I am trying to train an lstm with a sequence and get the sequence classification for the whole sequence.

I have sequences of varying length so I have one input neuron and I am feeding one item at a time. Isn't that the proper solution?

My issue is that I am training each of these inputs with a single ideal-output, but some of the sequences exist in other sequences with other ideal-outputs.

So when I train 0.74 with 1,0 and 0.83 with 1,0 and 0.32 with 1,0 it gets trained with the class 0, 1. But when I then train with 0.74 0.83 0.32 with 0,1 the training diverges to infinity because I assigned two different classes to the same input.

How am I supposed to train a lstm with a sequence when some of the elements treat across sequences? Unless there is another way to train a deep-network that has an lstm when the sequences are of varying lengths?


Train it one character at a time. It shouldn't diverge unless the characters are the same and have different ideal-outputs. In that case consider using one-hot vectors instead of scalar vectors. Meaning if a, b, and c are your characters then if a is the character 1, 0, 0 is the input.


Your Answer

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

Not the answer you're looking for? Browse other questions tagged or ask your own question.