This repository is reproduced code of "Generative Adversarial Positive-Unlabelled Learning"
The paper link is here
Input data is positive (orange color) and unlabeled (blue) data.
This algorithm trains the positive generator and the negative generator.