Positive-Unlabeled Learning with Uncertainty-aware Pseudo-label Selection

Proposing a novel uncertainty-aware pseudo-labeling framework, PUUPL, for Positive-Unlabeled Learning to address imbalanced datasets and model calibration issues.