准备工作
MNIST数据集介绍
MNIST(“Modified National Institute of Standards and Technology”)是事实上的计算机视觉“hello world”数据集。自 1999 年发布以来,这个经典的手写图像数据集一直作为基准分类算法的基础。随着新的机器学习技术的出现,MNIST 仍然是研究人员和学习者的可靠资源。我们的目标是从数万张手写图像的数据集中正确识别数字。
数据文件 train.csv 和 test.csv 包含从零到九的手绘数字的灰度图像。
每张图像高 28 像素,宽 28 像素,总共 784 像素。每个像素都有一个与之关联的像素值,表示该像素的亮度或暗度,数字越大表示越暗。该像素值是介于 0 和 255 之间的整数,包括 0 和 255。
训练数据集 (train.csv) 有 785 列。第一列称为“标签”,是用户绘制的数字。其余列包含相关图像的像素值。
训练集中的每个像素列都有一个类似 pixelx 的名称,其中 x 是 0 到 783 之间的整数,包括 0 到 783。要在图像上定位该像素,假设我们已将 x 分解为 x = i * 28 + j,其中 i 和 j 是 0 到 27 之间的整数,包括 0 和 27。然后 pixelx 位于 28 x 28 矩阵的第 i 行和第 j 列(索引为零)。
例如,pixel31 表示左数第四列、上数第二行的像素,如下面的 ascii 图表所示。
从视觉上看,如果我们省略“像素”前缀,像素组成图像如下:
000 001 002 003 ... 026 027 028 029 030 031 ... 054 055 056 057 058 059 ... 082 083 | | | | ... | | 728 729 730 731 ... 754 755 756 757 758 759 ... 782 783
测试数据集 (test.csv) 与训练集相同,只是它不包含“标签”列。
您的提交文件应采用以下格式:对于测试集中的 28000 张图像中的每一张,输出一行包含 ImageId 和您预测的数字。例如,如果您预测第一张图像是 3,第二张图像是 7,第三张图像是 8,那么您的提交文件将如下所示:
ImageId,Label 1,3 2,7 3,8 (27997 more lines)
本次比赛的评价指标是分类准确率,或者说测试图像被正确分类的比例。例如,0.97 的分类准确度表示您已正确分类除 3% 的图像之外的所有图像。
数据集下载:https://wwp.lanzoub.com/iIUFY08t575a
导入包
import numpy as np import pandas as pd import seaborn as sns import matplotlib.pyplot as plt %matplotlib inline
读取数据集
train = pd.read_csv('../input/digit-recognizer/train.csv')
test = pd.read_csv('../input/digit-recognizer/test.csv')
查看数据特征
train.head()
label | pixel0 | pixel1 | pixel2 | pixel3 | pixel4 | pixel5 | pixel6 | pixel7 | pixel8 | ... | pixel774 | pixel775 | pixel776 | pixel777 | pixel778 | pixel779 | pixel780 | pixel781 | pixel782 | pixel783 | |
---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
0 | 1 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | ... | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 |
1 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | ... | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 |
2 | 1 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | ... | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 |
3 | 4 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | ... | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 |
4 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | ... | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 |
5 rows × 785 columns
train.info()
<class 'pandas.core.frame.DataFrame'>
RangeIndex: 42000 entries, 0 to 41999
Columns: 785 entries, label to pixel783
dtypes: int64(785)
memory usage: 251.5 MB
train.isnull().sum()
label 0
pixel0 0
pixel1 0
pixel2 0
pixel3 0
..
