DeepLearning - HyperParameter 튜닝

from keras.layers import BatchNormalization

import keras_tuner

from tensorflow import keras

def build_model(hp):

    ip = Input(shape=(X.shape[1],))

    n = BatchNormalization()(ip)

    n = Dense(hp.Choice('units', [8, 16, 32, 64, 128, 256]),

              activation='elu')(n)

    n = Dropout(0.5)(n)

    n = BatchNormalization()(n)

    n = Dense(hp.Choice('units', [8, 16, 32, 64, 128, 256]),

              activation='elu')(n)

    n = BatchNormalization()(n)

    n = Dense(hp.Choice('units', [8, 16, 32, 64, 128, 256]),

              activation='elu')(n)

    n = Dropout(0.5)(n)

    n = BatchNormalization()(n)

    n = Dense(hp.Choice('units', [8, 16, 32, 64, 128, 256]),

              activation='elu')(n)

    n = Dense(1, activation='sigmoid')(n)

    model = Model(inputs=ip, outputs=n)

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

    return model

tuner3 = keras_tuner.RandomSearch(

    build_model, objective='val_accuracy', max_trials=5, directory='./tuner4')

tuner3.search(X, y, epochs=1500, validation_split=0.1)

best_model3 = tuner3.get_best_models()[0]

best_model3.evaluate(X_test, y_test)