如何将多层作为 keras 中的时间步长赋予 LSTM

Question

我想给一个 lstm 两个独立的神经网络作为 2 个时间步长。这是我的代码：

input1 = Input(shape=(self.state_size,1))
input2 = Input(shape=(self.state_size,1))

out1 = Conv1D(12, 5, padding="SAME", activation="relu")(input1)
out1 = Flatten()(out1)
out1 = Dense(12, activation="relu")(out1)

out2 = Conv1D(12, 5, padding="SAME", activation="relu")(input2)
out2 = Flatten()(out2)
out2 = Dense(12, activation="relu")(out2)

out = CuDNNLSTM(1)([out1,out2])

错误是：

ValueError: Input 0 is incompatible with layer cu_dnnlstm_1: expected ndim=3, found ndim=2

指的是：

out = CuDNNLSTM(1)([out1,out2])

我也试过：

out = CuDNNLSTM(1)(out1,out2)

我的输入形状是 (none,4,1)，我需要我的输出形状是 (none,1)。显然 CuDNNLSTM 的输入形状必须是 (none,2,12)，但我很难连接 out1 和 out2

Answer 1

你要stack中间维度的张量：

steps = Lambda(lambda x: K.stack(x, axis=1))([out1, out2])
out = CuDNNSLTM(1)(steps)

但我不确定包含两个步骤的序列是否会带来常规图层无法带来的出色效果。

如何将多层作为 keras 中的时间步长赋予 LSTM

How to give multiple layers to LSTM as timesteps in keras

lstm

keras

tensorflow

recurrent-neural-network