Back Propagation for Multi-Label Learning in Pytorch

How to implement BPMLL loss in pytorch?


Is there any official repo that has BPMLL loss implemented for pytorch?

Hi. If I understand correctly, and you are talking about Multi Label Classification, then Binary Cross Entropy and its more stable version should be a good start for your search :slight_smile:

1 Like

Yeah I tried with Binary Cross Entropy but my Macro AUC is not boosting up even after rigorous hyperparam tuning using raytune. The dataset is highly imbalanced. I referred the internet where they stated that BP_MLL loss is the perfect thing for multi-label classification.