As far as I know, the LSTM has a loop that inputs from the last hidden layer resulting the model to have a long short term memory. Then why use sliding windows?
Can y
