인공 신경망

 

 

 

 

7.1 인공신경망¶

패션 MNIST¶

In [1]:

from tensorflow import keras

(train_input, train_target), (test_input, test_target) = keras.datasets.fashion_mnist.load_data()

 

Downloading data from https://storage.googleapis.com/tensorflow/tf-keras-datasets/train-labels-idx1-ubyte.gz
32768/29515 [=================================] - 0s 0us/step
Downloading data from https://storage.googleapis.com/tensorflow/tf-keras-datasets/train-images-idx3-ubyte.gz
26427392/26421880 [==============================] - 0s 0us/step
Downloading data from https://storage.googleapis.com/tensorflow/tf-keras-datasets/t10k-labels-idx1-ubyte.gz
8192/5148 [===============================================] - 0s 0us/step
Downloading data from https://storage.googleapis.com/tensorflow/tf-keras-datasets/t10k-images-idx3-ubyte.gz
4423680/4422102 [==============================] - 0s 0us/step

In [2]:

print(train_input.shape, train_target.shape)

 

(60000, 28, 28) (60000,)

In [3]:

print(test_input.shape, test_target.shape)

 

(10000, 28, 28) (10000,)

In [4]:

import matplotlib.pyplot as plt
fig, axs = plt.subplots(1, 10, figsize=(10,10))
for i in range(10):
  axs[i].imshow(train_input[i], cmap = 'gray_r')
  axs[i].axis('off')
plt.show()

 

In [5]:

print([train_target[i] for i in range(10)])

 

[9, 0, 0, 3, 0, 2, 7, 2, 5, 5]

In [6]:

import numpy as np
print(np.unique(train_target, return_counts = True))

 

(array([0, 1, 2, 3, 4, 5, 6, 7, 8, 9], dtype=uint8), array([6000, 6000, 6000, 6000, 6000, 6000, 6000, 6000, 6000, 6000]))

 

로지스틱 회귀로 패션 아이템 분류하기¶

In [7]:

train_scaled = train_input/255.0
train_scaled = train_scaled.reshape(-1,28*28)

In [8]:

print(train_scaled.shape)

 

(60000, 784)

In [9]:

from sklearn.model_selection import cross_validate
from sklearn.linear_model import SGDClassifier
sc = SGDClassifier(loss = 'log', max_iter=5, random_state=42)
scores = cross_validate(sc, train_scaled, train_target, n_jobs=-1)
print(np.mean(scores['test_score']))

 

0.8195666666666668

 

인공 신경망¶

In [10]:

import tensorflow as tf

In [11]:

from tensorflow import keras

 

인공 신경망으로 모델 만들기¶

인공 신경망에서는 교차 검증을 잘 사용하지 않고 검증 세트를 별도로 덜어내서 사용¶

1. 딥러닝 분야의 데이터 셋은 충분히 크기 때문에 검증 점수가 안정적
2. 교차 검증을 수행하기에는 훈련 시간이 너무 오래 걸리기 때문

In [12]:

from sklearn.model_selection import train_test_split
train_scaled, val_scaled, train_target, val_target = train_test_split(train_scaled, train_target, test_size = 0.2, random_state = 42)

In [13]:

print(train_scaled.shape, train_target.shape)

 

(48000, 784) (48000,)

In [14]:

print(val_scaled.shape, val_target.shape)

 

(12000, 784) (12000,)

In [16]:

dense = keras.layers.Dense(10, activation='softmax', input_shape = (784,))

In [17]:

model = keras.Sequential(dense)

 

인공 신경망으로 패션 아이템 분류하기¶

In [24]:

model.compile(loss='sparse_categorical_crossentropy', metrics='accuracy')

In [25]:

print(train_target[:10])

 

[7 3 5 8 6 9 3 3 9 9]

In [26]:

model.fit(train_scaled, train_target, epochs=5)

 

Epoch 1/5
1500/1500 [==============================] - 5s 2ms/step - loss: 0.7805 - accuracy: 0.7343
Epoch 2/5
1500/1500 [==============================] - 2s 2ms/step - loss: 0.4886 - accuracy: 0.8385
Epoch 3/5
1500/1500 [==============================] - 2s 2ms/step - loss: 0.4674 - accuracy: 0.8449
Epoch 4/5
1500/1500 [==============================] - 2s 2ms/step - loss: 0.4517 - accuracy: 0.8486
Epoch 5/5
1500/1500 [==============================] - 2s 2ms/step - loss: 0.4318 - accuracy: 0.8579

Out[26]:

<tensorflow.python.keras.callbacks.History at 0x7f01eaaea850>

In [27]:

model.evaluate(val_scaled, val_target)

 

375/375 [==============================] - 1s 1ms/step - loss: 0.4521 - accuracy: 0.8491

Out[27]:

[0.452136754989624, 0.8490833044052124]

In [ ]:

 

출처 : 혼자 공부하는 머신러닝+딥러닝 github/박해선