pixel779 0
pixel780 0
pixel781 0
pixel782 0
pixel783 0
Length: 785, dtype: int64
sum(train.isnull().sum())
0
预处理训练集|测试集
#y_train 是数字标签 y_train = train['label'].copy()
#X_train 是各像素亮度值 X_train = train.drop('label',axis=1)
y_train.value_counts()
1 4684
7 4401
3 4351
9 4188
2 4177
6 4137
0 4132
4 4072
8 4063
5 3795
Name: label, dtype: int64
y_train = pd.get_dummies(y_train,prefix='Num')
y_train.head()
Num_0 | Num_1 | Num_2 | Num_3 | Num_4 | Num_5 | Num_6 | Num_7 | Num_8 | Num_9 | |
---|---|---|---|---|---|---|---|---|---|---|
0 | 0 | 1 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 |
1 | 1 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 |
2 | 0 | 1 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 |
3 | 0 | 0 | 0 | 0 | 1 | 0 | 0 | 0 | 0 | 0 |
4 | 1 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 |
#28×28一共784个像素,其中的数值表示亮度[0,255] X_train.describe()
pixel0 | pixel1 | pixel2 | pixel3 | pixel4 | pixel5 | pixel6 | pixel7 | pixel8 | pixel9 | ... | pixel774 | pixel775 | pixel776 | pixel777 | pixel778 | pixel779 | pixel780 | pixel781 | pixel782 | pixel783 | |
---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
count | 42000.0 | 42000.0 | 42000.0 | 42000.0 | 42000.0 | 42000.0 | 42000.0 | 42000.0 | 42000.0 | 42000.0 | ... | 42000.000000 | 42000.000000 | 42000.000000 | 42000.00000 | 42000.000000 | 42000.000000 | 42000.0 | 42000.0 | 42000.0 | 42000.0 |
mean | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | ... | 0.219286 | 0.117095 | 0.059024 | 0.02019 | 0.017238 | 0.002857 | 0.0 | 0.0 | 0.0 | 0.0 |
std | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | ... | 6.312890 | 4.633819 | 3.274488 | 1.75987 | 1.894498 | 0.414264 | 0.0 | 0.0 | 0.0 | 0.0 |
min | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | ... | 0.000000 | 0.000000 | 0.000000 | 0.00000 | 0.000000 | 0.000000 | 0.0 | 0.0 | 0.0 | 0.0 |
25% | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | ... | 0.000000 | 0.000000 | 0.000000 | 0.00000 | 0.000000 | 0.000000 | 0.0 | 0.0 | 0.0 | 0.0 |
50% | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | ... | 0.000000 | 0.000000 | 0.000000 | 0.00000 | 0.000000 | 0.000000 | 0.0 | 0.0 | 0.0 | 0.0 |
75% | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | ... | 0.000000 | 0.000000 | 0.000000 | 0.00000 | 0.000000 | 0.000000 | 0.0 | 0.0 | 0.0 | 0.0 |
max | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | ... | 254.000000 | 254.000000 | 253.000000 | 253.00000 | 254.000000 | 62.000000 | 0.0 | 0.0 | 0.0 | 0.0 |
8 rows × 784 columns
#from sklearn.preprocessing import Normalizer
X_train = X_train/255
X_train.head()
pixel0 | pixel1 | pixel2 | pixel3 | pixel4 | pixel5 | pixel6 | pixel7 | pixel8 | pixel9 | ... | pixel774 | pixel775 | pixel776 | pixel777 | pixel778 | pixel779 | pixel780 | pixel781 | pixel782 | pixel783 | |
---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | ... | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 |
1 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | ... | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 |
2 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | ... | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 |
3 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | ... | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 |
4 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | ... | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 |
