Implements multi-label support for BCEWithLogitsLoss and MOONBCEWithLogitsLoss
This MR implements support for multi-label tasks (tasks in which multiple outputs of a classifier or segmenter may be positive concurrently).
Two losses are currently available to support binary, multi-class and multi-label tasks:
- BCEWithLogitsLoss: stock loss at pytorch
- MOONBCEWithLogitsLoss: implementation derived from our paper at https://arxiv.org/abs/2407.14064
Tests have been added to the test suite.
Closes #68 (closed), biosignal/software/paper/euvip24-refine-cad-tb#2
Edited by André Anjos