Implementation of the Asymmetric Cross Entropy Loss for the Positive-Unlabeled Learning (PU-Learning) for PyTorch. The method is adapted from the Asymmetric Loss for Positive-Unlabeled Learning paper [1] - it is not one-to-one implementation.
[1] Wang, C., Pu, J., Xu, Z. and Zhang, J., 2021, July. Asymmetric Loss for Positive-Unlabeled Learning. In 2021 IEEE International Conference on Multimedia and Expo (ICME) (pp. 1-6). IEEE.