0
$\begingroup$

In many papers and articles related to Recurrent Neural networks(RNN) with specific (LSTM)Long short term memory technique I have seen that in training phase labels and ground truth data are fed with the input data.Whereas In most of the classification techique only the labels are required with the data in the training phase.

Please help me understand why we need both labels and ground truth in training phase.

Reference LSTM Paper

Improve this question
$\endgroup$

1 Answer 1

Reset to default
2
$\begingroup$

In that context, ground truth and gold labels are synonymous. LSTMs don't require any additional supervision compared to other supervised machine learning models.

Improve this answer
$\endgroup$
2
  • $\begingroup$ But I have seen that in LSTM programming codes they maintain dictionary of labels? $\endgroup$
    – Edison
    Commented Aug 1, 2016 at 5:37
  • $\begingroup$ @NaseerAhmed They often maintain a dictionary mapping word to word index, maybe this is what you read. $\endgroup$
    – Franck Dernoncourt
    Commented Aug 1, 2016 at 14:35

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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