I want to use prediction models like LSTM-AE to predict time-series data. The feature that the neural network should learn is in frequency between 40-60Hz. So, in order to learn the feature more effectively and removing the noises, the signal will be filtered using a bandpass filter and the result will be then passed to the network.
The problem is that, if I want to develop an end-to-end (i.e. omitting the bandpass filtering) solution, how can I do that?