심층 신경망

심층 신경망

2개의 층

입력층은 생각하지 않고 은닉층 출력층이 존재하는 경우이다

활성화 함수가 필요하다

가중치와 절편으로만 이루어져 있다면 단순 산술 연산이므로 2개의 연산을 쉽게 1개의 연산으로 바꿀 수도 있으므로 층이 2개 있는 것이 의미가 없어진다

그러므로 활성화 함수를 통해서 중간에 비선형함수로 바꾸어주는 과정을 통해 각 층에서의 선형 식이 합쳐지지 못하도록(각 가중치가 의미가 있게 되도록)만들어야 한다.

from tensorflow import keras

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

from sklearn.model_selection import train_test_split

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

train_scaled, val_scaled, train_target, val_target = train_test_split(
    train_scaled, train_target, test_size=0.2, random_state=42)

dense1 = keras.layers.Dense(100, activation='sigmoid', input_shape=(784,))
dense2 = keras.layers.Dense(10, activation='softmax')

심층 신경망 만들기

model = keras.Sequential([dense1, dense2])

model.summary()

층을 추가하는 다른 방법

model = keras.Sequential([
    keras.layers.Dense(100, activation='sigmoid', input_shape=(784,), name='hidden'),
    keras.layers.Dense(10, activation='softmax', name='output')
], name='패션 MNIST 모델')

model.summary()

model = keras.Sequential()
model.add(keras.layers.Dense(100, activation='sigmoid', input_shape=(784,)))
model.add(keras.layers.Dense(10, activation='softmax'))

model.summary()

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

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

렐루 활성화 함수

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

model.summary()

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

train_scaled = train_input / 255.0

train_scaled, val_scaled, train_target, val_target = train_test_split(
    train_scaled, train_target, test_size=0.2, random_state=42)

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

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

model.evaluate(val_scaled, val_target)

옵티마이저

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

sgd = keras.optimizers.SGD()
model.compile(optimizer=sgd, loss='sparse_categorical_crossentropy', metrics='accuracy')

sgd = keras.optimizers.SGD(learning_rate=0.1)

sgd = keras.optimizers.SGD(momentum=0.9, nesterov=True)

adagrad = keras.optimizers.Adagrad()
model.compile(optimizer=adagrad, loss='sparse_categorical_crossentropy', metrics='accuracy')

rmsprop = keras.optimizers.RMSprop()
model.compile(optimizer=rmsprop, loss='sparse_categorical_crossentropy', metrics='accuracy')

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

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

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

model.evaluate(val_scaled, val_target)