Hi, I would like to build an end-to-end ML pipeline for training a multi-label text classifier. When trying to achieve this, I see that the "Train Model" component does not allow the target label column to be a list of classes (Object type) which would be the case in Multi-label classification problems. Please correct me if I am wrong. Does the Train Model and Score Model components support multi-label classification?
Hi, thanks for reaching out. Yes, Train Model supports multi-classification but only a single column is supported. The selected column should contain the outcomes (classes). An example is provided in Designer > Show More Samples > Text Classification - Wikipedia SP 500 Dataset.
--please don't forget to Accept Answer if the reply is helpful. Thanks.--
@GiftA-MSFT Thanks for the prompt support! I checked the example, and I could see that there is only one class corresponding to each text. What if a text corresponds to two or more categories?
Looks like what you're trying to do is multi-label classification which isn't currently supported in AML.
@GiftA-MSFT
Oh I see, it would be a very useful feature to have. Any specific reason/challenges on why this has not been implemented yet?
There has been a similar request back in 2017: https://social.msdn.microsoft.com/Forums/en-US/d3f338dd-a917-45fc-92a6-1c634fa5d588/multi-label-classification-in-azure-ml?forum=MachineLearning
10 people are following this question.
