The litterature on multi-label learning uses the terms: multi-label learning and multi-label classification. I was wondering what the difference between these terms is, and when to use one over the other??

  • Sharing your research helps everyone. Tell us what you've tried and why it didn’t meet your needs. This demonstrates that you’ve taken the time to try to help yourself, it saves us from reiterating obvious answers, and most of all it helps you get a more specific and relevant answer. Also see How to Ask – gnat Jul 23 '15 at 17:15
  • @gnat The question is specific to multi-label learning, a a field within machine-learning, which is a field within artificial intelligence. The question is simple for those who know something about this field. Do you tell someone who asks a question about C#, or some other programming language - which you may not know anything about, to share what they know about that particular field? – Kenci Jul 23 '15 at 17:52
  • did you check meta guidance referred in prior comment? – gnat Jul 23 '15 at 18:12

Both are forms of supervised learning where the classification algorithm should learn from a set of instances/examples.

In multi-class problems each example is restricted to have only one class label.

Multi-label learning is a generalized version in which each instance can belong to multiple classes (e.g. a research paper can belong to both the health and science category).

Binary and multi-class problems can be posed as specific cases of multi-label problem. However, the generality of multi-label problems makes them more difficult than the others.

  • Thanks for your answer. Is the term "multi-label classification" used for binary problems (within multi-label problems), and "multi-label learning" for a multi-class problems (within multi-label problems)? For instance, Mulan (mulan.sourceforge.net) is "A Java Library for Multi-Label Learning", but you can use it for multi-label classification. – Kenci Jul 24 '15 at 11:35
  • @Kenci As far as I know binary classification problems are a special case of multi-class classification problems. Multi-class classification problems are a special case of multi-labelled classification. You can take a look at A Literature Survey on Algorithms for Multi-label Learning for more details. – manlio Jul 24 '15 at 16:42

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