Is there a way to get embedding for an ordered sequence of vectors?

I want to get embeddings to feed them further into net i.e. train it to arbitrary loss function simultaneously for embeddings and other parts so that similar items get similar embeddings.

Data is not descrete so it's not a word embedding.

  • $\begingroup$ Yes, there is. Let the embedding be a parameter used in predicting the next element; e.g., the state of an RNN. So you use the same model for all the time series except for that parameter. For details read A review of unsupervised feature learning and deep learning for time-series modeling. Note also that word embeddings are continuous, even though they are mappings from a discrete space. Welcome to the site and good luck. $\endgroup$ – Emre Aug 30 '18 at 16:27

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.