Keras，在每个时期获得一层的输出

Question

我做了什么？我实现了一个keras模型如下：train_X, test_X, train_Y, test_Y = train_test_split(X, Y, test_size=0.2, random_state=np.random.seed(7), shuffle=True)train_X = np.reshape(train_X, (train_X.shape[0], 1, train_X.shape[1]))test_X = np.reshape(test_X, (test_X.shape[0], 1, test_X.shape[1]))model = Sequential()model.add(LSTM(100, return_sequences=False, input_shape=(train_X.shape[1], train_X.shape[2])))model.add(Dense(train_Y.shape[1], activation='softmax'))model.compile(loss='mean_squared_error', optimizer='adam', metrics=['accuracy'])model.fit(train_X, train_Y, validation_split=.20,                        epochs=1000, batch_size=50)我想要的是？我想给出support vector machine(SVM)倒数第二层 (LSTM) 的输出，在任何epoch（即 1000）层中svm也要进行训练。但我不知道如何做到这一点？任何的想法？更新：我使用ModelCheckpoint如下：model = Sequential()model.add(LSTM(100, return_sequences=False, input_shape=(train_X.shape[1], train_X.shape[2])))model.add(Dense(train_Y.shape[1], activation='softmax'))model.compile(loss='mean_squared_error', optimizer='adam', metrics=['accuracy'])# checkpointfilepath="weights-{epoch:02d}-{val_acc:.2f}.hdf5"checkpoint = ModelCheckpoint(filepath, monitor='val_acc', verbose=1, save_best_only=True, mode='max')callbacks_list = [checkpoint]model.fit(train_X, train_Y, validation_split=.20,                    epochs=1000, batch_size=50, callbacks=callbacks_list, verbose=0)输出：Epoch 00991: val_acc did not improveEpoch 00992: val_acc improved from 0.93465 to 0.93900, saving model to weights-992-0.94.hdf5Epoch 00993: val_acc did not improveEpoch 00994: val_acc did not improveEpoch 00995: val_acc did not improveEpoch 00996: val_acc did not improveEpoch 00997: val_acc did not improveEpoch 00998: val_acc improved from 0.93900 to 0.94543, saving model to weights-998-0.94.hdf5Epoch 00999: val_acc did not improve问题：如@IonicSolutions 所说，如何加载所有这些模型以获得每个时期中 LSTM 层的输出？

Answer 1

在您的情况下最有效的方法取决于您如何准确设置和训练 SVM，但使用回调至少有两个选项：

您可以使用ModelCheckpoint回调来保存您在每个时期训练的模型的副本，然后加载所有这些模型以获得 LSTM 层的输出。

您还可以通过实现Callback基类来创建自己的回调。在回调中，可以访问模型，您可以使用on_epoch_end它在每个时期结束时提取 LSTM 输出。

编辑：要方便地访问倒数第二层，您可以执行以下操作：

# Create the model with the functional API

inp = Input((train_X.shape[1], train_X.shape[2],))

lstm = LSTM(100, return_sequences=False)(inp)

dense = Dense(train_Y.shape[1], activation='softmax')(lstm)

# Create the full model

model = Model(inputs=inp, outputs=dense)

# Create the model for access to the LSTM layer

access = Model(inputs=inp, outputs=lstm)

然后，您可以access在实例化它时传递给您的回调。最关键的事情，这里要注意的是，model与access共享同样的LSTM层，它们的权重会发生变化时训练model。