[시각 지능] MNIST

MNIST

 

https://ko.wikipedia.org/wiki/MNIST_%EB%8D%B0%EC%9D%B4%ED%84%B0%EB%B2%A0%EC%9D%B4%EC%8A%A4

 

MNIST 데이터베이스 는 손으로 쓴 숫자들로 이루어진 대형 데이터베이스이며, 다양한 화상 처리 시스템을 트레이닝하기 위해 일반적으로 사용된다. 이 데이터베이스는 또한 기계 학습 분야의 트레이닝 및 테스트에 널리 사용된다. NIST의 오리지널 데이터셋의 샘플을 재혼합하여 만들어졌다.

 

import tensorflow as tf
import numpy as np

from tensorflow.keras.models import Sequential
from tensorflow.keras.layers import Flatten, Dense
from tensorflow.keras.datasets import mnist
from tensorflow.keras.utils import to_categorical

(x_train, t_train), (x_test, t_test) = mnist.load_data()

print('')
print('x_train.shape = ', x_train.shape, ', t_train.shape = ', t_train.shape)
print('x_test.shape = ', x_test.shape, ', t_test.shape = ', t_test.shape)

Downloading data from https://storage.googleapis.com/tensorflow/tf-keras-datasets/mnist.npz
11493376/11490434 [==============================] - 0s 0us/step
11501568/11490434 [==============================] - 0s 0us/step

x_train.shape =  (60000, 28, 28) , t_train.shape =  (60000,)
x_test.shape =  (10000, 28, 28) , t_test.shape =  (10000,)

 

25개 이미지 출력

 

import matplotlib.pyplot  as plt

plt.figure(figsize = (6, 6))

for index in range(25):
  plt.subplot(5, 5, index + 1) # 5행 5열
  plt.imshow(x_train[index], cmap = 'gray')
  plt.axis('off')

plt.show()

plt.imshow(x_train[9], cmap = 'gray')
plt.colorbar()
plt.show()

 

데이터 전처리

 

# x_train, x_test 값 범위를 0 ~ 1로 초기화
x_train = x_train / 255.0
x_test = x_test / 255.0

print('train max = ', x_train[0].max(), ', train min = ', x_train[0].min())
print('text max = ', x_test[0].max(), ', test min = ', x_test[0].min())

train max =  1.0 , train min =  0.0
text max =  1.0 , test min =  0.0

# 정답 데이터 one-hot encoding
t_train = to_categorical(t_train, 10)
t_test = to_categorical(t_test, 10)

print('train label = ', t_train[0], ', decimal value = ', np.argmax(t_train[0]))
print('test label = ', t_test[0], ', decimal value = ', np.argmax(t_test[0]))

train label =  [0. 0. 0. 0. 0. 1. 0. 0. 0. 0.] , decimal value =  5
test label =  [0. 0. 0. 0. 0. 0. 0. 1. 0. 0.] , decimal value =  7

 

모델 구축 및 컴파일

 

from tensorflow.keras.optimizers import SGD

model = Sequential()

model.add(Flatten(input_shape = (28, 28, 1)))
model.add(Dense(100, activation = 'relu'))
model.add(Dense(10, activation = 'softmax'))

model.compile(optimizer = SGD(),
              loss = 'categorical_crossentropy',
              metrics = ['accuracy'])

model.summary()

Model: "sequential"
_________________________________________________________________
 Layer (type)                Output Shape              Param #   
=================================================================
 flatten (Flatten)           (None, 784)               0         
                                                                 
 dense (Dense)               (None, 100)               78500     
                                                                 
 dense_1 (Dense)             (None, 10)                1010      
                                                                 
=================================================================
Total params: 79,510
Trainable params: 79,510
Non-trainable params: 0
_________________________________________________________________

 

모델 학습

 

hist = model.fit(x_train, t_train, epochs = 30, validation_split = 0.2)

 

모델 (정확도) 평가

 

model.evaluate(x_test, t_test)

313/313 [==============================] - 1s 3ms/step - loss: 0.1099 - accuracy: 0.9688
[0.10988297313451767, 0.9688000082969666]

 

손실, 정확도

 

plt.title('Loss Trend')
plt.xlabel('epochs')
plt.ylabel('loss')
plt.grid()

plt.plot(hist.history['loss'], label = 'training loss')
plt.plot(hist.history['val_loss'], label = 'validation loss')
plt.legend(loc = 'best')

plt.show()

plt.title('Accuracy Trend')
plt.xlabel('epochs')
plt.ylabel('accuracy')
plt.grid()

plt.plot(hist.history['loss'], label = 'training accuracy')
plt.plot(hist.history['val_loss'], label = 'validation accuracy')
plt.legend(loc = 'best')

plt.show()

 

예측

 

pred = model.predict(x_test)

print(pred.shape) # 입력 : 10,000 / 출력 : 10
print(pred[:5])

(10000, 10)
[[2.47125226e-05 9.41667935e-08 4.55119327e-04 9.51758772e-03
  2.68577605e-08 1.28551646e-05 4.31200603e-10 9.89943504e-01
  1.48223171e-05 3.13286109e-05]
 [1.60143209e-05 3.11233511e-04 9.98242259e-01 7.63157965e-04
  5.11713616e-10 1.65575620e-05 2.92068580e-04 7.99445440e-11
  3.58617574e-04 9.49742174e-10]
 [9.39205347e-05 9.84965324e-01 2.76321894e-03 9.09237191e-04
  4.20741271e-04 3.95900395e-04 4.91019455e-04 6.11370383e-03
  3.54454620e-03 3.02398985e-04]
 [9.99855399e-01 1.17819205e-07 3.06073744e-05 1.49597452e-06
  7.35541164e-07 2.91437154e-05 1.38678743e-05 4.04107523e-05
  1.97497926e-07 2.80292970e-05]
 [7.03856858e-05 1.86522607e-06 1.27948762e-04 6.76886975e-06
  9.76610422e-01 6.29926399e-06 1.87695157e-04 7.62023556e-04
  1.49609084e-04 2.20768899e-02]]