'KerasClassifier'物件沒有屬性'summary'-嘗試從KerasClassifier構建的lstm模型中獲取摘要

似乎 KerasClassifier 對可定制模型做了一些包裝，但我不知道如何把它弄出來......

我想將我的 lstm 模型從幾乎沒有創建到 keras 包裝器，例如KerasClassifier：

model1 = Sequential()
model1.add(LSTM(units=60, activation='relu', input_shape=(60, 1),
        return_sequences=True, recurrent_dropout=0.1))
model1.add(LSTM(units=30))
model1.add(Dense(units=1, activation='sigmoid'))
model1.compile(optimizer='adam', loss='binary_crossentropy', metrics=['accuracy'])

到

def create_model():
    model = Sequential()
    model.add(LSTM(units=60, activation='relu', input_shape=(60, 1),
        return_sequences=True, recurrent_dropout=0.1))
    model.add(LSTM(units=30))
    model.add(Dense(units=1, activation='sigmoid'))
    model.compile(optimizer='adam', loss='binary_crossentropy', metrics=['accuracy'])
    return model

modelk = KerasClassifier(build_fn=create_model,
                         epochs=10,
                         batch_size=30,
                         verbose=0)

如果我model1.summary()使用model1回傳的第一種方法，我會得到類似：

Model: "sequential"
_________________________________________________________________
 Layer (type)                Output Shape              Param #   
=================================================================
 lstm (LSTM)                 (None, 60, 60)            14880     
                                                                 
 lstm_1 (LSTM)               (None, 30)                10920     
                                                                 
 dense (Dense)               (None, 1)                 31        
                                                                 
=================================================================
Total params: 25,831
Trainable params: 25,831
Non-trainable params: 0

但是如果我使用modelk從第二種方法回傳的“modelk.summary()”，我會收到如下錯誤：

'KerasClassifier' object has no attribute 'summary'

uj5u.com熱心網友回復：

嘗試modelk.build_fn().summary()：

from keras.models import Sequential
from keras.layers import Dense, LSTM
from keras.wrappers.scikit_learn import KerasClassifier
 
def create_model():
    model = Sequential()
    model.add(LSTM(units=60, activation='relu', input_shape=(60, 1),
        return_sequences=True, recurrent_dropout=0.1))
    model.add(LSTM(units=30))
    model.add(Dense(units=1, activation='sigmoid'))
    model.compile(optimizer='adam', loss='binary_crossentropy', metrics=['accuracy'])
    return model

modelk = KerasClassifier(build_fn=create_model,
                         epochs=10,
                         batch_size=30,
                         verbose=0)
print(modelk.build_fn().summary())

Model: "sequential_2"
_________________________________________________________________
 Layer (type)                Output Shape              Param #   
=================================================================
 lstm_4 (LSTM)               (None, 60, 60)            14880     
                                                                 
 lstm_5 (LSTM)               (None, 30)                10920     
                                                                 
 dense_2 (Dense)             (None, 1)                 31        
                                                                 
=================================================================
Total params: 25,831
Trainable params: 25,831
Non-trainable params: 0
_________________________________________________________________
None

您還可以做的是model.summary在內部使用，create_model并且在內部呼叫時將列印摘要model.fit：

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

def create_model(optimizer='rmsprop'):
    model = Sequential()
    model.add(LSTM(units=60, activation='relu', input_shape=(60, 1),
        return_sequences=True, recurrent_dropout=0.1))
    model.add(LSTM(units=30))
    model.add(Dense(units=1, activation='sigmoid'))
    model.compile(optimizer=optimizer, loss='binary_crossentropy', metrics=['accuracy'])
    print(model.summary())
    return model

modelk = KerasClassifier(build_fn=create_model,
                         epochs=10,
                         batch_size=25,
                         verbose=0)

optimizers = ['rmsprop', 'adam']
param_grid = dict(optimizer=optimizers)
grid = GridSearchCV(estimator=modelk, param_grid=param_grid)

X = np.random.random((50, 60, 1))
Y = np.random.random((50,))
grid_result = grid.fit(X, Y)

標籤：Python 张量流 喀拉斯 lstm

上一篇：在SQL表中的XML資料列上使用WHERE條件選擇資料

下一篇：Keras序列模型中的資料增強層