MNIST(「Modified National Institute of Standards and Technology」)是事實上的計算機視覺「hello world」資料集。自 1999 年釋出以來,這個經典的手寫影象資料集一直作為基準分類演演算法的基礎。隨著新的機器學習技術的出現,MNIST 仍然是研究人員和學習者的可靠資源。我們的目標是從數萬張手寫影象的資料集中正確識別數位。
資料檔案 train.csv 和 test.csv 包含從零到九的手繪數位的灰度影象。
每張影象高 28 畫素,寬 28 畫素,總共 784 畫素。每個畫素都有一個與之關聯的畫素值,表示該畫素的亮度或暗度,數位越大表示越暗。該畫素值是介於 0 和 255 之間的整數,包括 0 和 255。
訓練資料集 (train.csv) 有 785 列。第一列稱為「標籤」,是使用者繪製的數位。其餘列包含相關影象的畫素值。
訓練集中的每個畫素列都有一個類似 pixelx 的名稱,其中 x 是 0 到 783 之間的整數,包括 0 到 783。要在影象上定位該畫素,假設我們已將 x 分解為 x = i * 28 + j,其中 i 和 j 是 0 到 27 之間的整數,包括 0 和 27。然後 pixelx 位於 28 x 28 矩陣的第 i 行和第 j 列(索引為零)。
例如,pixel31 表示左數第四列、上數第二行的畫素,如下面的 ascii 圖表所示。
從視覺上看,如果我們省略「畫素」字首,畫素組成影象如下:
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
測試資料集 (test.csv) 與訓練集相同,只是它不包含「標籤」列。
您的提交檔案應採用以下格式:對於測試集中的 28000 張影象中的每一張,輸出一行包含 ImageId 和您預測的數位。例如,如果您預測第一張影象是 3,第二張影象是 7,第三張影象是 8,那麼您的提交檔案將如下所示:
ImageId,Label
1,3
2,7
3,8
(27997 more lines)
本次比賽的評價指標是分類準確率,或者說測試影象被正確分類的比例。例如,0.97 的分類準確度表示您已正確分類除 3% 的影象之外的所有影象。
資料集下載:https://wwp.lanzoub.com/iIUFY08t575a
import numpy as np
import pandas as pd
import seaborn as sns
import matplotlib.pyplot as plt
%matplotlib inline
train = pd.read_csv('../input/digit-recognizer/train.csv')
test = pd.read_csv('../input/digit-recognizer/test.csv')
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
#y_train 是數位標籤
y_train = train['label'].copy()
#X_train 是各畫素亮度值
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個畫素,其中的數值表示亮度[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.]],
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(時期):
# 當一個完整的資料集通過了神經網路一次並且返回了一次,這個過程稱為一次>epoch。(也就是說,所有訓練樣本在神經網路中都 進行了一次正向傳播 和一次反向傳播 )
# 再通俗一點,一個Epoch就是將所有訓練樣本訓練一次的過程。
# 然而,當一個Epoch的樣本(也就是所有的訓練樣本)數量可能太過龐大(對於計算機而言),就需要把它分成多個小塊,也就是就是分成多個Batch 來進行訓練。**
# Batch(批 / 一批樣本):
# 將整個訓練樣本分成若干個Batch。
# Batch_Size(批大小):
# 每批樣本的大小。
# Iteration(一次迭代):
# 訓練一個Batch就是一次Iteration(這個概念跟程式語言中的迭代器相似)。
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 |
此模型最終準確率為:0.98857
原創作者:孤飛-部落格園
原文地址:https://www.cnblogs.com/ranxi169/p/16540166.html
jupyter格式程式碼檢視|下載:https://www.kaggle.com/code/ranxi169/digit-recognizer-with-cnn-for-beginner/notebook