5 rows × 784 columns
X_train.describe()
pixel0 | pixel1 | pixel2 | pixel3 | pixel4 | pixel5 | pixel6 | pixel7 | pixel8 | pixel9 | ... | pixel774 | pixel775 | pixel776 | pixel777 | pixel778 | pixel779 | pixel780 | pixel781 | pixel782 | pixel783 | |
---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
count | 42000.0 | 42000.0 | 42000.0 | 42000.0 | 42000.0 | 42000.0 | 42000.0 | 42000.0 | 42000.0 | 42000.0 | ... | 42000.000000 | 42000.000000 | 42000.000000 | 42000.000000 | 42000.000000 | 42000.000000 | 42000.0 | 42000.0 | 42000.0 | 42000.0 |
mean | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | ... | 0.000860 | 0.000459 | 0.000231 | 0.000079 | 0.000068 | 0.000011 | 0.0 | 0.0 | 0.0 | 0.0 |
std | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | ... | 0.024756 | 0.018172 | 0.012841 | 0.006901 | 0.007429 | 0.001625 | 0.0 | 0.0 | 0.0 | 0.0 |
min | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | ... | 0.000000 | 0.000000 | 0.000000 | 0.000000 | 0.000000 | 0.000000 | 0.0 | 0.0 | 0.0 | 0.0 |
25% | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | ... | 0.000000 | 0.000000 | 0.000000 | 0.000000 | 0.000000 | 0.000000 | 0.0 | 0.0 | 0.0 | 0.0 |
50% | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | ... | 0.000000 | 0.000000 | 0.000000 | 0.000000 | 0.000000 | 0.000000 | 0.0 | 0.0 | 0.0 | 0.0 |
75% | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | ... | 0.000000 | 0.000000 | 0.000000 | 0.000000 | 0.000000 | 0.000000 | 0.0 | 0.0 | 0.0 | 0.0 |
max | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | ... | 0.996078 | 0.996078 | 0.992157 | 0.992157 | 0.996078 | 0.243137 | 0.0 | 0.0 | 0.0 | 0.0 |
8 rows × 784 columns
X_train = X_train.values.reshape(-1,28,28,1)
X_train
array([[[[0.],
[0.],
[0.],
...,
[0.],
[0.],
[0.]],
test.info()
<class 'pandas.core.frame.DataFrame'>
RangeIndex: 28000 entries, 0 to 27999
Columns: 784 entries, pixel0 to pixel783
dtypes: int64(784)
memory usage: 167.5 MB
test.isnull().sum()
pixel0 0
pixel1 0
pixel2 0
pixel3 0
pixel4 0
..
pixel779 0
pixel780 0
pixel781 0
pixel782 0
pixel783 0
Length: 784, dtype: int64
sum(test.isnull().sum())
0
test = test/255
test.head()
pixel0 | pixel1 | pixel2 | pixel3 | pixel4 | pixel5 | pixel6 | pixel7 | pixel8 | pixel9 | ... | pixel774 | pixel775 | pixel776 | pixel777 | pixel778 | pixel779 | pixel780 | pixel781 | pixel782 | pixel783 | |
---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | ... | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 |
1 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | ... | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 |
2 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | ... | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 |
3 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | ... | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 |
4 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | ... | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 |
5 rows × 784 columns
test.describe()
pixel0 | pixel1 | pixel2 | pixel3 | pixel4 | pixel5 | pixel6 | pixel7 | pixel8 | pixel9 | ... | pixel774 | pixel775 | pixel776 | pixel777 | pixel778 | pixel779 | pixel780 | pixel781 | pixel782 | pixel783 | |
---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
