Good day I'm trying to build a multi-label classifier but having some trouble achieving it. I know what I have to do theoretically - create a series of binary classifiers for each label that classifies A/not-A, B/not-B etc.
My question is how to actually do this. Do I simply build, train and deploy an app for each label, and then submit a query to each of these? This seems inefficient. On the other hand, I saw on the DASE model that we can provide multiple algorithms. Would it be possible to add an algorithm for each label, and then combine them when serving? If somebody could point me in the right direction with some code suggestions I would be really grateful. I have found no useful examples online. Regards, Mark