Adversarial Training with Complementary Labels: On the Benefit of Gradually Informative Attacks.
Jianan ZhouJianing ZhuJingfeng ZhangTongliang LiuGang NiuBo HanMasashi SugiyamaPublished in: NeurIPS (2022)
Keyphrases
- training examples
- training set
- labelled data
- multi agent
- pairwise
- test set
- training process
- information systems
- active learner
- supervised learning
- countermeasures
- training phase
- traffic analysis
- class labels
- fully labeled
- representative images
- image labeling
- information security
- training samples
- active learning
- training data