LSTM 순환 신경망 텐서플로로 구현하기

데이터 호출

 

from tensorflow.keras.layers import LSTM
from tensorflow.keras.models import Sequential
from tensorflow.keras.layers import Embedding
from tensorflow.keras.layers import Dense

 

import numpy as np
from tensorflow.keras.datasets import imdb

(x_train_all, y_train_all), (x_test, y_test) = imdb.load_data(skip_top=20, num_words=1000)

for i in range(len(x_train_all)):
    x_train_all[i] = [w for w in x_train_all[i] if w > 2]
    
x_train = x_train_all[random_index[:20000]]
y_train = y_train_all[random_index[:20000]]
x_val = x_train_all[random_index[20000:]]
y_val = y_train_all[random_index[20000:]]
maxlen=100  #타임 스텝의 길이가 100인 시퀀스 데이터를 만듭니다.
x_train_seq = sequence.pad_sequences(x_train, maxlen=maxlen)
x_val_seq = sequence.pad_sequences(x_val, maxlen=maxlen)

LSTM 순환 신경망 만들기

 

model_lstm = Sequential()
model_lstm.add(Embedding(1000, 32))
model_lstm.add(LSTM(8))
model_lstm.add(Dense(1, activation='sigmoid'))
model_lstm.summary()

Model: "sequential_3"
_________________________________________________________________
 Layer (type)                Output Shape              Param #   
=================================================================
 embedding_2 (Embedding)     (None, None, 32)          32000     
                                                                 
 lstm_2 (LSTM)               (None, 8)                 1312      
                                                                 
 dense_1 (Dense)             (None, 1)                 9         
                                                                 
=================================================================
Total params: 33,321
Trainable params: 33,321
Non-trainable params: 0
_________________________________________________________________

모델 훈련하기

 

model_lstm.compile(optimizer='adam', loss='binary_crossentropy', metrics=['accuracy'])

history = model_lstm.fit(x_train_seq, y_train, epochs=10, batch_size=32, 
                         validation_data=(x_val_seq, y_val))

Epoch 1/10
625/625 [==============================] - 6s 8ms/step - loss: 0.4474 - accuracy: 0.7947 - val_loss: 0.3754 - val_accuracy: 0.8362
Epoch 2/10
625/625 [==============================] - 5s 8ms/step - loss: 0.3344 - accuracy: 0.8604 - val_loss: 0.3557 - val_accuracy: 0.8408
Epoch 3/10
625/625 [==============================] - 5s 8ms/step - loss: 0.3116 - accuracy: 0.8686 - val_loss: 0.3571 - val_accuracy: 0.8416
Epoch 4/10
625/625 [==============================] - 5s 8ms/step - loss: 0.2954 - accuracy: 0.8756 - val_loss: 0.3573 - val_accuracy: 0.8432
Epoch 5/10
625/625 [==============================] - 5s 8ms/step - loss: 0.2809 - accuracy: 0.8821 - val_loss: 0.3708 - val_accuracy: 0.8410
Epoch 6/10
625/625 [==============================] - 5s 8ms/step - loss: 0.2681 - accuracy: 0.8887 - val_loss: 0.3861 - val_accuracy: 0.8394
Epoch 7/10
625/625 [==============================] - 5s 8ms/step - loss: 0.2563 - accuracy: 0.8921 - val_loss: 0.3869 - val_accuracy: 0.8372
Epoch 8/10
625/625 [==============================] - 5s 8ms/step - loss: 0.2433 - accuracy: 0.8978 - val_loss: 0.4016 - val_accuracy: 0.8344
Epoch 9/10
625/625 [==============================] - 5s 8ms/step - loss: 0.2325 - accuracy: 0.9032 - val_loss: 0.4087 - val_accuracy: 0.8316
Epoch 10/10
625/625 [==============================] - 5s 8ms/step - loss: 0.2252 - accuracy: 0.9071 - val_loss: 0.4177 - val_accuracy: 0.8336

손실 그래프와 정확도 그래프 그리기

 

import matplotlib.pyplot as plt
plt.plot(history.history['loss'])
plt.plot(history.history['val_loss'])
plt.show()


plt.plot(history.history['accuracy'])
plt.plot(history.history['val_accuracy'])
plt.show()

검증 세트 정확도 평가하기

loss, accuracy = model_lstm.evaluate(x_val_seq, y_val, verbose=0)
print(accuracy)

0.8335999846458435