版權宣告：本文為博主原創文章，未經博主允許不得轉載。 https://blog.csdn.net/weixin_44474718/article/details/86250535

適用於少量資料的實驗是非常有效的方法。

from keras.models import Sequential
from keras.layers import Dense
import numpy as np
from sklearn.model_selection import GridSearchCV
from keras.wrappers.scikit_learn import KerasClassifier


# 構建模型
 

def create_model(optimizer='adam', init='glorot_uniform'):
    # 構建模型
    model = Sequential()
    model.add(Dense(units=12, kernel_initializer=init, input_dim=8, activation='relu'))
    model.add(Dense(units=8, kernel_initializer=init, activation='relu'))
    model.add(Dense(units=1, kernel_initializer=
 
init, activation='sigmoid'))

    # 編譯模型
    model.compile(loss='binary_crossentropy', optimizer=optimizer, metrics=['accuracy'])

    return model

seed = 7
# 設定隨機數種子
np.random.seed(seed)

# 匯入資料
dataset = np.loadtxt('pima-indians-diabetes.csv', delimiter=',')
# 分割輸入x和輸出Y
x = dataset[:, 0 : 8]
Y = dataset[
 
:, 8]

#建立模型 for scikit-learn
model = KerasClassifier(build_fn=create_model, verbose=0)

# 構建需要調參的引數
param_grid = {}
param_grid['optimizer'] = ['rmsprop', 'adam']
param_grid['init'] = ['glorot_uniform', 'normal', 'uniform']
param_grid['epochs'] = [50, 100, 150, 200]
param_grid['batch_size'] = [5, 10, 20]

# 調參
grid = GridSearchCV(estimator=model, param_grid=param_grid)
results = grid.fit(x, Y)

# 輸出結果
print('Best: %f using %s' % (results.best_score_, results.best_params_))
means = results.cv_results_['mean_test_score']
stds = results.cv_results_['std_test_score']
params = results.cv_results_['params']

for mean, std, param in zip(means, stds, params):
    print('%f (%f) with: %r' % (mean, std, param))

筆者用了20分鐘！！！而且執行一次不一定能找到最優的引數組合，需要多次重複執行。