The Perils of Learning From Unlabeled Data: Backdoor Attacks on Semi-supervised Learning.
Virat ShejwalkarLingjuan LyuAmir HoumansadrPublished in: CoRR (2022)
Keyphrases
- semi supervised learning
- unlabeled data
- labeled and unlabeled data
- labeled data
- supervised learning
- semi supervised
- co training
- learning problems
- active learning
- learning algorithm
- unsupervised learning
- semi supervised classification
- multi view learning
- supervised learning algorithms
- learning models
- positive examples
- training data
- domain adaptation
- manifold regularization
- machine learning
- labeled examples
- semi supervised learning setting
- data points
- transfer learning
- learning frameworks
- semi supervised learning methods
- labeled training data
- learning tasks
- class labels
- labeled instances
- semi supervised learning algorithms
- label propagation
- partially labeled
- learning process
- reinforcement learning
- improve the classification accuracy
- prior knowledge
- semi supervised clustering
- training examples
- text classification
- unlabeled examples
- sample selection bias
- partially labeled data
- multiple instance learning
- unlabeled samples
- decision boundary
- training set
- multi label
- number of labeled examples