I understand how binary relevance works on a multi-label dataset: the data is split up into L data sets, where L is the number of labels. Each subset has a column where either a 0 or a 1 is assigned to an instance, indicating the presence or absence of that label on that instance. A separate classifier is trained on each data set.

But, in the case of a multi-class multi-label problem, where some labels may take on different, mutually exclusive class values, is each class treated as a separate label, or is the split still done the same way, except that instead of a "0" or "1", the class is assigned in the label column, thus requiring a multi-class classifier/ classification scheme?


1 Answer 1


I think binary relevance should be done still in the way that you proposed.

For an overview of different methods for multilabel classification look here: https://journal.r-project.org/archive/2017/RJ-2017-012/RJ-2017-012.pdf I don't know how they could be adapted for multiclass multi-label problems. I would also rather call them multitarget problems.


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.