如何通过GridsearchCV在Keras中对MLP模型进行超参数调整？

问题描述

我使用该函数创建了一个模型，并以1D数组的形式将两个嵌入作为模型的输入传递。在尝试对输入进行网格搜索时，它会引发错误，指出“找到的输入变量样本数量不一致”。我什至试图重塑输入，但徒劳无功，因为输入是一维的。

所以问题是如何对具有多个输入和单个输出的keras MLP模型进行超参数调整？

另一个问题是如何获取模型摘要？我尝试使用model.summary（），但不适用于通过函数创建的模型。

def create_baseline():
    
    activation = 'relu'
    dropout_rate = 0.2
    kernel_initializer = 'uniform'
    optimizer = 'adam'
    learning_rate = 0.001
    latent_dim = 50
    
    # Create model

    user_input = Input(shape = (1,),name = 'user-input')
    user_embedding = Embedding(num_users + 1,latent_dim,name = 'user-embedding')(user_input)
    user_vec = Flatten(name = 'user-flatten')(user_embedding)

    movie_input = Input(shape = (1,name = 'movie-input')
    movie_embedding = Embedding(num_movies + 1,name = 'movie-embedding')(movie_input)
    movie_vec = Flatten(name = 'movie-flatten')(movie_embedding)

    merged = concatenate([user_vec,movie_vec],axis = 1,name = 'user-movie-concat')
    merged_dropout = Dropout(dropout_rate)(merged)

    fc_1 = Dense(128,name = 'fc-1',kernel_initializer = kernel_initializer,activation = activation,activity_regularizer = l2(0.01))(merged_dropout)
    fc_1_dropout = Dropout(dropout_rate,name = 'fc-1-dropout')(fc_1)
    fc_2 = Dense(64,name = 'fc-2',activity_regularizer = l2(0.01))(fc_1_dropout)
    fc_2_dropout = Dropout(dropout_rate,name = 'fc-2-dropout')(fc_2)
    fc_3 = Dense(32,name = 'fc-3',activity_regularizer = l2(0.01))(fc_2_dropout)
    fc_3_dropout = Dropout(dropout_rate,name = 'fc-3-dropout')(fc_3)
    fc_4 = Dense(1,name = 'fc-4',activation = activation)(fc_3_dropout)

    model = Model([user_input,movie_input],fc_4)
    model.compile(loss = 'mean_squared_error',optimizer = optimizer,metrics = [rmse,'mae'])
    
    return model


from keras.callbacks import EarlyStopping
from keras.wrappers.scikit_learn import KerasRegressor

early_stop = EarlyStopping(monitor = 'val_loss',patience = 2,verbose = 0)

# Create a KerasRegressor
model = KerasRegressor(build_fn = create_baseline,epochs = 10,batch_size = 8,verbose = 0,callbacks = [early_stop])

# Defining the parameters to tune.

mlp_params = {'batch_size': [2,4,8,16,32,64],'learning_rate': [0.0001,0.001,0.01,0.1],'dropout_rate': [0.0,0.1,0.2,0.3,0.4,0.5],'optimizer': ['Adam','RMSprop','SGD','Adagrad'],'kernel_initializer': ['uniform','normal','glorot_normal','glorot_uniform','he_normal','he_uniform']}

# Using Grid Search to tune the hyperparameters. 

from sklearn.model_selection import GridSearchCV

gs_mlp_model = GridSearchCV(estimator = model,param_grid = mlp_params,scoring = 'neg_mean_squared_error',cv = 5,n_jobs = -1,verbose = 1)

gs_mlp_model.fit([X_train_1.UserID.values,X_train_1.MovieID.values],y_train_1.values)


# The error is:

---------------------------------------------------------------------------
ValueError                                Traceback (most recent call last)
<ipython-input-157-dcc10ddae7b6> in <module>
----> 1 gs_mlp_model.fit([X_train_1.UserID.values,y_train_1.values)

~\anaconda3\lib\site-packages\sklearn\model_selection\_search.py in fit(self,X,y,groups,**fit_params)
    648             refit_metric = 'score'
    649 
--> 650         X,groups = indexable(X,groups)
    651         fit_params = _check_fit_params(X,fit_params)
    652 

~\anaconda3\lib\site-packages\sklearn\utils\validation.py in indexable(*iterables)
    246     """
    247     result = [_make_indexable(X) for X in iterables]
--> 248     check_consistent_length(*result)
    249     return result
    250 

~\anaconda3\lib\site-packages\sklearn\utils\validation.py in check_consistent_length(*arrays)
    210     if len(uniques) > 1:
    211         raise ValueError("Found input variables with inconsistent numbers of"
--> 212                          " samples: %r" % [int(l) for l in lengths])
    213 
    214 

ValueError: Found input variables with inconsistent numbers of samples: [2,800167]

解决方法

暂无找到可以解决该程序问题的有效方法，小编努力寻找整理中！

如果你已经找到好的解决方法，欢迎将解决方案带上本链接一起发送给小编。

小编邮箱:dio#foxmail.com (将#修改为@）