count | 28000.0 | 28000.0 | 28000.0 | 28000.0 | 28000.0 | 28000.0 | 28000.0 | 28000.0 | 28000.0 | 28000.0 | ... | 28000.000000 | 28000.000000 | 28000.000000 | 28000.000000 | 28000.000000 | 28000.0 | 28000.0 | 28000.0 | 28000.0 | 28000.0 |
mean | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | ... | 0.000646 | 0.000287 | 0.000110 | 0.000044 | 0.000026 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 |
std | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | ... | 0.021464 | 0.014184 | 0.007112 | 0.004726 | 0.003167 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 |
min | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | ... | 0.000000 | 0.000000 | 0.000000 | 0.000000 | 0.000000 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 |
25% | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | ... | 0.000000 | 0.000000 | 0.000000 | 0.000000 | 0.000000 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 |
50% | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | ... | 0.000000 | 0.000000 | 0.000000 | 0.000000 | 0.000000 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 |
75% | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | ... | 0.000000 | 0.000000 | 0.000000 | 0.000000 | 0.000000 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 |
max | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | ... | 0.992157 | 0.996078 | 0.756863 | 0.733333 | 0.466667 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 |
8 rows × 784 columns
test = test.values.reshape(-1,28,28,1)
test
array([[[[0.],
[0.],
[0.],
...,
[0.],
[0.],
[0.]],
训练CNN Model
import tensorflow as tf
tf.__version__
'2.6.4'
cnn = tf.keras.models.Sequential()
2022-08-01 05:41:16.816392: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1510] Created device /job:localhost/replica:0/task:0/device:GPU:0 with 15403 MB memory: -> device: 0, name: Tesla P100-PCIE-16GB, pci bus id: 0000:00:04.0, compute capability: 6.0
#Convolution cnn.add(tf.keras.layers.Conv2D(filters=256,kernel_size=(5,5),activation='relu',input_shape=(28,28,1)))
#Max Pooling cnn.add(tf.keras.layers.MaxPool2D(pool_size=(3,3),strides=3))
cnn.add(tf.keras.layers.BatchNormalization())
cnn.add(tf.keras.layers.Conv2D(filters=128,kernel_size=(4,4),activation='relu'))
cnn.add(tf.keras.layers.MaxPool2D(pool_size=(2,2),strides=2))
#Flattening cnn.add(tf.keras.layers.Flatten())
#Full connection cnn.add(tf.keras.layers.Dense(units=256,activation='relu'))
#Output Layer cnn.add(tf.keras.layers.Dense(units=10,activation='softmax'))
#Compile cnn cnn.compile(optimizer='adam',loss='categorical_crossentropy')
# Epoch(时期): # 当一个完整的数据集通过了神经网络一次并且返回了一次,这个过程称为一次>epoch。(也就是说,所有训练样本在神经网络中都 进行了一次正向传播 和一次反向传播 ) # 再通俗一点,一个Epoch就是将所有训练样本训练一次的过程。 # 然而,当一个Epoch的样本(也就是所有的训练样本)数量可能太过庞大(对于计算机而言),就需要把它分成多个小块,也就是就是分成多个Batch 来进行训练。** # Batch(批 / 一批样本): # 将整个训练样本分成若干个Batch。 # Batch_Size(批大小): # 每批样本的大小。 # Iteration(一次迭代): # 训练一个Batch就是一次Iteration(这个概念跟程序语言中的迭代器相似)。 cnn.fit(X_train,y_train,batch_size=32,epochs=50)
2022-08-01 05:41:18.154328: I tensorflow/compiler/mlir/mlir_graph_optimization_pass.cc:185] None of the MLIR Optimization Passes are enabled (registered 2)
Epoch 1/50
2022-08-01 05:41:19.541340: I tensorflow/stream_executor/cuda/cuda_dnn.cc:369] Loaded cuDNN version 8005
1313/1313 [] - 13s 5ms/step - loss: 0.1159
Epoch 2/50
1313/1313 [] - 5s 4ms/step - loss: 0.0496
Epoch 3/50
1313/1313 [] - 6s 4ms/step - loss: 0.0367
