904 - val_loss: 0.0087 - val_accuracy: 0.9971
Epoch 40/50
85/85 [==============================] - 16s 193ms/step - loss: 0.0179 - accuracy: 0.9971 - val_loss: 0.0086 - val_accuracy: 0.9956
Epoch 41/50
85/85 [==============================] - 16s 192ms/step - loss: 0.0166 - accuracy: 0.9937 - val_loss: 0.0076 - val_accuracy: 0.9971
Epoch 42/50
85/85 [==============================] - 16s 192ms/step - loss: 0.0159 - accuracy: 0.9946 - val_loss: 0.0074 - val_accuracy: 0.9985
Epoch 43/50
85/85 [==============================] - 16s 192ms/step - loss: 0.0161 - accuracy: 0.9942 - val_loss: 0.0079 - val_accuracy: 0.9971
Epoch 44/50
85/85 [==============================] - 16s 192ms/step - loss: 0.0173 - accuracy: 0.9957 - val_loss: 0.0073 - val_accuracy: 0.9971
Epoch 45/50
85/85 [==============================] - 16s 192ms/step - loss: 0.0149 - accuracy: 0.9965 - val_loss: 0.0069 - val_accuracy: 0.9985
Epoch 46/50
85/85 [==============================] - 16s 192ms/step - loss: 0.0157 - accuracy: 0.9963 - val_loss: 0.0076 - val_accuracy: 0.9971
Epoch 47/50
85/85 [==============================] - 16s 192ms/step - loss: 0.0153 - accuracy: 0.9959 - val_loss: 0.0074 - val_accuracy: 0.9971
Epoch 48/50
85/85 [==============================] - 16s 192ms/step - loss: 0.0162 - accuracy: 0.9957 - val_loss: 0.0068 - val_accuracy: 0.9985
Epoch 49/50
85/85 [==============================] - 16s 193ms/step - loss: 0.0132 - accuracy: 0.9955 - val_loss: 0.0079 - val_accuracy: 0.9971
Epoch 50/50
85/85 [==============================] - 16s 192ms/step - loss: 0.0131 - accuracy: 0.9954 - val_loss: 0.0085 - val_accuracy: 0.9971

六、评估模型

1、Accuracy与Loss图

from matplotlib.ticker import MultipleLocator
plt.rcParams['savefig.dpi'] = 300 #图片像素
plt.rcParams['figure.dpi']  = 300 #分辨率

acc1     = history_model1.history['accuracy']
acc2     = history_model2.history['accuracy']
val_acc1 = history_model1.history['val_accuracy']
val_acc2 = history_model2.history['val_accuracy']

loss1     = history_model1.history['loss']
loss2     = history_model2.history['loss']
val_loss1 = history_model1.history['val_loss']
val_loss2 = history_model2.history['val_loss']

epochs_range = range(len(acc1))

plt.figure(figsize=(16, 4))
plt.subplot(1, 2, 1)

plt.plot(epochs_range, acc1, label='Training Accuracy-Adam')
plt.plot(epochs_range, acc2, label='Training Accuracy-SGD')
plt.plot(epochs_range, val_acc1, label='Validation Accuracy-Adam')
plt.plot(epochs_range, val_acc2, label='Validation Accuracy-SGD')
plt.legend(loc='lower right')
plt.title('Training and Validation Accuracy')
# 设置刻度间隔，x轴每1一个刻度
ax = plt.gca()
ax.xaxis.set_major_locator(MultipleLocator(1))

plt.subplot(1, 2, 2)
plt.plot(epochs_range, loss1, label='Training Loss-Adam')
plt.plot(epochs_range, loss2, label='Training Loss-SGD')
plt.plot(epochs_range, val_loss1, label='Validation Loss-Adam')
plt.plot(epochs_range, val_loss2, label='Validation Loss-SGD')
plt.legend(loc='upper right')
plt.title('Training and Validation Loss')
   
# 设置刻度间隔，x轴每1一个刻度
ax = plt.gca()
ax.xaxis.set_major_locator(MultipleLocator(1))

plt.show()

在这里插入图片描述

2、模型评估

def test_accuracy_report(model):
    score = model.evaluate(val_ds, verbose=0)
    print('Loss function: %s, accuracy:' % score[0], score[1])
    
test_accuracy_report(model2)

Loss function: 0.008488086983561516, accuracy: 0.9970588088035583