GradPU: Positive-Unlabeled Learning via Gradient Penalty and Positive Upweighting.
Songmin DaiXiaoqiang LiYue ZhouXichen YeTong LiuPublished in: AAAI (2023)
Keyphrases
- positive and unlabeled examples
- learning algorithm
- active learning
- unsupervised learning
- positive and negative
- prior knowledge
- online learning
- learning tasks
- reinforcement learning
- learning process
- training examples
- positive data
- knowledge acquisition
- learning systems
- learning experience
- mobile learning
- genetic algorithm