Epoch 4/50
1313/1313 [] - 5s 4ms/step - loss: 0.0289
Epoch 5/50
1313/1313 [] - 6s 4ms/step - loss: 0.0256
Epoch 6/50
1313/1313 [] - 5s 4ms/step - loss: 0.0220
Epoch 7/50
1313/1313 [] - 6s 4ms/step - loss: 0.0192
Epoch 8/50
1313/1313 [] - 5s 4ms/step - loss: 0.0167
Epoch 9/50
1313/1313 [] - 6s 4ms/step - loss: 0.0146
Epoch 10/50
1313/1313 [] - 5s 4ms/step - loss: 0.0121
Epoch 11/50
1313/1313 [] - 6s 4ms/step - loss: 0.0133
Epoch 12/50
1313/1313 [] - 5s 4ms/step - loss: 0.0142
Epoch 13/50
1313/1313 [] - 6s 4ms/step - loss: 0.0119
Epoch 14/50
1313/1313 [] - 6s 4ms/step - loss: 0.0125
Epoch 15/50
1313/1313 [] - 6s 4ms/step - loss: 0.0103
Epoch 16/50
1313/1313 [] - 6s 4ms/step - loss: 0.0103
Epoch 17/50
1313/1313 [] - 6s 4ms/step - loss: 0.0130
Epoch 18/50
1313/1313 [] - 6s 4ms/step - loss: 0.0118
Epoch 19/50
1313/1313 [] - 6s 4ms/step - loss: 0.0093
Epoch 20/50
1313/1313 [] - 6s 4ms/step - loss: 0.0075
Epoch 21/50
1313/1313 [] - 6s 4ms/step - loss: 0.0075
Epoch 22/50
1313/1313 [] - 6s 5ms/step - loss: 0.0129
Epoch 23/50
1313/1313 [] - 6s 4ms/step - loss: 0.0105
Epoch 24/50
1313/1313 [] - 6s 4ms/step - loss: 0.0087
Epoch 25/50
1313/1313 [] - 6s 4ms/step - loss: 0.0097
Epoch 26/50
1313/1313 [] - 6s 4ms/step - loss: 0.0117
Epoch 27/50
1313/1313 [] - 5s 4ms/step - loss: 0.0051
Epoch 28/50
1313/1313 [] - 6s 5ms/step - loss: 0.0086
Epoch 29/50
1313/1313 [] - 6s 4ms/step - loss: 0.0100
Epoch 30/50
1313/1313 [] - 6s 4ms/step - loss: 0.0087
Epoch 31/50
1313/1313 [] - 6s 4ms/step - loss: 0.0096
Epoch 32/50
1313/1313 [] - 6s 4ms/step - loss: 0.0065
Epoch 33/50
1313/1313 [] - 5s 4ms/step - loss: 0.0082
Epoch 34/50
1313/1313 [] - 6s 4ms/step - loss: 0.0110
Epoch 35/50
1313/1313 [] - 6s 4ms/step - loss: 0.0063
Epoch 36/50
1313/1313 [] - 6s 4ms/step - loss: 0.0107
Epoch 37/50
1313/1313 [] - 5s 4ms/step - loss: 0.0048
Epoch 38/50
1313/1313 [] - 6s 4ms/step - loss: 0.0076
Epoch 39/50
1313/1313 [] - 5s 4ms/step - loss: 0.0154
Epoch 40/50
1313/1313 [] - 6s 4ms/step - loss: 0.0095
Epoch 41/50
1313/1313 [] - 5s 4ms/step - loss: 0.0052
Epoch 42/50
1313/1313 [] - 6s 4ms/step - loss: 0.0057
Epoch 43/50
1313/1313 [] - 5s 4ms/step - loss: 0.0080
Epoch 44/50
1313/1313 [] - 6s 4ms/step - loss: 0.0085
Epoch 45/50
1313/1313 [] - 5s 4ms/step - loss: 0.0108
Epoch 46/50
1313/1313 [] - 6s 4ms/step - loss: 0.0062
Epoch 47/50
1313/1313 [] - 5s 4ms/step - loss: 0.0118
Epoch 48/50
1313/1313 [] - 6s 4ms/step - loss: 0.0078
Epoch 49/50
1313/1313 [] - 5s 4ms/step - loss: 0.0083
Epoch 50/50
1313/1313 [] - 6s 4ms/step - loss: 0.0044
<keras.callbacks.History at 0x7f35f40ac710>
pred = cnn.predict(test)
pred = np.argmax(pred,axis=1)
pred
array([2, 0, 9, ..., 3, 9, 2])
pred = pd.DataFrame(pred,columns=['Label'])
test_id = list(range(1,len(test)+1,1))
test_id = pd.DataFrame(test_id,columns=['ImageId'])
submission = pd.concat([test_id,pred],axis=1)
submission.describe()
ImageId | Label | |
---|---|---|
count | 28000.000000 | 28000.000000 |
mean | 14000.500000 | 4.453036 |
std | 8083.048105 | 2.896665 |
min | 1.000000 | 0.000000 |
25% | 7000.750000 | 2.000000 |
50% | 14000.500000 | 4.000000 |
75% | 21000.250000 | 7.000000 |
max | 28000.000000 | 9.000000 |
此模型最终准确率为:0.98857
原创作者:孤飞-博客园
原文地址:https://www.cnblogs.com/ranxi169/p/16540166.html
jupyter格式代码查看|下载:https://www.kaggle.com/code/ranxi169/digit-recognizer-with-cnn-for-beginner/notebook