Say I have a tensor of data with shape (30, 16000, 38) - where each tuple element corresponds to (datasets, samples per set, features). Now, I want to feed this data into a LSTM Neural Network and classify each datapoint as one of four classes (a multiclass problem). As I'm not doing prediction but rather one-to-one classification, does this render applying a sliding window on my samples per set unnecessary?

Stated more generally:

While doing LSTM classification without prediction, under what circumstances should I think about applying a sliding window to split the sequences in smaller timestep_look_back sets?


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.