Convex formulation of multiple instance learning from positive and unlabeled bags.
Han BaoTomoya SakaiIssei SatoMasashi SugiyamaPublished in: Neural Networks (2018)
Keyphrases
- multiple instance learning
- convex formulation
- positive bags
- supervised learning
- semi supervised learning
- multiple instance
- convex optimization
- potts model
- semi supervised
- class labels
- unlabeled data
- gradient method
- diverse density
- multi class
- labeled data
- unsupervised learning
- image annotation
- loss function
- active learning
- convex relaxation
- positive and negative
- training data
- training examples
- instance selection
- machine learning
- prior knowledge
- information extraction
- model selection
- learning problems
- learning tasks
- multi label