0
$\begingroup$

Can we use duplicate data as an input to SVM? The duplicate data that I mean is, let say we have 50 of same data (maybe being duplicate) from total of 100 data. Will this kind of data effect the performance result that we obtained from SVM? Do we have to avoid any duplicate data for SVM?

$\endgroup$
1
$\begingroup$

You can, yes as long as the repeated data has the same label there will be no problem. It is not ideal though. The performance will be lower than if you trained with a more diverse dataset of the same size, and just as good as if you dropped the duplicates and trained.

$\endgroup$
  • $\begingroup$ ,thank you for your comment. Yes, the replicate data will become a problem when it was labeled with wrong label. I have search again the information about replicate data after read your comment. Thanks again for your explanation. :) $\endgroup$ – user3847711 Aug 24 '14 at 15:36

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.