혼공 ML+DL #25

📌 LSTM 신경망

• LSTM 셀
• 은닉상태(h) x 셀(w) 4개 -> 셀(cell)상태
• 셀 상태 에서 순환되는 상태 값
  
  

📍 모델 생성

• LSTM(8) : 8개 뉴런, dropout 지정

model = keras.Sequential()

model.add(keras.layers.Embedding(500, 16, input_length=100))
model.add(keras.layers.LSTM(8,  dropout=0.3))
model.add(keras.layers.Dense(1, activation='sigmoid'))

• LSTM Param
  ( 16개 입력 x 8개 뉴런 완전연결
  +8개 뉴런 x 8개 은닉상태
  +8개 절편 ) x 4개 셀
  = 800

📍 모델 훈련

• 2개 층 연결

model.add(keras.layers.LSTM(8, dropout=0.3, return_sequences=True))

rmsprop = keras.optimizers.RMSprop(learning_rate=1e-4)
model.compile(optimizer=rmsprop, loss='binary_crossentropy', 
               metrics=['accuracy'])

checkpoint_cb = keras.callbacks.ModelCheckpoint('best-2rnn-model.h5', 
                                                save_best_only=True)
early_stopping_cb = keras.callbacks.EarlyStopping(patience=3,
                                                  restore_best_weights=True)

history = model.fit(train_seq, train_target, epochs=100, batch_size=64,
                     validation_data=(val_seq, val_target),
                     callbacks=[checkpoint_cb, early_stopping_cb])

import matplotlib.pyplot as plt
plt.plot(history.history['loss'])
plt.plot(history.history['val_loss'])
plt.xlabel('epoch')
plt.ylabel('loss')
plt.legend(['train', 'val'])
plt.show()

📌 GRU 신경망

• GRU 셀
• 뉴런마다 절편 2개씩 필요
• LSTM보다 효율적. 메모리 절감

📍 모델 생성&훈련

model4 = keras.Sequential()

model4.add(keras.layers.Embedding(500, 16, input_length=100))
model4.add(keras.layers.GRU(8))
model4.add(keras.layers.Dense(1, activation='sigmoid'))

• GRU Param
  ( 16개 입력 x 8개 뉴런 완전연결
  +8개 뉴런 x 8개 은닉상태
  +8개 뉴런 x 2개 절편) x 3개 셀
  = 624
  
  

rmsprop = keras.optimizers.RMSprop(learning_rate=1e-4)
model4.compile(optimizer=rmsprop, loss='binary_crossentropy', 
               metrics=['accuracy'])

checkpoint_cb = keras.callbacks.ModelCheckpoint('best-gru-model.h5', 
                                                save_best_only=True)
early_stopping_cb = keras.callbacks.EarlyStopping(patience=3,
                                                  restore_best_weights=True)

history = model4.fit(train_seq, train_target, epochs=100, batch_size=64,
                     validation_data=(val_seq, val_target),
                     callbacks=[checkpoint_cb, early_stopping_cb])

plt.plot(history.history['loss'])
plt.plot(history.history['val_loss'])
plt.xlabel('epoch')
plt.ylabel('loss')
plt.legend(['train', 'val'])
plt.show()

🔗 혼공 MLDL-25