SaR: Self-adaptive Refinement on Pseudo Labels for Multiclass-Imbalanced Semi-supervised Learning.
Zhengfeng LaiChao WangSen-Ching S. CheungChen-Nee ChuahPublished in: CVPR Workshops (2022)
Keyphrases
- semi supervised learning
- multi class
- label propagation
- semi supervised
- pairwise
- multi label classification
- binary classification problems
- unlabeled data
- training data
- labeled examples
- labeled data
- multiple classes
- binary classification
- active learning
- support vector machine
- unsupervised learning
- class imbalance
- learning problems
- class labels
- supervised learning
- multiclass classification
- machine learning
- cost sensitive
- co training
- pairwise constraints
- training examples
- class distribution
- binary classifiers
- base classifiers
- feature selection
- text categorization
- multiple instance learning
- transfer learning
- multi label
- object recognition
- training set
- learning models
- image reconstruction
- error correcting output codes
- loss function
- reinforcement learning
- image segmentation
- learning algorithm