Local Rademacher Complexity: Sharper risk bounds with and without unlabeled samples.
Luca OnetoAlessandro GhioSandro RidellaDavide AnguitaPublished in: Neural Networks (2015)
Keyphrases
- risk bounds
- data dependent
- semi supervised
- empirical risk minimization
- learning algorithm
- statistical learning theory
- semi supervised learning
- vc dimension
- stochastic processes
- labeled data
- parameter free
- labeled and unlabeled data
- mixture model
- unsupervised learning
- pairwise constraints
- semi supervised classification
- supervised learning
- machine learning
- training data
- uniform convergence
- probability distribution
- training set