seq2seq 中的 TimeDistributed(Dense) 与 Dense

鉴于下面的代码

encoder_inputs = Input(shape=(16, 70))
encoder = LSTM(latent_dim, return_state=True)
encoder_outputs, state_h, state_c = encoder(encoder_inputs)
# We discard `encoder_outputs` and only keep the states.
encoder_states = [state_h, state_c]

# Set up the decoder, using `encoder_states` as initial state.
decoder_inputs = Input(shape=(59, 93))
# We set up our decoder to return full output sequences,
# and to return internal states as well. We don't use the
# return states in the training model, but we will use them in inference.
decoder_lstm = LSTM(latent_dim, return_sequences=True, return_state=True)
decoder_outputs,_,_ = decoder_lstm(decoder_inputs,
                                     initial_state=encoder_states)
decoder_dense = TimeDistributed(Dense(93, activation='softmax'))
decoder_outputs = decoder_dense(decoder_outputs)

# Define the model that will turn
# `encoder_input_data` & `decoder_input_data` into `decoder_target_data`
model = Model([encoder_inputs, decoder_inputs], decoder_outputs)

如果我改变

decoder_dense = TimeDistributed(Dense(93, activation='softmax'))

to

decoder_dense = Dense(93, activation='softmax')

仍然有效，但是哪种方法更有效呢？

如果您的数据依赖于时间，例如Time Series数据或包含不同帧的数据Video，然后时间Distributed Dense分层比简单更有效Dense Layer.

Time Distributed Dense同样适用dense期间的每个时间步骤层GRU/LSTM细胞展开。这就是为什么误差函数将在predicted label sequence和actual label sequence.

Using return_sequences=False, the Dense图层只会在最后一个单元格中应用一次。通常情况下，当RNNs用于分类问题。

If return_sequences=True，那么Dense图层用于在每个时间步应用，就像TimeDistributedDense.

在您的模型中，两者都是相同的，但是如果您将第二个模型更改为return_sequences=False，那么Dense仅适用于最后一个单元格。

希望这可以帮助。快乐学习！