代码之家 › 专栏 › 技术社区 › Sreeram TP

多步预测LSTM模型

lstm time-series keras machine-learning python

1

Sreeram TP · 技术社区 · 6 年前

我试图在Keras中实现一个多步预测LSTM模型。数据的形状如下:

X : (5831, 48, 1)
y : (5831, 1, 12)

我尝试使用的模型是:

power_in = Input(shape=(X.shape[1], X.shape[2]))


power_lstm = LSTM(50, recurrent_dropout=0.4128,
                  dropout=0.412563, kernel_initializer=power_lstm_init, return_sequences=True)(power_in)

main_out = TimeDistributed(Dense(12, kernel_initializer=power_lstm_init))(power_lstm)

在训练模特时:

hist = forecaster.fit([X], y, epochs=325, batch_size=16, validation_data=([X_valid], y_valid), verbose=1, shuffle=False)

我得到以下错误:

ValueError: Error when checking target: expected time_distributed_16 to have shape (48, 12) but got array with shape (1, 12)

如何解决这个问题?

1 回复 | 直到 6 年前

1

today 6 年前

根据你的评论:

我的数据是t-48,t-47,t-46,t-1作为过去的数据和

您可能不需要使用 TimeDistributed resturn_sequences=True LSTM层的参数。完成后,LSTM层将过去的输入时间序列编码为一个形状向量 (50,)

# make sure the labels have are in shape (num_samples, 12)
y = np.reshape(y, (-1, 12))

power_in = Input(shape=(X.shape[1:],))
power_lstm = LSTM(50, recurrent_dropout=0.4128,
                  dropout=0.412563,
                  kernel_initializer=power_lstm_init)(power_in)

main_out = Dense(12, kernel_initializer=power_lstm_init)(power_lstm)

或者,如果您想使用 时间分布的 并且考虑到输出本身是一个序列,我们可以在我们的模型中通过在密集层之前使用另一个LSTM层(添加一个 RepeatVector

# make sure the labels have are in shape (num_samples, 12, 1)
y = np.reshape(y, (-1, 12, 1))

power_in = Input(shape=(48,1))
power_lstm = LSTM(50, recurrent_dropout=0.4128,
                  dropout=0.412563,
                  kernel_initializer=power_lstm_init)(power_in)

rep = RepeatVector(12)(power_lstm)
out_lstm = LSTM(32, return_sequences=True)(rep)
main_out = TimeDistributed(Dense(1))(out_lstm)

model = Model(power_in, main_out)
model.summary()

_________________________________________________________________
Layer (type)                 Output Shape              Param #   
=================================================================
input_3 (InputLayer)         (None, 48, 1)             0         
_________________________________________________________________
lstm_3 (LSTM)                (None, 50)                10400     
_________________________________________________________________
repeat_vector_2 (RepeatVecto (None, 12, 50)            0         
_________________________________________________________________
lstm_4 (LSTM)                (None, 12, 32)            10624     
_________________________________________________________________
time_distributed_1 (TimeDist (None, 12, 1)             33        
=================================================================
Total params: 21,057
Trainable params: 21,057
Non-trainable params: 0
_________________________________________________________________

当然,在这两种模型中,您可能需要调整超参数(例如LSTM层的数量、LSTM层的尺寸等),以便能够准确地比较它们并获得良好的结果。

实际上,在您的场景中,您不需要使用 时间分布的 Dense layer is applied on the last axis . 因此, TimeDistributed(Dense(...)) Dense(...) 是等价的。